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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
f9f5d544-0c0b-407b-9410-b778eaedcf7e | tener-adapting-transformer-encoder-for-name | 1911.04474 | null | https://arxiv.org/abs/1911.04474v3 | https://arxiv.org/pdf/1911.04474v3.pdf | TENER: Adapting Transformer Encoder for Named Entity Recognition | The Bidirectional long short-term memory networks (BiLSTM) have been widely used as an encoder in models solving the named entity recognition (NER) task. Recently, the Transformer is broadly adopted in various Natural Language Processing (NLP) tasks owing to its parallelism and advantageous performance. Nevertheless, the performance of the Transformer in NER is not as good as it is in other NLP tasks. In this paper, we propose TENER, a NER architecture adopting adapted Transformer Encoder to model the character-level features and word-level features. By incorporating the direction and relative distance aware attention and the un-scaled attention, we prove the Transformer-like encoder is just as effective for NER as other NLP tasks. | ['Bocao Deng', 'Xipeng Qiu', 'Xiaonan Li', 'Hang Yan'] | 2019-11-10 | null | null | null | null | ['chinese-named-entity-recognition'] | ['natural-language-processing'] | [-2.44400203e-01 -2.30850533e-01 2.40812432e-02 -2.95275480e-01
-6.54576540e-01 -3.57131362e-01 4.39681411e-01 1.39909178e-01
-1.07838118e+00 8.12622488e-01 6.18366957e-01 -3.29495311e-01
1.01560719e-01 -8.84249926e-01 -5.17632723e-01 -5.49410403e-01
-2.84720454e-02 1.73736155e-01 2.84382999e-01 -3.34931016e-01
2.15564072e-01 5.26959419e-01 -9.26299214e-01 1.98240146e-01
8.68928671e-01 7.36361861e-01 4.13338125e-01 6.63671017e-01
-6.75689399e-01 9.35367823e-01 -4.38026309e-01 -8.03345919e-01
-1.01696692e-01 -2.14221418e-01 -1.09320164e+00 -8.20034862e-01
-2.91180275e-02 -6.88834414e-02 -5.41862607e-01 9.29217041e-01
7.50217974e-01 3.90465140e-01 3.13999504e-01 -7.28959918e-01
-1.10217130e+00 7.31454551e-01 -3.55926782e-01 6.25592887e-01
2.07949266e-01 -2.38725424e-01 1.21731865e+00 -1.02969384e+00
3.08068007e-01 1.20113862e+00 9.13485944e-01 5.56483150e-01
-5.04731476e-01 -4.34729040e-01 7.24278167e-02 4.92720187e-01
-1.38873780e+00 -2.70015091e-01 4.24835145e-01 1.53828353e-01
1.70400178e+00 -9.11914334e-02 2.60541558e-01 9.62882459e-01
6.36075556e-01 8.51967216e-01 6.76766634e-01 -3.54400992e-01
-1.25293911e-01 -2.03542024e-01 3.09303880e-01 6.40527129e-01
-6.08975627e-02 6.49665520e-02 -1.85941145e-01 5.80038242e-02
6.43856168e-01 1.20877370e-01 -5.99778369e-02 3.83658886e-01
-1.21017838e+00 7.43014574e-01 9.36441302e-01 8.90125692e-01
-6.61393583e-01 2.68593669e-01 7.41309464e-01 6.32950589e-02
3.64183336e-01 4.38378930e-01 -7.51554787e-01 -3.14630806e-01
-5.99026918e-01 -3.56943727e-01 6.77568018e-01 8.89313817e-01
4.21832591e-01 3.50031525e-01 -3.94209385e-01 9.05394256e-01
2.23956585e-01 1.27560124e-01 9.41635787e-01 -1.85026988e-01
6.00918651e-01 6.32471800e-01 -1.91787541e-01 -8.31091523e-01
-4.47668970e-01 -4.90778327e-01 -1.31233764e+00 -2.54481196e-01
4.20751609e-02 -3.03431123e-01 -8.59653354e-01 1.85039651e+00
3.06031033e-02 2.85555154e-01 3.07638317e-01 6.62801087e-01
8.59332204e-01 1.15405703e+00 6.80580556e-01 7.15872422e-02
1.70854270e+00 -1.13556623e+00 -9.76897478e-01 -3.38121235e-01
8.40489089e-01 -7.09490538e-01 8.22346449e-01 -2.20807716e-01
-7.84316719e-01 -6.40200615e-01 -7.09722400e-01 -6.14803135e-01
-8.93077016e-01 1.91544026e-01 5.16298056e-01 4.88888681e-01
-9.64977682e-01 5.73251486e-01 -7.69062579e-01 -4.94485736e-01
2.43935004e-01 2.95241892e-01 -5.64028621e-01 1.77233800e-01
-1.67656898e+00 1.25863898e+00 5.84757805e-01 5.53265512e-01
-5.43973267e-01 -2.87677288e-01 -9.80152786e-01 4.75214422e-01
-1.39448568e-01 -6.87381625e-01 1.05487061e+00 -4.50831085e-01
-1.44352162e+00 5.97850561e-01 -3.10303986e-01 -5.08943677e-01
-2.47090086e-02 -2.23848760e-01 -6.01976156e-01 -2.95983374e-01
-1.89398555e-03 5.63864827e-01 1.95696756e-01 -4.14181709e-01
-5.51334918e-01 -3.92106742e-01 7.69300237e-02 3.94678116e-01
-6.27277434e-01 3.84700865e-01 -2.19635535e-02 -6.03425682e-01
-3.86110187e-01 -4.34980273e-01 -3.64786565e-01 -4.11635280e-01
-4.41703945e-01 -7.04652846e-01 5.87564349e-01 -9.20751333e-01
1.37859118e+00 -2.14265990e+00 -9.60987508e-02 -3.14548492e-01
1.14436168e-02 7.23442554e-01 -3.47001612e-01 7.04897881e-01
4.50009778e-02 5.00855207e-01 -1.77027702e-01 -1.81803837e-01
6.97496831e-02 1.93032667e-01 -2.46179521e-01 8.93168375e-02
3.76554608e-01 1.20259023e+00 -8.38709116e-01 -4.39937115e-01
1.19799644e-01 7.71298945e-01 -1.15243860e-01 3.83001268e-01
1.97285905e-01 2.41180152e-01 -4.51742798e-01 2.05949277e-01
5.20794451e-01 -5.89781068e-02 -1.66567892e-01 -2.62449056e-01
-2.75213391e-01 5.86712599e-01 -7.43238449e-01 1.61315560e+00
-7.69276261e-01 6.28977895e-01 -1.93917722e-01 -7.83555090e-01
1.00470889e+00 7.49311984e-01 -1.30187407e-01 -8.09938431e-01
2.36748859e-01 1.92789972e-01 3.73682790e-02 -5.08627295e-01
5.02486229e-01 -4.91279066e-01 -1.73741594e-01 2.06958294e-01
3.49542350e-01 4.75823551e-01 1.16689697e-01 -1.21292233e-01
1.23860443e+00 1.53952204e-02 6.74598098e-01 -1.51763767e-01
7.83067584e-01 -1.97694778e-01 7.98601687e-01 3.85345668e-01
-2.30942652e-01 1.84546024e-01 1.57041222e-01 -4.31603163e-01
-1.13508463e+00 -8.27169776e-01 -1.37044579e-01 1.35423684e+00
-4.96030897e-02 -3.17362636e-01 -6.99653864e-01 -7.37540901e-01
-4.30669427e-01 7.15942800e-01 -3.91470134e-01 -1.89091384e-01
-6.87209368e-01 -6.79592907e-01 1.23459649e+00 9.29910243e-01
1.01380110e+00 -1.42868090e+00 -3.00979555e-01 4.05178905e-01
-1.85481295e-01 -1.17093706e+00 -8.47736120e-01 3.46966833e-01
-9.19431448e-01 -7.15565860e-01 -8.82756650e-01 -1.32114351e+00
2.22275570e-01 3.29216048e-02 9.00285721e-01 -2.56427497e-01
3.88573632e-02 -1.81532249e-01 -4.70557481e-01 -1.06977046e-01
-4.00671698e-02 5.07394314e-01 -1.17116213e-01 -1.74615547e-01
7.18345642e-01 -5.42411387e-01 -3.67219090e-01 1.00003041e-01
-8.08122098e-01 -4.52467650e-02 8.18707228e-01 1.01608717e+00
3.38788450e-01 1.33284360e-01 8.15485537e-01 -7.49136925e-01
8.86568069e-01 -4.83805776e-01 -1.77633792e-01 4.18082654e-01
-3.59847605e-01 2.67235398e-01 1.17635429e+00 -2.39701807e-01
-1.37548780e+00 -3.71087432e-01 -9.03527915e-01 -1.95538672e-03
-2.77077407e-01 7.56499588e-01 -3.34942728e-01 1.21240564e-01
1.96465075e-01 5.10257840e-01 -9.24079895e-01 -8.60985756e-01
1.68739483e-01 8.24601948e-01 3.56893480e-01 -3.88643533e-01
4.63281751e-01 -6.86885938e-02 -3.49369019e-01 -7.36746609e-01
-1.08396471e+00 -3.91687721e-01 -6.86408162e-01 4.21119750e-01
1.33678842e+00 -8.09811592e-01 -6.73157513e-01 3.81570667e-01
-1.67831230e+00 -3.38776447e-02 -1.63469583e-01 7.07517803e-01
-7.58690014e-02 2.24857390e-01 -1.15501499e+00 -8.17979097e-01
-8.30780268e-01 -8.77010643e-01 8.34940195e-01 3.86532426e-01
2.40139619e-01 -1.30022585e+00 2.33037531e-01 1.94532964e-02
7.49970853e-01 -1.12362653e-01 1.06110513e+00 -1.03330231e+00
-1.39707059e-01 -7.96142071e-02 -5.57911396e-01 4.91797447e-01
-3.49697731e-02 -4.04440522e-01 -9.00879979e-01 -8.30873400e-02
1.55278482e-02 -7.64965117e-02 8.98072243e-01 -2.55312864e-02
9.81541872e-01 -4.37741756e-01 -1.42680109e-01 7.18152344e-01
1.58422768e+00 3.22012484e-01 7.50409544e-01 4.22920585e-01
8.28881979e-01 3.80647957e-01 1.23865999e-01 7.66697079e-02
5.20616651e-01 3.12882483e-01 1.48369312e-01 -1.32347777e-01
9.82289836e-02 -5.46784401e-01 4.80659336e-01 1.47905886e+00
-8.19655061e-02 -5.16504049e-01 -1.11780250e+00 7.76702404e-01
-1.69456804e+00 -1.00821435e+00 -8.74029100e-02 1.70120406e+00
7.52812803e-01 1.38756996e-02 -4.68389720e-01 -1.46585703e-01
1.11658835e+00 2.84718871e-01 -4.32855874e-01 -9.95150447e-01
-1.83038265e-01 4.51232165e-01 5.32385111e-01 1.80880278e-01
-9.92643058e-01 1.06459725e+00 6.08075809e+00 8.97366822e-01
-9.49223638e-01 4.82118249e-01 4.84179437e-01 5.98304629e-01
-2.05432519e-01 -1.21635042e-01 -8.78094912e-01 3.85415196e-01
1.42641318e+00 -2.98666596e-01 2.42975071e-01 6.81223154e-01
1.29641294e-01 4.81188238e-01 -8.41498256e-01 8.50852728e-01
-2.10215643e-04 -1.04669571e+00 2.06852123e-01 -1.59427166e-01
3.74509931e-01 4.47077960e-01 -3.04156631e-01 6.27457201e-01
2.88097501e-01 -1.16338468e+00 3.93421650e-01 3.08661878e-01
8.03395867e-01 -9.39901233e-01 1.37528908e+00 5.58130503e-01
-1.63576889e+00 -2.10598916e-01 -6.66904032e-01 -1.56151503e-01
5.12101710e-01 5.15515089e-01 -4.08638299e-01 6.27584815e-01
6.40393019e-01 6.76439643e-01 -2.44125739e-01 1.02800727e+00
-4.54170644e-01 6.73983693e-01 -2.22907439e-01 -2.57349759e-01
6.72577560e-01 -1.34042174e-01 2.66881287e-01 1.66564775e+00
4.39547539e-01 8.44567493e-02 -1.92028418e-01 4.09854889e-01
-5.87514877e-01 4.41107333e-01 -6.41314209e-01 -3.97210032e-01
4.97394025e-01 1.29125154e+00 -3.42955559e-01 -3.30908716e-01
-4.56444323e-01 1.23407471e+00 7.32434690e-01 2.23572150e-01
-8.39721918e-01 -1.00381505e+00 4.50360417e-01 -2.59231091e-01
5.29004753e-01 -2.62485355e-01 -2.07045212e-01 -1.21973860e+00
-1.44574240e-01 -3.64284456e-01 4.81473774e-01 -8.47919583e-01
-1.60119534e+00 1.09750366e+00 -5.50809205e-01 -8.78941178e-01
1.09694585e-01 -7.96981633e-01 -8.35486472e-01 9.99918640e-01
-1.93675447e+00 -1.21223581e+00 5.37966155e-02 6.04859352e-01
4.98809963e-01 -2.95596384e-02 1.14831102e+00 6.33887708e-01
-8.66031885e-01 5.36516309e-01 2.59897828e-01 7.07884252e-01
6.24409199e-01 -1.34509325e+00 6.54530764e-01 7.53318906e-01
1.82880104e-01 8.94254565e-01 1.64111912e-01 -3.26468468e-01
-1.17168975e+00 -1.25728250e+00 1.72243607e+00 -7.95634985e-02
6.91011429e-01 -3.41499835e-01 -9.08413589e-01 9.49913323e-01
6.88844502e-01 8.86330660e-03 7.08312988e-01 -5.41383997e-02
-4.71860528e-01 -1.44667551e-01 -1.08445823e+00 4.76633400e-01
1.07990050e+00 -7.71843851e-01 -1.15464222e+00 -3.92925739e-02
1.04043150e+00 -2.06874788e-01 -1.09714460e+00 2.52329141e-01
3.01513791e-01 -7.05698967e-01 8.67013872e-01 -6.59538269e-01
3.66686016e-01 -1.03697412e-01 -1.04089446e-01 -1.42413330e+00
-6.81094110e-01 -2.41426498e-01 8.05723146e-02 1.69987345e+00
3.33453804e-01 -8.95077825e-01 2.81704336e-01 1.39490068e-01
-2.82123983e-01 -6.63633585e-01 -1.10486937e+00 -6.70108855e-01
2.67273813e-01 -2.74925679e-01 6.52257264e-01 9.13911104e-01
9.62142497e-02 9.90296483e-01 -4.89147604e-01 -6.59059137e-02
1.50298715e-01 -1.58766061e-01 -8.12607817e-03 -1.02543938e+00
3.58134620e-02 -2.49481305e-01 -4.61983800e-01 -1.26580846e+00
4.04023588e-01 -1.07227182e+00 1.35568634e-01 -1.83367598e+00
2.43700877e-01 -3.03697050e-01 -8.24011385e-01 7.22690225e-01
-3.17324430e-01 -2.28406978e-03 1.49965897e-01 -3.61428410e-02
-6.17648602e-01 7.87033916e-01 1.10294878e+00 2.45601162e-02
1.62486881e-01 -3.38130176e-01 -6.14830971e-01 5.10097623e-01
8.78640473e-01 -6.46751642e-01 2.94765625e-02 -9.65592265e-01
3.54681820e-01 3.65154333e-02 -1.37465689e-02 -8.29438448e-01
5.30561447e-01 1.30606234e-01 4.59242284e-01 -5.37420571e-01
1.44829199e-01 -7.56697237e-01 -1.36167124e-01 4.10082012e-01
-4.01302904e-01 7.37547159e-01 1.31881580e-01 5.36554873e-01
-5.84375978e-01 -2.52628714e-01 6.13803506e-01 -3.44837129e-01
-1.03958428e+00 6.19337320e-01 -4.49774325e-01 2.32743904e-01
7.56259143e-01 -1.14352629e-02 -2.41226837e-01 -6.59232354e-03
-3.57690573e-01 2.12210521e-01 3.38533930e-02 4.19051319e-01
5.59007108e-01 -1.39841115e+00 -8.70169640e-01 1.05881512e-01
2.36163214e-02 -2.14255035e-01 3.26904982e-01 7.74477720e-01
-4.48617816e-01 6.90639257e-01 -2.84473568e-01 7.39723295e-02
-8.02650094e-01 5.94900429e-01 3.31334233e-01 -8.33499432e-01
-6.62991405e-01 7.98966527e-01 1.80380747e-01 -8.20215523e-01
5.52927563e-03 -1.13386542e-01 -5.53726912e-01 -1.98614120e-01
6.49339914e-01 3.74114156e-01 8.98803174e-02 -7.46287763e-01
-5.46329200e-01 5.00299096e-01 -2.81570051e-02 2.30532885e-01
1.51484716e+00 -2.40157917e-01 -3.66854817e-01 5.88635445e-01
1.35875654e+00 -1.45820931e-01 -6.26737773e-01 -2.79175311e-01
4.43620831e-01 7.75631145e-02 1.32783756e-01 -5.75731397e-01
-9.18345571e-01 1.45628476e+00 3.73971879e-01 1.50274619e-01
9.72767711e-01 -4.28999007e-01 1.54191542e+00 5.10325432e-01
3.78678024e-01 -1.01630235e+00 -4.52093482e-01 1.18803155e+00
5.22814035e-01 -9.13106263e-01 -6.45401239e-01 1.42274469e-01
-5.16242743e-01 1.25783503e+00 5.91822863e-01 -3.44378620e-01
6.28534853e-01 5.85032046e-01 -6.44912422e-02 5.16742244e-02
-9.37730908e-01 -3.32668096e-01 3.39329019e-02 4.32909518e-01
8.64741385e-01 -1.55768767e-01 -6.07205153e-01 6.57946348e-01
-1.46622257e-02 6.02547452e-02 2.17973650e-01 5.19210458e-01
-4.94169235e-01 -1.11847174e+00 -6.42474741e-02 4.47548926e-01
-9.12283182e-01 -6.18200064e-01 -1.16201505e-01 3.69628429e-01
1.22976750e-01 8.83025408e-01 1.19664855e-01 -3.45779657e-01
3.62345219e-01 2.20289648e-01 8.06017220e-02 -6.52218878e-01
-1.05227530e+00 -5.08134425e-01 2.00881034e-01 -3.34521234e-01
-2.64180183e-01 -2.54956931e-02 -1.63878238e+00 -4.51693803e-01
-3.87142241e-01 6.06451988e-01 6.03419542e-01 1.12583780e+00
5.25425911e-01 7.71610558e-01 5.43717384e-01 -3.91234249e-01
-5.82230031e-01 -1.18758857e+00 -5.25534689e-01 2.80365080e-01
1.89837322e-01 -3.11333627e-01 -3.41947496e-01 -3.30870986e-01] | [9.896418571472168, 9.633264541625977] |
0bdc365f-1ea7-416b-99da-b74245a714ca | understanding-important-features-of-deep | 1912.06077 | null | https://arxiv.org/abs/1912.06077v1 | https://arxiv.org/pdf/1912.06077v1.pdf | Understanding Important Features of Deep Learning Models for Transmission Electron Microscopy Image Segmentation | Cutting edge deep learning techniques allow for image segmentation with great speed and accuracy. However, application to problems in materials science is often difficult since these complex models may have difficultly learning physical parameters. In situ electron microscopy provides a clear platform for utilizing automated image analysis. In this work we consider the case of studying coarsening dynamics in supported nanoparticles, which is important for understanding e.g. the degradation of industrial catalysts. By systematically studying dataset preparation, neural network architecture, and accuracy evaluation, we describe important considerations in applying deep learning to physical applications, where generalizable and convincing models are required. | ['James P. Horwath', 'Remi Megret', 'Eric A. Stach', 'Dmitri N. Zakharov'] | 2019-12-12 | null | null | null | null | ['electron-microscopy-image-segmentation'] | ['computer-vision'] | [ 3.60132754e-01 -3.77643853e-01 -2.87939347e-02 1.03322044e-01
-5.42788446e-01 -3.32520038e-01 3.22757274e-01 2.07579777e-01
-4.69819605e-01 8.95308554e-01 -6.09620810e-01 -7.09266305e-01
-2.02225819e-01 -6.15993142e-01 -8.60825360e-01 -1.17056000e+00
1.78989217e-01 7.34543622e-01 2.82490049e-02 2.72716191e-02
4.82249886e-01 9.98133004e-01 -1.18466794e+00 3.61843109e-01
5.30750871e-01 1.12313640e+00 6.35105789e-01 8.04606616e-01
-9.36331525e-02 5.85220218e-01 -3.83728832e-01 2.85159592e-02
-8.79349262e-02 -8.02801549e-02 -9.54657555e-01 1.56152248e-01
3.14490527e-01 -1.60366297e-01 1.30100042e-01 8.95067751e-01
2.80502379e-01 2.69204170e-01 9.68157351e-01 -5.20359755e-01
-6.33381546e-01 4.06133026e-01 -2.89467305e-01 4.06026572e-01
-1.91408724e-01 3.24547857e-01 7.07808793e-01 -6.64808631e-01
5.60228229e-01 8.11712086e-01 7.70728111e-01 6.13088608e-01
-1.32537305e+00 -2.92311937e-01 -5.25808595e-02 3.48803967e-01
-8.43301117e-01 -5.53263247e-01 8.59956384e-01 -6.41257584e-01
9.79785323e-01 -2.49101818e-02 5.96812010e-01 8.92341375e-01
7.66365886e-01 5.32084286e-01 1.14718497e+00 -3.94836575e-01
3.57854962e-01 -2.13108599e-01 6.15430295e-01 6.76876247e-01
5.30934274e-01 3.81889492e-02 6.15155809e-02 2.38644555e-01
1.10086882e+00 -6.13281243e-02 -3.84421274e-02 -2.67449945e-01
-9.36252236e-01 5.96910238e-01 4.34905022e-01 4.36705023e-01
-4.49005216e-01 4.64334577e-01 3.32117915e-01 7.08719119e-02
3.00099194e-01 9.30704236e-01 -5.26387334e-01 -1.94137886e-01
-9.31558967e-01 2.15747014e-01 6.19870722e-01 3.79386067e-01
7.25104272e-01 2.29210958e-01 5.73742092e-01 5.23632705e-01
4.88549247e-02 2.90495813e-01 9.14581865e-02 -1.37875211e+00
1.73291191e-02 3.45610797e-01 2.72647738e-01 -5.02593637e-01
-7.22800314e-01 -9.71505605e-03 -9.90052462e-01 6.79927945e-01
5.43207169e-01 -7.12704509e-02 -8.77003312e-01 1.08254898e+00
7.80332386e-02 -3.16823691e-01 2.30774861e-02 6.69254005e-01
6.62786245e-01 6.77362204e-01 4.70128134e-02 -3.45240116e-01
1.08041811e+00 -7.13015258e-01 -5.48523605e-01 -6.95365444e-02
5.60883343e-01 -5.52097321e-01 1.05707324e+00 7.36214519e-01
-1.17196226e+00 -4.55514699e-01 -1.07527292e+00 -3.13583821e-01
-6.63456976e-01 1.88810796e-01 9.37462211e-01 4.37686503e-01
-6.39362454e-01 1.28968704e+00 -1.48934233e+00 -3.77206057e-01
7.37977982e-01 7.74753034e-01 -1.33774653e-01 1.31112576e-01
-5.96485138e-01 9.27251577e-01 1.52803004e-01 1.32068649e-01
-7.24695921e-01 -7.12138712e-01 -5.74489892e-01 1.07584380e-01
3.90054844e-02 -5.84884763e-01 1.42474973e+00 -5.92397511e-01
-1.65297997e+00 6.43726647e-01 -2.49347001e-01 -5.43972373e-01
3.90589833e-01 6.51544109e-02 -1.00416325e-01 6.26970083e-02
-1.79185703e-01 5.45648873e-01 7.11895108e-01 -1.25954509e+00
-1.46902114e-01 -3.48911107e-01 7.51003698e-02 -3.46293896e-01
4.37469687e-03 -2.07533747e-01 2.00134948e-01 -1.56915218e-01
-1.59307271e-01 -7.42674887e-01 -4.89271522e-01 -4.99081537e-02
-2.18476310e-01 -3.04362234e-02 1.06366873e+00 -5.60418367e-01
6.91386461e-01 -1.70825720e+00 -3.03516705e-02 1.09489009e-01
4.89345938e-01 2.55282819e-01 1.90692991e-01 3.09622318e-01
1.19942665e-01 8.21905211e-02 -2.21175119e-01 2.54745968e-02
-2.45557249e-01 1.95748180e-01 1.34548441e-01 4.78008568e-01
2.88779885e-01 9.25626636e-01 -5.77417612e-01 -1.88385233e-01
5.97633898e-01 4.64108646e-01 -5.11694610e-01 -3.10403500e-02
-4.89325672e-01 5.65279782e-01 -4.64869440e-01 5.37137389e-01
3.67841542e-01 -8.45787406e-01 1.11401930e-01 -5.37469804e-01
-4.47261184e-01 2.34825149e-01 -7.85577655e-01 1.26843572e+00
-6.91331863e-01 1.02264190e+00 5.12730420e-01 -1.32775652e+00
5.62593460e-01 9.57921706e-03 6.78173900e-01 -7.15927362e-01
4.83915150e-01 1.62153229e-01 2.12482318e-01 -6.68423951e-01
4.55196708e-01 -5.86624444e-01 3.87564927e-01 5.68637490e-01
-2.72524387e-01 -2.43927270e-01 3.51754963e-01 -2.08740443e-01
8.33782136e-01 -2.75809109e-01 7.79257435e-03 -7.73414850e-01
2.36428231e-01 2.86033362e-01 6.14455789e-02 7.21469104e-01
-4.40817438e-02 4.94461566e-01 3.38643461e-01 -8.37124169e-01
-1.63732886e+00 -6.62348688e-01 -3.31184775e-01 5.51132262e-01
1.56271935e-01 1.41419560e-01 -9.63812768e-01 -6.78082258e-02
3.65624502e-02 1.87145814e-01 -6.36654198e-01 2.96148434e-02
-4.94614184e-01 -1.03905606e+00 3.50908376e-02 8.05458069e-01
1.84474736e-01 -1.12913954e+00 -6.21244133e-01 6.28811836e-01
3.81913006e-01 -1.31993210e+00 1.45367056e-01 5.42092264e-01
-9.21516001e-01 -1.43347037e+00 -5.46167612e-01 -6.92653656e-01
5.49162984e-01 3.19246680e-01 9.57246602e-01 2.76112497e-01
-6.90728009e-01 5.41024029e-01 1.42509431e-01 -3.75389427e-01
-8.31613958e-01 4.92445558e-01 8.77443403e-02 -4.80058849e-01
1.13963455e-01 -5.37302256e-01 -6.89647794e-01 9.29387361e-02
-7.67043114e-01 1.20497800e-01 4.04418916e-01 8.32789540e-01
7.08372056e-01 2.41192907e-01 5.19051492e-01 -9.37505603e-01
8.60973001e-01 -1.41960338e-01 -8.34858835e-01 2.28089094e-01
-7.73419797e-01 9.83664021e-03 9.76549506e-01 -3.17373842e-01
-1.02216601e+00 7.58452993e-03 -4.97834325e-01 -8.69601890e-02
-4.52080458e-01 3.44102770e-01 3.34444679e-02 -4.75515515e-01
6.50597990e-01 -7.48970956e-02 3.00566405e-01 -3.57920647e-01
1.75275002e-02 4.74104702e-01 3.96707475e-01 -1.05189216e+00
1.55920729e-01 7.22178042e-01 2.54303843e-01 -1.49297142e+00
-6.24878883e-01 -1.68428078e-01 -8.17855477e-01 -2.44995251e-01
1.01475215e+00 -3.95143956e-01 -1.07462513e+00 5.38402021e-01
-1.21649003e+00 -9.11531150e-01 -2.88578987e-01 3.20217997e-01
-7.48201728e-01 2.73079485e-01 -1.04047143e+00 -5.11306643e-01
-3.65137368e-01 -1.64477074e+00 9.46971118e-01 2.37182140e-01
-2.19018534e-01 -1.41361213e+00 -1.45033211e-01 4.09900963e-01
4.87809539e-01 4.02997956e-02 1.10227454e+00 -2.11483553e-01
-5.40725946e-01 -3.31782848e-02 -2.27317244e-01 3.19851190e-01
4.18249905e-01 5.36905885e-01 -9.30332303e-01 -2.86873311e-01
-1.21235391e-02 -3.44739407e-01 9.18403327e-01 8.97697449e-01
1.56340241e+00 9.77727175e-02 -3.68427455e-01 5.14214694e-01
1.42220509e+00 3.31790566e-01 5.30721962e-01 4.50653970e-01
1.00584209e+00 2.71133721e-01 3.29939574e-02 1.21628955e-01
-1.46331772e-01 3.63583714e-01 1.42491177e-01 -2.87552416e-01
-1.90593183e-01 4.64032501e-01 6.77679926e-02 8.64061177e-01
-4.97865498e-01 -1.91204637e-01 -1.04084659e+00 1.45807758e-01
-1.42348337e+00 -9.80139434e-01 -2.52121598e-01 1.68786001e+00
5.17527819e-01 2.59224385e-01 1.17504261e-02 2.76048928e-01
7.11799383e-01 -5.88532463e-02 -7.99187005e-01 -5.02218485e-01
3.52999680e-02 3.99359614e-01 5.85438848e-01 6.79310977e-01
-9.85176325e-01 7.55027115e-01 7.05441713e+00 6.11450434e-01
-1.56531107e+00 1.61614711e-03 8.74047995e-01 1.51127100e-01
-1.39173105e-01 -2.40684882e-01 -7.47626364e-01 4.21030015e-01
8.34402323e-01 2.30562881e-01 7.67363846e-01 5.03798127e-01
4.93521571e-01 -2.29581922e-01 -1.21478319e+00 8.01928937e-01
-5.80103934e-01 -2.13167977e+00 4.98186201e-02 1.66026980e-01
8.43970358e-01 2.42817089e-01 4.06415612e-02 -1.85519159e-01
2.18279272e-01 -9.43167865e-01 4.27131802e-01 3.80798966e-01
7.58847952e-01 -5.74688733e-01 4.72626179e-01 8.15944374e-02
-8.65764558e-01 4.43818346e-02 -6.57361448e-01 -3.40310425e-01
1.31023973e-01 7.38486528e-01 -6.53687537e-01 6.12713918e-02
5.95979631e-01 6.58148468e-01 4.96749543e-02 7.54391134e-01
4.10866261e-01 5.10179341e-01 -4.26498532e-01 -2.22707257e-01
4.18263853e-01 -4.64600384e-01 2.72275545e-02 1.27488816e+00
2.50384927e-01 1.30874217e-01 -4.72415239e-02 1.06274283e+00
-1.75264806e-01 -2.64432430e-01 -5.71805954e-01 -3.01703185e-01
1.45652086e-01 1.30867362e+00 -1.37430906e+00 -1.64447337e-01
-3.51216674e-01 5.02363741e-01 3.99155468e-01 3.17567438e-01
-5.07710636e-01 -4.63237539e-02 6.64032936e-01 4.70225602e-01
3.63601983e-01 -7.36709237e-01 -5.41201532e-01 -9.26855862e-01
-3.51193309e-01 -5.07931530e-01 -2.48787314e-01 -6.41535342e-01
-1.12587917e+00 -3.58685246e-03 -3.18165511e-01 -3.87722582e-01
2.66181797e-01 -1.52871394e+00 -8.43024194e-01 3.15685183e-01
-1.41243672e+00 -9.43131208e-01 -3.35737050e-01 1.45741329e-01
6.88836277e-01 -8.83044899e-02 5.38367093e-01 2.17120871e-01
-9.08445001e-01 5.38132973e-02 7.00775504e-01 -2.27418005e-01
4.67262492e-02 -1.23523724e+00 5.00768304e-01 4.19642717e-01
-1.55366845e-02 4.12941396e-01 7.80211687e-01 -5.10565639e-01
-1.84755886e+00 -9.61680233e-01 2.87005436e-02 -2.84813136e-01
9.71822977e-01 -1.98901251e-01 -1.17518032e+00 4.55209404e-01
1.43498868e-01 -1.12172648e-01 5.10813475e-01 -6.98158294e-02
4.25073653e-01 5.63223511e-02 -1.08072495e+00 4.93884265e-01
5.75130999e-01 -4.99086678e-01 -1.28445961e-02 4.89407599e-01
1.75842628e-01 -2.86685288e-01 -1.13345671e+00 2.01498702e-01
6.52284682e-01 -7.54315674e-01 9.25936520e-01 -8.46408188e-01
5.61415493e-01 1.83625966e-01 3.92793901e-02 -1.06121874e+00
-5.35646439e-01 -5.10801494e-01 -4.82017659e-02 9.21169400e-01
5.56359708e-01 -4.12615627e-01 1.13060629e+00 7.71290064e-01
-3.05876613e-01 -9.27187979e-01 -6.21782899e-01 -8.66931260e-01
6.12800598e-01 -3.85935724e-01 4.32612538e-01 8.11808944e-01
-2.02285603e-01 2.89624214e-01 1.72392771e-01 8.17646831e-02
6.04615629e-01 1.51527092e-01 3.08366090e-01 -1.44690776e+00
1.01424620e-01 -6.66442513e-01 -2.23099664e-02 -8.77759099e-01
3.55196506e-01 -5.25494754e-01 -2.09588986e-02 -1.54177237e+00
2.85077304e-01 -5.79929173e-01 -2.02341974e-01 -4.85484116e-02
1.63523287e-01 1.20304361e-01 -1.43874735e-01 2.78236389e-01
-5.13814509e-01 2.92549163e-01 1.43453348e+00 -4.24302489e-01
2.66208444e-02 -7.87140951e-02 -4.58864301e-01 8.27700019e-01
1.16130412e+00 -1.48022428e-01 -2.12563425e-01 -6.32633150e-01
2.99646556e-01 -1.02069139e-01 4.29073006e-01 -1.18491864e+00
3.07497799e-01 -3.43581200e-01 4.89886403e-01 -3.36293846e-01
2.95038760e-01 -8.88893008e-01 1.30084753e-02 4.39494193e-01
-2.86488116e-01 5.86262085e-02 2.30054349e-01 3.48847777e-01
8.96257833e-02 -4.59805936e-01 1.11202049e+00 -6.82264924e-01
-6.33840561e-01 4.76438582e-01 -7.86139727e-01 -1.58404693e-01
7.83821881e-01 -3.45170587e-01 -3.59807700e-01 -3.66706843e-03
-1.05749428e+00 -7.68672004e-02 9.26270306e-01 -2.96991467e-01
5.55923998e-01 -1.00318170e+00 -2.05375589e-02 9.82111767e-02
-4.59558040e-01 2.56142497e-01 2.88663715e-01 6.85274661e-01
-9.71521378e-01 3.13736975e-01 -3.98563564e-01 -6.72710538e-01
-9.00383532e-01 5.76088369e-01 6.40687227e-01 -2.90197492e-01
-4.71615434e-01 5.81985056e-01 1.65423691e-01 -6.89300224e-02
-1.33029357e-01 -6.79615676e-01 8.44215080e-02 -1.83546707e-01
2.26702780e-01 6.33393824e-01 4.73460883e-01 -1.51774257e-01
-2.44072210e-02 5.94787419e-01 -3.86444747e-01 3.59299839e-01
1.63224423e+00 -3.41142267e-02 -1.54530123e-01 4.11414653e-01
1.11139953e+00 -5.10355651e-01 -1.67743599e+00 2.78071016e-01
4.53323536e-02 4.05109942e-01 2.74989069e-01 -1.29137948e-01
-1.12707794e+00 1.43964469e+00 5.00555992e-01 4.70773816e-01
6.84826791e-01 1.27986014e-01 9.29288030e-01 9.98122811e-01
1.64280951e-01 -1.31875944e+00 -4.98657208e-03 5.73898137e-01
4.83300328e-01 -1.35627544e+00 1.88388824e-01 -1.78745955e-01
1.16067149e-01 1.62194288e+00 4.93251801e-01 -2.24534750e-01
7.09872663e-01 6.52705252e-01 -1.27436981e-01 -5.81217647e-01
-6.74050391e-01 2.31123716e-01 -2.55176753e-01 6.29459500e-01
5.13239980e-01 1.00274354e-01 1.83754832e-01 3.49707484e-01
2.33934168e-02 -9.17277485e-02 7.54350305e-01 9.34964895e-01
-8.07660699e-01 -9.82231021e-01 -2.82325357e-01 7.76854217e-01
-5.79121649e-01 6.52730986e-02 -2.15606883e-01 8.28819036e-01
-1.01627700e-01 3.42618883e-01 2.88083225e-01 -4.55011353e-02
3.14111784e-02 -5.48184887e-02 1.07200301e+00 -3.96966547e-01
-1.74131140e-01 -1.84565485e-01 6.63405433e-02 -9.65907872e-02
-5.74179411e-01 -7.85521150e-01 -1.48326039e+00 -4.79128957e-01
-4.82145190e-01 1.17484942e-01 8.20158541e-01 1.25571442e+00
1.49585858e-01 6.16680086e-01 3.59200925e-01 -1.45151746e+00
-1.22454949e-01 -6.17652595e-01 -6.99285448e-01 1.37274265e-01
6.23396575e-01 -7.48475194e-01 -2.18057066e-01 3.05085629e-01] | [14.128927230834961, -2.8544318675994873] |
953ff21d-201a-4694-9552-0a82ca3ea40a | regression-forest-based-atlas-localization | 2005.03345 | null | https://arxiv.org/abs/2005.03345v1 | https://arxiv.org/pdf/2005.03345v1.pdf | Regression Forest-Based Atlas Localization and Direction Specific Atlas Generation for Pancreas Segmentation | This paper proposes a fully automated atlas-based pancreas segmentation method from CT volumes utilizing atlas localization by regression forest and atlas generation using blood vessel information. Previous probabilistic atlas-based pancreas segmentation methods cannot deal with spatial variations that are commonly found in the pancreas well. Also, shape variations are not represented by an averaged atlas. We propose a fully automated pancreas segmentation method that deals with two types of variations mentioned above. The position and size of the pancreas is estimated using a regression forest technique. After localization, a patient-specific probabilistic atlas is generated based on a new image similarity that reflects the blood vessel position and direction information around the pancreas. We segment it using the EM algorithm with the atlas as prior followed by the graph-cut. In evaluation results using 147 CT volumes, the Jaccard index and the Dice overlap of the proposed method were 62.1% and 75.1%, respectively. Although we automated all of the segmentation processes, segmentation results were superior to the other state-of-the-art methods in the Dice overlap. | ['Kensaku MORI', 'Takayuki Kitasaka', "Ken'ichi Karasawa", 'Masahiro Oda', 'Yukitaka Nimura', 'Michitaka Fujiwara', 'Daniel Rueckert', 'Kazunari Misawa', 'Natsuki Shimizu'] | 2020-05-07 | null | null | null | null | ['pancreas-segmentation', 'automated-pancreas-segmentation'] | ['medical', 'medical'] | [-4.48929787e-01 1.75844654e-01 1.47428215e-01 -4.68820661e-01
-5.02601564e-01 -7.72127986e-01 3.19236517e-01 5.24354219e-01
-3.83299649e-01 7.40087271e-01 3.16315889e-01 -7.75060132e-02
-1.50410622e-01 -6.99415326e-01 -4.69107538e-01 -9.53710556e-01
-2.51659870e-01 1.03115296e+00 5.07070184e-01 3.12469929e-01
3.01675379e-01 7.51113653e-01 -4.90383208e-01 -2.36745209e-01
1.14618063e+00 6.64658964e-01 -1.06787831e-02 4.65940148e-01
-6.12353265e-01 4.48342621e-01 -4.22265738e-01 -2.92694747e-01
6.10027730e-01 -7.49150276e-01 -2.53635615e-01 1.56564936e-01
1.09106019e-01 -1.51903600e-01 -9.90286842e-02 1.23384964e+00
3.00231397e-01 -2.47804403e-01 1.03399885e+00 -1.09672976e+00
-1.40260577e-01 1.05862629e+00 -7.22523451e-01 4.73702401e-02
1.98856592e-01 6.19476996e-02 9.90527794e-02 -2.65650392e-01
6.98795855e-01 9.30965066e-01 1.00874543e+00 -6.17764017e-04
-1.20250058e+00 -6.20784640e-01 -1.38406143e-01 -2.19768927e-01
-1.38388014e+00 4.83304799e-01 6.49735570e-01 -8.32019091e-01
5.04881553e-02 5.05168326e-02 1.23592257e+00 1.73484221e-01
8.24368954e-01 2.55541444e-01 1.59170806e+00 -3.34052742e-01
2.52380759e-01 1.04537979e-02 1.25006497e-01 8.40443671e-01
6.75494075e-01 -9.80002247e-03 3.44326645e-01 -2.34904408e-01
1.12655449e+00 4.37953919e-02 -2.31427848e-01 -9.48174298e-01
-1.77038193e+00 9.56794679e-01 6.48036301e-01 5.39680362e-01
-8.09713840e-01 -2.54248023e-01 3.79980296e-01 -3.78251374e-01
-1.40124530e-01 3.39038491e-01 -5.54777198e-02 2.98673213e-01
-1.42792857e+00 -1.05432041e-01 1.29388654e+00 1.11803448e+00
2.01111183e-01 -2.06321567e-01 -1.05131671e-01 1.99789330e-01
8.51556182e-01 3.88474733e-01 6.03589237e-01 -7.32097507e-01
-3.41033028e-03 7.79446840e-01 -2.58579463e-01 -7.86264718e-01
-7.68901646e-01 -4.10567284e-01 -9.72471118e-01 3.82086039e-01
1.12432253e+00 -3.66024733e-01 -1.26599038e+00 1.14177287e+00
6.43009663e-01 -3.60392481e-02 -7.96540454e-02 1.07713509e+00
8.73204648e-01 3.00821900e-01 4.49231476e-01 -3.99065942e-01
1.37076080e+00 -9.13108885e-01 -9.19443905e-01 6.31779075e-01
4.07738447e-01 -1.03677714e+00 2.07908943e-01 4.20917451e-01
-9.97634947e-01 -1.30721536e-02 -9.51814175e-01 6.28669024e-01
-9.48618874e-02 -1.40686974e-01 4.78438288e-01 7.77932107e-01
-9.18479443e-01 4.87613380e-01 -1.09099257e+00 -6.86124206e-01
4.68603313e-01 1.76515624e-01 -3.37061942e-01 1.43914431e-01
-4.66524631e-01 1.33541048e+00 4.43650603e-01 -1.58455089e-01
-6.14056945e-01 -6.14147246e-01 -7.67271817e-01 -1.31514773e-01
-1.14967316e-01 -8.21410000e-01 9.35445368e-01 -7.51608908e-01
-1.77152383e+00 6.65150404e-01 1.84799731e-01 -5.13504982e-01
1.15958774e+00 4.52992290e-01 -2.85140276e-02 3.53460521e-01
4.61124405e-02 5.72119355e-01 1.71209142e-01 -1.49969089e+00
-3.26499552e-01 -4.19566810e-01 -6.31082714e-01 2.14075029e-01
7.46067047e-01 -1.19240358e-02 -7.49537870e-02 -5.74333787e-01
9.11990583e-01 -1.01303792e+00 -7.40771472e-01 4.00514781e-01
-4.85142499e-01 1.70761228e-01 3.35100681e-01 -1.06834018e+00
7.72774041e-01 -1.83587360e+00 -1.76292598e-01 7.92548418e-01
3.88265312e-01 -2.16158539e-01 4.85227466e-01 -1.20318485e-02
1.31833911e-01 -1.69161871e-01 -6.80462301e-01 3.02040726e-01
-1.21823385e-01 1.33938208e-01 4.19781595e-01 9.30305958e-01
-6.64772928e-01 5.34061015e-01 -1.00032711e+00 -1.24164546e+00
4.86957431e-01 4.90797997e-01 -1.68587685e-01 1.06980436e-01
2.93136477e-01 7.67653346e-01 -2.79813766e-01 6.59026086e-01
9.86003935e-01 2.46763214e-01 3.55214596e-01 -5.55906117e-01
-3.70519072e-01 -4.35626119e-01 -1.27550519e+00 1.90685284e+00
4.49902564e-02 1.36627614e-01 1.02145627e-01 -5.63674033e-01
1.30296862e+00 5.06617665e-01 1.17466617e+00 -2.38742247e-01
2.82300681e-01 4.13672447e-01 4.18024927e-01 -3.27220351e-01
-1.64788708e-01 -4.81482625e-01 1.82544142e-01 2.17363000e-01
2.08626807e-01 -5.67537069e-01 3.51619184e-01 1.35178462e-01
7.53263652e-01 3.12600225e-01 7.35017598e-01 -9.44597304e-01
4.23770636e-01 5.50123155e-01 4.70247328e-01 4.62481886e-01
-6.81896925e-01 7.19498098e-01 4.05011475e-01 -5.03395438e-01
-1.01217425e+00 -1.52791595e+00 -7.25523770e-01 -1.95224509e-02
3.79593283e-01 6.61889762e-02 -1.05032027e+00 -9.73416448e-01
1.73404425e-01 7.21663177e-01 -4.27060574e-01 5.19578516e-01
-4.56287324e-01 -1.09294248e+00 2.70824343e-01 2.74107307e-01
4.43351507e-01 -7.31304526e-01 -5.99512696e-01 5.18584430e-01
4.84075584e-02 -7.47360229e-01 -4.63453740e-01 -1.95937585e-02
-1.52189052e+00 -1.22670138e+00 -1.20698833e+00 -7.93573201e-01
1.05131721e+00 -3.38746607e-01 1.08195019e+00 -3.03157330e-01
-2.87413746e-01 1.53754860e-01 -2.52625644e-01 -3.11911404e-01
-6.21655941e-01 -2.86309153e-01 -2.92962283e-01 -3.69108588e-01
1.62803724e-01 -5.85448563e-01 -9.23655868e-01 5.11294246e-01
-5.60363531e-01 -9.58123952e-02 8.04644108e-01 3.62144917e-01
9.12196398e-01 -2.50640154e-01 2.90554255e-01 -7.83479810e-01
3.62863481e-01 -6.80214345e-01 -8.84875715e-01 6.79348856e-02
-6.42746031e-01 9.74988714e-02 2.35614941e-01 -4.97352988e-01
-8.76544535e-01 6.91993296e-01 2.00466931e-01 -1.84968282e-02
-4.51153934e-01 2.42320240e-01 1.62898153e-01 -3.59094203e-01
4.86216962e-01 1.16582207e-01 4.21181619e-01 -2.88931161e-01
3.84536535e-01 2.05969140e-01 3.99031579e-01 -1.97367460e-01
4.20903534e-01 3.42770398e-01 4.49122310e-01 -2.47091860e-01
7.91896284e-02 -3.75612676e-01 -1.26918268e+00 -3.88340771e-01
1.26584792e+00 -4.61272120e-01 -4.45812285e-01 6.55662492e-02
-9.07374084e-01 7.51613313e-03 -4.15245056e-01 1.21534765e+00
-4.83435124e-01 6.40889704e-01 -5.74105382e-01 -4.56940353e-01
-5.18554091e-01 -1.51792002e+00 3.12957197e-01 3.57636064e-01
-1.81435302e-01 -1.11059475e+00 3.68365765e-01 -2.10783824e-01
6.88873410e-01 8.99271607e-01 6.99555218e-01 -9.16900635e-01
-5.04223883e-01 -2.99230725e-01 -1.57370329e-01 -1.43953368e-01
4.24489044e-02 1.04583554e-01 -1.77836433e-01 1.87655360e-01
4.61626463e-02 7.88870454e-01 3.22949797e-01 1.11216140e+00
6.77502871e-01 -2.80321300e-01 -4.91849929e-01 6.32225931e-01
1.69336390e+00 6.34683311e-01 4.15062249e-01 3.02459896e-01
5.44287443e-01 3.41218203e-01 3.06684583e-01 3.90238792e-01
4.67581511e-01 3.45040053e-01 4.47587013e-01 -2.84374952e-01
-3.26373279e-01 -3.24634314e-02 -1.26440883e-01 1.13277364e+00
-1.65066451e-01 2.73909569e-01 -1.06866372e+00 5.54842412e-01
-1.51839912e+00 -5.05131304e-01 -7.52201915e-01 2.37286663e+00
7.53783047e-01 -1.61617428e-01 3.95907134e-01 -3.80259752e-01
9.59463716e-01 -5.06117284e-01 -1.65933371e-01 -3.01568002e-01
1.76069230e-01 -5.29222414e-02 1.08332634e+00 5.77084959e-01
-1.06206989e+00 4.02992189e-01 6.66563129e+00 -3.02863470e-03
-1.01623201e+00 1.16893716e-01 3.66099417e-01 3.35615247e-01
6.16587773e-02 9.88100618e-02 -2.94240087e-01 6.86249673e-01
5.73995531e-01 -2.08908111e-01 1.30729824e-01 7.21125722e-01
1.31454822e-02 -5.74240506e-01 -7.16105759e-01 7.63260663e-01
5.85399568e-02 -7.64281332e-01 -1.54033974e-01 1.93951249e-01
8.04591477e-01 1.33505523e-01 -6.54047191e-01 -1.02858350e-01
4.65315551e-01 -7.40189314e-01 4.97346312e-01 1.08235526e+00
3.87069255e-01 -4.39991444e-01 1.24098217e+00 -3.17770354e-02
-1.11869705e+00 4.75246698e-01 -8.29440504e-02 5.08067787e-01
5.41534364e-01 8.68606150e-01 -1.20707393e+00 5.21276295e-01
3.32567781e-01 1.23838633e-01 -4.39034253e-01 2.17689013e+00
-2.13258892e-01 4.22986478e-01 -7.55947649e-01 -1.18193112e-01
1.69105083e-01 -8.39979827e-01 7.91283906e-01 1.44717014e+00
6.74052596e-01 -4.38031033e-02 2.62324244e-01 1.02159011e+00
2.91255891e-01 7.12819993e-01 -4.27514672e-01 5.00932038e-01
2.80092239e-01 1.55030859e+00 -1.46021402e+00 -6.16045356e-01
-2.53115714e-01 6.51951969e-01 -6.03116214e-01 -6.02027215e-02
-8.80857050e-01 -1.23422109e-01 -1.84377685e-01 3.31679076e-01
-1.01627514e-01 -1.54873058e-01 -5.78649819e-01 -9.52216983e-01
-3.07761669e-01 -3.29906493e-01 5.64129472e-01 -5.46323419e-01
-1.35863435e+00 8.24829817e-01 3.06393594e-01 -1.20794082e+00
-1.57166123e-01 -4.33475435e-01 -7.05522776e-01 8.66050303e-01
-9.92799222e-01 -9.83381391e-01 -6.33531392e-01 3.06334317e-01
2.23561838e-01 1.51366919e-01 8.80693614e-01 1.15394197e-01
-9.39419568e-02 2.20006883e-01 2.32295752e-01 2.26216391e-01
7.18637645e-01 -1.70099103e+00 -4.19064552e-01 6.73221469e-01
-1.89996004e-01 3.24281007e-01 6.17831290e-01 -9.50400114e-01
-9.55629528e-01 -8.59276712e-01 6.16036534e-01 -1.72913715e-01
5.45570433e-01 2.06803143e-01 -6.34654284e-01 7.85360456e-01
4.75955039e-01 3.27136517e-01 6.50948644e-01 -5.58659315e-01
2.61505514e-01 3.00363172e-03 -1.93857098e+00 3.11572999e-01
3.70328307e-01 5.35972595e-01 -7.98509598e-01 2.90375382e-01
-5.50922379e-02 -5.97516537e-01 -1.61212361e+00 4.13677841e-01
7.93889582e-01 -8.03492367e-01 7.09704220e-01 9.38740969e-02
-1.05717756e-01 -6.28861427e-01 8.86917934e-02 -1.41816282e+00
-2.81290889e-01 -4.22766149e-01 3.08385104e-01 9.69637752e-01
4.26985383e-01 -7.28538454e-01 6.19620979e-01 6.63853228e-01
-1.72666907e-01 -5.11287570e-01 -9.60908711e-01 -4.01529968e-01
3.68318230e-01 3.05393457e-01 5.27374208e-01 1.17305505e+00
1.77956760e-01 -3.70679796e-01 5.62934935e-01 1.33495986e-01
1.34351242e+00 2.12726891e-02 4.82775927e-01 -1.52319491e+00
2.98550725e-01 -7.02747941e-01 -7.46235132e-01 -3.00363332e-01
-4.90699440e-01 -1.04443777e+00 1.64021820e-01 -1.98713374e+00
3.94739598e-01 -5.37402093e-01 -2.78679341e-01 4.57283035e-02
1.49000570e-01 -2.25096513e-02 2.80889034e-01 2.93399394e-01
-2.24773243e-01 1.87288467e-02 1.72780871e+00 2.97858864e-01
-3.42476219e-01 5.37446812e-02 -1.93551406e-01 9.17629123e-01
9.21054482e-01 -7.46866822e-01 2.04007462e-01 2.62524754e-01
-6.34670436e-01 3.84895563e-01 9.23216417e-02 -1.22950685e+00
4.40110475e-01 1.01508535e-01 6.28562629e-01 -7.73119152e-01
-3.45455229e-01 -1.32639158e+00 5.82605183e-01 9.49273348e-01
-5.95046207e-03 3.25703532e-01 -2.07983330e-02 1.64447978e-01
-2.98925061e-02 -4.89366323e-01 1.09904051e+00 -5.76690316e-01
-7.50819873e-03 7.24747851e-02 -6.15419090e-01 -2.76435047e-01
1.33551109e+00 -2.50732362e-01 1.37327120e-01 -2.37132430e-01
-1.10858750e+00 -6.71412870e-02 4.85764056e-01 -3.87690574e-01
2.35514686e-01 -1.26679349e+00 -8.83488834e-01 7.97997415e-02
-2.31523231e-01 1.96321562e-01 3.95621210e-02 1.59403419e+00
-1.36188245e+00 1.58184901e-01 -6.92517340e-01 -8.90814602e-01
-9.60060239e-01 6.12328529e-01 7.09596634e-01 -3.48354757e-01
-8.51863027e-01 2.00982496e-01 -8.01899284e-03 -5.32658458e-01
1.17739879e-01 -8.76378477e-01 -2.34978452e-01 -8.78427848e-02
9.23548825e-04 4.71587926e-01 -3.21233124e-01 -8.71364594e-01
-3.48701924e-01 8.43807995e-01 4.33903545e-01 -1.24124289e-01
1.02130556e+00 -1.53577745e-01 -2.43567795e-01 1.09133624e-01
6.96919978e-01 1.38307810e-01 -1.16319025e+00 6.57536760e-02
9.01450589e-02 -3.04335326e-01 2.33286679e-01 -1.30857623e+00
-1.31446338e+00 5.60511827e-01 1.09106827e+00 6.75685182e-02
8.27605784e-01 -2.32708633e-01 6.17814481e-01 -4.96480763e-01
3.87611926e-01 -5.65168023e-01 -7.60411501e-01 7.07350671e-02
7.84877777e-01 -1.12822056e+00 2.24367782e-01 -5.57697177e-01
-8.05609226e-01 1.48116374e+00 2.23932847e-01 -5.46301365e-01
9.52225804e-01 6.38916910e-01 4.94018346e-01 -2.97463406e-02
3.02615464e-01 6.64641783e-02 3.73355746e-01 6.32570624e-01
5.02569973e-01 4.79673743e-01 -9.63761389e-01 5.26795506e-01
-2.03683868e-01 4.99119014e-02 4.36108261e-01 8.12631905e-01
-4.35984075e-01 -7.54316151e-01 -8.07784081e-01 3.55256587e-01
-5.96410036e-01 9.35674459e-02 -3.37208286e-02 1.10441947e+00
5.45315854e-02 4.40676570e-01 2.16944497e-02 1.85923427e-01
3.20777655e-01 1.72921166e-01 7.20401585e-01 5.12439152e-03
-7.41797447e-01 3.64741653e-01 -3.06797564e-01 -3.98134470e-01
-3.93586576e-01 -8.27911139e-01 -1.60931945e+00 1.03570800e-02
-3.78014594e-01 3.17384273e-01 1.35753310e+00 6.59676075e-01
-1.43734425e-01 6.05428278e-01 3.02489877e-01 -8.71769845e-01
-3.73631418e-01 -1.13568413e+00 -7.54574597e-01 4.41231072e-01
-1.28847584e-01 -7.30852187e-01 -3.26316506e-01 1.18661553e-01] | [14.43752670288086, -2.731698989868164] |
aa4fde92-b2a7-44d2-a2fc-af2977ffceaf | topic-modeling-of-hierarchical-corpora | 1409.3518 | null | http://arxiv.org/abs/1409.3518v2 | http://arxiv.org/pdf/1409.3518v2.pdf | Topic Modeling of Hierarchical Corpora | We study the problem of topic modeling in corpora whose documents are
organized in a multi-level hierarchy. We explore a parametric approach to this
problem, assuming that the number of topics is known or can be estimated by
cross-validation. The models we consider can be viewed as special
(finite-dimensional) instances of hierarchical Dirichlet processes (HDPs). For
these models we show that there exists a simple variational approximation for
probabilistic inference. The approximation relies on a previously unexploited
inequality that handles the conditional dependence between Dirichlet latent
variables in adjacent levels of the model's hierarchy. We compare our approach
to existing implementations of nonparametric HDPs. On several benchmarks we
find that our approach is faster than Gibbs sampling and able to learn more
predictive models than existing variational methods. Finally, we demonstrate
the large-scale viability of our approach on two newly available corpora from
researchers in computer security---one with 350,000 documents and over 6,000
internal subcategories, the other with a five-level deep hierarchy. | ['Geoffrey M. Voelker', 'Do-kyum Kim', 'Lawrence K. Saul'] | 2014-09-11 | null | null | null | null | ['computer-security'] | ['miscellaneous'] | [-1.94199950e-01 5.72834671e-01 -3.88916314e-01 -3.97985458e-01
-1.28463376e+00 -4.18932885e-01 9.20329392e-01 1.53651625e-01
-3.47305864e-01 8.66018236e-01 4.36123490e-01 -4.58948702e-01
1.39412642e-01 -8.90059531e-01 -6.93389118e-01 -7.99644172e-01
-3.30438763e-01 1.42251468e+00 6.83274209e-01 1.26201898e-01
3.69408488e-01 1.19802594e-01 -1.29073179e+00 1.62442490e-01
4.65289921e-01 4.93669242e-01 -1.04277685e-01 7.64569700e-01
-2.76068896e-01 3.19778591e-01 -6.02579892e-01 -3.74809951e-01
-2.06062555e-01 2.18940079e-01 -1.21241939e+00 1.46260157e-01
9.24510732e-02 -1.55072704e-01 1.30635425e-02 8.40228140e-01
1.48652583e-01 5.21908179e-02 1.18615651e+00 -1.31112421e+00
-4.94983718e-02 7.66120017e-01 -8.81880224e-01 2.79412448e-01
1.04539365e-01 -4.51574177e-01 1.20818436e+00 -6.24522567e-01
6.48724377e-01 1.71581507e+00 7.48324811e-01 3.63156050e-01
-1.95494258e+00 -4.18681413e-01 2.15120167e-01 -1.87320411e-01
-1.40202558e+00 -2.56331325e-01 3.35514665e-01 -7.26169646e-01
7.89205194e-01 -4.50550802e-02 4.57535535e-01 1.16289771e+00
2.85865545e-01 7.14264214e-01 1.11664832e+00 -4.68235403e-01
6.03205681e-01 4.46262211e-01 6.07851207e-01 4.70281780e-01
4.06084299e-01 -3.11448812e-01 -5.20135820e-01 -1.27190685e+00
7.23690510e-01 -2.54478633e-01 2.35838313e-02 -5.18171370e-01
-7.33377695e-01 1.37960362e+00 -4.19660449e-01 2.74351209e-01
-2.43125468e-01 2.04330072e-01 2.34303564e-01 -1.92885831e-01
8.81129861e-01 -1.41286924e-01 -5.24044394e-01 -3.00472409e-01
-1.49465656e+00 2.66493678e-01 1.56026113e+00 1.01082611e+00
7.51120389e-01 -3.77916068e-01 6.48921579e-02 6.91319466e-01
6.95659101e-01 -3.15722600e-02 1.93843573e-01 -9.94933784e-01
2.25056484e-01 -2.19700128e-01 4.98256087e-01 -5.00804067e-01
-2.91550532e-02 1.39337480e-01 -5.28857410e-01 -2.81904131e-01
3.22216183e-01 -9.08185169e-02 -7.31242776e-01 1.75490475e+00
5.07244587e-01 2.43269339e-01 -1.50723815e-01 5.02323620e-02
8.86194035e-02 1.06442618e+00 5.17998576e-01 -4.95523781e-01
1.57410133e+00 -3.24528784e-01 -6.05420470e-01 2.06338137e-01
3.39821249e-01 -6.20604396e-01 8.28493655e-01 6.91357613e-01
-1.22820103e+00 -1.62192300e-01 -7.03506529e-01 -2.21669257e-01
-1.84873626e-01 -3.29506457e-01 6.28437877e-01 1.05765951e+00
-1.44217968e+00 4.48238283e-01 -1.28849161e+00 -4.01416451e-01
1.60575449e-01 3.17018092e-01 1.05650634e-01 1.14544250e-01
-1.31759048e+00 5.71504176e-01 4.32232589e-01 -4.13900077e-01
-1.25486124e+00 -5.66101909e-01 -4.99697864e-01 1.74410760e-01
3.22977155e-02 -3.65063697e-01 1.49371707e+00 -5.80714270e-02
-1.36300910e+00 8.67154539e-01 -5.36460280e-01 -5.67996025e-01
4.68941182e-01 -1.75702676e-01 1.12981498e-01 2.21809119e-01
2.80599505e-01 6.01783156e-01 8.65413129e-01 -1.55172133e+00
-7.11353838e-01 -3.28438491e-01 -1.74781512e-02 -7.29980692e-02
-4.12031442e-01 1.41859859e-01 -6.49232745e-01 -3.36299807e-01
1.25535324e-01 -8.82944465e-01 -4.03935343e-01 -5.63557208e-01
-6.87053323e-01 -7.62375772e-01 6.82456672e-01 -7.38646865e-01
1.20170522e+00 -1.91708410e+00 1.50923833e-01 3.98450375e-01
2.13065758e-01 -5.45010805e-01 5.91693044e-01 6.79829180e-01
1.62241802e-01 6.37861729e-01 -3.21826041e-01 -8.84150326e-01
3.92168015e-01 3.96389693e-01 -6.60382569e-01 5.91029763e-01
-2.45101333e-01 1.53898001e-01 -4.72582579e-01 -1.06568301e+00
-2.82406569e-01 5.08710146e-01 -8.35673928e-01 -1.01324720e-02
-5.31497300e-01 -2.66307481e-02 -5.47601879e-01 2.21265629e-01
7.02145278e-01 -3.68004143e-01 5.83391368e-01 3.02603185e-01
-5.97547553e-02 4.66264188e-01 -1.19500959e+00 1.36240268e+00
-2.70484060e-01 5.20901918e-01 3.32525074e-01 -8.27266634e-01
6.09568179e-01 5.31082034e-01 5.76269686e-01 5.20418525e-01
-1.15639031e-01 -2.12513641e-01 -6.22639775e-01 6.99906945e-02
8.20590973e-01 -5.09570181e-01 -3.00938874e-01 6.73292994e-01
1.07990302e-01 -3.08221251e-01 2.16576621e-01 5.98699212e-01
8.70156705e-01 -3.39480750e-02 1.91172719e-01 -8.26320410e-01
-1.00695930e-01 1.50984272e-01 5.61634183e-01 1.09135330e+00
-4.80843075e-02 2.65492588e-01 1.26273072e+00 -1.07065424e-01
-1.17623591e+00 -1.34914589e+00 -8.32744062e-01 1.41796339e+00
-2.55911022e-01 -6.83941245e-01 -1.03776515e+00 -3.47535461e-01
-1.40606150e-01 9.13200200e-01 -7.45291591e-01 4.24295098e-01
-2.62950778e-01 -1.30337107e+00 5.33248544e-01 3.72811347e-01
1.26031369e-01 -5.63037753e-01 -1.74402684e-01 2.32000723e-01
-2.66490072e-01 -9.69425678e-01 4.91283555e-03 3.95282567e-01
-1.18668306e+00 -7.68744826e-01 -6.25401676e-01 -6.08904421e-01
4.11722869e-01 -3.58776987e-01 1.43393421e+00 -4.21442062e-01
-9.49659348e-02 5.54072857e-01 1.27051041e-01 -4.05509055e-01
-5.76111615e-01 3.50581497e-01 1.15488634e-01 -3.95329952e-01
6.11339211e-01 -6.02174878e-01 -1.18324690e-01 1.20932035e-01
-9.41488802e-01 -2.73772269e-01 2.00088382e-01 7.86217332e-01
4.40311283e-01 4.65509176e-01 6.00653067e-02 -1.36153257e+00
6.74126506e-01 -8.87555301e-01 -8.49796057e-01 9.31047648e-02
-4.31364298e-01 1.49656951e-01 1.69840790e-02 -3.15922022e-01
-1.33286321e+00 -2.11898386e-01 -3.06314807e-02 5.37126437e-02
-3.00752193e-01 3.69459629e-01 -2.03459978e-01 5.53522944e-01
4.00398701e-01 -6.37917072e-02 -3.34938496e-01 -6.01024866e-01
3.36756825e-01 6.51950300e-01 1.74325868e-01 -1.29684603e+00
5.62100947e-01 7.77648628e-01 -1.94317058e-01 -1.18174922e+00
-8.55999470e-01 -5.91909349e-01 -5.83581150e-01 3.60915095e-01
1.02237749e+00 -1.09614229e+00 -6.98413193e-01 1.54403806e-01
-1.33965588e+00 -4.72108752e-01 -1.09311812e-01 3.52092266e-01
-7.37792969e-01 5.00419855e-01 -1.25398290e+00 -1.15296805e+00
1.46918952e-01 -1.02145100e+00 1.38363457e+00 -1.17302686e-01
-2.59172052e-01 -1.38008809e+00 5.12371302e-01 2.17328757e-01
6.06797971e-02 -6.90781698e-02 1.37770832e+00 -8.63375366e-01
-3.93959761e-01 -2.93260086e-02 1.29242137e-01 1.06911004e-01
-3.98933023e-01 2.76282668e-01 -9.72976863e-01 -3.32626104e-01
1.79776341e-01 -2.22796857e-01 9.75121915e-01 7.47322261e-01
1.09700894e+00 -3.30441654e-01 -7.56545484e-01 -1.87598262e-02
1.35002708e+00 -2.02231258e-01 6.34400964e-01 2.68940240e-01
2.06061214e-01 7.07833827e-01 3.44081879e-01 6.75513744e-01
5.99511385e-01 4.67426598e-01 3.06543540e-02 1.41910940e-01
7.91784525e-01 -1.99123293e-01 2.64019221e-01 7.95002759e-01
-3.34877633e-02 -1.27764016e-01 -1.24829268e+00 7.86292076e-01
-1.74958861e+00 -1.04602420e+00 -2.68398732e-01 2.18463111e+00
1.36107683e+00 5.91170132e-01 3.65382761e-01 -2.37726629e-01
1.10082376e+00 1.11699425e-01 -1.50729254e-01 -4.13440824e-01
3.75685126e-01 3.60994101e-01 3.65061134e-01 8.62476408e-01
-1.36927986e+00 9.06099558e-01 8.05178738e+00 1.01657712e+00
-1.19190834e-01 5.47478676e-01 9.06187594e-01 4.92857061e-02
-3.97783369e-01 3.07931662e-01 -1.50397539e+00 4.78471786e-01
1.51819706e+00 -1.75012257e-02 -2.58386791e-01 1.13102591e+00
-1.98381413e-02 -3.99591446e-01 -1.03920031e+00 2.16005012e-01
-1.95299819e-01 -1.08513153e+00 -3.88492495e-02 6.98314011e-01
9.84934032e-01 -5.01540974e-02 8.11986178e-02 4.59704131e-01
1.19258714e+00 -8.19118738e-01 4.05240119e-01 3.95856708e-01
4.44487751e-01 -9.03991163e-01 5.34503698e-01 6.82373822e-01
-8.04206014e-01 2.48432755e-01 -7.59606004e-01 2.81255692e-01
2.77855754e-01 8.48813176e-01 -8.70908380e-01 1.22676946e-01
8.13606441e-01 1.17592566e-01 -5.47611304e-02 8.07864547e-01
1.08459417e-03 1.15619433e+00 -6.45898700e-01 8.61590877e-02
2.98982173e-01 -1.13272086e-01 3.43464643e-01 1.38021970e+00
6.42481968e-02 1.27201930e-01 3.15868378e-01 6.73351884e-01
3.33423652e-02 -2.17749104e-01 -2.87772447e-01 2.12259635e-01
5.66024303e-01 9.78468120e-01 -8.49312961e-01 -6.67882144e-01
-1.48225471e-01 5.09964943e-01 1.72141388e-01 4.17392284e-01
-7.28164315e-01 -3.57562676e-02 4.96035010e-01 2.58908689e-01
5.09623289e-01 -4.38050449e-01 2.55062319e-02 -1.16361523e+00
-5.05304098e-01 -5.27534127e-01 4.99303520e-01 -3.88534993e-01
-1.48791635e+00 4.15675879e-01 7.52189040e-01 -3.39693040e-01
-6.35519505e-01 -5.00209689e-01 -4.94830847e-01 9.04016376e-01
-1.00190187e+00 -9.12682652e-01 3.84418964e-01 4.18499827e-01
4.67387915e-01 1.48268700e-01 1.00089025e+00 -1.92853287e-01
-3.16310197e-01 -5.57634458e-02 4.78633046e-01 -9.66285467e-02
3.83698642e-01 -1.78695333e+00 4.71357048e-01 4.51157451e-01
1.76682264e-01 8.99650395e-01 1.11599445e+00 -6.47728443e-01
-7.10026622e-01 -7.52383888e-01 9.92277265e-01 -7.29981661e-01
9.71919119e-01 -8.30117166e-01 -1.00654471e+00 1.14858305e+00
2.80932248e-01 -4.68395472e-01 1.15252936e+00 7.51553416e-01
-3.79404545e-01 5.19242823e-01 -1.06479192e+00 1.72338113e-01
3.71935755e-01 -5.34386635e-01 -9.12608147e-01 6.65721536e-01
6.54463351e-01 -6.49887249e-02 -1.09469664e+00 -1.05665572e-01
3.65185380e-01 -9.61335003e-01 8.56173396e-01 -7.87687480e-01
2.34969273e-01 2.28350148e-01 -3.15299153e-01 -8.44639003e-01
-2.14414641e-01 -7.42134333e-01 -1.35128856e-01 1.40793478e+00
3.55127007e-01 -5.06644666e-01 1.09159565e+00 9.10997212e-01
2.40847126e-01 -6.06309891e-01 -1.10915387e+00 -5.02896726e-01
7.46554732e-01 -4.68714982e-01 2.55871952e-01 8.09921563e-01
1.58230290e-02 1.96355164e-01 -2.51762301e-01 3.69769722e-01
1.14070821e+00 -8.59765261e-02 5.81541538e-01 -1.77650547e+00
-6.23851180e-01 -1.03008941e-01 -1.74632579e-01 -1.17430651e+00
5.64650416e-01 -2.85241604e-01 2.15604901e-01 -1.50142562e+00
7.02764153e-01 -4.67149407e-01 6.65888116e-02 2.39497855e-01
1.81916028e-01 -9.35194567e-02 -4.22930628e-01 5.24391532e-01
-5.47250271e-01 4.23647791e-01 3.48019928e-01 1.23658389e-01
-2.63092797e-02 1.37170255e-01 -5.96072435e-01 1.08067787e+00
6.70819879e-01 -8.54368925e-01 -2.21705869e-01 5.37842475e-02
3.86530310e-02 2.84407794e-01 2.11850017e-01 -6.00950778e-01
3.79568219e-01 -2.70821694e-02 2.01519907e-01 -1.04878581e+00
6.54315412e-01 -3.21161449e-01 -1.41190393e-02 2.03408152e-01
-4.90789443e-01 -1.97673872e-01 5.51930033e-02 9.55511212e-01
-2.46725366e-01 -4.45134133e-01 7.07604051e-01 -2.17145652e-01
-2.58675665e-01 2.23396495e-01 -8.46269250e-01 2.29542211e-01
8.28269303e-01 2.49450520e-01 -9.97416005e-02 -4.94357467e-01
-1.19286656e+00 1.45065367e-01 4.11300778e-01 -2.50915825e-01
1.06824696e-01 -9.44191575e-01 -5.55255294e-01 -2.95405030e-01
-2.97891229e-01 -1.45597318e-02 1.15695976e-01 6.16719365e-01
-2.72059798e-01 5.70307314e-01 3.12187105e-01 -7.69979298e-01
-1.00631869e+00 3.86762470e-01 -5.96458577e-02 -7.86290467e-01
-2.31391832e-01 7.97638714e-01 4.24287230e-01 -3.04907233e-01
3.61647576e-01 -9.48812813e-02 3.24565955e-02 2.35592753e-01
2.79481322e-01 3.42044950e-01 -3.10394585e-01 -3.95147890e-01
-1.53936729e-01 6.73577338e-02 -4.93915319e-01 -8.03174198e-01
1.32871556e+00 -2.33165741e-01 -4.49002683e-01 1.00260806e+00
1.07855225e+00 -4.67045158e-02 -1.22601247e+00 -2.02903464e-01
3.92390609e-01 -8.16838294e-02 -1.06841791e-02 -1.43451244e-01
-2.87419647e-01 1.06984997e+00 6.36760741e-02 9.00755942e-01
5.19497693e-01 4.07403618e-01 3.02491128e-01 4.01572496e-01
6.78747296e-01 -1.20002842e+00 -1.21314220e-01 4.11458790e-01
3.69062811e-01 -8.40188861e-01 2.17523158e-01 -4.69040662e-01
-4.03490961e-01 9.37741518e-01 5.86132146e-02 -1.77733719e-01
1.14955437e+00 3.76749814e-01 -6.19203329e-01 -2.25032389e-01
-1.15230405e+00 2.15501919e-01 -1.15873545e-01 4.40534562e-01
6.26846552e-01 1.42849773e-01 -9.82990339e-02 5.64161360e-01
-4.47786599e-01 -3.56146187e-01 6.56816065e-01 9.95849013e-01
-7.83526719e-01 -1.17446864e+00 -5.30938864e-01 3.99760723e-01
-8.97081316e-01 -3.47110122e-01 -1.25024365e-02 1.09234214e+00
-3.28326404e-01 9.50903654e-01 2.95486957e-01 3.08179557e-01
-3.75458598e-01 3.47331703e-01 2.34702721e-01 -9.80593204e-01
-1.08392060e-01 5.83226025e-01 1.69033572e-01 -2.99894333e-01
-4.28372204e-01 -1.17529321e+00 -9.04513776e-01 -4.84357357e-01
-2.98320234e-01 9.55588341e-01 8.04626524e-01 8.16423297e-01
-1.96144860e-02 -6.29825294e-02 3.78932416e-01 -8.51213336e-01
-8.95300329e-01 -1.15037012e+00 -1.19491124e+00 -2.40183115e-01
3.63356471e-02 -7.05363393e-01 -6.39690101e-01 4.22632694e-01] | [10.282029151916504, 6.837799549102783] |
9d8d5fde-c5eb-4e52-ba42-89c17131a55f | caibc-capturing-all-round-information-beyond | 2209.05773 | null | https://arxiv.org/abs/2209.05773v1 | https://arxiv.org/pdf/2209.05773v1.pdf | CAIBC: Capturing All-round Information Beyond Color for Text-based Person Retrieval | Given a natural language description, text-based person retrieval aims to identify images of a target person from a large-scale person image database. Existing methods generally face a \textbf{color over-reliance problem}, which means that the models rely heavily on color information when matching cross-modal data. Indeed, color information is an important decision-making accordance for retrieval, but the over-reliance on color would distract the model from other key clues (e.g. texture information, structural information, etc.), and thereby lead to a sub-optimal retrieval performance. To solve this problem, in this paper, we propose to \textbf{C}apture \textbf{A}ll-round \textbf{I}nformation \textbf{B}eyond \textbf{C}olor (\textbf{CAIBC}) via a jointly optimized multi-branch architecture for text-based person retrieval. CAIBC contains three branches including an RGB branch, a grayscale (GRS) branch and a color (CLR) branch. Besides, with the aim of making full use of all-round information in a balanced and effective way, a mutual learning mechanism is employed to enable the three branches which attend to varied aspects of information to communicate with and learn from each other. Extensive experimental analysis is carried out to evaluate our proposed CAIBC method on the CUHK-PEDES and RSTPReid datasets in both \textbf{supervised} and \textbf{weakly supervised} text-based person retrieval settings, which demonstrates that CAIBC significantly outperforms existing methods and achieves the state-of-the-art performance on all the three tasks. | ['Yifeng Li', 'Tian Wang', 'Chao Liu', 'Xili Wan', 'Jingyi Xue', 'Aichun Zhu', 'Zijie Wang'] | 2022-09-13 | null | null | null | null | ['person-retrieval', 'nlp-based-person-retrival'] | ['computer-vision', 'computer-vision'] | [ 2.96063181e-02 -5.94385862e-01 -7.90910125e-02 -4.02093887e-01
-8.30141187e-01 -4.21405166e-01 6.23134553e-01 -4.95349383e-03
-6.17223978e-01 6.37970507e-01 -2.02161670e-01 1.09463491e-01
-2.37546071e-01 -7.03171074e-01 -2.31472000e-01 -8.17684293e-01
2.08548397e-01 7.33907700e-01 -2.08971262e-01 -2.86061019e-01
2.81214155e-02 3.44630957e-01 -1.65476191e+00 2.70419806e-01
7.96514630e-01 1.20243716e+00 -5.17423674e-02 4.94952768e-01
-2.23742519e-02 7.12212861e-01 -6.57823622e-01 -8.10354412e-01
4.22410458e-01 -3.61787319e-01 -5.61050773e-01 1.45683989e-01
5.13850272e-01 -3.78174812e-01 -7.38336563e-01 9.97905374e-01
8.11986029e-01 2.30663747e-01 9.91653025e-01 -1.31875563e+00
-8.04313362e-01 5.31069040e-02 -8.40645313e-01 2.00179055e-01
6.53998673e-01 1.56788334e-01 1.10413373e+00 -9.11874652e-01
2.25255564e-01 1.45279503e+00 3.22199732e-01 5.68737686e-01
-1.00398922e+00 -8.62353206e-01 3.12181979e-01 2.49964967e-01
-1.73519003e+00 -1.75974369e-01 8.14009011e-01 -2.60269284e-01
5.64701736e-01 6.02880597e-01 7.83520460e-01 9.93597150e-01
-1.04862630e-01 1.29806232e+00 1.54699385e+00 -4.61050868e-01
-3.50907505e-01 3.61801386e-01 1.27722353e-01 9.54866767e-01
2.47242793e-01 2.25490153e-01 -8.58128607e-01 -1.07054397e-01
6.69052064e-01 3.77393395e-01 -4.39893067e-01 9.57370251e-02
-9.86661077e-01 4.91210222e-01 7.47587979e-01 2.43441448e-01
4.91370335e-02 1.18079565e-01 4.48253602e-02 3.71561795e-01
3.78206521e-01 1.64367363e-01 -2.15075240e-02 2.62953877e-01
-8.44475448e-01 2.92230099e-01 5.82995355e-01 9.99045134e-01
8.83463085e-01 -2.22548023e-01 -3.18226129e-01 1.09629762e+00
4.96454656e-01 1.08586705e+00 4.24020320e-01 -2.73576766e-01
7.02086806e-01 7.52294302e-01 8.75353366e-02 -1.20915854e+00
-2.60554045e-01 -1.47223920e-01 -1.03465712e+00 -5.50438650e-02
4.76868123e-01 6.38495609e-02 -9.53206301e-01 1.46278811e+00
5.41302897e-02 -4.25441712e-01 -1.49278387e-01 1.39669979e+00
1.20026875e+00 4.84622478e-01 6.85181320e-02 2.50216037e-01
1.45285213e+00 -8.84899080e-01 -3.78131241e-01 -2.96679974e-01
2.00823069e-01 -8.35029662e-01 9.84380901e-01 4.71982747e-01
-8.50950360e-01 -6.50591195e-01 -1.09912467e+00 -1.48756698e-01
-6.87675536e-01 4.76784587e-01 6.38043106e-01 7.68821657e-01
-1.06407475e+00 1.01849966e-01 -1.61575243e-01 -3.99127066e-01
1.69563890e-01 4.18174714e-01 -4.00875390e-01 -5.28147042e-01
-1.27225161e+00 5.61962068e-01 3.66636664e-01 5.77555478e-01
-5.06823480e-01 -1.55149428e-02 -6.96696997e-01 -2.21453786e-01
4.55997437e-01 -7.63951480e-01 5.33715248e-01 -1.05666482e+00
-1.10163760e+00 1.22160757e+00 6.82164636e-03 9.12964419e-02
8.35238516e-01 -3.13320279e-01 -5.33224225e-01 6.00682557e-01
1.63651586e-01 8.49266291e-01 1.12699175e+00 -1.35534561e+00
-5.08717418e-01 -7.98186541e-01 1.15127474e-01 5.48974097e-01
-5.37216961e-01 1.34574398e-01 -1.18082762e+00 -9.02680218e-01
1.16300397e-01 -1.07320976e+00 3.27910930e-01 1.84688434e-01
-6.79281950e-01 -3.61505330e-01 5.11701465e-01 -5.48000872e-01
1.02118635e+00 -1.76413548e+00 2.79211193e-01 5.11429012e-01
2.96493441e-01 1.26408443e-01 -1.42015470e-03 3.53428185e-01
1.50512978e-01 -1.18485942e-01 2.02399477e-01 -4.61429656e-01
7.80794695e-02 2.90946029e-02 3.61183211e-02 5.62578917e-01
-3.98165099e-02 9.23399448e-01 -5.41205883e-01 -9.03410554e-01
3.75784367e-01 5.50760567e-01 -8.43252093e-02 2.31759235e-01
1.98881820e-01 2.80577898e-01 -6.98637426e-01 1.07288933e+00
4.90015715e-01 -3.66464823e-01 -1.29312992e-01 -1.87448621e-01
2.18033493e-01 -4.83424515e-01 -1.12460399e+00 1.41893327e+00
-3.99592333e-02 4.65714306e-01 -9.05376375e-02 -8.68621826e-01
1.02342749e+00 1.65724978e-01 6.06442630e-01 -1.02512228e+00
2.02616557e-01 7.55217373e-02 -4.07776326e-01 -3.94998282e-01
4.77160066e-01 2.84175742e-02 -9.56339538e-02 5.65636218e-01
-1.55730233e-01 9.48739499e-02 1.91361204e-01 3.11399788e-01
4.98107910e-01 1.28392935e-01 -1.36490107e-01 -2.19327778e-01
7.96333253e-01 -2.03546584e-01 2.14353353e-01 9.29069579e-01
-2.80297846e-01 6.83251023e-01 8.93752053e-02 -4.34125245e-01
-6.09848619e-01 -8.09457421e-01 7.91933089e-02 1.24435091e+00
6.33607209e-01 -4.04212743e-01 -4.96051669e-01 -5.60689569e-01
9.78244692e-02 -4.88082506e-02 -7.38252997e-01 -1.36988804e-01
-3.28848600e-01 -1.03105879e+00 7.80388474e-01 3.83633256e-01
1.09726822e+00 -9.46091652e-01 -5.17108321e-01 -2.94540942e-01
-5.21049261e-01 -8.99462104e-01 -7.38044262e-01 -1.37439743e-01
-2.45279968e-01 -1.24059415e+00 -1.28388286e+00 -6.26507521e-01
8.80750895e-01 4.77829605e-01 1.10135067e+00 5.72608173e-01
-6.57366455e-01 8.13244760e-01 -3.70637208e-01 -5.06699383e-01
5.07690191e-01 -2.04360411e-01 -4.49652597e-02 2.61198133e-01
7.59998739e-01 1.81708336e-01 -1.06456900e+00 5.72499335e-01
-7.95717359e-01 3.86941545e-02 7.28881061e-01 9.59178865e-01
5.50662994e-01 2.55728006e-01 9.72181559e-02 -3.29455227e-01
4.86372322e-01 5.92245981e-02 -4.45169270e-01 7.11554646e-01
-6.99995577e-01 -2.66921490e-01 3.14243972e-01 -4.09244597e-01
-1.04656971e+00 -2.05974624e-01 4.00565743e-01 -4.03034449e-01
1.06453069e-03 3.36316586e-01 -2.19024882e-01 -1.78502083e-01
4.05221313e-01 4.49190259e-01 -1.80022866e-01 -3.67579788e-01
6.12035766e-02 7.10274339e-01 4.69604969e-01 -9.47728693e-01
8.40456307e-01 5.47412515e-01 -1.77746639e-01 -7.07591832e-01
-8.31588984e-01 -7.93447316e-01 -6.23147666e-01 -6.11894846e-01
8.89731586e-01 -1.08168006e+00 -9.17127848e-01 7.75547028e-01
-6.49697840e-01 -3.37665789e-02 2.74560958e-01 3.09122682e-01
-1.66263238e-01 3.94803226e-01 -7.20215619e-01 -1.08419538e+00
-5.67257524e-01 -1.13881898e+00 1.36727989e+00 5.33001661e-01
2.07060844e-01 -7.07921982e-01 -5.15018106e-01 9.51148331e-01
5.90167232e-02 2.82141343e-02 6.07618392e-01 -5.22916079e-01
-7.04628468e-01 -5.85939527e-01 -7.93791652e-01 1.52728289e-01
2.00899318e-02 -3.66026193e-01 -9.70491469e-01 -5.70355833e-01
-3.98578256e-01 -6.59242690e-01 9.63397086e-01 1.69833153e-02
1.04073167e+00 -9.59455818e-02 -3.02642107e-01 4.04394656e-01
1.39890885e+00 -7.23962262e-02 5.10338902e-01 4.74882960e-01
7.63664126e-01 7.24371910e-01 5.77046812e-01 4.24827009e-01
7.00035334e-01 7.35565960e-01 -1.23695638e-02 -4.52435046e-01
-2.01990515e-01 -5.18696718e-02 1.21722296e-01 2.62738794e-01
-4.69474822e-01 -3.01586568e-01 -8.41613710e-01 2.07320109e-01
-1.81607425e+00 -8.26693654e-01 1.28156081e-01 2.18799734e+00
9.48289812e-01 -1.22750990e-01 1.75068513e-01 4.98419181e-02
6.31790400e-01 6.15071543e-02 -4.64715272e-01 4.18484241e-01
-3.66215050e-01 -9.10831317e-02 2.99131453e-01 -4.65084575e-02
-1.27938390e+00 9.47530746e-01 5.09394407e+00 8.47667158e-01
-1.16105664e+00 -3.14124495e-01 9.03072536e-01 -1.79452077e-01
-6.97461292e-02 -3.72180909e-01 -8.86079848e-01 5.01400530e-01
3.34415101e-02 9.72981527e-02 5.10767341e-01 4.73257422e-01
-2.28393748e-01 -4.00388211e-01 -1.07308805e+00 1.61363065e+00
5.42190492e-01 -6.14133656e-01 3.06257069e-01 5.64240813e-02
5.14262795e-01 -3.87917370e-01 3.29749674e-01 3.79697531e-01
1.68097794e-01 -1.20102119e+00 8.49921227e-01 8.13056886e-01
9.59410191e-01 -7.51779020e-01 7.42483258e-01 1.26894519e-01
-1.47595370e+00 6.04956187e-02 -2.78215736e-01 3.76698494e-01
-1.71694174e-01 2.82592475e-01 -1.70904517e-01 8.07200730e-01
1.31777787e+00 6.74538612e-01 -1.00355136e+00 9.07732844e-01
-1.91650033e-01 4.93875556e-02 -2.34575257e-01 -2.37685278e-01
1.07650436e-01 -3.28803331e-01 3.64318281e-01 1.22082400e+00
2.97259036e-02 3.59977216e-01 6.55174077e-01 6.89364910e-01
-3.78214121e-02 1.85499862e-01 -1.97199568e-01 5.43001704e-02
2.38353923e-01 1.06851554e+00 -6.94720566e-01 -4.20926183e-01
-3.93465430e-01 1.15876162e+00 1.68106288e-01 8.14353108e-01
-6.37524068e-01 -3.17359358e-01 1.16269417e-01 -5.88365830e-02
1.01615265e-01 8.35553557e-02 -4.64659138e-03 -1.37749434e+00
2.58117557e-01 -1.06426406e+00 8.52806330e-01 -1.02874827e+00
-1.63465130e+00 6.35027349e-01 1.39460430e-01 -1.16661835e+00
-2.35445425e-02 -6.65976644e-01 4.76250611e-03 1.01168537e+00
-1.80863893e+00 -1.57498753e+00 -5.65152049e-01 1.18214691e+00
2.99110681e-01 -5.33732355e-01 5.73053360e-01 4.53417420e-01
-6.87129974e-01 8.25721264e-01 -1.32371664e-01 5.35315096e-01
1.08547473e+00 -1.39650857e+00 -4.25108552e-01 7.39038765e-01
1.61290765e-01 9.03045118e-01 4.31386650e-01 -7.23231435e-01
-1.70067334e+00 -7.92452931e-01 5.25811672e-01 -2.99774587e-01
3.41024965e-01 -1.84073359e-01 -5.12796760e-01 4.27901953e-01
-1.04429215e-01 -1.78072006e-02 8.02947462e-01 9.79197472e-02
-6.07399285e-01 -3.92456502e-01 -9.81992662e-01 6.83690965e-01
7.83369541e-01 -7.70997167e-01 -3.49349499e-01 3.86125147e-01
1.05554722e-02 -2.21289858e-01 -7.00192928e-01 1.73219100e-01
7.29668796e-01 -1.08362043e+00 1.26869619e+00 -1.66684210e-01
1.27906948e-01 -2.68152297e-01 -1.51538208e-01 -7.06009507e-01
6.92038685e-02 -4.85866934e-01 4.83239442e-02 1.16632271e+00
1.10413313e-01 -5.65900505e-01 6.14339471e-01 8.97479475e-01
4.64738250e-01 -5.18949211e-01 -7.07633495e-01 -5.61111271e-01
-9.84957293e-02 -2.48858631e-01 5.20695090e-01 6.96927965e-01
-2.42192551e-01 1.22055508e-01 -6.87323272e-01 -8.76993462e-02
9.05110896e-01 3.00035655e-01 8.67716014e-01 -1.32689619e+00
-2.48840019e-01 -4.78583068e-01 -2.68343270e-01 -1.33430016e+00
-1.44710407e-01 -8.21174026e-01 4.46226709e-02 -1.25388896e+00
7.23378360e-01 -7.33209848e-01 -5.96309423e-01 4.86798763e-01
-4.88504052e-01 7.33147144e-01 4.74517554e-01 6.85760677e-01
-9.90739703e-01 5.47775626e-01 1.22256994e+00 -6.07160211e-01
9.80236158e-02 -4.05938625e-02 -6.96263850e-01 4.22677547e-01
2.57820755e-01 4.27349843e-02 -3.08300972e-01 -5.08194864e-01
2.00937346e-01 4.68177237e-02 7.68032372e-01 -7.84050643e-01
4.73301142e-01 1.98085591e-01 1.05245268e+00 -7.86387026e-01
7.69333839e-01 -8.30692530e-01 -2.77878404e-01 1.71042487e-01
-2.22987175e-01 1.72308445e-01 -1.48800984e-01 7.32958078e-01
-2.52104491e-01 8.15618560e-02 4.97624874e-01 -4.35086817e-01
-6.24432445e-01 5.28446674e-01 -4.72184643e-02 -2.30256975e-01
6.42810464e-01 -4.04229045e-01 -3.31652164e-01 -6.39137387e-01
-4.44700778e-01 4.98049080e-01 3.36919874e-01 5.06641269e-01
8.48928392e-01 -1.34498608e+00 -6.61490977e-01 2.25564390e-01
5.22545516e-01 -2.90613353e-01 2.70752579e-01 7.83358157e-01
-3.32560569e-01 6.83353424e-01 -4.39527258e-03 -7.41332829e-01
-1.44664562e+00 3.04255277e-01 4.69801366e-01 -2.66492665e-01
-4.32084143e-01 9.48642075e-01 2.28183657e-01 -2.69274712e-01
4.78266180e-01 4.38097179e-01 -2.31966883e-01 1.31983608e-01
6.61088288e-01 3.23902667e-01 -8.46638680e-02 -1.09259832e+00
-3.11299056e-01 7.53585398e-01 -3.03272605e-01 -2.07154602e-01
8.94191921e-01 -3.05127770e-01 -2.64685690e-01 2.04704642e-01
9.94518936e-01 -4.19218570e-01 -9.43318903e-01 -6.09647870e-01
-1.76621854e-01 -6.71453953e-01 2.03113183e-02 -1.12086272e+00
-1.25433457e+00 6.62347019e-01 8.69566560e-01 -1.38726115e-01
1.15073740e+00 -6.80516288e-02 7.40623057e-01 6.27461910e-01
6.87478185e-01 -1.49131012e+00 6.37218356e-01 8.72765761e-03
8.70031118e-01 -1.61938000e+00 4.37777966e-01 -3.12835813e-01
-6.47226453e-01 1.09368980e+00 8.03525269e-01 1.98790357e-01
5.44588327e-01 -3.60461980e-01 7.88580105e-02 -4.23954815e-01
-3.12610835e-01 -5.60096502e-01 1.00664878e+00 3.72461230e-01
2.85903752e-01 1.05879962e-01 -1.16418168e-01 3.25913280e-01
-1.31941810e-01 -1.84825450e-01 -2.94051707e-01 9.13893878e-01
-1.88037544e-01 -1.03190231e+00 -7.81030357e-01 4.19121236e-01
-4.47168529e-01 -1.28688172e-01 -7.83890545e-01 9.05399382e-01
1.02658793e-01 1.34747684e+00 -2.45578766e-01 -2.31700167e-01
6.48038788e-03 -1.82950392e-01 6.64922476e-01 -1.56062618e-01
-6.41182363e-01 4.96206105e-01 -1.72806278e-01 -3.67910504e-01
-6.60398662e-01 -5.17113149e-01 -8.90761614e-01 -3.64233285e-01
-4.28737491e-01 -1.92216281e-02 2.19329223e-01 1.04004133e+00
-1.81025609e-01 1.18367210e-01 4.94199991e-01 -5.74773967e-01
-1.66234985e-01 -8.13315392e-01 -7.36002564e-01 7.23974228e-01
1.30040035e-01 -6.99130356e-01 -4.12414446e-02 -1.43461570e-01] | [14.567193984985352, 0.8137559294700623] |
8d0cfd18-4774-45f1-874f-9330eab6a8d8 | weighing-features-of-lung-and-heart-regions | 2105.12430 | null | https://arxiv.org/abs/2105.12430v1 | https://arxiv.org/pdf/2105.12430v1.pdf | Weighing Features of Lung and Heart Regions for Thoracic Disease Classification | Chest X-rays are the most commonly available and affordable radiological examination for screening thoracic diseases. According to the domain knowledge of screening chest X-rays, the pathological information usually lay on the lung and heart regions. However, it is costly to acquire region-level annotation in practice, and model training mainly relies on image-level class labels in a weakly supervised manner, which is highly challenging for computer-aided chest X-ray screening. To address this issue, some methods have been proposed recently to identify local regions containing pathological information, which is vital for thoracic disease classification. Inspired by this, we propose a novel deep learning framework to explore discriminative information from lung and heart regions. We design a feature extractor equipped with a multi-scale attention module to learn global attention maps from global images. To exploit disease-specific cues effectively, we locate lung and heart regions containing pathological information by a well-trained pixel-wise segmentation model to generate binarization masks. By introducing element-wise logical AND operator on the learned global attention maps and the binarization masks, we obtain local attention maps in which pixels are $1$ for lung and heart region and $0$ for other regions. By zeroing features of non-lung and heart regions in attention maps, we can effectively exploit their disease-specific cues in lung and heart regions. Compared to existing methods fusing global and local features, we adopt feature weighting to avoid weakening visual cues unique to lung and heart regions. Evaluated by the benchmark split on the publicly available chest X-ray14 dataset, the comprehensive experiments show that our method achieves superior performance compared to the state-of-the-art methods. | ['Jiang Liu', 'Junling Liu', 'Yuguang Yan', 'Yitian Zhao', 'Yanwu Xu', 'Jiansheng Fang'] | 2021-05-26 | null | null | null | null | ['thoracic-disease-classification'] | ['computer-vision'] | [ 4.23477083e-01 1.65540054e-01 -4.82722431e-01 -2.84698159e-01
-1.01358891e+00 -2.10207626e-01 2.03209236e-01 2.65558511e-01
-4.50336099e-01 4.16138202e-01 9.38242525e-02 -4.29967523e-01
-1.34026229e-01 -8.80265176e-01 -6.03253663e-01 -7.63742447e-01
1.53240249e-01 2.99944282e-01 5.84583938e-01 2.01807141e-01
-3.72028872e-02 5.60407221e-01 -1.10142756e+00 3.25568885e-01
7.81610787e-01 1.10982978e+00 6.67999208e-01 7.28480697e-01
-4.43615019e-02 8.30873191e-01 -3.07984978e-01 -1.00990525e-02
1.79176629e-01 -7.52840161e-01 -9.17877913e-01 2.67456293e-01
1.89433739e-01 -5.21922946e-01 -4.13551062e-01 1.11917746e+00
5.16296744e-01 -1.10700697e-01 5.81178904e-01 -3.68022770e-01
-6.12470984e-01 2.44858608e-01 -7.82300472e-01 7.28910327e-01
-8.80383626e-02 3.37206990e-01 1.05849600e+00 -7.41647542e-01
5.01604736e-01 7.28679836e-01 5.25896430e-01 4.83696789e-01
-6.33471251e-01 -5.11154652e-01 1.81861684e-01 1.52768016e-01
-1.23979259e+00 1.07941255e-01 7.83032179e-01 -3.11022907e-01
3.51293385e-01 6.75795972e-01 5.98702610e-01 5.32828629e-01
4.08921242e-01 1.00241256e+00 1.02576387e+00 -3.52616489e-01
-1.33088514e-01 -3.27970088e-02 4.83204052e-02 1.09555376e+00
3.91166881e-02 -1.41750410e-01 4.82822359e-02 -4.52361777e-02
1.24500334e+00 6.79189980e-01 -5.44721246e-01 -3.72536510e-01
-1.40157330e+00 6.86128259e-01 1.16031015e+00 5.79953849e-01
-6.14603758e-01 1.71260238e-01 2.61645526e-01 -3.70397449e-01
2.65266001e-01 2.19969317e-01 -3.40765178e-01 4.16392237e-01
-9.45083439e-01 -1.91449821e-01 2.60361880e-02 3.70984316e-01
6.61246181e-01 -4.08778280e-01 -6.74347997e-01 9.67004299e-01
3.00528944e-01 2.91018337e-01 8.04177880e-01 -4.81979012e-01
5.55612683e-01 9.52578485e-01 -2.87500232e-01 -6.76541686e-01
-5.56814194e-01 -8.06840599e-01 -1.07036245e+00 -6.40729815e-02
2.86544085e-01 6.57613501e-02 -1.48620152e+00 1.30460262e+00
4.91552413e-01 5.76023832e-02 -4.22919422e-01 1.29582918e+00
9.37653601e-01 3.87674481e-01 3.08358103e-01 -1.69930488e-01
1.55816615e+00 -1.09410417e+00 -5.15186310e-01 -2.35218003e-01
7.53303707e-01 -4.42808241e-01 1.36492181e+00 -9.93441492e-02
-1.03091562e+00 -6.33260608e-01 -8.78186524e-01 -1.28443062e-01
-9.21963677e-02 4.40733224e-01 4.69490498e-01 5.46368122e-01
-6.38733745e-01 3.80733311e-01 -1.08153951e+00 -2.45009378e-01
7.57036686e-01 3.47100168e-01 -2.13113666e-01 -1.95754409e-01
-1.08069468e+00 6.82330787e-01 3.23453009e-01 1.93685308e-01
-9.68772769e-01 -6.19778574e-01 -7.26160884e-01 2.02631816e-01
5.91162980e-01 -6.80899143e-01 1.08166778e+00 -7.77583539e-01
-1.22341228e+00 1.05859053e+00 2.48758141e-02 -1.57427564e-01
3.73459816e-01 -4.92756367e-02 -1.08257547e-01 5.66730142e-01
3.32011372e-01 5.21080613e-01 6.66466832e-01 -1.11567783e+00
-6.99720681e-01 -3.35183233e-01 -9.14008021e-02 3.49805921e-01
-4.77825075e-01 9.61701497e-02 -1.01345682e+00 -9.49206054e-01
4.78926748e-01 -5.23443878e-01 -6.48819447e-01 2.47538254e-01
-5.78634381e-01 2.07710415e-02 8.85216832e-01 -8.90914381e-01
1.47850025e+00 -2.00120878e+00 -8.55016559e-02 3.75301838e-01
4.80272800e-01 1.59612790e-01 2.76937425e-01 -3.65636736e-01
-1.43869504e-01 2.60936230e-01 -5.88064790e-01 -3.42304036e-02
-3.43195677e-01 2.31563568e-01 9.56402943e-02 5.40960789e-01
3.38095754e-01 1.12722754e+00 -8.64289939e-01 -1.13757002e+00
4.31000710e-01 2.97914743e-01 -4.56345290e-01 3.51591170e-01
1.43378405e-02 6.96214199e-01 -9.69663978e-01 1.01711416e+00
3.68686765e-01 -6.78016484e-01 -5.90060875e-02 -6.69984072e-02
2.04208598e-01 8.43887925e-02 -6.94882095e-01 1.66475296e+00
-5.34168124e-01 9.30896252e-02 2.59694695e-01 -1.05712450e+00
5.94525635e-01 2.82147139e-01 5.69379210e-01 -7.10269094e-01
2.92410612e-01 1.69381663e-01 1.72831953e-01 -7.27548838e-01
-2.03295931e-01 -3.74381006e-01 -7.07603768e-02 4.12271857e-01
-1.02292143e-01 -1.11505136e-01 -1.74782619e-01 5.14663532e-02
1.36664712e+00 -1.44113421e-01 3.12967241e-01 -1.53789431e-01
8.42994213e-01 -1.18050292e-01 5.73508024e-01 8.58033836e-01
-3.30343395e-01 1.22070515e+00 2.89133430e-01 -4.19137329e-01
-6.30397797e-01 -1.10608423e+00 -3.77296895e-01 1.12162876e+00
2.61985421e-01 -1.27355233e-01 -6.59324348e-01 -1.35474825e+00
-2.88170755e-01 1.28834754e-01 -9.53960717e-01 -6.15405254e-02
-9.47443128e-01 -9.10711288e-01 2.41356716e-01 8.02550018e-01
6.65499389e-01 -1.21218848e+00 -8.66130471e-01 1.46290213e-01
-6.00949302e-02 -7.35646725e-01 -5.78005850e-01 3.82315189e-01
-9.78983045e-01 -1.12350059e+00 -1.36430645e+00 -8.67555141e-01
1.07512474e+00 1.29655123e-01 1.08907247e+00 6.03742957e-01
-8.10087085e-01 1.37557849e-01 -2.75256068e-01 -9.88528281e-02
3.01816352e-02 2.49882251e-01 -7.10660934e-01 1.98884532e-02
-1.75282821e-01 -7.32530951e-02 -1.05906451e+00 3.60725582e-01
-9.89833355e-01 9.81624126e-02 1.18941939e+00 1.07593334e+00
1.08718276e+00 4.86563742e-02 2.42971301e-01 -1.17558241e+00
3.03471982e-01 -5.82302511e-01 -1.48125038e-01 4.04790074e-01
-1.16742156e-01 -2.43624136e-01 6.67877078e-01 -3.11613772e-02
-9.02828753e-01 2.43202984e-01 -3.00679624e-01 -4.78497177e-01
-2.83707082e-01 4.79065478e-01 -1.59163117e-01 -1.48268715e-01
3.32046062e-01 2.78572351e-01 -4.40883279e-01 -4.67426747e-01
8.73741135e-02 5.53654790e-01 6.42180443e-01 -3.45417023e-01
5.50585866e-01 4.85484809e-01 7.94624072e-03 -4.41498578e-01
-1.05999577e+00 -6.27831042e-01 -6.83537304e-01 -1.94538936e-01
1.32123041e+00 -6.05027437e-01 -1.37424484e-01 5.49385771e-02
-7.42919445e-01 -2.01147690e-01 -4.36673790e-01 6.00106895e-01
-3.71427983e-01 5.46345592e-01 -6.67122245e-01 -4.14719999e-01
-4.55093205e-01 -1.41113448e+00 1.36752951e+00 3.12473953e-01
1.70805529e-02 -8.77077222e-01 -7.23192096e-02 4.52588946e-01
2.82939970e-01 2.43358180e-01 1.09216177e+00 -4.97478634e-01
-7.64680564e-01 -3.36057514e-01 -6.27862930e-01 3.32453698e-01
2.60457635e-01 -2.54663914e-01 -7.98174918e-01 -1.54143870e-01
1.86287329e-01 -2.10547045e-01 1.16731405e+00 6.86708093e-01
1.84368098e+00 -2.08491180e-02 -6.65526927e-01 8.95303905e-01
1.20431173e+00 5.80267459e-02 2.85624474e-01 1.51570410e-01
9.91656721e-01 3.50922555e-01 6.29661679e-01 3.17274094e-01
2.47588590e-01 2.95022488e-01 6.68054581e-01 -8.30498040e-01
-1.96461022e-01 -3.29394601e-02 -2.66363263e-01 5.38048506e-01
-9.31324288e-02 -5.85733242e-02 -1.11282670e+00 7.40295410e-01
-1.48326099e+00 -5.20797849e-01 -4.41394262e-02 2.10693979e+00
9.68061745e-01 2.70289004e-01 -1.33341879e-01 6.56485483e-02
7.37161696e-01 2.69409806e-01 -5.47324777e-01 1.34114817e-01
2.22300842e-01 6.86380327e-01 5.49970925e-01 2.37566441e-01
-1.50752175e+00 4.04164135e-01 5.99700546e+00 1.03955173e+00
-1.23416400e+00 3.63143027e-01 1.09335923e+00 -7.32072443e-02
-2.63598025e-01 -1.92517504e-01 -5.14018536e-01 5.60412884e-01
1.58953458e-01 4.36188847e-01 -2.27108642e-01 8.65141749e-01
-6.65879995e-02 -1.25787616e-01 -8.31716537e-01 7.33363569e-01
-1.05159776e-02 -1.18290472e+00 -8.47237557e-02 -3.24809887e-02
7.62759268e-01 3.36063914e-02 7.34201297e-02 2.30155736e-01
-1.41392007e-01 -1.20796061e+00 2.82707334e-01 4.78973925e-01
9.21335340e-01 -5.12187123e-01 9.27874088e-01 3.70366514e-01
-1.31083155e+00 -8.30084011e-02 -3.75835687e-01 3.42359275e-01
-9.32257324e-02 5.08035004e-01 -7.59751499e-01 5.50325513e-01
8.84401143e-01 5.43586254e-01 -7.71628082e-01 1.04851210e+00
-2.77337313e-01 7.97023237e-01 -3.60989571e-01 2.14778334e-01
3.60372603e-01 9.96117592e-02 1.47143841e-01 1.16914344e+00
2.87776440e-01 3.28346014e-01 3.12700182e-01 9.60972309e-01
-2.54479468e-01 3.55650842e-01 -1.81479573e-01 3.05245966e-01
-1.06781684e-01 1.46768975e+00 -1.16206956e+00 -6.46550655e-01
-5.27852774e-01 9.00816143e-01 1.14569813e-01 2.57784247e-01
-1.10003376e+00 -3.63866478e-01 -1.44095212e-01 4.28200632e-01
2.67903566e-01 2.25199014e-01 -5.21279812e-01 -9.97098446e-01
-1.01974525e-01 -4.76357788e-01 7.15339661e-01 -6.33369148e-01
-1.02028954e+00 6.30133450e-01 -2.30085880e-01 -1.24043059e+00
8.93733725e-02 -4.51291621e-01 -9.91714597e-01 9.53930199e-01
-1.55092442e+00 -1.11801016e+00 -7.46534467e-01 5.94872773e-01
4.30238008e-01 2.13702843e-01 5.51493227e-01 3.67960781e-01
-6.93230927e-01 6.22856975e-01 -5.76136671e-02 3.05995703e-01
5.13260901e-01 -1.35100353e+00 -4.49227914e-02 7.81659901e-01
-1.83296472e-01 5.28337181e-01 3.08451615e-02 -5.99214077e-01
-8.14940155e-01 -1.26214182e+00 4.59884644e-01 -3.96951228e-01
2.78782159e-01 4.44363691e-02 -1.13590658e+00 5.11950374e-01
4.06963984e-03 7.68170834e-01 5.49491227e-01 -3.44378531e-01
2.40945086e-01 2.32448839e-02 -9.99580383e-01 3.05344135e-01
8.44321549e-01 -4.94244397e-01 -5.96841514e-01 6.13850653e-01
4.09961194e-01 -5.96421361e-01 -9.00663495e-01 8.72858047e-01
2.45169044e-01 -9.32641149e-01 1.23503745e+00 -2.41641119e-01
4.93488759e-01 -3.56851995e-01 1.69381171e-01 -7.07938075e-01
-4.76307184e-01 -8.66126046e-02 5.25838360e-02 8.68489325e-01
2.21953750e-01 -3.31389070e-01 1.11909914e+00 2.46744990e-01
-4.92319822e-01 -1.26436579e+00 -9.04802322e-01 -2.34839946e-01
1.12907276e-01 -1.62982777e-01 4.05767769e-01 7.44536281e-01
-5.03704190e-01 -1.03668170e-02 9.01739225e-02 9.21789333e-02
3.12506706e-01 2.94405758e-01 4.32914734e-01 -8.08185220e-01
-6.13458157e-01 -7.30351686e-01 -2.47916013e-01 -1.17088318e+00
-3.45994085e-01 -9.57719147e-01 7.17893764e-02 -1.76084268e+00
6.80209816e-01 -5.36277473e-01 -1.00906432e+00 5.33049285e-01
-8.94677103e-01 6.32135510e-01 4.83482070e-02 4.20095980e-01
-6.60236478e-01 3.65070462e-01 1.90684688e+00 -1.07849665e-01
7.90356565e-03 6.91098124e-02 -5.41255951e-01 9.50691044e-01
6.50264740e-01 -4.53113884e-01 -2.31056437e-01 -3.57709676e-01
-2.15955690e-01 1.76266328e-01 5.69894314e-01 -1.15890586e+00
7.86545873e-02 -1.01639047e-01 8.15265477e-01 -8.82428586e-01
9.42780823e-02 -7.77801037e-01 -4.37921256e-01 6.78937435e-01
-3.65179598e-01 -3.44279140e-01 -1.00531399e-01 5.50390422e-01
-4.84532803e-01 -2.71424562e-01 9.37670052e-01 -5.74093044e-01
-5.39084613e-01 5.60569584e-01 -2.14969724e-01 4.40008901e-02
1.23767018e+00 -2.14321747e-01 3.52279395e-02 5.21752834e-02
-7.42115140e-01 3.15724283e-01 3.18550974e-01 -5.95371658e-03
6.57347739e-01 -1.14917696e+00 -5.81276655e-01 5.33768415e-01
1.86130598e-01 5.49965203e-01 5.20696938e-01 1.22836721e+00
-8.81526589e-01 4.31656778e-01 -4.28413786e-02 -8.96545053e-01
-1.16823328e+00 6.00001872e-01 5.87215781e-01 -7.70217359e-01
-6.75155461e-01 1.15921712e+00 1.01582122e+00 -2.50729263e-01
4.94804941e-02 -9.06920135e-01 -1.82453468e-01 -4.09073412e-01
2.22667947e-01 -7.71690756e-02 5.98641001e-02 -5.62061548e-01
-4.88612950e-01 9.83666360e-01 -2.00021446e-01 3.88392955e-01
8.32858026e-01 -8.67482424e-02 -5.42880967e-03 -9.45594069e-03
1.20127106e+00 2.86091510e-02 -1.09303772e+00 -2.94077635e-01
-2.59180695e-01 -4.45798367e-01 3.18884909e-01 -7.99510479e-01
-1.36666822e+00 1.36385691e+00 7.52375841e-01 6.35745376e-02
1.47319937e+00 4.04295146e-01 9.42910850e-01 -8.76481310e-02
-2.32615292e-01 -8.27614188e-01 1.83451310e-01 1.14748344e-01
7.01439202e-01 -1.37855887e+00 1.87729895e-01 -4.65721071e-01
-5.24024248e-01 9.13654327e-01 8.59858751e-01 -2.22143814e-01
6.13596082e-01 -2.80971918e-02 1.34277597e-01 -3.35568190e-01
-2.80873924e-01 -3.39800775e-01 5.86330771e-01 2.63856441e-01
5.36526799e-01 7.25330487e-02 -2.96428055e-01 1.01156092e+00
4.20448661e-01 -2.84858376e-01 -1.99483424e-01 1.14715981e+00
-7.09881306e-01 -9.30570126e-01 -7.37793148e-01 8.59341681e-01
-9.53152239e-01 -8.07018653e-02 -2.52620697e-01 9.41086829e-01
5.01937270e-01 3.99349660e-01 -7.41058812e-02 -1.51980251e-01
1.30818203e-01 -2.29345962e-01 1.76153690e-01 -8.94733965e-01
-7.86021113e-01 4.23366278e-01 -3.87552112e-01 -4.44080412e-01
-2.02958062e-01 -3.93097162e-01 -1.66757178e+00 3.75261933e-01
-4.91934925e-01 -7.78514519e-02 1.81603849e-01 9.11856353e-01
-2.47826185e-02 1.12271392e+00 6.37172222e-01 -6.88331962e-01
-2.88906485e-01 -7.37732947e-01 -4.62202698e-01 5.60029924e-01
2.71025658e-01 -6.35644615e-01 -8.95834938e-02 -5.43484986e-02] | [15.098577499389648, -2.1361100673675537] |
15dbf58b-5dfd-4d7c-925b-f36e83a287f4 | a-transformer-based-approach-for-translating | null | null | https://www.dre.vanderbilt.edu/~schmidt/PDF/A_Transformer_based_Approach_for_TranslatingNatural_Language_to_Bash_Commands.pdf | https://www.dre.vanderbilt.edu/~schmidt/PDF/A_Transformer_based_Approach_for_TranslatingNatural_Language_to_Bash_Commands.pdf | A Transformer-based Approach for Translating Natural Language to Bash Commands | This paper explores the translation of natural language into Bash Commands, which developers commonly use to accomplish command-line tasks in a terminal. In our approach a terminal takes a command as a sentence in plain English and translates it into the corresponding string of Bash Commands. The paper analyzes the performance of several architectures on this translation problem using the data from the NLC2CMD competition at the NeurIPS 2020 conference. The approach presented in this paper is the best performing architecture on this problem to date and improves the current state-of-the-art accuracy on this translation task from 13.8% to 53.2%. | ['Douglas C. Schmidt', 'Jules White', 'Zhongwei Teng', 'Quchen Fu'] | 2021-12-14 | null | null | null | 20th-ieee-international-conference-on-machine | ['code-translation'] | ['computer-code'] | [ 3.70906144e-01 6.81970716e-02 -1.85636356e-01 -8.01252306e-01
-1.18018425e+00 -6.92610085e-01 7.58318603e-01 -3.67776491e-02
-7.11608529e-01 8.02740216e-01 1.03583194e-01 -9.08277512e-01
5.48092127e-01 -4.57776010e-01 -7.34748721e-01 7.56803080e-02
2.01829538e-01 9.09750581e-01 1.44639805e-01 -8.66126001e-01
1.21354684e-01 4.25743818e-01 -7.33945012e-01 8.42082798e-01
6.76529169e-01 7.83315837e-01 2.39818051e-01 7.53844738e-01
-4.54887062e-01 4.92388517e-01 -9.84112918e-01 -6.94178581e-01
1.84528366e-01 -6.22081995e-01 -1.32546437e+00 -9.10878122e-01
4.75947052e-01 -2.97872663e-01 2.36071572e-02 1.08834398e+00
5.08915484e-01 -5.77365756e-01 4.32054251e-01 -1.18892300e+00
-4.83030587e-01 1.16664779e+00 -4.05557714e-02 3.96524251e-01
1.09706032e+00 6.31728470e-02 8.26603591e-01 -8.56183410e-01
8.11011910e-01 1.00853765e+00 5.30576348e-01 7.39345431e-01
-1.24556124e+00 -5.58332562e-01 -3.86698276e-01 -6.76551014e-02
-1.33628118e+00 -5.91426611e-01 1.18645348e-01 -2.89735675e-01
1.90684903e+00 4.91664290e-01 5.50532639e-01 1.23585033e+00
8.63634586e-01 6.64618313e-01 9.81256783e-01 -5.90290964e-01
3.60961370e-02 -7.96830878e-02 2.58877516e-01 5.79327822e-01
-1.86769471e-01 -7.96405971e-02 -7.48339891e-01 -3.53185199e-02
4.93085861e-01 -6.99132502e-01 -1.34053171e-01 3.41921836e-01
-1.79769683e+00 5.15258074e-01 3.59985918e-01 5.32576978e-01
-3.39926362e-01 1.32759139e-01 6.37183130e-01 1.07524991e+00
3.63906473e-01 8.83424163e-01 -4.77105796e-01 -5.36124289e-01
-1.00356460e+00 5.14699042e-01 1.24351501e+00 1.56938231e+00
9.03363526e-02 -9.25148055e-02 -1.09868929e-01 4.82604563e-01
6.01311997e-02 5.50439954e-01 6.02908075e-01 -5.67539752e-01
1.31999469e+00 1.66773900e-01 1.66341320e-01 -2.95838416e-01
-5.73886752e-01 -4.12637353e-01 -5.34666777e-01 2.80312896e-01
2.37164289e-01 -5.73602259e-01 -7.68917382e-01 1.38399231e+00
-3.85659873e-01 -6.44410789e-01 1.58797622e-01 7.97979653e-01
6.23818278e-01 9.20878649e-01 -2.60770053e-01 -2.88491081e-02
9.66308475e-01 -8.79454017e-01 -7.78575003e-01 -3.62033963e-01
7.59252012e-01 -1.05735445e+00 1.01136577e+00 4.28896755e-01
-1.40372443e+00 -5.01166880e-01 -1.31769860e+00 -7.80685470e-02
-4.81186271e-01 3.30860138e-01 4.18092817e-01 4.47617739e-01
-1.54800057e+00 5.66443145e-01 -9.30590630e-01 -7.64296651e-01
1.25967339e-02 5.14650822e-01 -3.73653561e-01 2.12238580e-01
-1.36894381e+00 1.59892702e+00 5.65412164e-01 -1.04068704e-01
-7.13902116e-01 -6.09688461e-01 -5.11754155e-01 -1.86315998e-01
-2.23373875e-01 -6.40394807e-01 2.09092021e+00 -6.70569062e-01
-1.71071017e+00 9.05369937e-01 -1.61810070e-01 -1.01836073e+00
6.97499156e-01 -3.20706636e-01 -6.20705247e-01 -1.19313627e-01
2.65996963e-01 8.38859320e-01 6.11149013e-01 -3.89922947e-01
-7.59918988e-01 -2.72623003e-01 1.52002856e-01 1.05897784e-01
1.17847890e-01 7.16704726e-01 -2.65436172e-01 -4.15820450e-01
-6.14218377e-02 -9.42806661e-01 8.00453424e-02 -3.22579622e-01
-5.25204659e-01 -3.31370771e-01 4.08363104e-01 -7.81959236e-01
1.12235916e+00 -1.84800565e+00 5.23301125e-01 -6.67895228e-02
6.68208599e-02 1.89908803e-01 -3.34790170e-01 9.74448025e-01
-2.15154648e-01 8.85987747e-03 -1.53702483e-01 -2.99049973e-01
8.51525143e-02 -1.31258965e-01 -6.04376078e-01 3.07990313e-01
3.68628800e-01 1.12121081e+00 -6.64526284e-01 -2.01043949e-01
-3.06447018e-02 1.19162858e-01 -2.62286484e-01 3.52149397e-01
-4.33918387e-01 3.70171554e-02 -4.14146602e-01 3.06509852e-01
3.21705073e-01 3.41257483e-01 8.13916922e-02 6.06928617e-02
-3.96241009e-01 1.04377770e+00 -3.24330389e-01 2.30442309e+00
-7.07270741e-01 1.03011894e+00 -9.26377773e-02 -8.42659593e-01
9.53501344e-01 5.03626764e-01 7.04538673e-02 -5.59312105e-01
2.61602253e-01 6.49367094e-01 3.03636849e-01 -3.13824356e-01
3.70696217e-01 -7.49456435e-02 -3.89304429e-01 4.67977911e-01
3.71737748e-01 -6.25080287e-01 3.04862052e-01 7.16312528e-02
1.37001324e+00 3.08896065e-01 5.39206088e-01 -4.92574245e-01
3.95544291e-01 6.07158601e-01 -3.36014211e-01 7.49576449e-01
2.76688397e-01 5.95090389e-01 4.22890663e-01 -4.76043165e-01
-1.05271876e+00 -8.45923662e-01 1.45547882e-01 1.15561461e+00
-5.60150623e-01 -6.32676005e-01 -1.23567283e+00 -5.40654778e-01
-4.33410287e-01 1.33652675e+00 -2.97941685e-01 -3.21533419e-02
-7.17676818e-01 -1.92598820e-01 1.08240652e+00 5.16708791e-01
4.49272752e-01 -1.17161798e+00 -8.74913871e-01 4.84026045e-01
-4.21973914e-01 -1.23782444e+00 -5.00511348e-01 7.10888207e-01
-7.23659635e-01 -3.70667428e-01 -7.31903315e-01 -1.05350959e+00
3.28839630e-01 -1.27001539e-01 1.34155881e+00 -3.08084399e-01
-1.34936450e-02 -3.52054626e-01 -6.38128519e-01 -6.62600994e-01
-9.88883913e-01 8.13658834e-01 -8.13266709e-02 -6.28786564e-01
7.03416109e-01 -3.33101660e-01 4.34719414e-01 -5.43367043e-02
-5.89674473e-01 4.53063965e-01 7.52291560e-01 7.57777452e-01
4.79404703e-02 -3.93684089e-01 1.76526040e-01 -9.50787842e-01
1.03970814e+00 -1.83428839e-01 -5.89693487e-01 -1.73379686e-02
-5.36779106e-01 1.97007313e-01 1.18611538e+00 -7.36157298e-02
-7.50966728e-01 3.51322651e-01 -7.10062683e-01 1.83069751e-01
-5.12378752e-01 9.30596888e-01 1.52139768e-01 -1.03520341e-01
1.07434750e+00 4.66104329e-01 2.03285366e-02 -5.10680318e-01
2.71612495e-01 1.17088926e+00 8.72716010e-01 -3.79736453e-01
8.10558558e-01 -2.68285245e-01 -2.78295577e-01 -5.37936270e-01
-4.71882820e-01 -2.41278931e-01 -8.75283301e-01 2.01194599e-01
8.39187741e-01 -7.38533795e-01 3.17540392e-02 4.81306642e-01
-1.99864829e+00 -3.93574357e-01 7.77286366e-02 4.14462477e-01
-9.28470254e-01 -2.12402388e-01 -7.03657627e-01 -9.88654196e-02
-7.20023692e-01 -1.14332974e+00 1.14490354e+00 -4.11330312e-01
-7.62512803e-01 -7.17353344e-01 4.33214381e-03 -2.64490452e-02
8.12981904e-01 -1.39622107e-01 1.15702105e+00 -8.37792575e-01
-1.39982849e-01 -5.05675197e-01 1.10808909e-01 1.71949849e-01
-1.21760346e-01 -2.29676291e-01 -4.81457740e-01 -2.82025367e-01
9.65958927e-03 -4.22447354e-01 3.48204970e-01 -1.71507359e-01
5.05144954e-01 -1.86050743e-01 -3.11586738e-01 4.99591500e-01
1.27678370e+00 4.71326262e-01 6.29119396e-01 4.18654025e-01
2.40725115e-01 3.92249078e-01 4.90932256e-01 7.96412379e-02
2.58889556e-01 1.02138877e+00 3.55902940e-01 -7.39777237e-02
-1.09637871e-01 -1.50350541e-01 9.32744145e-01 8.16934168e-01
1.74157143e-01 -3.16816717e-01 -1.39475560e+00 1.90822080e-01
-1.62197137e+00 -6.40379786e-01 -1.53371349e-01 1.79574084e+00
1.16811812e+00 5.15209615e-01 4.09538066e-03 -1.43282339e-01
4.02668923e-01 -1.32765144e-01 -1.81334823e-01 -9.44927275e-01
1.65544659e-01 5.85233986e-01 3.82381618e-01 5.72637737e-01
-9.06968772e-01 1.24330413e+00 7.44425869e+00 5.41780531e-01
-1.32994115e+00 3.01502682e-02 5.82952201e-02 9.91929024e-02
2.95822807e-02 -3.22313786e-01 -9.38909829e-01 3.61281306e-01
1.74593949e+00 -4.93584156e-01 8.30630600e-01 6.09972000e-01
1.00168847e-01 1.82004243e-01 -2.12185836e+00 8.92829239e-01
3.69357258e-01 -1.37007427e+00 1.84632406e-01 -3.28231364e-01
9.89471525e-02 7.74264753e-01 -1.80027306e-01 4.01983172e-01
1.60578027e-01 -1.30438685e+00 1.14038157e+00 3.17857713e-01
1.22739589e+00 -3.83111119e-01 1.03843880e+00 5.46389163e-01
-7.03839779e-01 3.04018945e-01 -2.05679715e-01 -2.52277404e-01
2.54913330e-01 -1.56663224e-01 -1.12536824e+00 5.87553024e-01
4.31338042e-01 5.94816566e-01 -7.22775102e-01 7.07762957e-01
-2.86929578e-01 6.13166988e-01 -4.48320508e-01 -3.96394700e-01
4.49577570e-01 -1.96508132e-02 4.88642097e-01 1.65247083e+00
3.66070360e-01 -7.20213577e-02 -1.72525510e-01 8.15129161e-01
-1.27252519e-01 8.56645107e-02 -9.08210456e-01 -3.26590240e-01
4.47231978e-01 8.42765391e-01 -1.81554243e-01 -3.83148968e-01
-6.81855023e-01 1.42112124e+00 3.31013829e-01 1.81105360e-01
-8.31639826e-01 -1.03378808e+00 4.40365642e-01 2.23738980e-02
-1.58135340e-01 -3.03400904e-01 -3.40007395e-01 -1.04654717e+00
2.35861734e-01 -1.29905188e+00 -1.24329530e-01 -1.11799502e+00
-1.01965058e+00 1.16840494e+00 3.58859375e-02 -1.39088571e+00
-8.61008763e-01 -7.60577381e-01 -5.76801717e-01 1.29630697e+00
-9.14088547e-01 -9.98824835e-01 1.03361987e-01 3.90686393e-01
8.41110110e-01 -5.86445451e-01 1.39894855e+00 2.46937498e-01
-6.20323233e-02 7.72154450e-01 7.37019852e-02 3.88642013e-01
5.91406763e-01 -1.25037408e+00 1.54553759e+00 9.20601189e-01
2.32236922e-01 9.48431969e-01 8.39017332e-01 -3.18204999e-01
-1.73985732e+00 -9.30638492e-01 1.58410585e+00 -5.49688995e-01
7.33806431e-01 -4.59581375e-01 -4.14503783e-01 1.05500543e+00
8.84307861e-01 -5.75117528e-01 3.16036791e-01 -3.19472849e-01
-2.86368906e-01 -4.69807684e-02 -9.28427875e-01 6.37466073e-01
8.56346905e-01 -7.06166923e-01 -1.10119450e+00 5.10334790e-01
8.54178846e-01 -9.75894094e-01 -1.01960266e+00 -2.89929807e-02
5.42588234e-01 -4.99321401e-01 4.04966354e-01 -1.05904067e+00
7.05584526e-01 -2.06859648e-01 -2.48572275e-01 -1.76460755e+00
-9.52581093e-02 -9.37537193e-01 4.41226721e-01 8.54680836e-01
1.15814269e+00 -5.43942571e-01 5.02215624e-01 3.45179170e-01
-5.22661150e-01 -6.86778843e-01 -1.07769823e+00 -7.98762262e-01
5.07429659e-01 -4.66875345e-01 5.93691230e-01 5.88978946e-01
5.67206144e-01 9.17435050e-01 -5.35575897e-02 -3.77955973e-01
7.53725618e-02 5.71975149e-02 8.08383524e-01 -8.85857522e-01
-2.25292787e-01 -3.85232747e-01 -4.54083294e-01 -1.32193100e+00
2.32728034e-01 -1.41660345e+00 3.19584668e-01 -1.58546877e+00
-2.86790103e-01 1.07811108e-01 2.64520973e-01 7.52196431e-01
4.96079981e-01 3.19363773e-02 4.51905787e-01 1.68669045e-01
-2.57837355e-01 4.98561934e-02 6.42998934e-01 -3.30446631e-01
1.17244765e-01 7.42868632e-02 -5.29976189e-01 3.95306885e-01
1.06158853e+00 -5.47224164e-01 -1.40482202e-01 -1.17577004e+00
2.72009134e-01 4.25380051e-01 -2.01651573e-01 -1.08716702e+00
2.64271647e-01 9.13068354e-02 1.47946075e-01 -5.12301683e-01
1.38092786e-01 -8.13959777e-01 1.19129859e-01 6.20662808e-01
-8.33848238e-01 8.29391599e-01 2.07262710e-01 -2.36604407e-01
-4.58841532e-01 -3.22149187e-01 6.18408799e-01 -1.59097865e-01
-5.07096648e-01 -2.83589393e-01 -9.12072718e-01 9.17622298e-02
7.22932100e-01 3.15447479e-01 -5.62741995e-01 -4.01690453e-01
-4.55098152e-01 8.54872540e-02 3.79614294e-01 6.07979655e-01
6.04547083e-01 -1.13014174e+00 -1.00371230e+00 2.76622683e-01
1.47520497e-01 -4.76992100e-01 -8.49715889e-01 1.04607677e+00
-1.12848485e+00 1.02229857e+00 -7.51537323e-01 -4.13586646e-01
-1.12994206e+00 1.14931822e-01 3.99292320e-01 -5.57819046e-02
-5.26143014e-01 8.37114394e-01 -8.75412762e-01 -5.58267415e-01
2.11948961e-01 -6.83358729e-01 -6.51586652e-02 -2.65831560e-01
5.42711020e-01 1.32234227e-02 7.38887250e-01 -3.90787065e-01
-5.85845470e-01 3.82219881e-01 -3.90806884e-01 -5.60232818e-01
1.05485380e+00 3.60419631e-01 -4.58218664e-01 5.15694559e-01
1.36366117e+00 -4.23267215e-01 -1.31023452e-01 -5.91831543e-02
2.49813616e-01 1.10813037e-01 -1.47206888e-01 -1.31807005e+00
-5.07565260e-01 1.04109383e+00 2.65084863e-01 1.55287132e-01
9.64968324e-01 -1.32995129e-01 1.00479841e+00 1.09735560e+00
9.39659774e-01 -8.58510077e-01 -7.01914489e-01 9.51908767e-01
1.11782789e+00 -7.76834726e-01 -3.09779704e-01 -2.38746569e-01
-6.00679874e-01 1.58697224e+00 4.77833211e-01 -2.03860968e-01
1.20541550e-01 8.26576293e-01 3.91362786e-01 9.68427137e-02
-1.06360734e+00 1.87679484e-01 -1.28640160e-01 7.05364168e-01
8.60942841e-01 1.68058828e-01 -6.73264802e-01 3.31502408e-01
-8.48051071e-01 2.99367577e-01 8.07261109e-01 1.08609986e+00
-1.80345386e-01 -1.32867837e+00 -8.15396532e-02 6.26231194e-01
-6.61156237e-01 -6.57752872e-01 -9.72530544e-01 6.03430748e-01
-2.89156228e-01 9.21679854e-01 -1.99781060e-01 -7.39470124e-01
5.89945316e-01 3.25718403e-01 6.22879326e-01 -1.09530520e+00
-1.10884702e+00 -2.43865997e-01 6.04285479e-01 -7.93201029e-01
6.55844957e-02 -6.91843808e-01 -1.43037224e+00 -3.56064200e-01
1.46136284e-01 3.29787165e-01 9.68275487e-01 8.09089482e-01
2.63346106e-01 4.79334325e-01 2.24907219e-01 -7.02983320e-01
-1.18104720e+00 -1.19802868e+00 -1.35848671e-01 -6.93423972e-02
3.74325424e-01 1.35507271e-01 1.79539565e-02 2.65371561e-01] | [11.543485641479492, 10.348734855651855] |
3d8924b0-4beb-4b30-9028-166218670f4c | comparative-analysis-of-non-blind-deblurring | 2205.03464 | null | https://arxiv.org/abs/2205.03464v1 | https://arxiv.org/pdf/2205.03464v1.pdf | Comparative Analysis of Non-Blind Deblurring Methods for Noisy Blurred Images | Image blurring refers to the degradation of an image wherein the image's overall sharpness decreases. Image blurring is caused by several factors. Additionally, during the image acquisition process, noise may get added to the image. Such a noisy and blurred image can be represented as the image resulting from the convolution of the original image with the associated point spread function, along with additive noise. However, the blurred image often contains inadequate information to uniquely determine the plausible original image. Based on the availability of blurring information, image deblurring methods can be classified as blind and non-blind. In non-blind image deblurring, some prior information is known regarding the corresponding point spread function and the added noise. The objective of this study is to determine the effectiveness of non-blind image deblurring methods with respect to the identification and elimination of noise present in blurred images. In this study, three non-blind image deblurring methods, namely Wiener deconvolution, Lucy-Richardson deconvolution, and regularized deconvolution were comparatively analyzed for noisy images featuring salt-and-pepper noise. Two types of blurring effects were simulated, namely motion blurring and Gaussian blurring. The said three non-blind deblurring methods were applied under two scenarios: direct deblurring of noisy blurred images and deblurring of images after denoising through the application of the adaptive median filter. The obtained results were then compared for each scenario to determine the best approach for deblurring noisy images. | ['Poorna Banerjee Dasgupta'] | 2022-05-06 | null | null | null | null | ['blind-image-deblurring'] | ['computer-vision'] | [ 2.86023766e-01 -7.32930303e-01 5.18506408e-01 1.02348059e-01
-4.54931818e-02 -6.96577787e-01 5.09564936e-01 -6.08832479e-01
-3.46677303e-01 9.79211330e-01 5.67458868e-01 -2.81345308e-01
-4.43725675e-01 -1.41361415e-01 -3.38154018e-01 -9.09465313e-01
1.55410767e-01 -2.80076742e-01 -2.26716742e-01 9.11516175e-02
3.83840919e-01 4.41584408e-01 -1.45746934e+00 -4.59700003e-02
1.12736893e+00 7.20943570e-01 8.76862347e-01 1.15827310e+00
7.72086009e-02 8.59235227e-01 -8.90429258e-01 1.12767868e-01
2.13721648e-01 -6.18340552e-01 -6.32134616e-01 5.47175407e-01
7.83949345e-02 -6.18332028e-01 -6.62972331e-01 1.60942054e+00
4.11698222e-01 3.06964755e-01 5.15610218e-01 -6.11520052e-01
-1.21555328e+00 8.96933228e-02 -7.11806357e-01 7.06412554e-01
4.16294456e-01 2.30281264e-01 -7.31002390e-02 -9.86359060e-01
2.66087085e-01 1.15715015e+00 6.24719203e-01 2.63018996e-01
-1.14096594e+00 -1.88338280e-01 -6.93388224e-01 5.24471521e-01
-1.49414563e+00 -5.75752616e-01 6.55157804e-01 -5.32164037e-01
1.60735071e-01 5.67043900e-01 2.01344997e-01 4.64895844e-01
4.86094922e-01 2.57455826e-01 1.71659040e+00 -5.54634988e-01
1.71739355e-01 -1.40639599e-02 2.88732797e-01 1.99582919e-01
6.91745758e-01 4.29455906e-01 -1.03097588e-01 -2.26056874e-01
8.13973188e-01 1.42506540e-01 -1.11587346e+00 6.04237355e-02
-1.18626428e+00 1.47913620e-01 4.06056017e-01 5.55921435e-01
-6.22184396e-01 -1.61751166e-01 2.04822794e-02 2.74076134e-01
4.52544868e-01 4.55661029e-01 -1.80321038e-01 -7.66405538e-02
-1.18650222e+00 -2.48414837e-02 6.36303544e-01 4.42484677e-01
7.00090349e-01 1.18316580e-02 -2.38532007e-01 9.71701443e-01
2.76688039e-01 8.04444313e-01 8.45173061e-01 -9.15433466e-01
2.06872746e-02 -1.00640200e-01 8.01213503e-01 -9.33998227e-01
1.48092017e-01 -2.28981555e-01 -1.09989989e+00 4.51250494e-01
4.73243952e-01 -3.18526238e-01 -1.17516863e+00 1.15354621e+00
-1.19031347e-01 4.74151760e-01 3.59466791e-01 1.40184069e+00
6.50713503e-01 6.46473169e-01 -3.74398649e-01 -5.11579454e-01
1.52598953e+00 -7.33150780e-01 -1.38508320e+00 -2.60245681e-01
-3.21347386e-01 -1.43244326e+00 5.19302964e-01 3.10221434e-01
-1.06530261e+00 -7.73731470e-01 -1.03566635e+00 1.59089297e-01
-1.32817701e-01 1.95265442e-01 -1.07368931e-01 7.36927748e-01
-1.17327499e+00 4.58399355e-01 -5.34773827e-01 -2.37408191e-01
1.01056129e-01 1.42166475e-02 -2.94986248e-01 -6.54624879e-01
-1.22887731e+00 1.40805662e+00 -1.36070955e-03 6.22595131e-01
-8.02327693e-01 -4.03214991e-01 -7.59869516e-01 -1.45823546e-02
-6.53933287e-02 -8.45231533e-01 1.22194660e+00 -1.01518583e+00
-1.22095931e+00 4.70160127e-01 -6.17663026e-01 -3.64682108e-01
6.59379721e-01 -4.05770510e-01 -5.40148437e-01 1.77380770e-01
-1.99848749e-02 -2.52634972e-01 1.52623248e+00 -1.55912173e+00
-1.60368562e-01 -2.66325265e-01 -4.85715836e-01 4.55640018e-01
4.41870630e-01 1.72941104e-01 -1.98717326e-01 -9.67293143e-01
1.62750274e-01 -5.83001912e-01 -6.91881701e-02 -3.17123741e-01
-2.33061045e-01 6.48342073e-01 1.15711427e+00 -1.20402026e+00
1.21308994e+00 -2.44845867e+00 -5.11680655e-02 -2.26800352e-01
2.47572616e-01 4.97371882e-01 5.94510930e-03 1.99877098e-01
-3.22137624e-01 -1.22869335e-01 -4.77114737e-01 -4.92245238e-03
-6.06088698e-01 1.08229645e-01 -1.89347118e-01 9.81722891e-01
-2.80278832e-01 6.03178740e-01 -1.01094723e+00 -2.49543618e-02
5.43001473e-01 5.85262001e-01 3.22775453e-01 4.64442849e-01
5.04136145e-01 5.66011965e-01 2.26891208e-02 4.08646494e-01
1.36789334e+00 2.72629578e-02 -3.44716907e-01 -4.96307284e-01
-3.78666788e-01 -4.94640857e-01 -1.09114528e+00 1.03595257e+00
-3.71198386e-01 1.03088462e+00 6.52603209e-01 -6.00163043e-01
7.10191965e-01 6.48387730e-01 1.38818219e-01 1.17154578e-02
8.10887367e-02 4.46107864e-01 1.43110892e-02 -1.03323936e+00
8.90758157e-01 -2.70033389e-01 5.56192219e-01 2.93452024e-01
-3.54718596e-01 -4.08805549e-01 -9.90234390e-02 -1.83347061e-01
8.94925833e-01 -2.60347396e-01 3.47462565e-01 -4.33810771e-01
7.09008098e-01 -1.51691169e-01 6.28504204e-03 9.31437314e-01
-3.95010054e-01 1.01169705e+00 -3.77690226e-01 -1.24736786e-01
-9.58173394e-01 -8.67418408e-01 -3.31517518e-01 2.69738078e-01
8.08322370e-01 5.10070801e-01 -1.07738185e+00 7.46515095e-02
-2.01643944e-01 6.67387962e-01 -3.59338790e-01 -9.58968699e-02
-2.59576589e-01 -1.07155764e+00 4.59134877e-01 -2.20718086e-01
9.91775990e-01 -7.75725663e-01 -3.89466494e-01 1.61048815e-01
-3.87915790e-01 -8.09692979e-01 -8.84874701e-01 -2.10775122e-01
-8.05002272e-01 -1.33137810e+00 -1.54433668e+00 -7.14357972e-01
1.06435323e+00 1.15315497e+00 4.42036122e-01 4.00158837e-02
-2.69894123e-01 3.28898609e-01 -2.05587327e-01 1.12645216e-01
-8.00675809e-01 -9.32653129e-01 8.56929198e-02 3.26267749e-01
2.36232370e-01 -1.15804426e-01 -9.02739286e-01 6.33763492e-01
-1.19095612e+00 -2.02750340e-01 6.71339393e-01 9.08775568e-01
-1.00518592e-01 8.16858590e-01 1.26895353e-01 -1.97104126e-01
1.26327598e+00 -4.22919780e-01 -3.71485144e-01 2.05183968e-01
-5.76458752e-01 -3.87810245e-02 3.46799463e-01 -5.79301000e-01
-1.62005424e+00 -3.90727758e-01 5.74998796e-01 -6.32344961e-01
-4.36054677e-01 4.96930212e-01 1.00320667e-01 -2.84504920e-01
8.57248127e-01 6.10748947e-01 4.26332504e-01 -7.71600604e-01
4.78760093e-01 1.35474944e+00 1.07596684e+00 8.13371912e-02
7.26363719e-01 3.64673287e-01 -3.06990296e-01 -1.20366883e+00
-2.21762910e-01 -7.19607055e-01 -1.97484478e-01 -2.79877424e-01
7.30403662e-01 -7.98227847e-01 -2.76533812e-01 1.44362593e+00
-1.42306542e+00 1.22715309e-01 2.59150397e-02 8.27022076e-01
-2.21017584e-01 9.68705297e-01 -8.13368857e-01 -9.02123153e-01
-2.12860733e-01 -1.41390049e+00 1.46242395e-01 5.27852714e-01
-1.14696287e-01 -1.02079976e+00 -1.53546646e-01 5.18824577e-01
9.87685502e-01 -2.35949516e-01 4.50605661e-01 8.20993036e-02
-3.11832219e-01 -1.64631426e-01 -6.03550196e-01 7.70736933e-01
1.07702792e+00 -4.45796698e-01 -9.66818273e-01 -3.13523591e-01
8.56601477e-01 4.91433829e-01 7.61581361e-01 9.74575758e-01
6.14653885e-01 -5.40223420e-01 2.61005480e-02 4.61424649e-01
1.50338233e+00 3.64135414e-01 1.02639842e+00 3.77731115e-01
4.95164990e-01 3.52259248e-01 3.68224472e-01 2.25582480e-01
-3.08284640e-01 4.24370944e-01 1.52297929e-01 -2.38278449e-01
-6.57715440e-01 3.46684039e-01 1.99123725e-01 4.58570629e-01
-3.09635729e-01 -2.89106101e-01 -5.88595867e-01 7.14340210e-01
-1.49223256e+00 -1.12072194e+00 -4.94107246e-01 2.26786089e+00
9.17421103e-01 -5.34441352e-01 -6.23913705e-01 1.18034713e-01
1.43118787e+00 2.10777432e-01 -5.00693381e-01 -8.88138264e-02
-3.04628760e-01 -8.59830976e-02 8.94130230e-01 1.04506731e+00
-9.77664053e-01 4.14413214e-01 6.31847191e+00 5.84042609e-01
-1.11315382e+00 -2.40507517e-02 4.95236814e-01 5.28841734e-01
8.99803638e-02 2.57057771e-02 -2.85550561e-02 8.48705590e-01
6.79880679e-01 -3.84852856e-01 1.12076795e+00 2.89067537e-01
1.00646627e+00 -7.72617221e-01 -2.77238548e-01 1.30756164e+00
8.31745937e-02 -8.86283338e-01 -3.05558980e-01 -1.28234297e-01
8.78678262e-01 -1.90688431e-01 3.20975594e-02 -5.80834866e-01
2.43396863e-01 -1.04185474e+00 5.79321206e-01 9.67552066e-01
8.07818472e-01 -6.67543113e-02 1.17889738e+00 2.33769655e-01
-5.70836484e-01 3.22077200e-02 -2.26144686e-01 -3.36352974e-01
2.98026055e-01 9.00185823e-01 -5.27453303e-01 5.12154460e-01
5.18446565e-01 5.67451358e-01 -2.81146437e-01 1.48688757e+00
-2.76879579e-01 4.99018967e-01 1.73456132e-01 4.72115487e-01
-6.85615242e-02 -5.11708856e-01 9.29197073e-01 1.02156317e+00
7.92482913e-01 3.52071911e-01 -7.14244306e-01 8.99016261e-01
1.61185652e-01 -5.44832945e-01 -3.25476885e-01 2.02234104e-01
6.18825078e-01 7.88410962e-01 -3.93978834e-01 -4.89774674e-01
-2.32605636e-01 1.39504743e+00 -8.20851326e-01 9.71558213e-01
-4.26498890e-01 -4.97058898e-01 8.11471939e-01 -3.19229402e-02
3.36467803e-01 -1.25757188e-01 -5.12400746e-01 -1.20585001e+00
-3.36545110e-01 -8.65936875e-01 -1.88882530e-01 -1.47727048e+00
-1.29635012e+00 6.40483677e-01 1.01795718e-01 -1.22046483e+00
3.92379835e-02 -3.15988570e-01 -4.72547144e-01 1.80923522e+00
-1.50380421e+00 -5.03859758e-01 -6.44824624e-01 3.48051429e-01
7.29966879e-01 8.89699347e-03 2.85904646e-01 7.33600631e-02
-2.09104478e-01 -3.10924858e-01 7.15381742e-01 -2.28667095e-01
8.35695684e-01 -1.15677297e+00 1.18937276e-01 1.36921382e+00
-6.15789771e-01 7.87219405e-01 1.36183238e+00 -7.76423573e-01
-1.18579018e+00 -7.85755515e-01 6.35801494e-01 5.28203323e-02
5.14003038e-01 5.64487278e-01 -1.20579565e+00 1.47571802e-01
5.92020094e-01 5.86813055e-02 -1.53849076e-03 -8.66530180e-01
1.69862062e-01 6.40857145e-02 -1.45185101e+00 2.88397431e-01
1.49523959e-01 -5.06863415e-01 -1.04784024e+00 -2.03147888e-01
3.60772312e-01 -4.14692879e-01 -5.43441892e-01 1.29730299e-01
2.54414469e-01 -8.98312926e-01 1.04226220e+00 1.81291014e-01
3.42387408e-01 -9.34523880e-01 1.12400174e-01 -1.76043248e+00
-6.28533959e-01 -9.04139996e-01 -4.27735969e-02 9.35060620e-01
-2.02496797e-01 -6.95884764e-01 1.85859263e-01 5.23268282e-01
-9.30517539e-03 1.14037909e-01 -5.49776435e-01 -6.05535209e-01
-4.75542992e-01 1.47353426e-01 4.26953167e-01 8.97956789e-01
-2.26429030e-01 -4.69825640e-02 -7.24125862e-01 5.50437272e-01
9.68147814e-01 -2.58034259e-01 3.67004395e-01 -6.81395590e-01
-4.44807559e-02 -2.16698766e-01 -1.80499718e-01 -1.28275561e+00
-4.46760446e-01 -1.06488965e-01 4.72181588e-01 -1.59825075e+00
1.72321379e-01 -5.82127646e-02 -4.61938940e-02 -1.69517875e-01
-6.91127658e-01 1.58374667e-01 5.05922772e-02 9.98077989e-01
3.64727080e-01 1.14971377e-01 1.43958759e+00 -9.12295207e-02
-1.50114998e-01 1.90476522e-01 -5.91488123e-01 6.33019388e-01
6.24073267e-01 -1.66494608e-01 -3.56264234e-01 -5.20416796e-01
-4.76286471e-01 3.68189156e-01 5.39541423e-01 -7.96027541e-01
3.73272747e-01 -1.59728199e-01 2.60946929e-01 -2.88287371e-01
6.02277294e-02 -8.55389655e-01 6.87574029e-01 4.88413393e-01
-2.06522495e-02 -2.06014276e-01 1.33706108e-01 6.58310831e-01
-5.07606983e-01 -7.66521513e-01 1.00570071e+00 -3.16794872e-01
-8.52073133e-01 -4.58209962e-01 -8.48743796e-01 -4.70897019e-01
7.83344090e-01 -5.58428824e-01 -6.63056016e-01 -6.69566274e-01
-6.92685783e-01 -4.61736202e-01 7.67599165e-01 -2.74782092e-03
7.97042727e-01 -7.71501780e-01 -7.18652070e-01 2.17768744e-01
-5.31006753e-01 -3.15721542e-01 6.44235551e-01 9.76728976e-01
-9.70742762e-01 1.84772566e-01 -1.58433154e-01 -1.67960614e-01
-1.47595727e+00 7.46363521e-01 5.84232688e-01 2.56096721e-01
-1.26500070e-01 5.38163781e-01 1.73018724e-01 4.36743975e-01
-1.43578500e-01 -2.73528308e-01 -2.51773924e-01 -3.84042144e-01
9.43056345e-01 9.15486932e-01 4.31552948e-03 -9.43611860e-01
-1.09665722e-01 5.85170090e-01 1.64333254e-01 -1.58294320e-01
7.67071068e-01 -8.99195969e-01 -6.08924150e-01 2.52819601e-02
1.03432906e+00 2.51367986e-01 -1.31311429e+00 -1.37587115e-01
-1.22254729e-01 -9.04480755e-01 6.59106910e-01 -9.98709321e-01
-7.08589017e-01 4.11823541e-01 8.54945898e-01 4.29701209e-01
1.51935458e+00 -2.39505455e-01 4.75142360e-01 -2.91541517e-01
1.55612873e-02 -4.18887824e-01 -3.24867845e-01 2.44894654e-01
9.34084773e-01 -1.09586930e+00 2.10963503e-01 -3.38003427e-01
-3.94606292e-01 1.15275133e+00 -1.15071617e-01 1.02247372e-01
5.52413523e-01 1.92740619e-01 3.94036800e-01 8.66575316e-02
-2.21902002e-02 -1.25554204e-01 3.02638948e-01 6.51542902e-01
2.05433354e-01 -6.42944500e-02 -4.96347815e-01 2.14645490e-01
2.43064493e-01 4.69170988e-01 1.02939451e+00 7.43779540e-01
-7.08075941e-01 -5.10805488e-01 -1.64390647e+00 1.12735309e-01
-6.49092197e-01 -2.81127334e-01 4.23255414e-02 1.42313719e-01
1.57050967e-01 1.49521863e+00 -1.51273951e-01 -3.13099846e-02
7.77066126e-02 -3.88189554e-01 3.04353327e-01 -1.75907239e-01
-5.94695620e-02 1.12046294e-01 -2.12143183e-01 1.06730469e-01
-6.05630100e-01 -4.63256925e-01 -6.45410657e-01 -4.53817427e-01
-6.32869244e-01 3.83733332e-01 8.67536187e-01 8.76913488e-01
1.02939159e-01 2.50975311e-01 5.33766508e-01 -1.00161648e+00
-6.18424118e-01 -1.31133854e+00 -8.75670612e-01 5.16832590e-01
1.05258203e+00 -3.43476683e-01 -1.06207430e+00 6.83924079e-01] | [11.614234924316406, -2.7362046241760254] |
3d1215e6-c56f-444a-9405-5d54ec53b6ba | stochastic-partial-swap-enhanced-model | null | null | http://openaccess.thecvf.com//content/ICCV2021/html/Huang_Stochastic_Partial_Swap_Enhanced_Model_Generalization_and_Interpretability_for_Fine-Grained_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Huang_Stochastic_Partial_Swap_Enhanced_Model_Generalization_and_Interpretability_for_Fine-Grained_ICCV_2021_paper.pdf | Stochastic Partial Swap: Enhanced Model Generalization and Interpretability for Fine-Grained Recognition | Learning mid-level representation for fine-grained recognition is easily dominated by a limited number of highly discriminative patterns, degrading its robustness and generalization capability. To this end, we propose a novel Stochastic Partial Swap (SPS) scheme to address this issue. Our method performs element-wise swapping for partial features between samples to inject noise during training. It equips a regularization effect similar to Dropout, which promotes more neurons to represent the concepts. Furthermore, it also exhibits other advantages: 1) suppressing over-activation to some part patterns to improve feature representativeness, and 2) enriching pattern combination and simulating noisy cases to enhance classifier generalization. We verify the effectiveness of our approach through comprehensive experiments across four network backbones and three fine-grained datasets. Moreover, we demonstrate its ability to complement high-level representations, allowing a simple model to achieve performance comparable to the top-performing technologies in fine-grained recognition, indoor scene recognition, and material recognition while improving model interpretability. | ['DaCheng Tao', 'Xinchao Wang', 'Shaoli Huang'] | 2021-01-01 | null | null | null | iccv-2021-1 | ['material-recognition', 'scene-recognition'] | ['computer-vision', 'computer-vision'] | [ 3.90785754e-01 -2.89324373e-01 -2.24451900e-01 -6.30159557e-01
-4.33024228e-01 -6.17688358e-01 6.70003057e-01 -4.29045521e-02
-2.12008134e-01 7.29809701e-01 1.97844788e-01 -2.53123101e-02
-3.66031200e-01 -9.13973033e-01 -8.44503462e-01 -8.14420104e-01
6.95935860e-02 -9.02549457e-03 1.09502293e-01 2.80649848e-02
1.39668509e-01 1.01784909e+00 -1.90330887e+00 7.11874723e-01
1.06650519e+00 1.45296693e+00 3.20391238e-01 1.37191117e-01
-1.65028825e-01 7.29501367e-01 -6.67382836e-01 -1.36517361e-01
1.50602505e-01 1.10828057e-01 -4.83888865e-01 1.69775605e-01
8.11562955e-01 -4.27384414e-02 -4.50497270e-01 8.92155170e-01
2.84772962e-01 2.35061496e-01 7.12449789e-01 -1.04713571e+00
-7.88346708e-01 2.54111022e-01 -4.37013507e-01 2.54334230e-02
1.92647204e-01 2.19640136e-01 9.52827513e-01 -9.36615586e-01
1.21709861e-01 1.35669315e+00 7.98004866e-01 6.38829947e-01
-1.27963877e+00 -7.01449811e-01 6.96707904e-01 4.52697259e-06
-1.56922650e+00 -4.81920481e-01 7.06884921e-01 -2.13617250e-01
9.68686521e-01 5.14058411e-01 4.17383879e-01 1.39839208e+00
-1.70772061e-01 8.70622277e-01 1.20677328e+00 -1.28115669e-01
2.72179425e-01 -1.62218884e-02 4.10196751e-01 8.06220710e-01
3.40969622e-01 7.99243227e-02 -5.30439079e-01 -1.27433032e-01
7.78557003e-01 3.29216838e-01 -3.33857894e-01 -3.46296012e-01
-1.06651664e+00 4.31351483e-01 7.27766275e-01 4.80934709e-01
-3.32491904e-01 1.28738075e-01 1.03824422e-01 1.75197527e-01
1.56053692e-01 5.49517035e-01 -3.87387633e-01 4.10071388e-02
-9.31188226e-01 1.28454730e-01 5.66653192e-01 8.49626541e-01
9.14423287e-01 2.68026918e-01 -7.14475870e-01 1.15321386e+00
2.11808439e-02 4.18773890e-01 6.39743268e-01 -6.62882566e-01
5.11776209e-01 8.52970600e-01 -1.10131912e-01 -1.02060437e+00
-3.06617796e-01 -9.14596081e-01 -1.23112977e+00 4.94634081e-03
2.45918930e-01 3.27698678e-01 -1.14802372e+00 1.75092471e+00
-2.75034964e-01 3.00241321e-01 -2.35706255e-01 8.16469908e-01
7.55070806e-01 4.44155186e-01 1.22881524e-01 3.33670437e-01
1.28092468e+00 -1.12885261e+00 -2.00343817e-01 -3.44381928e-01
3.57315093e-01 -3.15421909e-01 1.47329319e+00 3.33852619e-01
-6.05568111e-01 -8.92274678e-01 -1.12760723e+00 1.17346622e-01
-6.86598539e-01 3.12800229e-01 8.00755620e-01 7.33499885e-01
-7.89396286e-01 8.21696639e-01 -6.66503012e-01 -1.35031953e-01
8.16431224e-01 4.82166380e-01 -4.58388686e-01 -4.18057352e-01
-8.72037053e-01 4.82499778e-01 2.30222628e-01 2.37391263e-01
-8.57615352e-01 -8.44091475e-01 -8.32594693e-01 4.46517080e-01
-5.72979674e-02 -7.31694341e-01 6.55725420e-01 -8.21457982e-01
-1.35238767e+00 4.95815367e-01 -3.20817947e-01 -2.72756904e-01
2.55775839e-01 -1.48385555e-01 -5.30023575e-01 -1.74720764e-01
-9.38569233e-02 7.05842614e-01 1.02822971e+00 -1.27759361e+00
-4.60055649e-01 -3.52750778e-01 4.43694964e-02 2.58351937e-02
-7.87866056e-01 -4.15319681e-01 -4.42523509e-01 -9.69146550e-01
1.61101416e-01 -6.43739998e-01 -3.72482002e-01 -8.27520266e-02
-4.58221406e-01 1.79325193e-02 7.97155440e-01 -1.81824729e-01
1.21054745e+00 -2.49307632e+00 -9.71796960e-02 3.66284966e-01
2.23570481e-01 3.05221677e-01 -6.28541768e-01 3.25658888e-01
4.52489816e-02 1.96922362e-01 -2.48530746e-01 -3.97964716e-01
5.76956272e-02 4.06799406e-01 -3.79215211e-01 2.70238966e-01
5.83675742e-01 1.04145861e+00 -7.41619885e-01 -1.03310481e-01
3.33455652e-01 5.89904428e-01 -5.46399415e-01 5.39997593e-02
-2.14122180e-02 2.36454710e-01 -6.83855593e-01 9.82442796e-01
7.22572327e-01 -4.41576153e-01 7.05784559e-02 -5.73207200e-01
1.03248574e-01 7.52641186e-02 -1.31438923e+00 1.49160016e+00
-7.67144859e-01 2.45084465e-01 -7.88735375e-02 -8.38747859e-01
1.18202639e+00 -1.75972983e-01 1.82913728e-02 -9.14290309e-01
-5.21526448e-02 1.18875355e-01 -2.36279353e-01 -1.85535163e-01
5.21414876e-01 1.67095944e-01 -2.52777368e-01 1.15736753e-01
1.35718763e-01 9.90655720e-02 2.57269312e-02 -1.42262518e-01
1.13916397e+00 2.19786484e-02 4.80165184e-02 -3.46747220e-01
5.52729785e-01 -4.83061939e-01 5.66200674e-01 1.24728155e+00
-3.13474238e-02 5.70104301e-01 2.58934814e-02 -4.78316128e-01
-6.72391832e-01 -1.11806691e+00 -1.44513488e-01 1.10706615e+00
2.14359805e-01 -3.96629870e-01 -4.81831729e-01 -8.29033256e-01
3.14669728e-01 6.01948500e-01 -7.24002361e-01 -4.96435255e-01
-4.50081110e-01 -7.46875942e-01 7.79762089e-01 8.04375589e-01
8.31153691e-01 -8.92912507e-01 -1.59167409e-01 -1.72911044e-02
-1.51964441e-01 -9.15460229e-01 -2.54702598e-01 6.35917664e-01
-8.34513426e-01 -8.12812030e-01 -6.16173327e-01 -7.40748823e-01
8.54530334e-01 4.17740941e-01 1.06146848e+00 3.40800919e-02
-2.66616613e-01 -1.23977093e-02 -2.41509184e-01 1.08403265e-01
3.79374027e-02 2.96641856e-01 1.45299032e-01 2.58146554e-01
2.55108625e-01 -7.10220337e-01 -5.22545934e-01 4.91492510e-01
-8.05216849e-01 -2.63713390e-01 8.63293469e-01 1.22306180e+00
6.09610736e-01 1.64067626e-01 5.34336448e-01 -6.89674199e-01
7.39797413e-01 -1.89452454e-01 -3.15900892e-01 4.36418086e-01
-4.53963637e-01 1.58659458e-01 1.19199288e+00 -5.02011657e-01
-1.00138426e+00 -1.01669177e-01 2.33184919e-02 -5.68771601e-01
-5.85552633e-01 2.24156469e-01 -2.73222148e-01 -4.79362935e-01
6.62439525e-01 4.46375042e-01 -3.22078168e-01 -8.95009220e-01
3.25912178e-01 6.11394703e-01 4.42923576e-01 -8.52339029e-01
8.87099028e-01 3.72190177e-01 -3.59210409e-02 -8.02837908e-01
-9.44145262e-01 -3.08794200e-01 -4.88411069e-01 1.43008158e-01
2.92236239e-01 -9.14953053e-01 -7.43174672e-01 5.24179757e-01
-7.36801147e-01 -2.13681683e-01 -5.17299891e-01 2.25553811e-01
-2.03681350e-01 2.10843042e-01 -6.19549990e-01 -6.72702551e-01
-5.44260070e-02 -9.91921008e-01 1.31658411e+00 4.02241319e-01
-1.18381403e-01 -7.32838809e-01 -4.42705542e-01 1.59051597e-01
6.02708876e-01 -4.23843376e-02 8.65780771e-01 -4.42431271e-01
-6.16931021e-01 -1.83640778e-01 -5.54053843e-01 4.61152494e-01
2.64720351e-01 -1.58287197e-01 -1.28016722e+00 -5.24659932e-01
-3.47342134e-01 -5.62735796e-01 1.42795122e+00 1.11334048e-01
1.88068914e+00 -1.62978739e-01 -5.46602607e-01 9.13072288e-01
1.22391176e+00 -1.53526902e-01 6.45017803e-01 3.10437649e-01
8.39217305e-01 5.67141056e-01 4.62857008e-01 3.41928333e-01
2.64555477e-02 6.58263087e-01 3.53126913e-01 -3.18268120e-01
-3.73024523e-01 -3.92375410e-01 7.83793479e-02 5.36042333e-01
9.30101424e-02 -3.31196368e-01 -6.27553165e-01 3.26009184e-01
-1.69710815e+00 -8.66225600e-01 2.53936201e-01 1.94906354e+00
4.89517063e-01 1.42552182e-01 -1.92381099e-01 1.61753237e-01
5.31884551e-01 3.89948964e-01 -5.48284411e-01 -2.62607306e-01
-5.10868847e-01 4.23722446e-01 4.17025447e-01 1.91325873e-01
-1.12503219e+00 9.82727945e-01 6.05048943e+00 1.31621349e+00
-1.22203350e+00 -2.56372094e-01 8.30054939e-01 -9.60419625e-02
-3.47253680e-01 -5.41576028e-01 -9.27084684e-01 5.27395427e-01
5.39092839e-01 5.01335621e-01 5.34470737e-01 8.21554780e-01
1.32154115e-02 2.66716897e-01 -1.06514502e+00 1.12531042e+00
2.38353126e-02 -1.58499110e+00 5.18747449e-01 -2.02628329e-01
8.06565702e-01 -6.93027601e-02 2.48052210e-01 3.47445905e-01
1.15324527e-01 -1.37695348e+00 7.44200587e-01 6.07516527e-01
1.00496292e+00 -5.89644611e-01 5.31754851e-01 2.34633520e-01
-1.36187220e+00 -4.72178429e-01 -4.70720440e-01 -1.25345558e-01
-2.43649751e-01 7.62542784e-01 -4.87061948e-01 6.75382853e-01
6.93971694e-01 6.42459393e-01 -8.82399738e-01 1.10059214e+00
-1.41758859e-01 5.15979588e-01 -4.71140414e-01 -4.21982035e-02
4.14636850e-01 5.80403060e-02 1.70865014e-01 1.33545589e+00
2.41196930e-01 -3.04627120e-01 2.43805796e-01 8.80563438e-01
-2.05083042e-01 -3.37347835e-01 -4.17849272e-01 1.07304364e-01
7.00924516e-01 1.24279714e+00 -6.58492684e-01 -1.90407485e-01
-3.12647730e-01 1.15658796e+00 5.71871519e-01 5.88337660e-01
-5.45897245e-01 -4.66901898e-01 8.02084029e-01 -4.50817309e-02
5.56007802e-01 -1.28369436e-01 -4.13551927e-01 -1.15602064e+00
3.21112096e-01 -9.82701719e-01 3.47284734e-01 -4.48145509e-01
-1.58716035e+00 8.39238822e-01 -3.56242865e-01 -1.09877896e+00
1.30953804e-01 -7.03896046e-01 -4.49591607e-01 9.69689727e-01
-1.69735348e+00 -1.36504209e+00 -6.42600715e-01 7.35978067e-01
3.11795533e-01 -3.18679586e-02 9.84291673e-01 5.90976179e-01
-7.37357259e-01 9.90683436e-01 7.28499889e-02 -4.61893789e-02
4.46362376e-01 -1.00352538e+00 4.86674607e-01 7.17384517e-01
8.73340964e-02 8.01869214e-01 2.24931374e-01 -4.49001402e-01
-1.30230618e+00 -1.44850004e+00 6.38936996e-01 -1.84765249e-01
4.19221878e-01 -7.44548023e-01 -8.64481747e-01 3.27413946e-01
-5.40233910e-01 3.43687207e-01 6.73317552e-01 3.63427520e-01
-8.24216783e-01 -5.32313585e-01 -1.25595844e+00 6.40789747e-01
1.34407365e+00 -6.02021396e-01 -3.88100356e-01 7.13005438e-02
5.32750547e-01 -4.51174155e-02 -7.26090848e-01 7.54489839e-01
6.41225755e-01 -9.45852757e-01 1.10799587e+00 -5.30321658e-01
2.35392720e-01 -4.11937416e-01 -5.18175900e-01 -1.27429783e+00
-8.53459954e-01 -2.72194982e-01 -3.14691901e-01 1.30584896e+00
3.72460753e-01 -7.72590399e-01 1.14182663e+00 1.90033451e-01
-2.90684342e-01 -9.66496885e-01 -9.91679132e-01 -1.14367533e+00
-1.74693495e-01 -3.09330583e-01 9.79528487e-01 8.31181109e-01
-5.36113977e-01 3.71915214e-02 -3.80182862e-01 2.84657925e-01
5.90309322e-01 4.82528985e-01 4.58243340e-01 -1.18360674e+00
-3.73210251e-01 -6.51024640e-01 -3.67737412e-01 -1.49869359e+00
1.10620536e-01 -7.63791680e-01 -9.20778215e-02 -1.22873592e+00
8.69512558e-02 -7.72783041e-01 -6.82110667e-01 7.08336651e-01
-4.21339273e-02 6.58159196e-01 1.11348666e-01 2.06793666e-01
-6.56765878e-01 6.55201256e-01 1.30823410e+00 -3.41754377e-01
-8.78280103e-02 -8.00709054e-02 -8.53140056e-01 4.78718519e-01
7.21748173e-01 -1.38107181e-01 -3.14052343e-01 -4.51788723e-01
-2.82764405e-01 -5.11443377e-01 6.22719705e-01 -1.23486412e+00
6.42214939e-02 -1.07673571e-01 9.21770036e-01 -3.14654589e-01
4.72093821e-01 -8.25101674e-01 7.06391037e-02 3.29055190e-01
-2.79350579e-01 -4.36509311e-01 4.20295119e-01 5.18178225e-01
-4.71032709e-01 -4.47032079e-02 7.85592020e-01 -1.40602320e-01
-1.00995290e+00 4.49024320e-01 -2.35358924e-01 -2.18749344e-01
7.40348041e-01 -6.30954385e-01 -5.37286401e-01 7.84794986e-02
-4.98910308e-01 1.09943658e-01 6.08928382e-01 5.58094859e-01
5.71367741e-01 -1.49088740e+00 -3.72799695e-01 7.15319514e-01
3.29803348e-01 -2.35725328e-01 4.65176761e-01 3.94301653e-01
-1.18125752e-01 5.98332524e-01 -2.11966708e-01 -5.73735714e-01
-8.65885794e-01 4.68060911e-01 4.23156738e-01 -3.32199484e-01
-5.06574333e-01 1.11806142e+00 3.68159801e-01 -6.46968544e-01
4.18602526e-01 -3.13910663e-01 -7.50388205e-02 -2.08383515e-01
5.73429883e-01 3.34014744e-01 2.96732962e-01 -2.91948020e-01
-6.74852848e-01 6.01389706e-01 -2.64517814e-01 6.36341631e-01
1.33953643e+00 -5.64143211e-02 1.50518402e-01 2.99029112e-01
1.24427736e+00 8.49756971e-02 -1.44117260e+00 -1.12150066e-01
-1.08205035e-01 -6.32307291e-01 -7.26908073e-03 -1.14987266e+00
-9.24148917e-01 6.80240929e-01 5.70374608e-01 2.03137308e-01
1.27822149e+00 -7.28352815e-02 5.69354355e-01 6.64890468e-01
4.48911667e-01 -7.83676326e-01 -3.93615253e-02 6.35171592e-01
7.98582017e-01 -9.76890504e-01 -2.36210018e-01 -5.88455558e-01
-1.43621400e-01 1.07817900e+00 7.36035466e-01 -1.30218938e-01
5.22555530e-01 4.18836415e-01 -1.33739114e-01 -1.24213077e-01
-5.71962655e-01 -8.66732150e-02 4.69994038e-01 7.82022417e-01
3.20969284e-01 1.79148749e-01 3.29537123e-01 7.50210643e-01
-6.82240576e-02 -1.16087206e-01 -1.17081873e-01 7.01150537e-01
-3.01111579e-01 -1.16676509e+00 -2.18537748e-01 7.91440904e-01
-1.44211620e-01 -3.00066352e-01 -3.28335673e-01 5.72428823e-01
2.49521792e-01 8.38756084e-01 1.87127173e-01 -5.82213998e-01
5.68851233e-01 -1.81527331e-01 5.35057783e-01 -5.11405766e-01
-7.76891887e-01 -4.16112751e-01 1.80823371e-01 -9.39380944e-01
-1.52504131e-01 -2.95943797e-01 -7.07826912e-01 -3.42765033e-01
-1.68833807e-01 7.34307393e-02 3.18539828e-01 8.32160056e-01
7.66814590e-01 8.34332108e-01 6.43954992e-01 -9.50508833e-01
-6.82090461e-01 -7.10826337e-01 -7.38431871e-01 4.90922332e-01
3.14252943e-01 -8.93777788e-01 -3.83435905e-01 -4.71250683e-01] | [9.606565475463867, 2.1203911304473877] |
717807e4-27ce-4f9d-aee3-e8c423316d97 | metaviewer-towards-a-unified-multi-view | 2303.06329 | null | https://arxiv.org/abs/2303.06329v1 | https://arxiv.org/pdf/2303.06329v1.pdf | MetaViewer: Towards A Unified Multi-View Representation | Existing multi-view representation learning methods typically follow a specific-to-uniform pipeline, extracting latent features from each view and then fusing or aligning them to obtain the unified object representation. However, the manually pre-specify fusion functions and view-private redundant information mixed in features potentially degrade the quality of the derived representation. To overcome them, we propose a novel bi-level-optimization-based multi-view learning framework, where the representation is learned in a uniform-to-specific manner. Specifically, we train a meta-learner, namely MetaViewer, to learn fusion and model the view-shared meta representation in outer-level optimization. Start with this meta representation, view-specific base-learners are then required to rapidly reconstruct the corresponding view in inner-level. MetaViewer eventually updates by observing reconstruction processes from uniform to specific over all views, and learns an optimal fusion scheme that separates and filters out view-private information. Extensive experimental results in downstream tasks such as classification and clustering demonstrate the effectiveness of our method. | ['Yilong Yin', 'Xiaoming Xi', 'Yuling Ma', 'Haoliang Sun', 'Ren Wang'] | 2023-03-11 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Wang_MetaViewer_Towards_a_Unified_Multi-View_Representation_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Wang_MetaViewer_Towards_a_Unified_Multi-View_Representation_CVPR_2023_paper.pdf | cvpr-2023-1 | ['multi-view-learning'] | ['computer-vision'] | [ 1.95131022e-02 -1.05403498e-01 -3.59780341e-01 -6.76322281e-01
-1.40069926e+00 -7.08798230e-01 6.27226830e-01 -1.05063289e-01
2.83271730e-01 1.73179716e-01 4.94351327e-01 3.11285496e-01
-7.12783486e-02 -5.88270009e-01 -6.94010079e-01 -8.14901829e-01
3.78025949e-01 6.62458599e-01 -9.69652086e-02 3.00759017e-01
4.21496890e-02 1.86649680e-01 -1.68197834e+00 8.85603666e-01
7.09224939e-01 8.21098447e-01 4.72732663e-01 6.83992207e-01
-4.35384475e-02 8.08111548e-01 -2.77653337e-01 -1.87172025e-01
3.28780472e-01 -3.20598811e-01 -7.09192753e-01 7.76306391e-01
4.77034718e-01 -2.76704013e-01 -5.74692413e-02 9.34223413e-01
2.82110155e-01 1.25101194e-01 6.23158753e-01 -9.99992251e-01
-3.34429264e-01 5.29278219e-01 -9.33790743e-01 -1.52788147e-01
2.64941812e-01 3.52694429e-02 1.16470480e+00 -1.24860060e+00
7.79634655e-01 1.34053195e+00 1.47231042e-01 3.13643098e-01
-1.44385827e+00 -4.17806506e-01 7.61836767e-01 -1.16402417e-01
-1.13744330e+00 -6.73394561e-01 8.97697508e-01 -6.01875603e-01
5.45976937e-01 1.39965072e-01 6.01373851e-01 1.04006290e+00
1.27650753e-01 1.00141752e+00 1.16317403e+00 9.74322483e-03
2.96988934e-02 1.47498757e-01 3.15367281e-02 9.20716345e-01
9.31118950e-02 -5.26490621e-02 -8.13954294e-01 -2.06169158e-01
5.69156706e-01 5.34640551e-01 -3.08467835e-01 -1.05169368e+00
-1.36995018e+00 7.43831396e-01 3.64584655e-01 4.50055636e-02
-5.75184941e-01 -2.05318406e-02 2.87922710e-01 3.24155837e-01
6.28343880e-01 1.57501608e-01 -6.55771375e-01 2.36477718e-01
-1.03296804e+00 2.01644823e-01 4.96008277e-01 1.01373136e+00
9.96350527e-01 -1.62133425e-01 -2.01783869e-02 7.78298497e-01
5.76891780e-01 1.71266869e-01 1.96483865e-01 -1.17971206e+00
6.91676319e-01 7.95602977e-01 -1.31659344e-01 -6.03881717e-01
-2.87486881e-01 -6.49383545e-01 -7.93654621e-01 3.88490081e-01
1.83721874e-02 3.58529426e-02 -1.00799513e+00 1.74088311e+00
7.25235641e-01 1.28791288e-01 1.32206410e-01 8.61759126e-01
9.79130983e-01 5.41406035e-01 -3.70929718e-01 -3.98037136e-01
1.40883994e+00 -1.40479720e+00 -4.74175215e-01 -1.14074238e-01
5.12624204e-01 -6.38799608e-01 6.73714340e-01 4.80999261e-01
-1.24997675e+00 -6.45102561e-01 -1.17341936e+00 -1.06390871e-01
-3.35544012e-02 2.20447063e-01 3.57436329e-01 1.58063501e-01
-7.92170584e-01 4.52474385e-01 -1.14442968e+00 -1.03078276e-01
5.68132162e-01 2.76893407e-01 -7.92621672e-01 -3.78897160e-01
-2.29874238e-01 6.27486169e-01 4.70901877e-01 -1.21138476e-01
-1.47971511e+00 -8.39894712e-01 -1.02519464e+00 8.01653191e-02
6.82558596e-01 -1.42331541e+00 1.04604149e+00 -8.79550159e-01
-1.30739343e+00 8.64132464e-01 -5.01452208e-01 1.33480117e-01
3.32638413e-01 -3.38730991e-01 -2.00601056e-01 1.27398893e-01
3.58792245e-01 3.33697498e-01 1.33400071e+00 -2.15182447e+00
-7.77083099e-01 -8.02179515e-01 2.81250209e-01 6.20394766e-01
3.45154889e-02 -2.92517394e-01 -9.80457127e-01 -4.85149443e-01
7.26433992e-01 -7.37271249e-01 -1.57943889e-01 -5.61129525e-02
-3.17732066e-01 8.93716365e-02 8.14772189e-01 -5.59737623e-01
1.04866087e+00 -2.25535226e+00 8.46912801e-01 1.05202794e-01
7.27324843e-01 -3.70589674e-01 -1.53080150e-01 3.74922842e-01
-2.82161355e-01 -4.01752979e-01 -2.27190077e-01 -1.05016017e+00
-3.42194885e-01 3.74778867e-01 -2.12044090e-01 6.95025861e-01
2.00860128e-02 6.39181793e-01 -1.01798129e+00 -4.92863148e-01
4.81095016e-01 3.82087499e-01 -8.62292528e-01 6.46495104e-01
-1.14461228e-01 8.45848978e-01 -5.40813684e-01 7.90336549e-01
8.12569261e-01 -6.58478796e-01 3.50434512e-01 -7.33651280e-01
1.12848066e-01 1.30254894e-01 -1.32528365e+00 2.31971169e+00
-8.16484988e-01 8.44336301e-02 5.98185003e-01 -1.04356492e+00
4.82410312e-01 2.19413340e-01 8.32339466e-01 1.03345197e-02
-6.69334903e-02 -1.70781776e-01 -3.83552343e-01 -4.88006055e-01
2.73029953e-01 -1.27353892e-01 -2.40613613e-02 7.10773408e-01
6.04603529e-01 -5.48827322e-03 -2.26591557e-01 4.09170270e-01
7.22544849e-01 5.63748837e-01 4.98061180e-01 7.78589547e-02
5.71248293e-01 -4.22327757e-01 5.41482270e-01 4.60825384e-01
2.60974050e-01 1.10872591e+00 2.37261072e-01 -2.95763791e-01
-6.59772456e-01 -1.23035264e+00 2.13071987e-01 1.31020975e+00
1.65425912e-01 -9.69900310e-01 -4.19893533e-01 -1.12314928e+00
-7.15399906e-02 5.48451781e-01 -6.98340595e-01 -9.87851620e-02
-3.73610288e-01 -3.25752139e-01 -3.78206164e-01 5.74180067e-01
8.36490989e-02 -4.82199967e-01 -4.14773166e-01 1.43597290e-01
-3.18906367e-01 -9.63415086e-01 -4.11059558e-01 4.18858796e-01
-1.05739701e+00 -1.12297261e+00 -3.41314673e-01 -1.99913904e-01
1.00905669e+00 8.16777050e-01 1.22387981e+00 2.64547765e-02
1.20253868e-01 6.85328603e-01 -2.18270063e-01 -6.56115413e-02
-2.18081042e-01 7.40091875e-02 -6.10584132e-02 4.26287174e-01
-2.07089052e-01 -8.32529128e-01 -5.13752401e-01 1.00674130e-01
-9.04073954e-01 3.17211330e-01 4.85493153e-01 1.01516235e+00
8.52442980e-01 -1.94505304e-01 2.70640731e-01 -1.15127802e+00
-1.24557875e-01 -5.72424829e-01 -6.52702093e-01 4.74473268e-01
-5.04601061e-01 2.86255926e-01 4.79540676e-01 1.37249697e-02
-1.19449759e+00 4.17090207e-01 2.87620693e-01 -1.19386363e+00
-1.08322546e-01 5.90103030e-01 -7.51841247e-01 3.80945235e-01
2.11718202e-01 3.46398026e-01 -8.54364317e-03 -6.07547998e-01
8.21908832e-01 1.65847197e-01 5.56966960e-01 -6.19495332e-01
9.40862477e-01 8.89079809e-01 -1.23251766e-01 -4.12417889e-01
-1.31839645e+00 -6.12852454e-01 -8.79496813e-01 -2.07732126e-01
1.05452776e+00 -1.61123276e+00 -4.47331756e-01 2.45892838e-01
-9.63670433e-01 1.88384861e-01 -4.21998948e-01 5.95762253e-01
-7.59193480e-01 2.19715253e-01 -1.55360401e-01 -5.54295719e-01
-1.20745242e-01 -1.39098191e+00 1.63520420e+00 -3.05236038e-02
7.87558034e-02 -1.03344917e+00 1.31578460e-01 8.05904269e-01
-1.61584273e-01 1.26398027e-01 6.91074908e-01 -4.93048310e-01
-9.42523479e-01 1.30635411e-01 -8.50237012e-02 3.23757499e-01
3.83604348e-01 -1.47004142e-01 -1.33264768e+00 -6.43260121e-01
3.26520175e-01 -3.79140258e-01 1.12863791e+00 3.64332348e-01
1.37936270e+00 -1.13584682e-01 -5.00071049e-01 1.23528767e+00
1.44668400e+00 -3.79228115e-01 -3.34039815e-02 5.51182888e-02
1.00948858e+00 5.84046662e-01 6.02607071e-01 6.20674372e-01
8.33824515e-01 5.41643679e-01 7.55216897e-01 2.89719384e-02
-3.15848775e-02 -2.00855017e-01 4.41291213e-01 1.09558010e+00
-1.23942547e-01 -8.71882662e-02 -3.08660150e-01 4.26728010e-01
-1.96227765e+00 -1.03198051e+00 3.01098168e-01 2.13491178e+00
3.92877787e-01 -1.02065720e-01 1.03052631e-01 -3.23077977e-01
4.68731195e-01 7.28319526e-01 -5.91837168e-01 2.59323388e-01
8.39631483e-02 -3.58150482e-01 1.11886553e-01 4.46000457e-01
-1.17005336e+00 8.26701283e-01 5.31604099e+00 3.76055539e-01
-1.09357071e+00 3.71877640e-01 3.16286564e-01 -6.29575610e-01
-6.67239070e-01 3.62532198e-01 -5.54323077e-01 1.02345720e-01
4.29038554e-01 9.05305222e-02 4.97088283e-01 7.66779661e-01
-1.00457512e-01 1.36340747e-03 -1.56627285e+00 1.37164855e+00
3.94872189e-01 -1.45870888e+00 2.69396305e-01 1.87283084e-01
9.38439965e-01 2.05650732e-01 -7.19091147e-02 2.91185975e-01
5.72222531e-01 -7.04720914e-01 8.15892160e-01 6.80923462e-01
5.43156564e-01 -7.94555008e-01 2.55452394e-01 4.46612835e-01
-1.53307593e+00 -1.04452208e-01 -3.02142084e-01 4.89207119e-01
4.42122310e-01 4.73646104e-01 -4.19420213e-01 1.43532133e+00
5.58225632e-01 1.25395334e+00 -5.29500127e-01 4.99069005e-01
-6.82378747e-03 2.36035958e-01 -9.09849182e-02 1.03154778e+00
1.32316677e-02 -3.53253096e-01 6.70789182e-01 7.32462347e-01
3.16155374e-01 8.13497975e-02 6.66106164e-01 6.57620907e-01
-4.85567562e-02 -3.85186613e-01 -6.62554622e-01 2.15240702e-01
3.30902457e-01 1.69187880e+00 -4.95427638e-01 -5.65890312e-01
-8.41633916e-01 1.14143527e+00 6.92850769e-01 3.53666693e-01
-7.98657060e-01 5.12482107e-01 7.70414710e-01 1.44604966e-02
7.36080945e-01 -7.86484871e-03 -2.53060818e-01 -1.85027730e+00
1.43934920e-01 -1.10082233e+00 7.63552845e-01 -6.62185252e-01
-1.31238914e+00 4.66676176e-01 1.85443476e-01 -1.65896690e+00
-2.94632077e-01 -2.16888756e-01 -4.17723805e-01 7.96778262e-01
-1.26340377e+00 -1.74477661e+00 -4.00135070e-01 7.81694114e-01
9.29306805e-01 -3.54926765e-01 7.08224654e-01 7.56537542e-02
-5.52448928e-01 3.76792312e-01 2.87694316e-02 -2.66923070e-01
7.56826639e-01 -1.29339027e+00 -4.31254581e-02 6.22642875e-01
3.49319249e-01 7.27047861e-01 3.17181617e-01 -4.29554492e-01
-1.76335454e+00 -1.23556495e+00 1.38924673e-01 -8.08617234e-01
3.17683280e-01 -4.18875784e-01 -8.60443354e-01 9.92637694e-01
2.73513108e-01 4.11526442e-01 8.17357302e-01 3.36862683e-01
-7.09847271e-01 -3.60328794e-01 -8.87715816e-01 3.73703331e-01
1.06414902e+00 -8.24237049e-01 -6.55169249e-01 2.57651180e-01
6.12529814e-01 -4.96043533e-01 -1.07015717e+00 4.15587276e-01
6.26437068e-01 -1.25237107e+00 1.09810269e+00 -7.27574944e-01
5.48438668e-01 -6.26015902e-01 -5.81678152e-01 -1.47836590e+00
-5.34758270e-01 -4.08611804e-01 -5.87229490e-01 1.39728427e+00
1.76847607e-01 -3.70936364e-01 5.93699634e-01 1.02205224e-01
-3.94018203e-01 -8.96740675e-01 -7.16269553e-01 -3.37631732e-01
-3.36345375e-01 -2.45748505e-01 5.65769613e-01 9.83183801e-01
-4.43786174e-01 7.12229848e-01 -4.05381769e-01 7.02417791e-01
9.57668543e-01 9.14457619e-01 1.19408965e+00 -1.15856946e+00
-7.24668980e-01 -1.28297031e-01 -9.32136923e-02 -1.13224304e+00
2.96458989e-01 -1.10776687e+00 -1.80555284e-01 -1.63645685e+00
6.71506763e-01 -7.58018941e-02 -4.62255895e-01 1.66815400e-01
-5.23096442e-01 -2.85323441e-01 3.47099006e-01 3.85219693e-01
-9.12589967e-01 7.30604410e-01 1.36351466e+00 -1.72232449e-01
-5.88211343e-02 1.03386924e-01 -9.16627884e-01 8.95550668e-01
5.80311194e-02 -5.59904575e-01 -7.23355412e-01 -5.56012630e-01
4.65461090e-02 3.30425590e-01 3.35993290e-01 -7.80493021e-01
8.17106515e-02 -2.07838073e-01 7.65341043e-01 -1.09373784e+00
7.12930202e-01 -1.08408380e+00 3.27380955e-01 -2.95717530e-02
-1.52301013e-01 1.11800075e-01 -3.06508720e-01 7.88034737e-01
-3.53871882e-01 3.12201045e-02 7.06086934e-01 -4.14701164e-01
-4.14967120e-01 6.01115048e-01 1.38728663e-01 -1.19351648e-01
1.03463471e+00 -1.80999309e-01 -8.36525485e-02 -4.68025535e-01
-1.07949555e+00 4.31771278e-01 7.41510749e-01 4.67724532e-01
5.49668610e-01 -1.31252301e+00 -6.54663742e-01 4.57244158e-01
4.69506681e-01 5.02190232e-01 5.70449889e-01 8.84812057e-01
1.43481508e-01 -1.22353420e-01 1.87289566e-02 -1.13487458e+00
-1.24660552e+00 8.87471318e-01 1.57860935e-01 -5.33236325e-01
-7.22261429e-01 8.27403069e-01 9.00516927e-01 -6.82561994e-01
-7.05764955e-03 -2.15509281e-01 -1.48177192e-01 3.99572819e-01
4.63429630e-01 2.88087040e-01 5.71843237e-03 -7.46780813e-01
-3.41080576e-01 7.09351182e-01 -3.68914783e-01 -3.00948828e-01
1.68164706e+00 -5.44377148e-01 -2.49072999e-01 8.41366291e-01
1.31107986e+00 1.36313781e-01 -1.62029779e+00 -3.17227364e-01
-3.65628481e-01 -6.34833932e-01 2.14584097e-02 -3.85527253e-01
-1.53868926e+00 7.42886305e-01 3.11637253e-01 -1.87045991e-01
1.21587884e+00 4.92902339e-01 3.46937954e-01 2.55376697e-01
3.44219565e-01 -7.51448154e-01 1.87159881e-01 2.63021380e-01
1.05486619e+00 -1.34851325e+00 5.37703335e-01 -4.21281904e-01
-7.15357184e-01 8.58149350e-01 8.83576751e-01 1.98658966e-02
8.75547886e-01 1.86485112e-01 1.42253533e-01 -5.97473800e-01
-1.35783863e+00 -6.79552332e-02 3.92227471e-01 4.34510618e-01
2.39064768e-01 -1.07346170e-01 4.13975120e-01 4.58785146e-01
4.66474220e-02 -3.44920993e-01 1.01383246e-01 1.08561265e+00
-6.66241348e-02 -1.03533840e+00 -3.27901483e-01 5.19004703e-01
-1.96349278e-01 2.10797161e-01 -1.38523608e-01 3.97404492e-01
1.36718988e-01 7.38485336e-01 -8.30971636e-03 -4.43445414e-01
1.66036189e-01 6.90902099e-02 6.58830702e-01 -1.00952923e+00
-7.03864932e-01 5.43509781e-01 1.24767590e-02 -9.27458107e-01
-7.28886366e-01 -9.05073404e-01 -9.20774341e-01 1.47459060e-01
-4.70378846e-01 -6.64888620e-02 4.08718109e-01 1.07427275e+00
6.24662697e-01 7.26790786e-01 9.48681653e-01 -1.13308787e+00
-6.33034408e-01 -7.72765040e-01 -4.88753557e-01 6.18882537e-01
6.16683543e-01 -8.78842533e-01 -1.77458555e-01 2.38606155e-01] | [8.553354263305664, 4.463045597076416] |
988ab1f3-340e-4661-8d5a-52d89571ba41 | ceres-pretraining-of-graph-conditioned-1 | 2204.04303 | null | https://arxiv.org/abs/2204.04303v1 | https://arxiv.org/pdf/2204.04303v1.pdf | CERES: Pretraining of Graph-Conditioned Transformer for Semi-Structured Session Data | User sessions empower many search and recommendation tasks on a daily basis. Such session data are semi-structured, which encode heterogeneous relations between queries and products, and each item is described by the unstructured text. Despite recent advances in self-supervised learning for text or graphs, there lack of self-supervised learning models that can effectively capture both intra-item semantics and inter-item interactions for semi-structured sessions. To fill this gap, we propose CERES, a graph-based transformer model for semi-structured session data. CERES learns representations that capture both inter- and intra-item semantics with (1) a graph-conditioned masked language pretraining task that jointly learns from item text and item-item relations; and (2) a graph-conditioned transformer architecture that propagates inter-item contexts to item-level representations. We pretrained CERES using ~468 million Amazon sessions and find that CERES outperforms strong pretraining baselines by up to 9% in three session search and entity linking tasks. | ['Chao Zhang', 'Tuo Zhao', 'Bing Yin', 'Qingyu Yin', 'Chen Luo', 'Rui Feng'] | 2022-04-08 | null | https://aclanthology.org/2022.naacl-main.16 | https://aclanthology.org/2022.naacl-main.16.pdf | naacl-2022-7 | ['session-search'] | ['natural-language-processing'] | [ 3.04909348e-01 3.27694625e-01 -7.53578126e-01 -7.10115910e-01
-4.59831536e-01 -8.35481822e-01 7.44864941e-01 5.64811766e-01
-1.09553300e-01 3.10212702e-01 5.94364405e-01 -3.68182153e-01
-2.14347348e-01 -8.27219784e-01 -8.98318052e-01 1.35080501e-01
-3.06832850e-01 9.83140171e-01 1.88862726e-01 -5.30466020e-01
-3.26455444e-01 -1.13128945e-01 -1.11349821e+00 7.91728616e-01
7.43872225e-01 1.00758672e+00 4.56410386e-02 5.92264652e-01
-6.42302275e-01 8.93034399e-01 -1.67527407e-01 -6.97428167e-01
1.01687955e-02 -6.35485411e-01 -9.24722731e-01 2.49643117e-01
6.31822705e-01 2.97414921e-02 -5.41550636e-01 6.30546749e-01
5.49515225e-02 2.90652603e-01 7.14317679e-01 -1.21341646e+00
-1.31085718e+00 1.25536335e+00 -5.03138542e-01 2.13482618e-01
6.03237212e-01 -5.50125614e-02 1.87625074e+00 -6.22270405e-01
8.56887281e-01 1.06253076e+00 5.97072482e-01 3.70905817e-01
-1.60058784e+00 -5.10810196e-01 4.59838092e-01 -1.54689059e-01
-1.09865344e+00 -1.61169663e-01 6.60416484e-01 -2.52687365e-01
1.61898208e+00 8.50742683e-02 7.01995552e-01 1.49634004e+00
8.32222402e-02 8.47842813e-01 6.74822628e-01 -1.84867293e-01
-2.79118001e-01 3.60251784e-01 7.19104230e-01 9.54027891e-01
4.72972244e-01 -7.36889020e-02 -9.95331466e-01 -2.07529187e-01
8.12627435e-01 1.56504720e-01 2.45808035e-01 -6.90633357e-01
-8.36165309e-01 9.22773182e-01 8.07561457e-01 4.27788377e-01
-2.55582541e-01 2.16179103e-01 4.67441112e-01 7.07754672e-01
5.94398022e-01 6.31264627e-01 -6.46016598e-01 3.85607094e-01
-7.61545360e-01 1.08601682e-01 1.22009826e+00 1.49783754e+00
7.88453817e-01 -4.20497730e-02 -1.49650425e-01 8.68772566e-01
6.30862057e-01 4.65170622e-01 3.77562582e-01 -1.98926494e-01
7.11746991e-01 8.90678227e-01 -2.30611756e-01 -9.13592994e-01
-4.58653659e-01 -7.62649775e-01 -6.13067269e-01 -8.85505855e-01
1.52679920e-01 2.16753736e-01 -9.86123204e-01 1.79700601e+00
-2.39351038e-02 9.34526771e-02 -1.91553026e-01 6.11484468e-01
1.52878165e+00 3.31505686e-01 5.93002379e-01 5.02375420e-03
1.45191681e+00 -1.31878185e+00 -8.37878942e-01 -6.16369128e-01
8.64901245e-01 -4.83494848e-01 1.27121031e+00 -2.66633600e-01
-1.14523375e+00 -5.23354113e-01 -8.62379849e-01 -4.41633314e-01
-8.70523572e-01 -1.74740046e-01 1.11483359e+00 1.77813366e-01
-1.09955037e+00 3.30525666e-01 -6.36526644e-01 -7.52603412e-01
3.47912580e-01 5.46344697e-01 -3.61995608e-01 6.85982108e-02
-1.39044201e+00 5.08094847e-01 1.65261388e-01 -2.85347342e-01
-5.35895646e-01 -1.12734127e+00 -1.26470256e+00 3.84738237e-01
4.86188859e-01 -1.07302678e+00 1.19828260e+00 -8.07295799e-01
-1.04498875e+00 1.14629376e+00 -3.16570014e-01 -6.76155865e-01
-2.31780663e-01 -2.04259679e-01 -8.65912020e-01 -2.68738955e-01
1.21034041e-01 2.38968357e-01 6.16917133e-01 -1.10745192e+00
-2.06436992e-01 -5.56299448e-01 1.90928474e-01 5.36779225e-01
-4.41642761e-01 -9.37149003e-02 -9.06736791e-01 -6.34839833e-01
-2.30521727e-02 -9.27724183e-01 -3.51766646e-02 -3.86284560e-01
-6.34355128e-01 -6.31147504e-01 3.82516444e-01 -4.16829288e-01
1.40527391e+00 -1.94047356e+00 2.20243841e-01 3.08181167e-01
4.47211385e-01 -2.84648649e-02 -5.05964220e-01 9.04135466e-01
-7.50468671e-02 3.83294597e-02 3.58698636e-01 -8.35413218e-01
3.03698570e-01 2.50985354e-01 -3.53579283e-01 -5.97759262e-02
-8.78720172e-03 1.63759577e+00 -9.85890806e-01 -3.57166588e-01
-6.72604740e-02 2.90824711e-01 -4.94550973e-01 4.95147169e-01
-5.54720640e-01 2.59504877e-02 -4.81733590e-01 7.33161569e-01
1.88298061e-01 -1.14240110e+00 6.01680994e-01 -4.51614529e-01
6.82519376e-01 9.31881309e-01 -7.80364394e-01 2.24025226e+00
-6.05976701e-01 1.74374893e-01 -3.93345058e-02 -6.84966862e-01
4.71372128e-01 1.22403152e-01 6.10098600e-01 -1.11496842e+00
-2.41233528e-01 -1.48537718e-02 -4.92508680e-01 -3.43751192e-01
8.63023221e-01 2.28100330e-01 -4.90536869e-01 9.57548797e-01
7.65160263e-01 6.61076754e-02 2.36082390e-01 1.16989446e+00
1.21125627e+00 7.06160665e-02 1.10957466e-01 -9.47959721e-02
1.65475890e-01 -1.36647835e-01 -5.78030944e-02 8.92321408e-01
3.89681548e-01 1.81502551e-01 2.95156598e-01 -2.12236539e-01
-6.53977811e-01 -1.25090146e+00 4.53214735e-01 1.92107618e+00
2.87871361e-01 -8.86324644e-01 -5.90508394e-02 -1.12517428e+00
6.09834731e-01 5.93865454e-01 -8.08847547e-01 -3.52568835e-01
-1.91034734e-01 -1.71267956e-01 1.73964590e-01 8.09962094e-01
1.14166029e-01 -1.15490305e+00 5.21908343e-01 2.07239673e-01
-4.28007990e-02 -1.32587385e+00 -1.18750966e+00 3.61585170e-01
-7.51016200e-01 -1.10936022e+00 -1.84643090e-01 -9.89398956e-01
5.92797160e-01 5.34403563e-01 2.06714487e+00 1.68304741e-01
-2.00434417e-01 7.82474041e-01 -2.09404781e-01 -3.05415113e-02
-1.14844620e-01 5.07568121e-01 -1.05141528e-01 -2.31475681e-01
8.48659754e-01 -7.13495970e-01 -7.32145846e-01 3.07960421e-01
-6.44385338e-01 4.50560637e-02 5.20822585e-01 6.05134487e-01
5.02155781e-01 -5.34491539e-01 5.74272335e-01 -1.95621836e+00
8.82019222e-01 -9.88303542e-01 -3.28199476e-01 7.23249257e-01
-1.05794358e+00 1.01139292e-01 2.57274479e-01 -4.02583957e-01
-9.66580093e-01 -7.71916881e-02 2.57208228e-01 -1.43618166e-01
5.54940552e-02 7.97991574e-01 1.51703954e-01 7.61118755e-02
8.39042902e-01 1.62097842e-01 -2.59603947e-01 -5.68129063e-01
9.60219562e-01 4.40738827e-01 2.49221310e-01 -3.94051969e-01
7.84006596e-01 1.45810083e-01 -3.49927276e-01 -3.97957087e-01
-1.49231446e+00 -1.05342007e+00 -4.05509651e-01 2.64497340e-01
7.71939576e-01 -1.26596725e+00 -7.18052983e-01 -8.30783136e-03
-6.38573885e-01 -6.69758379e-01 -6.60117567e-01 1.40373707e-01
-3.20717245e-01 1.88639507e-01 -1.03302109e+00 -3.49699736e-01
-6.19076550e-01 -3.72288436e-01 1.26709509e+00 1.52052298e-01
-3.95084351e-01 -1.61797380e+00 1.47147477e-01 6.71590269e-01
6.19773209e-01 -3.13192546e-01 8.63007069e-01 -1.08547056e+00
-4.38094109e-01 -1.06397152e-01 -4.23193842e-01 -1.05921201e-01
3.76657456e-01 -6.87055171e-01 -5.16412497e-01 -3.94101083e-01
-6.15606368e-01 -8.08204353e-01 9.96136248e-01 -2.89530475e-02
1.01376963e+00 -3.43189389e-01 -6.09912574e-01 5.85559249e-01
1.20236778e+00 -4.21440870e-01 7.16593713e-02 -2.03006476e-01
1.02626753e+00 3.22599143e-01 1.88540906e-01 5.99353351e-02
9.36667621e-01 6.91866696e-01 1.45080000e-01 -1.52362958e-01
-5.18007517e-01 -1.03304172e+00 1.73844635e-01 8.78642499e-01
3.57481629e-01 -5.69448292e-01 -3.25748712e-01 6.18044078e-01
-2.03235483e+00 -9.04361784e-01 -1.50779426e-01 1.91416156e+00
1.09407341e+00 1.00483827e-01 3.58440161e-01 -7.91639805e-01
1.74019098e-01 4.36778128e-01 -9.01309907e-01 -2.37152770e-01
-2.70138770e-01 2.76594281e-01 5.18767238e-01 4.64087963e-01
-1.04490232e+00 1.19986451e+00 6.42852497e+00 4.41066206e-01
-6.11798286e-01 8.65323767e-02 2.44904449e-03 9.08858888e-03
-8.97872746e-01 -1.79451212e-01 -9.22968388e-01 3.23709011e-01
1.01839268e+00 -3.56052727e-01 6.08536005e-01 7.47205079e-01
-3.59963745e-01 4.85756397e-01 -1.47874236e+00 9.25602019e-01
3.67369771e-01 -1.44243693e+00 2.66075820e-01 1.86777607e-01
9.44731414e-01 1.86192289e-01 2.59013146e-01 8.64052057e-01
1.05204439e+00 -1.09086776e+00 1.22653998e-01 3.06135714e-01
7.77466893e-01 -5.31119704e-02 2.96897620e-01 7.16180131e-02
-1.64040124e+00 1.25349119e-01 -5.16047850e-02 4.32047814e-01
1.89603463e-01 2.10710868e-01 -9.35333490e-01 8.12496424e-01
5.09638250e-01 1.34021926e+00 -8.72003973e-01 6.18423522e-01
-3.54286790e-01 7.98525035e-01 -2.95970321e-01 -3.01423464e-02
3.26394349e-01 -2.97564775e-01 1.48171261e-01 1.40554440e+00
-1.12050667e-01 -4.50737821e-03 3.25929880e-01 8.64331663e-01
-6.90397501e-01 2.18084320e-01 -8.76716793e-01 -8.19489300e-01
4.27283704e-01 1.31226993e+00 -4.75422144e-01 -3.72099668e-01
-8.52840066e-01 1.26765323e+00 6.94380879e-01 7.64040828e-01
-5.35567462e-01 -3.21819112e-02 6.38475478e-01 4.88909215e-01
4.32983637e-01 -2.61045784e-01 -1.21733665e-01 -1.48424804e+00
-3.17449898e-01 -7.69459426e-01 1.12047756e+00 -7.30277121e-01
-1.99336994e+00 5.64863384e-01 -2.18431935e-01 -6.37361348e-01
-5.62989652e-01 -3.32951427e-01 -4.10727322e-01 8.27480137e-01
-1.43951929e+00 -1.74798167e+00 -1.46784320e-01 1.05079067e+00
6.67763889e-01 -3.14924717e-01 1.09450376e+00 2.61877477e-01
-1.04797043e-01 9.47441399e-01 -2.29412869e-01 2.27842778e-01
7.65702784e-01 -1.54459417e+00 6.86979234e-01 1.59470633e-01
9.96294796e-01 1.13543510e+00 4.57462162e-01 -8.37273717e-01
-1.68329692e+00 -1.13062179e+00 1.34345067e+00 -8.80060852e-01
1.11249840e+00 -8.71797383e-01 -9.41721141e-01 1.44496560e+00
5.42588949e-01 2.23972738e-01 1.23233163e+00 1.24338579e+00
-9.97893035e-01 1.30564151e-02 -6.43361986e-01 3.67417663e-01
1.74196315e+00 -1.24363112e+00 -7.54270375e-01 9.38350737e-01
1.13041031e+00 -4.39621121e-01 -1.09342611e+00 1.13398343e-01
2.77311772e-01 -4.35606033e-01 1.09133375e+00 -1.35980642e+00
1.41486630e-01 2.99460500e-01 -2.20167101e-03 -1.27361131e+00
-5.57851732e-01 -7.22106278e-01 -8.21210504e-01 1.12255752e+00
9.19234455e-01 -4.11770999e-01 1.03606856e+00 6.14402831e-01
-4.08054665e-02 -5.83770931e-01 -1.25087321e-01 -6.83980823e-01
-3.34039778e-01 -3.00842263e-02 5.31124592e-01 1.09608150e+00
3.07219535e-01 1.21985447e+00 -3.31712157e-01 -4.90656868e-02
7.26861894e-01 6.50129974e-01 6.00894988e-01 -1.28562176e+00
-7.13927448e-01 -1.47970259e-01 -7.35288672e-03 -1.43766320e+00
4.71511692e-01 -1.28217089e+00 -3.95948082e-01 -2.04524565e+00
5.43155909e-01 -4.36240017e-01 -6.24076247e-01 6.15148067e-01
-3.37820537e-02 2.04731271e-01 -2.62000769e-01 1.25408635e-01
-1.48674965e+00 3.96852821e-01 1.02597106e+00 -1.93117589e-01
-2.89688528e-01 1.10423371e-01 -1.03586912e+00 1.48761481e-01
2.82422423e-01 -2.13090077e-01 -1.05900240e+00 -7.96605408e-01
6.98190749e-01 1.63399875e-02 -6.04078174e-03 -1.33294955e-01
6.07928157e-01 1.03694305e-01 1.64702997e-01 -4.51242179e-01
4.37768459e-01 -8.08317721e-01 2.34218761e-02 -1.05124533e-01
-9.95900273e-01 -1.90837517e-01 8.65985826e-02 1.15617418e+00
-1.33825809e-01 4.34609205e-01 -1.07713416e-01 -2.13386953e-01
-7.27128208e-01 7.19857156e-01 2.03278497e-01 3.22480530e-01
4.78398293e-01 7.19287097e-02 -4.58120525e-01 -6.86029196e-01
-1.18855512e+00 5.57110429e-01 6.21189401e-02 9.50567603e-01
2.26464212e-01 -1.39194608e+00 -3.21605295e-01 2.04418838e-01
5.26386321e-01 -3.36115330e-01 2.71724313e-01 2.83697724e-01
2.99584568e-01 7.59845793e-01 2.45279849e-01 -4.34484005e-01
-1.32070279e+00 7.35185504e-01 3.07333916e-02 -8.18533957e-01
-3.79954606e-01 1.29477465e+00 2.25542709e-01 -6.36532187e-01
4.44591522e-01 -9.84948948e-02 -3.35786849e-01 1.01936590e-02
1.49196818e-01 -3.29166561e-01 4.48157564e-02 -4.67786878e-01
-2.65873224e-01 2.67265737e-01 -5.52228510e-01 1.88041344e-01
1.45271599e+00 -3.35645527e-01 -3.75310332e-02 6.52418613e-01
1.38074076e+00 -6.14755675e-02 -8.17911685e-01 -1.07242155e+00
2.70276934e-01 -2.29399011e-01 -8.27950835e-02 -1.04202855e+00
-1.02083969e+00 3.59984964e-01 -1.60893742e-02 6.27750337e-01
7.38645077e-01 6.61630869e-01 9.49532926e-01 5.48171520e-01
4.83454973e-01 -7.91010261e-01 5.16625702e-01 4.87882316e-01
8.62710118e-01 -1.44182050e+00 1.24416091e-01 -7.69180477e-01
-9.26804066e-01 5.17294943e-01 8.03544343e-01 -1.36295289e-01
1.07570577e+00 -7.10782930e-02 -3.15602690e-01 -8.69819164e-01
-1.14119935e+00 -3.87575328e-01 9.50273871e-01 4.04443622e-01
9.32397604e-01 -8.99298023e-03 -3.25058997e-02 8.34261298e-01
-2.32932717e-02 -9.89135206e-02 -3.15753192e-01 6.23338938e-01
1.88968368e-02 -1.35408831e+00 8.26363385e-01 7.44769096e-01
-7.04246238e-02 -4.59615320e-01 -7.19645619e-01 5.11966467e-01
-3.36382985e-01 9.52557504e-01 1.46146566e-01 -4.54275548e-01
5.12558937e-01 2.59912252e-01 4.11435753e-01 -1.31868565e+00
-1.10157299e+00 -1.71246812e-01 4.48818058e-01 -9.84665990e-01
-2.02451929e-01 -4.28919524e-01 -1.07513797e+00 -2.22853377e-01
-3.04119468e-01 4.82244194e-01 5.04184425e-01 7.63771355e-01
8.72607052e-01 6.62052453e-01 4.08061147e-01 -1.87069878e-01
-5.89767218e-01 -9.86817241e-01 -8.97099972e-01 9.30153489e-01
-4.58405241e-02 -3.31178576e-01 -2.18698055e-01 -6.72149435e-02] | [10.92813777923584, 7.354452133178711] |
09ad7d57-afdc-4ec8-acba-d9e883199853 | tinyml-design-contest-for-life-threatening | 2305.05105 | null | https://arxiv.org/abs/2305.05105v2 | https://arxiv.org/pdf/2305.05105v2.pdf | TinyML Design Contest for Life-Threatening Ventricular Arrhythmia Detection | The first ACM/IEEE TinyML Design Contest (TDC) held at the 41st International Conference on Computer-Aided Design (ICCAD) in 2022 is a challenging, multi-month, research and development competition. TDC'22 focuses on real-world medical problems that require the innovation and implementation of artificial intelligence/machine learning (AI/ML) algorithms on implantable devices. The challenge problem of TDC'22 is to develop a novel AI/ML-based real-time detection algorithm for life-threatening ventricular arrhythmia over low-power microcontrollers utilized in Implantable Cardioverter-Defibrillators (ICDs). The dataset contains more than 38,000 5-second intracardiac electrograms (IEGMs) segments over 8 different types of rhythm from 90 subjects. The dedicated hardware platform is NUCLEO-L432KC manufactured by STMicroelectronics. TDC'22, which is open to multi-person teams world-wide, attracted more than 150 teams from over 50 organizations. This paper first presents the medical problem, dataset, and evaluation procedure in detail. It further demonstrates and discusses the designs developed by the leading teams as well as representative results. This paper concludes with the direction of improvement for the future TinyML design for health monitoring applications. | ['Yiyu Shi', 'Lichuan Ping', 'Xiaowei Xu', 'Liqi Liao', 'Cong Liu', 'Dawei Li', 'Zhenge Jia'] | 2023-05-09 | null | null | null | null | ['arrhythmia-detection'] | ['medical'] | [ 1.65909857e-01 5.68239614e-02 4.80777360e-02 1.33038372e-01
-7.36491323e-01 -5.20821571e-01 -2.52384692e-01 1.63099289e-01
9.52034593e-02 8.34180892e-01 -8.90820548e-02 -7.68622816e-01
-2.66668081e-01 -1.74176663e-01 -2.37890452e-01 -3.96192312e-01
-5.02308965e-01 5.02261341e-01 -5.76318920e-01 4.08665717e-01
-1.33224607e-01 5.92764795e-01 -1.00992858e+00 5.28893471e-01
6.46046817e-01 1.24805868e+00 1.35695999e-02 9.25894856e-01
7.91948140e-01 7.25129366e-01 -1.12712312e+00 2.51472294e-01
5.40236495e-02 -7.48482764e-01 -6.45805717e-01 -5.13741076e-01
-1.42585710e-01 2.98152976e-02 3.34786326e-01 4.15762067e-01
1.42724121e+00 -1.10967910e+00 6.57808483e-01 -1.31605899e+00
2.08577961e-01 6.24234974e-01 -1.81060150e-01 3.31667751e-01
5.52088499e-01 -2.24071462e-02 3.00042391e-01 -6.71836078e-01
5.07677615e-01 6.40307784e-01 1.28719604e+00 5.70230067e-01
-6.89120829e-01 -6.31528080e-01 -6.86229587e-01 -4.36966568e-02
-1.62549424e+00 -1.63761154e-01 6.74216926e-01 -7.33266056e-01
1.38315558e+00 7.10786998e-01 1.14396739e+00 8.89781952e-01
9.39061761e-01 6.27730727e-01 8.40819359e-01 -8.20332289e-01
5.06140292e-01 6.42658472e-02 1.27222672e-01 4.96607453e-01
7.78133869e-01 -9.89626125e-02 -1.19788781e-01 -8.55276883e-01
9.08941329e-01 -3.06628317e-01 -4.04791862e-01 -1.46676779e-01
-1.53294563e+00 3.42529416e-01 -3.95995289e-01 7.44637370e-01
-5.19493461e-01 2.22275093e-01 7.18122542e-01 3.51639599e-01
1.74003080e-01 7.38861442e-01 -1.08221388e+00 -6.21299148e-01
-4.22048926e-01 9.66187790e-02 9.84244347e-01 9.98399436e-01
-2.96710372e-01 2.55617559e-01 -1.34362310e-01 8.40993822e-01
4.88978982e-01 5.57087183e-01 5.44861853e-01 -7.55170286e-01
2.60751247e-01 3.82532239e-01 -1.00288540e-02 -8.88151050e-01
-1.00578833e+00 -5.13556302e-01 -1.22480857e+00 1.22861583e-02
-1.64787933e-01 -8.49415481e-01 -5.23596883e-01 1.08576751e+00
-4.71734628e-03 1.22126766e-01 3.05005990e-04 4.97169733e-01
1.00517225e+00 4.01714534e-01 -6.44529834e-02 -8.60197723e-01
1.52665520e+00 -1.42969593e-01 -9.23297167e-01 -1.28208444e-01
9.34282959e-01 -6.69578373e-01 2.64604330e-01 8.71152580e-01
-9.95167136e-01 -4.52316642e-01 -1.44599926e+00 6.27882481e-01
3.07945848e-01 4.17103291e-01 5.29887915e-01 1.36689854e+00
-8.54151070e-01 4.11537975e-01 -9.65701878e-01 -3.06211591e-01
6.32311523e-01 6.14737511e-01 1.51215985e-01 4.61527169e-01
-1.11177778e+00 1.03133667e+00 -4.41032052e-02 -2.15825200e-01
-6.35179937e-01 -1.08488607e+00 -4.32948261e-01 -5.82798243e-01
-4.57613021e-01 -9.85769928e-01 1.00780702e+00 -5.54545522e-01
-1.43661785e+00 1.06413269e+00 4.73631233e-01 -5.33799112e-01
2.32609227e-01 -6.58489466e-02 -7.43492723e-01 -9.66305435e-02
6.46810979e-02 1.82568967e-01 5.58131993e-01 -7.94448256e-01
-5.03771126e-01 -3.69183332e-01 -4.16929752e-01 -1.15064286e-01
-1.33604825e-01 2.84952730e-01 -1.27625197e-01 -1.18100572e+00
1.70369667e-03 -1.04482687e+00 -2.57070482e-01 -2.68156469e-01
-3.89504194e-01 -4.24973406e-02 7.88612366e-01 -4.36468929e-01
1.67416263e+00 -1.89666879e+00 1.51872501e-01 1.58001915e-01
2.90030152e-01 3.56655955e-01 4.20750231e-01 3.54485333e-01
-3.87420356e-01 9.16718245e-02 -2.11024120e-01 9.45714638e-02
-4.32647586e-01 -4.89576869e-02 1.31193757e-01 7.82883942e-01
-5.13667107e-01 9.09534693e-01 -2.81297982e-01 -4.05734271e-01
4.37152207e-01 3.34046125e-01 -1.09577194e-01 -1.40010744e-01
2.32473806e-01 5.40080667e-01 -7.36856401e-01 9.07040596e-01
5.24795771e-01 -2.23758578e-01 4.78849500e-01 -1.93338022e-01
2.41447501e-02 -3.13965641e-02 -1.22978687e+00 1.89420152e+00
-3.00523102e-01 4.09139097e-01 -1.46055654e-01 -1.06412911e+00
9.66490924e-01 1.11688507e+00 1.23106527e+00 -4.79103714e-01
4.87410694e-01 5.96699715e-01 6.51837885e-02 -7.60904491e-01
-6.18100822e-01 -3.28357637e-01 -3.45463604e-01 7.46753931e-01
-1.99199468e-01 -4.06953037e-01 -3.44230592e-01 -6.62135780e-02
1.59643888e+00 -2.87168622e-01 9.16305900e-01 -8.35091293e-01
4.06715810e-01 -3.00624102e-01 1.21296299e+00 6.73517406e-01
-3.99277180e-01 7.85562336e-01 1.78779542e-01 -9.51073945e-01
-8.88958752e-01 -7.40452111e-01 -6.65068209e-01 -1.80948198e-01
-1.88179433e-01 -8.44187319e-01 -8.56794894e-01 -2.98796505e-01
-1.54413342e-01 1.66542634e-01 -1.80026278e-01 -2.24830396e-02
-6.16011918e-01 -9.95580018e-01 1.01543438e+00 6.82874799e-01
2.11021394e-01 -1.23090506e+00 -1.50394762e+00 8.37811828e-01
-1.99606031e-01 -1.03030467e+00 1.03591047e-02 1.69105932e-01
-1.25451446e+00 -9.95390892e-01 -7.87254333e-01 -9.93323565e-01
1.18320480e-01 -7.63201296e-01 1.15031826e+00 2.02750444e-01
-1.42806566e+00 7.31119156e-01 -4.16849881e-01 -1.24252713e+00
-3.87892932e-01 -8.44171569e-02 5.28841615e-01 -4.38841492e-01
8.27339217e-02 -5.76600373e-01 -6.93765104e-01 2.68361598e-01
1.11352820e-02 -1.13616616e-01 4.73325849e-01 4.96516943e-01
3.69936854e-01 3.95905860e-02 1.26774156e+00 -5.20309925e-01
6.17278099e-01 -2.81131804e-01 -3.29456806e-01 3.08208078e-01
-6.73044384e-01 -5.91415644e-01 5.34543633e-01 -1.65407538e-01
-3.35068703e-01 2.70481467e-01 -1.98895901e-01 -1.23800868e-02
2.03559473e-02 1.61578387e-01 -3.52553099e-01 1.00496769e-01
8.94309461e-01 -2.89381832e-01 -1.17606178e-01 -2.67741531e-01
-4.48947608e-01 1.35903561e+00 3.97956699e-01 -4.05861825e-01
4.53794487e-02 1.58302054e-01 -1.10665396e-01 -8.58249903e-01
5.58714271e-02 -2.47663960e-01 -2.80748099e-01 -3.19689840e-01
8.80550086e-01 -1.01418495e+00 -6.36737347e-01 8.70594561e-01
-1.36554468e+00 -1.99666724e-01 -3.29165876e-01 7.80030787e-01
-6.12811565e-01 -1.43448152e-02 -6.27656400e-01 -7.04895973e-01
-1.30320930e+00 -1.09790087e+00 1.00116432e+00 -2.32021198e-01
-9.61411774e-01 -8.97414029e-01 2.68251926e-01 1.55085370e-01
3.24450016e-01 7.10075378e-01 1.00411296e+00 -3.23872268e-01
-1.02233060e-01 -5.18571794e-01 5.71254611e-01 4.72760916e-01
4.59061593e-01 -2.37581000e-01 -8.48811388e-01 -4.90280628e-01
6.41640723e-01 6.98570758e-02 -2.52282411e-01 1.09359264e+00
1.12776983e+00 6.69078007e-02 -1.04239714e+00 1.60611004e-01
1.25945020e+00 9.71733570e-01 1.09755039e+00 2.20678806e-01
4.59016442e-01 -2.44893998e-01 5.32508016e-01 9.97613490e-01
-7.09911808e-02 6.34618878e-01 -6.72595724e-02 -1.03885017e-01
1.15805812e-01 7.02574551e-01 2.64246762e-02 1.05311835e+00
-2.44438604e-01 -1.69144019e-01 -1.31931782e+00 2.22721189e-01
-1.39492631e+00 -3.66916776e-01 -4.11546022e-01 1.93655968e+00
6.68883979e-01 4.61027771e-02 2.33954727e-03 8.80506754e-01
5.34182608e-01 -5.06663680e-01 -5.15355527e-01 -4.86864448e-01
5.41957021e-02 6.64372385e-01 2.88476180e-02 3.65434028e-02
-1.17688906e+00 -6.25325888e-02 6.53812027e+00 6.93880856e-01
-1.16667366e+00 2.08789647e-01 8.57719541e-01 1.53283879e-01
-5.51998019e-02 -3.97109985e-01 -3.39902878e-01 4.75782186e-01
1.19731176e+00 -2.26638570e-01 -2.80475050e-01 6.72578096e-01
1.08089410e-02 6.46409541e-02 -1.17067504e+00 1.60356510e+00
2.29683533e-01 -1.62495124e+00 -5.12113690e-01 -1.59196764e-01
5.23920476e-01 -9.37093422e-03 -2.69281626e-01 -8.27021003e-02
-7.30106115e-01 -9.53691125e-01 6.61034882e-01 4.15451854e-01
1.34024429e+00 -7.02433228e-01 8.40355873e-01 2.84045339e-01
-1.20523643e+00 -3.81022930e-01 -3.94521579e-02 -1.34672955e-01
1.70785531e-01 1.02046442e+00 -7.37213850e-01 6.71792448e-01
9.98642564e-01 9.43214357e-01 -2.18767270e-01 1.15233433e+00
3.27485532e-01 7.43085027e-01 -4.22044575e-01 -8.27813298e-02
-6.59882665e-01 3.03989828e-01 7.91122437e-01 9.76911008e-01
3.96124989e-01 6.36660874e-01 -2.30713174e-01 2.60830611e-01
2.11068258e-01 1.80445854e-02 -6.30965471e-01 3.33709478e-01
6.83841646e-01 1.09280908e+00 -8.32274199e-01 -4.30018872e-01
-1.43084526e-01 4.90265518e-01 -7.24684536e-01 -2.23451063e-01
-1.13333058e+00 -6.35579884e-01 2.74408191e-01 3.45001519e-01
-4.16743368e-01 2.93503553e-01 -1.21799147e+00 -6.12237155e-01
2.24023506e-01 -1.46504581e+00 3.63462329e-01 -5.88356733e-01
-8.17066371e-01 8.78899574e-01 -2.74268270e-01 -1.86621296e+00
-1.79242253e-01 -5.49012721e-01 -4.29166168e-01 6.63066626e-01
-2.82301992e-01 -7.84085512e-01 -2.52178703e-02 2.61286348e-01
4.69858646e-01 -6.59991205e-01 1.56388021e+00 7.03658938e-01
-2.70777613e-01 5.68405509e-01 -1.13192409e-01 -2.00125143e-01
4.11469549e-01 -9.12075758e-01 2.33444959e-01 2.12973997e-01
-3.34189683e-01 5.22544980e-01 5.27234375e-01 -6.20347202e-01
-1.66746008e+00 -1.06572449e+00 8.35053444e-01 -7.22955167e-01
-5.61385695e-03 -2.69068778e-01 -3.05157840e-01 5.98311365e-01
2.22132504e-01 9.99566689e-02 7.69672275e-01 -3.85442257e-01
5.63685238e-01 -3.95786852e-01 -1.21501851e+00 6.05387203e-02
7.45512664e-01 -3.21878791e-01 -3.89826447e-01 5.34852624e-01
2.67508030e-01 -5.87963462e-01 -1.36587524e+00 1.19416690e+00
7.31967032e-01 -4.87336308e-01 9.17801797e-01 -1.49533883e-01
-2.43274406e-01 -2.25186273e-01 1.17483482e-01 -8.15311491e-01
-5.07644154e-02 -1.20038438e+00 2.21708547e-02 8.38236034e-01
4.86865073e-01 -6.44938469e-01 5.66937327e-01 3.67861092e-01
-5.34252405e-01 -1.30161238e+00 -1.27304149e+00 -7.92343378e-01
-2.67953798e-02 -7.69527674e-01 2.74361461e-01 8.13728929e-01
6.21428847e-01 3.68355304e-01 -4.22542661e-01 -1.04487903e-01
5.39275110e-01 -2.42223620e-01 2.28599384e-01 -1.39273417e+00
-3.04463476e-01 8.71343613e-02 -1.11246657e+00 -3.20010960e-01
-6.69288993e-01 -7.28968978e-01 -2.29337484e-01 -1.53509057e+00
6.02999181e-02 -9.10473883e-01 -2.30109647e-01 5.09836733e-01
1.17721088e-01 2.11747915e-01 -2.12145612e-01 6.42022043e-02
-3.59039664e-01 -1.79339498e-01 9.04337347e-01 -2.10951060e-01
-3.06984156e-01 3.31310481e-01 -4.74357247e-01 7.35217333e-01
1.09178162e+00 -6.05320215e-01 -3.76119554e-01 -2.26200923e-01
1.30126268e-01 4.97530490e-01 1.04575679e-01 -1.65278172e+00
2.53637612e-01 5.91668487e-01 5.35532534e-01 -6.39510036e-01
2.34493706e-02 -8.63932848e-01 8.00210655e-01 1.18459976e+00
2.98593581e-01 2.98695534e-01 5.22808552e-01 -6.13928912e-03
7.54018128e-02 9.48677734e-02 8.05324256e-01 1.67325720e-01
6.37625232e-02 1.25766814e-01 -1.04724896e+00 4.45618443e-02
1.67824054e+00 -3.40543270e-01 1.04320221e-01 1.45880029e-01
-8.86164784e-01 -7.83914253e-02 -2.13434622e-01 5.04887402e-01
8.69322777e-01 -1.25250983e+00 -8.78957272e-01 3.19546789e-01
6.31870776e-02 -1.19931668e-01 3.07876468e-01 1.02056932e+00
-1.01308250e+00 6.46655202e-01 -2.14388266e-01 -1.05752921e+00
-1.56318200e+00 1.73878506e-01 7.58375585e-01 -6.46318123e-02
-1.06599474e+00 7.32332826e-01 -3.61491978e-01 -1.40977949e-01
3.01632494e-01 -4.18274611e-01 -9.08506811e-02 -2.25932956e-01
5.18904030e-01 6.60418868e-01 5.12012720e-01 6.32076785e-02
-9.11515594e-01 8.81079674e-01 3.04363012e-01 2.65769567e-02
1.20283318e+00 9.12989303e-02 -2.17337087e-01 5.47417879e-01
1.08193994e+00 -4.94349152e-01 -6.92036748e-02 6.28270268e-01
8.45998675e-02 2.21332178e-01 -2.18174696e-01 -1.14691782e+00
-1.01628554e+00 6.58976555e-01 1.41261697e+00 8.62100162e-03
1.38771939e+00 -3.51467095e-02 7.29744017e-01 1.89688116e-01
1.00963402e+00 -1.11754441e+00 -3.25593501e-02 -4.15446758e-02
1.11970854e+00 -5.50184309e-01 2.80913293e-01 -2.74557412e-01
-5.32750547e-01 1.17613792e+00 2.27226347e-01 2.10367024e-01
1.28111589e+00 9.75695312e-01 3.84600431e-01 -2.35172480e-01
-6.67416334e-01 9.91189539e-01 -1.71310287e-02 7.73469090e-01
7.57291973e-01 3.78799409e-01 -7.73544252e-01 1.12009382e+00
-1.65549200e-02 6.82719171e-01 2.29860485e-01 1.53997684e+00
-2.78602660e-01 -1.27544260e+00 -5.20992041e-01 8.84330451e-01
-7.04568088e-01 -3.54177924e-03 -2.39067256e-01 3.81743848e-01
5.87500572e-01 1.12738574e+00 -1.85511991e-01 -3.56369704e-01
4.74400669e-01 -2.89977193e-01 6.21234894e-01 -5.84184408e-01
-8.28653216e-01 1.94715008e-01 1.80993780e-01 -3.74399483e-01
-1.10113189e-01 -7.56014466e-01 -1.56599152e+00 2.41170943e-01
-2.74048239e-01 2.61475116e-01 8.02277505e-01 5.13383210e-01
6.60782874e-01 1.00817180e+00 6.34446502e-01 -3.71176124e-01
1.49015477e-02 -1.09915900e+00 -8.94967556e-01 -3.43285084e-01
2.04638898e-01 -4.32718575e-01 -5.42334002e-03 2.17258453e-01] | [14.270562171936035, 3.264667272567749] |
7ebe1ec4-8c86-4091-85d3-a9334f1be114 | beyond-512-tokens-siamese-multi-depth | 2004.12297 | null | https://arxiv.org/abs/2004.12297v2 | https://arxiv.org/pdf/2004.12297v2.pdf | Beyond 512 Tokens: Siamese Multi-depth Transformer-based Hierarchical Encoder for Long-Form Document Matching | Many natural language processing and information retrieval problems can be formalized as the task of semantic matching. Existing work in this area has been largely focused on matching between short texts (e.g., question answering), or between a short and a long text (e.g., ad-hoc retrieval). Semantic matching between long-form documents, which has many important applications like news recommendation, related article recommendation and document clustering, is relatively less explored and needs more research effort. In recent years, self-attention based models like Transformers and BERT have achieved state-of-the-art performance in the task of text matching. These models, however, are still limited to short text like a few sentences or one paragraph due to the quadratic computational complexity of self-attention with respect to input text length. In this paper, we address the issue by proposing the Siamese Multi-depth Transformer-based Hierarchical (SMITH) Encoder for long-form document matching. Our model contains several innovations to adapt self-attention models for longer text input. In order to better capture sentence level semantic relations within a document, we pre-train the model with a novel masked sentence block language modeling task in addition to the masked word language modeling task used by BERT. Our experimental results on several benchmark datasets for long-form document matching show that our proposed SMITH model outperforms the previous state-of-the-art models including hierarchical attention, multi-depth attention-based hierarchical recurrent neural network, and BERT. Comparing to BERT based baselines, our model is able to increase maximum input text length from 512 to 2048. We will open source a Wikipedia based benchmark dataset, code and a pre-trained checkpoint to accelerate future research on long-form document matching. | ['Liu Yang', 'Michael Bendersky', 'Mingyang Zhang', 'Marc Najork', 'Cheng Li'] | 2020-04-26 | null | null | null | null | ['2048'] | ['playing-games'] | [ 3.00159156e-01 -6.92367107e-02 -1.45266116e-01 -4.38161105e-01
-1.19226933e+00 -2.43986711e-01 6.52294219e-01 6.05925143e-01
-7.04115629e-01 1.59807414e-01 5.09023130e-01 -4.56966996e-01
-2.11336151e-01 -8.04953456e-01 -6.45461321e-01 -1.61160260e-01
3.91815454e-01 7.46936023e-01 4.71884340e-01 -5.20976603e-01
6.12669945e-01 -1.02307327e-01 -1.52498567e+00 7.87680328e-01
9.86925542e-01 8.62181425e-01 6.99619949e-01 7.20543921e-01
-7.08136022e-01 6.29087627e-01 -4.21604872e-01 -5.20103872e-01
5.20699732e-02 -4.16530639e-01 -1.31046164e+00 -3.04079354e-01
7.15998173e-01 -2.01554850e-01 -3.93264234e-01 1.03599203e+00
7.03425467e-01 3.15223783e-01 3.81660432e-01 -9.00486708e-01
-9.42296684e-01 9.44164872e-01 -4.81021553e-01 3.10220569e-01
6.05233610e-01 -4.20234710e-01 1.38943553e+00 -9.15369093e-01
3.31822425e-01 1.57052433e+00 7.23898768e-01 5.32688975e-01
-8.16233277e-01 -5.63129663e-01 9.71529633e-02 4.99984771e-01
-1.08804917e+00 -3.25542897e-01 5.45562267e-01 -2.00738311e-01
1.54543209e+00 3.53904188e-01 1.12240471e-01 9.63067293e-01
4.09674495e-01 9.47165191e-01 6.77229464e-01 -7.40048647e-01
-1.72004342e-01 3.75494808e-02 7.20869482e-01 5.09273827e-01
-4.51205671e-02 -2.71005481e-01 -1.96396336e-01 -1.30944222e-01
2.59969860e-01 8.66543055e-02 -1.29346177e-01 1.19192705e-01
-1.15572774e+00 1.04821110e+00 5.47960401e-01 8.60132694e-01
-6.94661811e-02 -1.47368980e-03 7.52335906e-01 6.12401664e-01
4.57312793e-01 7.25112379e-01 -3.47830772e-01 9.79440436e-02
-1.14873993e+00 3.53232443e-01 8.31021726e-01 1.02502322e+00
6.52506471e-01 -4.34986651e-01 -6.78150833e-01 1.19854569e+00
3.34536374e-01 3.41924280e-01 1.18363285e+00 -4.89586443e-01
9.37714338e-01 7.49192238e-01 -1.97270319e-01 -1.11376417e+00
-3.98541510e-01 -3.77603024e-01 -9.35644507e-01 -6.08065307e-01
1.95157647e-01 2.68089652e-01 -7.97670364e-01 1.43077672e+00
-5.13540804e-02 -1.13397740e-01 2.48855770e-01 8.87270272e-01
1.21391785e+00 9.67778683e-01 -1.11634038e-01 2.27625161e-01
1.63173163e+00 -1.55156875e+00 -7.07064986e-01 -4.73261297e-01
1.03194618e+00 -9.89324629e-01 1.36215615e+00 -2.80889541e-01
-1.16899967e+00 -7.64796317e-01 -8.35617840e-01 -5.99349976e-01
-6.79386973e-01 -9.94951949e-02 1.95326686e-01 3.67676467e-01
-1.08759558e+00 6.63371027e-01 -4.76410955e-01 -1.04431117e+00
-6.62118644e-02 3.56694937e-01 -2.19006479e-01 -1.36895224e-01
-1.70977783e+00 8.63590956e-01 3.71757716e-01 -5.61402440e-02
-1.91564858e-01 -7.04054654e-01 -9.08205986e-01 4.65004891e-01
1.91889063e-01 -1.05281699e+00 1.41295528e+00 -8.47259343e-01
-1.34149766e+00 1.03778636e+00 -3.18212628e-01 -7.05528617e-01
1.70468509e-01 -2.42998093e-01 -4.49325204e-01 1.15678877e-01
3.14813286e-01 6.13995373e-01 6.65243328e-01 -6.52276874e-01
-4.66371894e-01 -5.53937733e-01 2.68600911e-01 4.43188310e-01
-7.74544835e-01 5.33523619e-01 -6.07396603e-01 -8.71797323e-01
-5.63457645e-02 -8.22125077e-01 -1.23709567e-01 -3.26114416e-01
-2.82065630e-01 -6.28556371e-01 7.80563295e-01 -6.87618077e-01
1.44424987e+00 -1.91768742e+00 1.02270968e-01 -2.28712425e-01
-9.41808447e-02 2.80436009e-01 -4.55743760e-01 8.37604463e-01
-2.03599911e-02 1.91893533e-01 -2.56247073e-01 -4.67114002e-01
1.75461069e-01 -7.72516504e-02 -4.94083941e-01 8.69676620e-02
-1.83603257e-01 1.24455476e+00 -7.45756984e-01 -6.54894054e-01
-2.14320287e-01 1.43196642e-01 -5.76068401e-01 5.28098166e-01
-2.27077112e-01 -1.50108948e-01 -4.20750201e-01 2.86597908e-01
4.05703634e-01 -4.88155901e-01 -1.71790957e-01 -1.10813685e-01
1.06803991e-01 7.02523172e-01 -7.79650688e-01 2.15203810e+00
-7.62420237e-01 5.71876943e-01 -1.06880970e-01 -1.21263063e+00
8.59979272e-01 3.83449316e-01 2.59784341e-01 -1.06256509e+00
4.08842564e-02 2.96957105e-01 -4.55822349e-02 -6.66129589e-01
9.13230479e-01 -6.81923470e-03 -2.56601900e-01 7.46842146e-01
-1.48617595e-01 -2.94236583e-03 4.36991155e-01 4.37746048e-01
1.15103698e+00 -2.82612920e-01 1.19822629e-01 -3.81112397e-01
8.61520827e-01 1.25522884e-02 5.31344395e-03 9.83281612e-01
2.21103042e-01 6.99048877e-01 8.60990807e-02 -1.29617944e-01
-1.04604518e+00 -3.16054434e-01 -5.43083176e-02 1.52454901e+00
2.76247859e-01 -5.53203762e-01 -9.50523436e-01 -6.31436408e-01
1.73674107e-01 4.63429838e-01 -5.45817971e-01 -2.81660080e-01
-8.28548729e-01 -4.92849201e-01 6.90233350e-01 6.13077104e-01
6.26074672e-01 -1.34622860e+00 -3.46098125e-01 4.62933362e-01
-3.09258521e-01 -1.22544849e+00 -1.02720368e+00 -4.18706015e-02
-8.75440300e-01 -7.60157585e-01 -1.08463037e+00 -1.18636155e+00
4.72408891e-01 5.52375853e-01 1.14268732e+00 3.42006832e-01
-1.23866029e-01 3.11239243e-01 -6.78932905e-01 -7.31651187e-02
-3.87869686e-01 7.31985211e-01 -2.80669630e-01 -4.32642475e-02
5.24069130e-01 -5.19833043e-02 -5.79979002e-01 3.20198655e-01
-1.19942999e+00 -6.35615736e-03 5.86919546e-01 1.18648553e+00
9.22050253e-02 -3.61219645e-01 6.52383447e-01 -1.07709610e+00
1.09091854e+00 -5.20371914e-01 -2.83339888e-01 5.35514116e-01
-6.94550455e-01 3.65022957e-01 8.56907070e-01 -3.05451840e-01
-1.01039791e+00 -4.70871508e-01 -3.59231174e-01 -1.70748428e-01
4.02353518e-02 9.46188986e-01 1.11961983e-01 1.29058123e-01
4.86492217e-01 2.46570021e-01 -1.70623481e-01 -7.16449678e-01
2.97397554e-01 1.31999028e+00 4.56022948e-01 -5.16296446e-01
5.39727569e-01 4.20117825e-02 -4.81062084e-01 -5.16272664e-01
-1.01965582e+00 -1.03831458e+00 -5.74370742e-01 2.94862896e-01
7.81627476e-01 -6.52587414e-01 -4.80155557e-01 3.55492473e-01
-1.42139912e+00 -1.80936173e-01 5.59104718e-02 2.05989420e-01
-3.04535449e-01 6.82168245e-01 -8.27517807e-01 -4.18569326e-01
-1.02245951e+00 -1.05554640e+00 1.42116511e+00 -3.87267545e-02
-3.30122888e-01 -1.14822733e+00 1.19984843e-01 7.47498274e-01
6.82519078e-01 -5.47080278e-01 1.22068381e+00 -9.85436201e-01
-2.00038239e-01 -3.30857545e-01 -4.05352861e-01 -5.67756929e-02
-1.36892036e-01 -5.97091317e-01 -6.59859598e-01 -6.22722149e-01
-8.05838779e-02 -5.02048492e-01 1.21613455e+00 3.27496640e-02
1.19864893e+00 -4.26702142e-01 -3.35466355e-01 4.22610521e-01
1.14249325e+00 7.66469538e-02 6.62942767e-01 7.02263713e-01
6.61390901e-01 8.55182230e-01 5.67119718e-01 8.45721141e-02
6.25233710e-01 9.67780828e-01 5.67324944e-02 -1.30700305e-01
-2.16003239e-01 -2.89360702e-01 2.43564263e-01 1.28606105e+00
6.83480740e-01 -5.27808547e-01 -9.55746472e-01 6.00336194e-01
-2.16290402e+00 -1.03064668e+00 -1.47143841e-01 2.01994538e+00
7.29463398e-01 6.01557875e-03 -4.42511514e-02 2.08041072e-01
8.27904344e-01 1.72738001e-01 -4.19703096e-01 -7.30573297e-01
-6.22507483e-02 1.23519421e-01 2.55907327e-01 5.90100169e-01
-1.06544530e+00 1.13538873e+00 5.28784895e+00 9.79643226e-01
-1.05941617e+00 2.48012245e-01 2.86534846e-01 2.37320125e-01
-3.84120643e-01 -1.58647019e-02 -1.09977245e+00 5.76798081e-01
1.10693812e+00 -4.13265079e-01 2.18608961e-01 5.96717894e-01
-2.31755264e-02 2.36485526e-01 -1.14106071e+00 9.59583938e-01
4.38020408e-01 -1.24082494e+00 3.46149564e-01 -3.67444634e-01
5.96044481e-01 3.90037000e-02 -1.41931519e-01 6.12871170e-01
-7.28324205e-02 -1.00161922e+00 5.31181633e-01 2.04587474e-01
6.45400286e-01 -3.84930074e-01 1.05903196e+00 5.18856049e-01
-1.30186260e+00 -1.92277804e-01 -5.66447139e-01 5.96305355e-02
8.29576850e-02 2.97102123e-01 -5.46500504e-01 7.08970964e-01
8.04571092e-01 8.90814543e-01 -7.18863308e-01 9.55536127e-01
2.93256313e-01 3.61357629e-01 -1.18014395e-01 -3.72638375e-01
6.09105885e-01 1.02079436e-01 3.20677578e-01 1.51524210e+00
3.16924512e-01 -6.90861493e-02 1.33396566e-01 7.06628680e-01
-3.13000798e-01 5.02042353e-01 -3.60611558e-01 -3.09393182e-02
3.00298899e-01 1.04145157e+00 -4.70610172e-01 -5.65502346e-01
-6.81372166e-01 1.25643754e+00 5.71291268e-01 2.78530270e-01
-4.10303533e-01 -9.11442757e-01 1.52831629e-01 2.17179909e-01
2.18144596e-01 1.21309273e-01 -1.63693920e-01 -1.24019980e+00
1.49817631e-01 -9.94208872e-01 7.64313817e-01 -7.09117770e-01
-1.44380295e+00 7.89211869e-01 -1.49245992e-01 -1.01833093e+00
-4.00052756e-01 -4.18925822e-01 -6.40356660e-01 9.55039144e-01
-1.59890831e+00 -1.11807227e+00 -2.25274637e-01 5.67091763e-01
1.04460287e+00 -3.01858157e-01 8.93488050e-01 5.54181695e-01
-5.27783513e-01 8.89586210e-01 3.74352098e-01 1.76482454e-01
8.83025229e-01 -1.06403303e+00 7.26195693e-01 5.84207833e-01
2.68498570e-01 7.70847976e-01 3.86613697e-01 -4.83176231e-01
-1.44657040e+00 -1.11356091e+00 1.58882356e+00 -2.67884165e-01
7.43426085e-01 -4.59944636e-01 -1.31427538e+00 5.05174756e-01
5.69033444e-01 -2.66888827e-01 4.58889157e-01 -4.03348282e-02
-3.62745523e-01 -2.04011798e-01 -8.48273337e-01 5.17428398e-01
8.71859670e-01 -8.39324057e-01 -9.35519218e-01 4.32203025e-01
1.00713766e+00 -3.37624848e-01 -8.39354098e-01 4.36416626e-01
4.94034559e-01 -6.64990783e-01 8.35173965e-01 -8.02836716e-01
4.59781229e-01 6.11336082e-02 -3.32399202e-03 -1.07879186e+00
-5.01731634e-01 -5.77818334e-01 8.11226442e-02 1.25599170e+00
4.38101798e-01 -5.17328322e-01 6.51316881e-01 3.87334198e-01
-3.01796407e-01 -8.38295043e-01 -6.88748777e-01 -8.01310003e-01
4.21003670e-01 -2.09679589e-01 6.05332553e-01 9.32902455e-01
3.21962267e-01 7.79327571e-01 -1.94605380e-01 -1.66645303e-01
2.03452364e-01 5.52777946e-01 5.68572938e-01 -1.13740671e+00
-2.70027280e-01 -7.33080685e-01 -2.38740474e-01 -1.53286374e+00
6.69395030e-01 -1.41423213e+00 -4.41586524e-02 -1.81522787e+00
5.83448589e-01 -1.46330714e-01 -2.29936808e-01 3.21568638e-01
-4.33670491e-01 4.40423340e-02 1.70242801e-01 3.33491117e-01
-7.56383061e-01 7.12854207e-01 9.20035660e-01 -5.03175557e-01
8.65984559e-02 -9.62033644e-02 -6.40536964e-01 2.16708511e-01
6.77902460e-01 -6.85759366e-01 -1.22664317e-01 -8.81605327e-01
2.02314347e-01 1.13638155e-01 3.55283916e-02 -6.83221877e-01
8.00075531e-01 2.91195124e-01 -1.95363969e-01 -5.88894665e-01
7.61252269e-03 -7.50957549e-01 -4.37841624e-01 5.44587910e-01
-9.94545579e-01 5.60221791e-01 1.72168583e-01 3.57844025e-01
-6.15335584e-01 -7.00703740e-01 5.92268348e-01 -2.85548449e-01
-5.14101148e-01 2.45649487e-01 -3.62982541e-01 3.50894034e-01
5.33469141e-01 -1.57755256e-01 -4.21756595e-01 -4.51707035e-01
-1.86838418e-01 6.15503013e-01 1.53292209e-01 9.13263142e-01
6.06069446e-01 -1.22435927e+00 -9.00956929e-01 9.41422358e-02
3.84466976e-01 -2.27598488e-01 1.44122794e-01 6.17095351e-01
-3.34970236e-01 1.15775895e+00 1.63857743e-01 -3.15631688e-01
-1.36478066e+00 6.54471457e-01 2.14179516e-01 -7.44133890e-01
-5.64662814e-01 7.14644611e-01 9.36794952e-02 -6.80517256e-01
4.54924941e-01 -3.05405766e-01 -3.72403741e-01 1.76216830e-02
5.42011082e-01 2.54136205e-01 4.09777045e-01 -6.18706405e-01
-3.14478159e-01 9.42028463e-01 -4.62371379e-01 1.20974347e-01
1.00076830e+00 -3.04709435e-01 -3.51108074e-01 2.83501625e-01
1.75006735e+00 -4.25017655e-01 -2.34564185e-01 -6.84154034e-01
4.83936667e-01 -1.95803359e-01 2.11343125e-01 -5.06290019e-01
-8.72559309e-01 1.10040295e+00 3.73622537e-01 2.69209862e-01
8.73597264e-01 3.96982506e-02 1.46797645e+00 7.56936550e-01
-3.38467546e-02 -1.14137506e+00 2.42103145e-01 9.55464900e-01
1.03475904e+00 -1.34590745e+00 -3.27643812e-01 -1.38221234e-02
-3.31002623e-01 1.19514048e+00 7.40990877e-01 1.67991929e-02
5.15060186e-01 5.64309955e-02 1.15174957e-01 -1.74747318e-01
-8.88282299e-01 -2.66443729e-01 8.16363156e-01 3.22685577e-02
8.41749609e-01 -5.29413998e-01 -4.36317533e-01 5.68874359e-01
-2.92934150e-01 -2.25141630e-01 1.53374329e-01 8.44787598e-01
-5.02879143e-01 -1.27666092e+00 -3.03898096e-01 6.93471491e-01
-7.26292908e-01 -6.10025942e-01 -3.98698419e-01 2.83000380e-01
-5.42299986e-01 1.03751504e+00 2.63419628e-01 -1.49423376e-01
4.58522648e-01 1.41012266e-01 2.39012033e-01 -8.87137711e-01
-1.17204857e+00 -2.34532341e-01 6.20596297e-02 -4.61771488e-01
-3.74011487e-01 -3.95506501e-01 -1.21889293e+00 -1.55461028e-01
-4.95491892e-01 3.90101403e-01 5.51109016e-01 1.03518999e+00
5.30822814e-01 3.82935435e-01 4.90152448e-01 -5.79178452e-01
-8.16981316e-01 -1.28796613e+00 -2.91972667e-01 7.01742470e-01
2.19919324e-01 -1.50369734e-01 -2.36944795e-01 -2.30167106e-01] | [11.143756866455078, 8.228445053100586] |
0a8c582a-e04f-46e9-95b4-2e5a0fb1adf4 | hybrik-x-hybrid-analytical-neural-inverse | 2304.05690 | null | https://arxiv.org/abs/2304.05690v1 | https://arxiv.org/pdf/2304.05690v1.pdf | HybrIK-X: Hybrid Analytical-Neural Inverse Kinematics for Whole-body Mesh Recovery | Recovering whole-body mesh by inferring the abstract pose and shape parameters from visual content can obtain 3D bodies with realistic structures. However, the inferring process is highly non-linear and suffers from image-mesh misalignment, resulting in inaccurate reconstruction. In contrast, 3D keypoint estimation methods utilize the volumetric representation to achieve pixel-level accuracy but may predict unrealistic body structures. To address these issues, this paper presents a novel hybrid inverse kinematics solution, HybrIK, that integrates the merits of 3D keypoint estimation and body mesh recovery in a unified framework. HybrIK directly transforms accurate 3D joints to body-part rotations via twist-and-swing decomposition. The swing rotations are analytically solved with 3D joints, while the twist rotations are derived from visual cues through neural networks. To capture comprehensive whole-body details, we further develop a holistic framework, HybrIK-X, which enhances HybrIK with articulated hands and an expressive face. HybrIK-X is fast and accurate by solving the whole-body pose with a one-stage model. Experiments demonstrate that HybrIK and HybrIK-X preserve both the accuracy of 3D joints and the realistic structure of the parametric human model, leading to pixel-aligned whole-body mesh recovery. The proposed method significantly surpasses the state-of-the-art methods on various benchmarks for body-only, hand-only, and whole-body scenarios. Code and results can be found at https://jeffli.site/HybrIK-X/ | ['Cewu Lu', 'Lixin Yang', 'Zhicun Chen', 'Chao Xu', 'Siyuan Bian', 'Jiefeng Li'] | 2023-04-12 | null | null | null | null | ['3d-human-pose-estimation', '3d-human-reconstruction'] | ['computer-vision', 'computer-vision'] | [-2.24400833e-01 8.44305307e-02 -3.46134007e-01 3.26445736e-02
-8.28684270e-01 -3.79728138e-01 2.29603872e-01 -5.02721548e-01
1.25346467e-01 4.07250166e-01 3.23914438e-01 4.40112650e-01
1.13725476e-01 -6.02353156e-01 -8.61750722e-01 -4.82538998e-01
2.69028872e-01 7.39331126e-01 3.67246531e-02 -3.87526125e-01
-2.61218518e-01 2.49352306e-01 -1.26221561e+00 -1.37259001e-02
6.99990451e-01 9.71149087e-01 -3.54131341e-01 6.04561269e-01
3.81081998e-01 1.65158808e-01 -1.91062495e-01 -5.28189600e-01
5.04404604e-01 -3.61778468e-01 -4.99518842e-01 2.54823536e-01
8.49911034e-01 -6.67950928e-01 -3.91019076e-01 6.92511380e-01
7.21570671e-01 -1.80946961e-01 6.06676221e-01 -1.23302376e+00
-3.13587070e-01 -1.50236432e-02 -1.04569590e+00 -7.27626204e-01
7.69672692e-01 3.39053005e-01 7.43187964e-01 -1.18090475e+00
9.49083388e-01 1.47417223e+00 1.20596552e+00 4.78064448e-01
-1.39924574e+00 -8.34110498e-01 -2.58952733e-02 -2.66587317e-01
-1.54552579e+00 -4.23801303e-01 1.06781185e+00 -5.67148745e-01
4.68436688e-01 3.09951991e-01 1.29595089e+00 9.71041441e-01
3.31629425e-01 7.54755378e-01 1.00055349e+00 -2.25720480e-01
-3.95736396e-01 -7.13799775e-01 -2.26091802e-01 1.13953376e+00
4.16670918e-01 1.88075602e-01 -8.23757231e-01 -2.08251715e-01
1.36815131e+00 1.88719071e-02 -4.88484472e-01 -9.37898755e-01
-1.59361470e+00 4.11323667e-01 3.85990798e-01 -4.15056318e-01
-4.18576092e-01 6.24166548e-01 3.10739845e-01 -2.14549035e-01
4.75070715e-01 -1.04495622e-02 -4.32726853e-02 2.34345868e-02
-1.15221179e+00 7.57870793e-01 6.05071902e-01 9.92879629e-01
5.52847385e-01 2.64737219e-01 -8.48833844e-02 6.39766514e-01
6.52008832e-01 9.71234024e-01 3.94497477e-02 -1.42468727e+00
4.74747628e-01 5.79768658e-01 1.19982630e-01 -1.14698040e+00
-4.81363535e-01 -3.79890978e-01 -1.12465870e+00 3.43881816e-01
6.03491008e-01 5.69972629e-03 -1.10887778e+00 1.51793969e+00
1.10984027e+00 -9.32418182e-02 -3.43811274e-01 1.39343703e+00
1.03799486e+00 3.85868728e-01 -2.21112445e-01 1.47804126e-01
1.53407526e+00 -1.04857194e+00 -7.54847705e-01 -2.64202803e-01
5.71354590e-02 -6.82483971e-01 9.34773207e-01 3.85374218e-01
-1.48301709e+00 -5.14491558e-01 -9.30896759e-01 -4.35281575e-01
2.45492026e-01 2.62706757e-01 3.97110432e-01 3.04748207e-01
-7.07612455e-01 6.28669620e-01 -9.95612264e-01 -1.83519065e-01
1.35302022e-01 3.06713670e-01 -6.50444567e-01 -6.09788932e-02
-9.69830215e-01 7.71646202e-01 -9.54102203e-02 4.07526702e-01
-7.84926057e-01 -9.42440331e-01 -1.23283899e+00 -4.16313976e-01
5.36649883e-01 -1.55397010e+00 1.25469995e+00 -5.19469023e-01
-1.76092684e+00 1.11742604e+00 -7.39035243e-03 1.68224975e-01
1.01683772e+00 -6.03902459e-01 1.41847834e-01 3.61966968e-01
2.43856162e-02 6.07182503e-01 1.09576011e+00 -1.53632402e+00
8.96395966e-02 -6.04300916e-01 -3.87623131e-01 4.33452606e-01
2.50409901e-01 -4.94331688e-01 -1.08518863e+00 -1.05057406e+00
4.91097093e-01 -1.12993085e+00 -8.09982195e-02 8.39327395e-01
-6.71222150e-01 3.18001628e-01 4.79360610e-01 -1.33428466e+00
1.02998555e+00 -1.74522221e+00 7.01156676e-01 2.63008863e-01
4.00935829e-01 -7.24668279e-02 1.58244029e-01 3.12794119e-01
7.03007430e-02 -2.80877292e-01 -3.34136158e-01 -6.85903132e-01
5.70541546e-02 4.07870471e-01 2.92637274e-02 8.78000200e-01
-1.16787232e-01 1.13824594e+00 -7.32960105e-01 -8.28268766e-01
3.49634230e-01 7.28439510e-01 -6.20544732e-01 7.39215836e-02
1.86272077e-02 5.77582777e-01 -3.84452105e-01 1.31189442e+00
6.27968073e-01 -2.28297815e-01 2.51009494e-01 -8.02823782e-01
1.11069091e-01 -3.43526483e-01 -1.44807553e+00 2.14670920e+00
-1.86003327e-01 -1.37151182e-01 6.76452935e-01 -4.65018034e-01
6.89770997e-01 3.90051454e-01 6.67222202e-01 -5.25693953e-01
1.88584834e-01 3.53795707e-01 -4.93572026e-01 -1.60347879e-01
2.68053085e-01 -2.71337241e-01 -1.43897548e-01 3.89183424e-02
-1.21157229e-01 -4.33902502e-01 -2.19439894e-01 -6.17422685e-02
3.96621644e-01 9.96544659e-01 3.42043310e-01 -8.09294078e-03
1.86803088e-01 7.77257681e-02 7.32285082e-01 1.27816290e-01
3.29251364e-02 1.04559290e+00 1.89573035e-01 -4.34719086e-01
-1.08778834e+00 -1.43336511e+00 7.82331601e-02 5.15520453e-01
4.34071749e-01 -5.64176023e-01 -8.28514934e-01 -1.31572366e-01
5.34152627e-01 1.44467667e-01 -5.21086872e-01 -6.00257814e-02
-8.53550673e-01 -3.50010782e-01 5.08316934e-01 6.05368972e-01
4.86325324e-01 -6.28933311e-01 -7.82509089e-01 1.65850028e-01
-6.10549986e-01 -9.59205866e-01 -7.97046602e-01 -4.62555528e-01
-1.10000992e+00 -1.25361109e+00 -1.32754755e+00 -2.89467156e-01
6.54630721e-01 -1.25559539e-01 1.27152085e+00 2.15307623e-01
-5.22207320e-01 5.31501591e-01 -3.48672159e-02 9.28511992e-02
-2.09319875e-01 -2.18389213e-01 2.44707957e-01 -2.69041628e-01
-6.11806870e-01 -7.35743463e-01 -1.04518890e+00 5.67401528e-01
-4.85123605e-01 6.52033985e-01 4.37148184e-01 7.99376905e-01
1.07533872e+00 -4.69959557e-01 -2.14115120e-02 -2.79687494e-01
-4.58635427e-02 -2.96612643e-02 -3.14006895e-01 9.29618031e-02
-3.30912948e-01 -1.93995684e-01 1.55046806e-01 -4.86989886e-01
-9.86253321e-01 4.05457944e-01 -2.98189968e-01 -8.51900399e-01
2.20187232e-01 2.20671207e-01 -3.04733515e-01 -1.68061048e-01
3.10910225e-01 3.13925505e-01 4.19313908e-01 -8.25578451e-01
4.84594882e-01 9.25023481e-02 9.33689296e-01 -9.38442171e-01
9.75435376e-01 7.43538439e-01 2.70230979e-01 -6.87300980e-01
-6.30832911e-01 -3.30656409e-01 -8.15849543e-01 -6.31214201e-01
6.97402716e-01 -1.14760220e+00 -7.83292770e-01 9.26857591e-01
-1.02820730e+00 -3.95752668e-01 -3.04010868e-01 3.99122149e-01
-7.24511385e-01 8.71653438e-01 -8.79705250e-01 -4.68788862e-01
-8.54491413e-01 -9.76344526e-01 1.78173959e+00 -1.61971897e-01
-6.08220935e-01 -6.02501690e-01 1.34991825e-01 6.41928852e-01
-2.83804893e-01 9.94169891e-01 6.81559563e-01 4.95684803e-01
-4.69647467e-01 -2.54599035e-01 2.12491736e-01 2.71706264e-02
-2.78571378e-02 1.57738104e-01 -5.18292427e-01 -4.42958951e-01
-5.36257386e-01 -4.06331182e-01 4.19168621e-01 6.20768785e-01
5.43626070e-01 -3.54633242e-01 -3.54676902e-01 1.07769632e+00
1.32842839e+00 -6.20044887e-01 6.25490248e-01 2.20743179e-01
1.09913087e+00 6.58768415e-01 8.07882726e-01 6.95474803e-01
5.82673013e-01 1.15608430e+00 6.39796138e-01 -2.55989581e-01
-6.42438412e-01 -7.58134365e-01 3.65342557e-01 9.63191807e-01
-5.98774791e-01 2.71339238e-01 -8.69828701e-01 3.74133140e-01
-1.76318514e+00 -5.51236272e-01 -3.92165631e-01 2.16063261e+00
1.07903373e+00 -1.83415219e-01 3.79669964e-01 5.03530689e-02
5.12954056e-01 8.17167386e-02 -6.90959811e-01 1.99894056e-01
1.41591445e-01 2.56017838e-02 4.02888924e-01 5.30585289e-01
-7.37152576e-01 8.28576088e-01 5.82257223e+00 7.23784924e-01
-9.67468679e-01 2.06328207e-03 1.20803274e-01 -1.49735972e-01
-1.89471260e-01 -1.33080542e-01 -7.57720172e-01 2.26866663e-01
1.11512311e-01 1.94932804e-01 6.27190694e-02 7.81771004e-01
1.35254949e-01 -6.59615127e-03 -9.25152659e-01 1.25846660e+00
4.63734530e-02 -1.21147394e+00 8.65955353e-02 2.10117444e-01
6.14416063e-01 -3.59265387e-01 -1.68347195e-01 -9.30388346e-02
-1.58561370e-03 -8.52508068e-01 1.38570869e+00 8.50303352e-01
1.34564245e+00 -5.17452836e-01 1.52031541e-01 3.00449371e-01
-1.53009057e+00 4.25313175e-01 1.02189451e-01 9.58991237e-03
6.06420994e-01 5.76989651e-01 -2.57968009e-01 8.82705331e-01
8.74180794e-01 8.34606171e-01 -2.27313668e-01 9.36778426e-01
-3.18259150e-01 2.90964037e-01 -5.40044606e-01 6.86146259e-01
-2.84419745e-01 -2.89494932e-01 7.81474113e-01 9.10524786e-01
3.99282366e-01 1.20622039e-01 3.72127771e-01 9.17027950e-01
1.13358781e-01 1.13062501e-01 -2.01823235e-01 3.02589953e-01
1.51454210e-01 1.29636526e+00 -5.48700213e-01 -3.98106396e-01
-3.10534984e-02 1.17078662e+00 2.79071052e-02 3.02889764e-01
-8.51568699e-01 1.45582005e-01 7.86187768e-01 7.36912727e-01
1.04405068e-01 -4.26702291e-01 -1.84764758e-01 -1.29949021e+00
3.09939057e-01 -9.75939333e-01 2.86791742e-01 -1.05168283e+00
-9.18426812e-01 5.25493063e-02 3.36295158e-01 -1.30663621e+00
-3.65711898e-01 -4.67822552e-01 -1.68537825e-01 7.79929101e-01
-1.22035992e+00 -1.77751100e+00 -6.11871123e-01 6.89142585e-01
3.85495722e-01 6.18061244e-01 6.50038064e-01 1.54556200e-01
-3.40914905e-01 5.22166014e-01 -2.13722914e-01 1.75099358e-01
8.71953845e-01 -1.06650758e+00 5.36379457e-01 4.71845210e-01
-2.78292179e-01 4.68620658e-01 6.38065040e-01 -1.09124565e+00
-1.93694198e+00 -8.20284784e-01 3.07370424e-01 -3.76510739e-01
3.71109471e-02 -2.83826351e-01 -6.43492579e-01 6.78578973e-01
-2.72307992e-01 2.87897021e-01 3.10517490e-01 -2.51213670e-01
-3.42935920e-01 -3.79856094e-03 -1.02933335e+00 7.24296272e-01
1.26719487e+00 -1.77174628e-01 -5.77005804e-01 -2.19542421e-02
4.44525838e-01 -1.22298026e+00 -1.32516503e+00 6.38691604e-01
1.27741587e+00 -7.35698998e-01 1.64455724e+00 -1.29146412e-01
5.55692255e-01 -5.99892259e-01 3.79081480e-02 -9.82013702e-01
-2.13165015e-01 -6.83852255e-01 -7.39712596e-01 9.98360574e-01
-4.38183784e-01 -2.42876276e-01 9.89872038e-01 5.57743192e-01
-9.35495645e-02 -9.13155794e-01 -1.02916467e+00 -6.12473011e-01
4.90834052e-03 -2.74074078e-01 3.89611185e-01 8.03107560e-01
-3.58635217e-01 -2.28684917e-02 -9.44532633e-01 1.48103103e-01
1.12766874e+00 4.83724147e-01 1.22986805e+00 -1.08574688e+00
-2.92591870e-01 -5.76824509e-02 -1.95663422e-01 -1.24396861e+00
-1.11615427e-01 -5.28399825e-01 1.06075309e-01 -1.63209021e+00
3.87444568e-04 -1.34839758e-01 4.30236876e-01 6.47342861e-01
-9.31107923e-02 6.89982116e-01 2.46441990e-01 3.83606017e-01
-1.21804483e-01 7.21310854e-01 1.91555679e+00 -5.28993346e-02
-4.65895087e-02 -1.98207498e-01 -2.60864556e-01 1.08465075e+00
2.70972252e-01 -1.74631953e-01 -6.36795685e-02 -3.26434582e-01
2.43193999e-01 4.81805652e-01 1.01332903e+00 -9.94268119e-01
8.58545490e-03 -1.28129438e-01 8.44397366e-01 -8.54824305e-01
8.50296080e-01 -8.08890939e-01 9.38540578e-01 7.72051990e-01
2.55374134e-01 -2.42221262e-02 4.03061248e-02 5.67551970e-01
5.40564507e-02 3.14261734e-01 7.08749413e-01 -3.38865191e-01
-5.30853927e-01 5.28613746e-01 1.36104226e-01 1.03392065e-01
7.31999457e-01 -6.94826245e-01 1.56094134e-01 -5.08470953e-01
-7.68832207e-01 2.80382782e-01 1.01294851e+00 3.04926783e-01
8.91532958e-01 -1.58075666e+00 -7.56949663e-01 1.28307104e-01
3.02378330e-02 5.52201748e-01 5.31579018e-01 1.15063286e+00
-8.94189358e-01 -4.93556075e-02 -1.92355946e-01 -8.68120670e-01
-1.55462039e+00 2.00702339e-01 5.31900942e-01 -2.18729928e-01
-1.04501796e+00 4.20870185e-01 3.50614101e-01 -6.94179833e-01
7.84900114e-02 -4.09898788e-01 3.33340287e-01 -1.36182487e-01
1.59159839e-01 6.08455062e-01 -2.58431703e-01 -1.15150106e+00
-3.55587006e-01 1.52129734e+00 5.89490056e-01 2.28036251e-02
1.07991672e+00 -2.48879477e-01 1.85966492e-03 2.80676764e-02
9.42263484e-01 2.18426108e-01 -1.54065609e+00 -9.07680467e-02
-7.69280791e-01 -5.84465325e-01 -2.20060512e-01 -7.40919530e-01
-1.19656444e+00 8.25311363e-01 3.60726386e-01 -8.59314919e-01
9.50277686e-01 -8.65830034e-02 1.22203982e+00 -1.12159438e-01
5.15866637e-01 -9.38922822e-01 1.02774277e-01 1.13819592e-01
1.33459759e+00 -8.49237084e-01 7.04017460e-01 -1.01290751e+00
-4.28686529e-01 1.04027772e+00 7.17426240e-01 -1.01952218e-01
3.90967369e-01 1.33218810e-01 4.26416174e-02 -3.57473165e-01
-1.24693833e-01 1.08822115e-01 7.36578345e-01 5.78547776e-01
2.25616872e-01 3.05152889e-02 -6.66806027e-02 5.88865221e-01
-2.61611015e-01 8.07083473e-02 -1.44633308e-01 9.83720839e-01
-4.22546118e-02 -1.01709306e+00 -1.00960040e+00 -1.01950737e-02
-3.06560487e-01 1.74798802e-01 -3.14622372e-01 1.07947528e+00
1.16110668e-01 1.91081628e-01 -3.06537062e-01 -2.20697239e-01
6.11927986e-01 -2.80021094e-02 8.61259162e-01 -2.75510073e-01
-3.56001526e-01 5.05172551e-01 1.27023369e-01 -1.03741324e+00
-4.33052570e-01 -4.51805234e-01 -1.31163979e+00 -4.24059480e-01
-1.22395590e-01 -4.02802587e-01 3.99484873e-01 4.72031116e-01
3.19588542e-01 3.85209560e-01 -9.42619964e-02 -1.61921024e+00
-6.00518644e-01 -5.23779511e-01 -6.63388789e-01 6.31755888e-01
3.53571296e-01 -1.11732411e+00 -2.77549010e-02 3.12289536e-01] | [7.080688953399658, -1.1747326850891113] |
e3dc7503-db4f-404c-b9e5-b5956dff3942 | what-s-behind-the-mask-estimating-uncertainty | 2211.15211 | null | https://arxiv.org/abs/2211.15211v1 | https://arxiv.org/pdf/2211.15211v1.pdf | What's Behind the Mask: Estimating Uncertainty in Image-to-Image Problems | Estimating uncertainty in image-to-image networks is an important task, particularly as such networks are being increasingly deployed in the biological and medical imaging realms. In this paper, we introduce a new approach to this problem based on masking. Given an existing image-to-image network, our approach computes a mask such that the distance between the masked reconstructed image and the masked true image is guaranteed to be less than a specified threshold, with high probability. The mask thus identifies the more certain regions of the reconstructed image. Our approach is agnostic to the underlying image-to-image network, and only requires triples of the input (degraded), reconstructed and true images for training. Furthermore, our method is agnostic to the distance metric used. As a result, one can use $L_p$-style distances or perceptual distances like LPIPS, which contrasts with interval-based approaches to uncertainty. Our theoretical guarantees derive from a conformal calibration procedure. We evaluate our mask-based approach to uncertainty on image colorization, image completion, and super-resolution tasks, demonstrating high quality performance on each. | ['Daniel Freedman', 'Michael Elad', 'Regev Cohen', 'Gilad Kutiel'] | 2022-11-28 | null | null | null | null | ['colorization'] | ['computer-vision'] | [ 7.15374172e-01 5.55813134e-01 4.71260659e-02 -5.33990204e-01
-1.04889989e+00 -5.18604159e-01 3.70611757e-01 -2.44749226e-02
-4.53330606e-01 7.72080898e-01 -1.19256303e-01 -1.92008503e-02
-2.46878102e-01 -6.74266398e-01 -8.46335769e-01 -6.37346745e-01
-2.61572421e-01 3.59022260e-01 2.65596300e-01 8.65715817e-02
1.21971905e-01 5.40633857e-01 -1.34393060e+00 2.60176212e-01
5.86420536e-01 1.30093229e+00 1.60741601e-02 9.25776184e-01
1.00972682e-01 4.50325012e-01 -5.72409809e-01 -3.74116987e-01
4.97207254e-01 -5.67991436e-01 -6.27794981e-01 2.56401777e-01
4.73880649e-01 -1.83204696e-01 -1.48031265e-01 1.46907628e+00
3.14950854e-01 -2.59054806e-02 6.36689961e-01 -1.04585159e+00
-6.29807889e-01 7.04619169e-01 -7.39367783e-01 1.06447816e-01
7.27222264e-02 -9.32477340e-02 6.95283771e-01 -8.81396770e-01
6.01213515e-01 1.16138792e+00 5.90104222e-01 6.49010479e-01
-1.72537386e+00 -4.23612058e-01 -2.17349641e-02 -2.24791050e-01
-1.57019114e+00 -6.25398099e-01 4.99680459e-01 -3.89043361e-01
2.41519839e-01 2.53262818e-01 2.28058159e-01 6.74411058e-01
1.80373654e-01 1.62465885e-01 1.43788970e+00 -5.28800130e-01
4.21638995e-01 4.67817694e-01 -3.33614141e-01 5.77197790e-01
2.45111018e-01 3.37890297e-01 -5.38660049e-01 -3.17983069e-02
1.18568349e+00 -3.05545181e-01 -5.21505594e-01 -4.55198497e-01
-1.01399708e+00 5.18285990e-01 4.53553259e-01 2.80369282e-01
-4.98384722e-02 4.81393844e-01 -7.25173801e-02 1.50065660e-01
6.22890174e-01 3.67266238e-01 -6.16009720e-02 2.87881523e-01
-1.10695040e+00 -5.59929833e-02 5.83575010e-01 8.17519963e-01
7.56819487e-01 5.11385314e-03 -1.02668770e-01 6.44654334e-01
9.48343799e-02 2.07389340e-01 -1.05984420e-01 -1.52935123e+00
4.83110398e-02 4.43782285e-02 3.39419365e-01 -1.08136773e+00
-3.01098257e-01 -5.37070811e-01 -8.94780874e-01 9.70192313e-01
6.64029717e-01 1.08103991e-01 -9.92246747e-01 2.07633305e+00
9.33522955e-02 1.04821891e-01 4.22972851e-02 8.22313249e-01
2.27575123e-01 2.37927541e-01 -3.14445466e-01 -3.82237464e-01
1.20833397e+00 -2.73260444e-01 -6.19999707e-01 -1.74931139e-01
-2.02355549e-01 -5.50630450e-01 9.55645025e-01 5.87092638e-01
-1.48733675e+00 -4.99874860e-01 -1.36326897e+00 7.75692761e-02
-6.60973713e-02 -1.65961772e-01 9.27667394e-02 8.74894857e-01
-1.47297752e+00 9.78480995e-01 -5.99939823e-01 -1.78503335e-01
5.01216173e-01 4.11937207e-01 -3.35209727e-01 -1.41478330e-01
-8.79553199e-01 9.19489861e-01 5.04848659e-01 -2.08336487e-02
-1.01659548e+00 -6.77935839e-01 -7.88745224e-01 -4.75462899e-03
2.14600027e-01 -4.41680908e-01 8.84891510e-01 -1.24595594e+00
-1.27922189e+00 1.09405136e+00 5.05909286e-02 -5.84592342e-01
8.29124331e-01 2.52004683e-01 -3.32452685e-01 4.52191889e-01
2.86760889e-02 9.61590052e-01 1.25228298e+00 -1.88052213e+00
-5.00015557e-01 -2.96179593e-01 6.14718199e-02 1.36690140e-01
-2.00576186e-02 -7.73777887e-02 -6.63590133e-01 -5.85982919e-01
3.33611757e-01 -7.24375904e-01 -3.61555606e-01 6.91957116e-01
-4.20864552e-01 4.61294711e-01 2.88401306e-01 -5.17818511e-01
7.58997202e-01 -2.38855863e+00 1.22654056e-02 5.67773342e-01
5.67676008e-01 -2.70555466e-01 -1.65942580e-01 -8.82426575e-02
-2.20171809e-01 2.45141923e-01 -8.42405200e-01 -5.65730751e-01
-2.06898853e-01 1.68888628e-01 -6.81256428e-02 8.64143431e-01
2.56992728e-01 5.63635290e-01 -7.97344029e-01 -7.92386889e-01
2.08944380e-01 6.27658844e-01 -2.74595678e-01 1.61243692e-01
-1.78031564e-01 5.62463522e-01 1.62354469e-01 3.35354120e-01
9.30874527e-01 -2.79392660e-01 8.37496892e-02 -2.78258920e-01
4.90747206e-02 -4.83148903e-01 -1.22128177e+00 1.70421731e+00
-3.57425779e-01 5.99965990e-01 4.32617605e-01 -6.16167426e-01
7.31761992e-01 3.51604551e-01 6.05031371e-01 -4.50149864e-01
-2.78175343e-02 1.23118177e-01 -1.04317911e-01 -1.93581954e-02
3.11040848e-01 -4.05251056e-01 1.35282502e-01 3.47051084e-01
-1.82782441e-01 -3.85306120e-01 -8.54952261e-02 3.48789096e-01
9.31331217e-01 9.85005498e-02 1.04643375e-01 -4.25296128e-01
2.68220931e-01 -1.35549009e-01 5.20766437e-01 8.51509511e-01
-2.97195703e-01 1.16704667e+00 7.35251069e-01 1.13083236e-01
-1.18499303e+00 -1.44591844e+00 -4.86036837e-01 4.72398758e-01
2.86570758e-01 8.16383883e-02 -1.10155106e+00 -4.42780644e-01
-1.34149253e-01 5.36518872e-01 -9.99605179e-01 7.36916661e-02
-2.81118184e-01 -4.79825199e-01 5.40162802e-01 3.06419790e-01
3.87605458e-01 -7.11894929e-01 -4.76314902e-01 2.18287967e-02
1.90058053e-02 -1.17443764e+00 -6.62052810e-01 2.40399480e-01
-8.29666317e-01 -1.06917107e+00 -7.24066079e-01 -5.00600994e-01
1.06175065e+00 1.68530643e-01 1.17661989e+00 -3.66371050e-02
-3.50705326e-01 4.75420356e-01 1.94867291e-02 -1.02839604e-01
-6.81147993e-01 -6.55211985e-01 -3.68231013e-02 1.91790581e-01
-2.97641128e-01 -6.29654884e-01 -8.04618001e-01 3.24282557e-01
-1.15165377e+00 -9.62850749e-02 5.93527257e-01 8.20289910e-01
8.01128149e-01 4.35950518e-01 2.93226480e-01 -8.83972168e-01
5.02251565e-01 -2.30499372e-01 -8.27990353e-01 2.52571970e-01
-8.14968348e-01 1.96807414e-01 4.01866704e-01 -2.99340963e-01
-9.51057255e-01 2.96378344e-01 2.50138700e-01 -6.23154044e-01
5.43209584e-03 2.93755233e-01 -2.07208887e-01 -3.45548213e-01
1.03315079e+00 -9.15548876e-02 3.01907539e-01 -4.62325402e-02
6.73195899e-01 2.86370635e-01 1.11469400e+00 -5.19193172e-01
9.27307725e-01 9.52351153e-01 1.78828806e-01 -4.56371963e-01
-6.39140844e-01 -1.53976604e-01 -4.99080390e-01 -4.07636374e-01
9.87782300e-01 -6.52734518e-01 -6.77953064e-01 1.10446721e-01
-1.13133025e+00 -1.57364845e-01 -6.61675394e-01 4.36287940e-01
-7.68238842e-01 4.76608247e-01 -5.16671479e-01 -9.74827170e-01
5.72366714e-02 -1.13828087e+00 8.38948369e-01 2.37844456e-02
2.71770544e-02 -8.99368167e-01 -2.10610405e-01 -1.08798314e-02
4.96667564e-01 5.97429812e-01 8.87187898e-01 -1.62672654e-01
-7.81018436e-01 -1.20874166e-01 -6.98621035e-01 6.58455193e-01
1.53139606e-01 -1.91057771e-01 -1.14613771e+00 -3.37719798e-01
4.30623204e-01 -2.73905486e-01 8.91162813e-01 7.57570565e-01
1.35677111e+00 -1.57417610e-01 -5.65832369e-02 6.02893174e-01
1.67056334e+00 -5.47583885e-02 8.23243558e-01 -2.06477754e-03
3.68219763e-01 8.71242106e-01 3.46351653e-01 2.87202924e-01
-4.79232706e-02 6.27361596e-01 7.21097112e-01 -3.04198533e-01
-2.62157738e-01 9.47883818e-03 1.58845872e-01 1.93258941e-01
1.49065107e-01 -1.32261604e-01 -5.00988781e-01 4.37797725e-01
-1.61604667e+00 -8.59388471e-01 1.55491307e-01 2.57399988e+00
9.64393437e-01 3.01309496e-01 -2.09441930e-01 4.54425402e-02
1.02352405e+00 1.34550393e-01 -8.04840028e-01 -1.36921003e-01
-3.14463288e-01 1.95285365e-01 6.92772448e-01 8.04236472e-01
-8.67175758e-01 4.25428510e-01 7.15399218e+00 8.82569432e-01
-7.96848595e-01 1.32796675e-01 1.23914206e+00 -1.31571293e-01
-5.11197448e-01 -1.56679466e-01 -2.58054197e-01 2.37213641e-01
7.74936616e-01 -1.04229361e-01 6.89835548e-01 5.94342649e-01
7.35289678e-02 -4.49652761e-01 -1.42592549e+00 1.10398829e+00
1.19973749e-01 -1.36860824e+00 -2.45368108e-01 1.19869985e-01
7.46416628e-01 -2.29394168e-01 4.54619676e-01 -4.41139936e-01
4.68929857e-01 -1.48981607e+00 7.82704175e-01 5.54138124e-01
1.31512833e+00 -7.60920167e-01 3.76744747e-01 -1.98244937e-02
-9.67595518e-01 2.49444008e-01 -3.66795361e-01 5.30859530e-01
4.75819670e-02 9.90462899e-01 -6.07905626e-01 4.13423181e-01
7.90093482e-01 2.38268495e-01 -3.49692851e-01 7.50142038e-01
3.56030278e-02 1.60139173e-01 -2.18081728e-01 5.67967832e-01
-2.20928192e-01 -1.06672101e-01 4.12798017e-01 1.03083479e+00
3.64391208e-01 -4.52394672e-02 2.36054529e-02 1.26766300e+00
-1.61276057e-01 -2.46740133e-01 -5.57403326e-01 3.13024044e-01
4.23537791e-01 1.28800392e+00 -1.00154603e+00 -2.13673115e-01
-2.09937096e-01 1.10771263e+00 1.57925069e-01 3.64257663e-01
-7.30756402e-01 -2.84175098e-01 5.03285766e-01 2.67828312e-02
1.22730374e-01 -1.12976074e-01 -5.95958054e-01 -7.91261435e-01
1.83653653e-01 -9.11370873e-01 2.93089896e-01 -9.30578232e-01
-1.30354726e+00 8.01061928e-01 2.04011485e-01 -1.27602160e+00
-9.53897536e-02 -6.27045274e-01 -3.22190076e-01 1.15257215e+00
-1.43192434e+00 -7.22439468e-01 -3.36256713e-01 6.28176868e-01
9.84908119e-02 2.42567375e-01 6.34721398e-01 1.86886132e-01
-1.17451392e-01 4.31738615e-01 1.50294706e-01 4.45477851e-02
8.01372111e-01 -1.41022313e+00 2.34742209e-01 1.16001487e+00
8.30820650e-02 3.91818404e-01 8.88826787e-01 -4.22812909e-01
-9.51789081e-01 -9.80846643e-01 2.72776663e-01 -5.83368480e-01
4.88800108e-01 -2.45774090e-01 -8.39053333e-01 4.23995703e-01
1.81158587e-01 3.16973180e-01 3.32867891e-01 -3.18738580e-01
-5.09484470e-01 -2.30339289e-01 -1.63026631e+00 6.36560678e-01
9.27125812e-01 -6.76896572e-01 -3.39822993e-02 2.69288510e-01
7.36089051e-01 -5.12624681e-01 -9.56268668e-01 3.69339317e-01
4.48441923e-01 -1.31838608e+00 9.62099493e-01 -9.08210501e-02
3.84653419e-01 -6.56534672e-01 -4.17449683e-01 -1.19417310e+00
-1.20381825e-01 -6.23862505e-01 1.54320523e-01 9.53884900e-01
5.07622004e-01 -5.37607491e-01 9.20933604e-01 8.21809351e-01
-8.22394490e-02 -4.73627806e-01 -1.25214255e+00 -8.71014118e-01
1.10397581e-02 -6.46370530e-01 3.60226482e-01 6.49947643e-01
-2.02436879e-01 -3.61102730e-01 -5.09896398e-01 4.38466311e-01
1.22783601e+00 -2.20073506e-01 2.81698883e-01 -1.03935230e+00
-6.13329947e-01 -5.09790719e-01 -4.17023480e-01 -6.67018890e-01
-1.50353387e-01 -7.54787266e-01 5.25932372e-01 -1.21845484e+00
2.64060438e-01 -5.31663358e-01 -4.95456100e-01 3.10107060e-02
3.22669968e-02 6.49112403e-01 -1.66971274e-02 1.41131401e-01
-2.09197626e-01 7.95848817e-02 1.05899608e+00 -1.16417773e-01
6.63912967e-02 -1.87683657e-01 -7.99321353e-01 6.32853270e-01
8.28258097e-01 -6.30906463e-01 -4.39776421e-01 -3.32537353e-01
2.22429350e-01 2.99335092e-01 6.20310664e-01 -9.82467651e-01
2.28153929e-01 -6.27584308e-02 6.77260876e-01 -1.64776072e-01
6.07722700e-01 -8.85028362e-01 3.80411118e-01 4.02639627e-01
-5.94328344e-01 -6.79038987e-02 1.15339398e-01 7.47603118e-01
4.85992432e-03 -2.87603199e-01 1.24939489e+00 -3.64000946e-01
-2.88263351e-01 2.83805072e-01 2.63017393e-03 4.56027910e-02
9.20038879e-01 -3.12292665e-01 -3.46190393e-01 -7.14207947e-01
-8.35637927e-01 -5.22201806e-02 7.14035630e-01 -7.33843744e-02
8.59777868e-01 -1.09091008e+00 -7.40832686e-01 1.10264100e-01
1.72433317e-01 5.19784912e-02 1.46187142e-01 6.64015055e-01
-4.87179905e-01 -1.83620796e-04 -9.76632237e-02 -8.68488789e-01
-1.12825990e+00 7.07097828e-01 5.74849129e-01 1.03733633e-02
-4.42367524e-01 8.31554532e-01 3.92673641e-01 -2.75047347e-02
3.84736270e-01 -2.29208067e-01 2.80827343e-01 -3.58283222e-01
5.78109443e-01 2.01128989e-01 -2.62255222e-01 -4.36646014e-01
-1.96555197e-01 6.78758740e-01 7.34172612e-02 -9.11060870e-01
9.20999110e-01 -3.02952915e-01 -3.57236832e-01 4.24230635e-01
1.10627389e+00 1.42939657e-01 -1.79380071e+00 -3.03689390e-01
4.12990935e-02 -6.60624564e-01 5.48364483e-02 -8.02830994e-01
-1.23746264e+00 6.74035311e-01 9.10894871e-01 2.86758214e-01
1.26946282e+00 2.16838762e-01 8.65936503e-02 6.58305734e-02
3.06286603e-01 -1.11479199e+00 1.45583495e-01 -2.40360305e-01
1.06330836e+00 -1.35384309e+00 2.69144356e-01 -3.83369684e-01
-5.39596140e-01 9.24715042e-01 4.05551106e-01 -4.12846021e-02
7.39549100e-01 4.79442686e-01 2.39899576e-01 1.19121019e-02
-4.12716776e-01 -7.01079294e-02 2.98836738e-01 8.12958837e-01
1.55199826e-01 -1.01895556e-01 1.29192948e-01 3.12043913e-02
4.17204201e-02 -3.13444771e-02 6.91581011e-01 6.70571804e-01
-4.73806262e-01 -1.00176895e+00 -6.61264181e-01 2.56175309e-01
-4.65756506e-01 -1.95662260e-01 -8.77552107e-02 5.95154405e-01
1.12541504e-01 1.02590132e+00 1.29015401e-01 -1.67512283e-01
8.79131630e-03 -3.21633995e-01 7.40621269e-01 -5.57676971e-01
-1.52764603e-01 1.57650352e-01 -6.93425983e-02 -6.56529725e-01
-6.04417443e-01 -4.84229743e-01 -1.23803926e+00 -1.97029963e-01
-2.93663263e-01 1.34245649e-01 7.92863846e-01 5.01477718e-01
1.82699114e-01 3.78107667e-01 7.85278440e-01 -8.74970734e-01
-4.33444858e-01 -4.91108954e-01 -7.84020126e-01 2.36026227e-01
4.43395346e-01 -2.79173940e-01 -5.19516587e-01 1.05903588e-01] | [11.572304725646973, -1.776789665222168] |
ec433fd1-9f07-48c9-b167-f5618fcd1b3b | verbal-and-nonverbal-clues-for-real-life | null | null | https://aclanthology.org/D15-1281 | https://aclanthology.org/D15-1281.pdf | Verbal and Nonverbal Clues for Real-life Deception Detection | null | ["Ver{\\'o}nica P{\\'e}rez-Rosas", 'Mohamed Abouelenien', 'Mihai Burzo', 'Yao Xiao', 'Rada Mihalcea', 'CJ Linton'] | 2015-09-01 | null | null | null | emnlp-2015-9 | ['deception-detection'] | ['miscellaneous'] | [-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] | [-7.398003101348877, 3.6464171409606934] |
54b2f87a-061c-4d7e-864e-cabd82c772d6 | dkm-differentiable-k-means-clustering-layer | 2108.12659 | null | https://arxiv.org/abs/2108.12659v4 | https://arxiv.org/pdf/2108.12659v4.pdf | DKM: Differentiable K-Means Clustering Layer for Neural Network Compression | Deep neural network (DNN) model compression for efficient on-device inference is becoming increasingly important to reduce memory requirements and keep user data on-device. To this end, we propose a novel differentiable k-means clustering layer (DKM) and its application to train-time weight clustering-based DNN model compression. DKM casts k-means clustering as an attention problem and enables joint optimization of the DNN parameters and clustering centroids. Unlike prior works that rely on additional regularizers and parameters, DKM-based compression keeps the original loss function and model architecture fixed. We evaluated DKM-based compression on various DNN models for computer vision and natural language processing (NLP) tasks. Our results demonstrate that DKM delivers superior compression and accuracy trade-off on ImageNet1k and GLUE benchmarks. For example, DKM-based compression can offer 74.5% top-1 ImageNet1k accuracy on ResNet50 DNN model with 3.3MB model size (29.4x model compression factor). For MobileNet-v1, which is a challenging DNN to compress, DKM delivers 63.9% top-1 ImageNet1k accuracy with 0.72 MB model size (22.4x model compression factor). This result is 6.8% higher top-1accuracy and 33% relatively smaller model size than the current state-of-the-art DNN compression algorithms. Additionally, DKM enables compression of DistilBERT model by 11.8x with minimal (1.1%) accuracy loss on GLUE NLP benchmarks. | ['Mohammad Rastegari', 'Saurabh Adya', 'Keivan A. Vahid', 'Minsik Cho'] | 2021-08-28 | dkm-differentiable-k-means-clustering-layer-1 | https://openreview.net/forum?id=J_F_qqCE3Z5 | https://openreview.net/pdf?id=J_F_qqCE3Z5 | iclr-2022-4 | ['neural-network-compression', 'neural-network-compression'] | ['methodology', 'miscellaneous'] | [-5.31964861e-02 -2.30364930e-02 -3.44803333e-01 -6.18039370e-01
-7.58170068e-01 -2.32615113e-01 4.14167136e-01 -8.81825536e-02
-1.18896043e+00 3.77386272e-01 -5.89559227e-03 -6.10039473e-01
-1.26409560e-01 -6.29788220e-01 -1.09069347e+00 -4.90990549e-01
1.37161300e-01 7.71798432e-01 -7.23869400e-03 2.35359907e-01
-7.57026002e-02 4.13282871e-01 -1.44127858e+00 2.12574691e-01
6.25341058e-01 1.36213052e+00 6.68886662e-01 9.28920686e-01
6.22442402e-02 8.66221964e-01 -6.35177493e-01 -6.34798586e-01
4.42721665e-01 1.63319111e-01 -7.79429734e-01 -5.21381140e-01
9.26654935e-01 -8.53939056e-01 -8.71984303e-01 1.20825005e+00
6.09272957e-01 1.13777436e-01 6.49606049e-01 -1.27287257e+00
-4.09805804e-01 8.86252701e-01 -3.84538144e-01 1.94477364e-01
-6.67976201e-01 1.93740219e-01 7.96469629e-01 -5.83734393e-01
2.96105802e-01 1.35214150e+00 7.87143707e-01 1.04371154e+00
-1.39651775e+00 -1.19089067e+00 -1.78291664e-01 3.79541725e-01
-1.56475306e+00 -7.81674385e-01 4.46083099e-02 -4.23580175e-03
1.56344783e+00 1.16236165e-01 7.22975135e-01 8.20262134e-01
1.65194571e-01 8.00833583e-01 3.05794209e-01 -5.67242689e-02
4.23747987e-01 -3.05964649e-01 1.59771547e-01 5.79166114e-01
2.69148111e-01 -1.18498720e-01 -4.52358454e-01 1.40374437e-01
6.60047591e-01 2.25476995e-01 2.33845357e-02 3.56797099e-01
-7.78870165e-01 6.12789273e-01 5.71200669e-01 -1.76524594e-01
-2.69163698e-01 1.11641908e+00 6.10733211e-01 2.61126101e-01
3.05090249e-01 9.96316597e-02 -6.60314977e-01 -8.62414300e-01
-1.44985938e+00 4.64630365e-01 6.37767136e-01 1.31006479e+00
3.86853069e-01 2.09376648e-01 1.11373588e-01 1.21022570e+00
5.01349926e-01 7.45849013e-01 6.28256083e-01 -1.37637508e+00
8.27978909e-01 3.47533911e-01 -6.10677719e-01 -6.07993901e-01
-2.12854400e-01 -3.42423737e-01 -1.33696115e+00 4.17617559e-02
1.30019352e-01 -6.52278066e-02 -1.24402487e+00 1.85595143e+00
-1.34101838e-01 3.28265995e-01 2.12518334e-01 6.86885178e-01
8.47839475e-01 8.31160367e-01 1.74206167e-01 2.26464644e-01
1.26538682e+00 -1.00887942e+00 -4.66262311e-01 -2.25797921e-01
9.50849354e-01 -5.55690050e-01 9.86841798e-01 3.41180444e-01
-1.43567038e+00 -5.19622028e-01 -9.76456761e-01 -5.20901084e-01
3.09273135e-03 5.90849086e-04 3.37259620e-01 4.42630976e-01
-1.42766809e+00 7.43975699e-01 -1.25887239e+00 -7.74184242e-03
9.98053193e-01 7.34298706e-01 -1.08950645e-01 -5.60977519e-01
-7.71900058e-01 7.38753498e-01 5.92502415e-01 -2.46737361e-01
-1.10542858e+00 -1.56775641e+00 -5.67989647e-01 4.07942832e-01
-1.81999207e-01 -7.09499478e-01 1.57498991e+00 -3.87390554e-01
-1.35984004e+00 7.41242349e-01 -1.23169325e-01 -1.38145268e+00
4.88983572e-01 -3.08331996e-01 -3.17541927e-01 1.98356465e-01
-4.14135456e-01 1.44588447e+00 5.23569584e-01 -6.07301474e-01
-7.14540780e-01 -2.26766661e-01 -2.44649544e-01 3.33266743e-02
-6.73584819e-01 -1.95772544e-01 -9.40895081e-01 -6.33381546e-01
-2.63380967e-02 -9.69204903e-01 -1.47115454e-01 3.59275550e-01
-1.83924198e-01 -2.18992770e-01 1.04264951e+00 -7.33892560e-01
1.26891959e+00 -2.05117464e+00 -1.75270885e-01 1.05603017e-01
7.08802223e-01 8.48095179e-01 -2.00381204e-01 -1.97021827e-01
3.13143760e-01 2.42128015e-01 -2.47243360e-01 -9.69583333e-01
2.44767025e-01 6.52199149e-01 -8.95588938e-03 2.07890838e-01
-2.20621318e-01 1.21504712e+00 -4.13299829e-01 -4.51016039e-01
3.39782387e-01 1.00361514e+00 -1.15463734e+00 -1.31000593e-01
-3.09343666e-01 -3.96882415e-01 2.87548602e-01 5.47009051e-01
9.09870744e-01 -1.44658491e-01 7.08937347e-02 -4.71061736e-01
3.36653650e-01 3.73868406e-01 -5.41987479e-01 1.92150104e+00
-5.92054069e-01 1.14182401e+00 2.03034822e-02 -9.26894367e-01
6.99211359e-01 1.11623757e-01 3.74545932e-01 -9.01142359e-01
2.47178495e-01 1.24675460e-01 6.84667975e-02 4.86251079e-02
5.87225914e-01 2.44270861e-01 1.91457644e-01 8.92500877e-02
4.89953309e-01 1.07680291e-01 1.97626621e-01 3.17201972e-01
1.30836535e+00 -7.02046573e-01 -2.41162330e-01 -4.34383601e-01
-1.45211443e-01 -2.59853005e-01 5.09688795e-01 6.85364604e-01
-1.80486590e-01 4.72545594e-01 2.36447141e-01 -4.28672433e-01
-1.43721545e+00 -1.03549254e+00 -4.45130706e-01 8.41549575e-01
-2.36073926e-01 -7.24964797e-01 -1.08566511e+00 -2.98796415e-01
1.04972526e-01 7.57236660e-01 -5.62803634e-02 -3.35295528e-01
-5.10772407e-01 -8.02936196e-01 1.07787514e+00 6.93842232e-01
1.03201044e+00 -6.89183176e-01 -5.21114171e-01 1.63073942e-01
-1.69262514e-01 -1.35902286e+00 -5.08405626e-01 1.47828192e-01
-1.38315547e+00 -7.00383484e-01 -6.43946052e-01 -8.46350193e-01
5.25999129e-01 4.49914075e-02 1.39030743e+00 1.90009959e-02
-1.68361425e-01 8.56878534e-02 -5.38764708e-02 -3.79375249e-01
-5.12719691e-01 4.65759128e-01 2.38163963e-01 -5.93047261e-01
7.38567054e-01 -9.50636864e-01 -7.58929670e-01 7.67355785e-02
-9.77547884e-01 3.56039435e-01 5.95544398e-01 5.66821456e-01
9.66093779e-01 1.72302291e-01 1.97188795e-01 -5.88136017e-01
3.78760725e-01 -3.20204198e-01 -7.96653092e-01 1.44101426e-01
-9.46907222e-01 1.48679480e-01 8.14760327e-01 -6.02477908e-01
-4.79038775e-01 -1.52667210e-01 -4.73534405e-01 -1.12325156e+00
8.12391862e-02 2.77432233e-01 -2.11426854e-01 1.57676153e-02
5.07756710e-01 2.82130927e-01 1.58948109e-01 -6.97404504e-01
2.39047766e-01 7.15105355e-01 9.44737196e-01 -4.65962738e-01
3.07420015e-01 2.21508414e-01 -1.02634072e-01 -7.63419986e-01
-3.29489142e-01 -2.45035440e-01 -1.62048817e-01 4.38509025e-02
8.81964862e-01 -1.19314492e+00 -1.06336689e+00 6.05258048e-01
-1.05333769e+00 -9.46805537e-01 -3.28716487e-01 6.29473269e-01
-4.05536681e-01 1.20893762e-01 -1.07085991e+00 -4.17923450e-01
-1.05228972e+00 -1.19739234e+00 7.65506923e-01 -1.29446611e-01
4.91628312e-02 -8.22442889e-01 -4.79101062e-01 3.74743760e-01
6.42024934e-01 -2.30379701e-01 8.73823822e-01 -5.00453234e-01
-4.58812684e-01 -1.61829636e-01 -6.25225961e-01 8.32303584e-01
-4.07025188e-01 -2.04657286e-01 -9.58426416e-01 -5.02156079e-01
8.82368907e-03 -3.20776969e-01 1.18104303e+00 6.93274200e-01
1.72560155e+00 -6.71645999e-01 -1.44091681e-01 1.29933465e+00
1.55546451e+00 2.71372586e-01 8.26799750e-01 2.93287989e-02
1.04098260e+00 -3.20092857e-01 -5.70714884e-02 4.12504494e-01
4.55966204e-01 6.79413378e-01 6.40715063e-01 1.81432307e-01
-4.55566257e-01 -3.24879289e-01 3.04094911e-01 1.39353311e+00
4.90086563e-02 -3.84096891e-01 -1.06334639e+00 4.00511950e-01
-1.68724787e+00 -6.89390719e-01 2.73780435e-01 1.90519965e+00
1.00267065e+00 3.81649703e-01 -1.23911537e-01 9.27656367e-02
4.50287282e-01 -8.31617489e-02 -1.04853797e+00 -3.13544959e-01
-1.42189665e-02 2.99089372e-01 1.15838861e+00 5.94181955e-01
-9.07695532e-01 8.96200538e-01 5.23493099e+00 1.45610392e+00
-8.67655277e-01 2.50570595e-01 8.59635949e-01 -8.13965023e-01
-7.51563236e-02 -4.39126611e-01 -1.34934282e+00 9.05449688e-01
1.65551353e+00 -9.44057107e-02 8.29592049e-01 1.03884327e+00
5.82469851e-02 3.16133261e-01 -1.25134337e+00 1.75586581e+00
-2.46604428e-01 -1.71200705e+00 4.14360166e-01 4.18143123e-01
6.63138330e-01 8.01222384e-01 8.94581378e-02 4.35984045e-01
4.13920194e-01 -1.16637325e+00 7.40856707e-01 4.12300736e-01
1.27450848e+00 -1.23994219e+00 6.53515637e-01 4.08475637e-01
-1.05926490e+00 -3.36717851e-02 -9.43403184e-01 1.28547758e-01
1.79402724e-01 6.57800496e-01 -8.05650413e-01 -2.28256598e-01
1.13177168e+00 8.62995446e-01 -3.00077349e-01 7.79932201e-01
3.95231158e-01 9.75684822e-01 -7.94010043e-01 1.38850912e-01
3.70782971e-01 -5.38837947e-02 1.10694841e-01 1.42118788e+00
2.76027352e-01 1.43312007e-01 -3.01743776e-01 6.68229938e-01
-9.84753430e-01 -3.75797927e-01 -1.88880533e-01 2.05399483e-01
9.48996484e-01 7.84828544e-01 -4.43535715e-01 -7.04002798e-01
-5.01862727e-02 9.80576694e-01 4.54832494e-01 2.09862202e-01
-1.05458045e+00 -2.20578268e-01 1.37473142e+00 1.78937033e-01
3.36972535e-01 -2.21541151e-01 -3.46594274e-01 -9.11144674e-01
3.05344071e-02 -7.35354066e-01 9.83285382e-02 -5.12948453e-01
-9.42026079e-01 5.55918813e-01 9.81829092e-02 -9.31190550e-01
-6.26310110e-02 -5.32023370e-01 -8.04388747e-02 4.94408339e-01
-1.48369551e+00 -7.80108213e-01 -2.40970641e-01 4.85779226e-01
7.28300929e-01 -4.83140945e-01 8.01525176e-01 1.02954328e+00
-5.60220003e-01 1.34679198e+00 6.51410580e-01 2.42530078e-01
2.71457970e-01 -9.54573989e-01 9.87480164e-01 7.01100707e-01
4.87960987e-02 4.02273983e-01 2.18320251e-01 -3.66085112e-01
-1.38821888e+00 -1.61234963e+00 9.60012674e-01 -8.35329443e-02
2.87110984e-01 -5.07829368e-01 -7.20609605e-01 6.07134521e-01
-1.38802618e-01 1.35246649e-01 5.11737943e-01 -4.08769846e-01
-4.83415842e-01 -5.24353981e-01 -1.40960622e+00 6.80535674e-01
1.34843493e+00 -4.51304138e-01 2.13639531e-02 4.47906494e-01
1.19030523e+00 -5.34737885e-01 -1.16475880e+00 3.49568665e-01
5.57074726e-01 -6.45417213e-01 1.20916903e+00 -5.12842774e-01
6.49976373e-01 1.41549379e-01 -7.01609433e-01 -9.15397465e-01
-1.81255624e-01 -4.52695251e-01 -7.66238987e-01 1.13342094e+00
3.36563259e-01 -4.04080927e-01 1.14487243e+00 1.14175880e+00
-3.89474064e-01 -1.01661754e+00 -1.23161685e+00 -8.80264044e-01
1.55340478e-01 -1.06601167e+00 6.37280047e-01 5.79221964e-01
-5.93020260e-01 8.14143792e-02 -3.05072546e-01 -1.70709446e-01
8.33785474e-01 -7.48668373e-01 6.00399971e-01 -1.04848468e+00
-2.23070934e-01 -7.81470299e-01 -7.58611441e-01 -1.59862280e+00
3.01621765e-01 -1.21447706e+00 -3.00802737e-01 -1.41713750e+00
1.49355963e-01 -6.45851135e-01 -3.26459348e-01 5.68056524e-01
4.75001425e-01 4.27179664e-01 6.82745397e-01 3.87865156e-01
-7.75265396e-01 5.81010342e-01 7.00879216e-01 -3.54484081e-01
-6.27321973e-02 -2.14560255e-01 -4.23837125e-01 6.01017892e-01
1.01147687e+00 -5.74776232e-01 -6.76178098e-01 -1.24382353e+00
-2.15762362e-01 -3.25942129e-01 3.80122989e-01 -1.49949408e+00
6.88687384e-01 3.44766647e-01 2.96961367e-01 -7.78906465e-01
5.82543254e-01 -9.47957158e-01 1.97040707e-01 6.81068540e-01
-3.33226770e-01 2.18116343e-01 5.02417266e-01 4.05774623e-01
-8.59784260e-02 -5.02139628e-02 1.06373549e+00 3.42528969e-02
-7.00213790e-01 7.10020065e-01 -1.65174067e-01 3.62650663e-01
6.27823293e-01 -2.10799664e-01 -3.05785745e-01 -1.85947075e-01
-3.83054405e-01 2.97333002e-01 2.46561572e-01 1.91700414e-01
9.27798390e-01 -1.41093016e+00 -6.71646237e-01 2.56996244e-01
-3.80164832e-01 6.10865891e-01 4.46153760e-01 4.65246707e-01
-1.10208607e+00 4.95634973e-01 -3.30383964e-02 -7.22756088e-01
-1.40530348e+00 9.36910138e-02 3.92355144e-01 -2.03942984e-01
-7.87288427e-01 1.14013231e+00 -1.20719515e-01 -3.13403726e-01
8.35496783e-01 -8.57047021e-01 3.73495132e-01 -3.85785311e-01
7.82837629e-01 6.89267993e-01 3.46805364e-01 -3.27869445e-01
-2.34361380e-01 2.23849654e-01 -3.70718926e-01 -3.36620305e-03
1.52383983e+00 1.47179335e-01 7.74799660e-02 -3.69427092e-02
1.92020309e+00 -9.39536214e-01 -1.69274664e+00 -3.91251922e-01
-2.91607440e-01 -1.99148282e-01 5.36019623e-01 -6.19967818e-01
-1.59407389e+00 1.00601172e+00 8.50885153e-01 -4.84599382e-01
1.20940602e+00 -9.11317468e-02 1.48468995e+00 7.58371592e-01
1.88903734e-01 -1.17756355e+00 -3.54249150e-01 7.61579633e-01
5.30555487e-01 -1.00739932e+00 9.67887137e-03 2.76438683e-01
-2.32624575e-01 8.00743759e-01 6.18258715e-01 -8.11726302e-02
1.00448251e+00 5.87038994e-01 -1.76163346e-01 -7.45580867e-02
-1.04956651e+00 4.89463955e-01 9.04705524e-02 4.49014395e-01
-9.25104134e-03 2.57029951e-01 2.71473408e-01 5.45059264e-01
-8.31940889e-01 3.11732106e-02 -6.88018277e-02 5.24879813e-01
-3.29027802e-01 -6.61362827e-01 1.62021786e-01 1.06631684e+00
-5.68830848e-01 -6.06594026e-01 2.74764180e-01 5.20436287e-01
-4.64724898e-02 7.23063648e-01 6.86431587e-01 -8.69344294e-01
2.52692908e-01 -1.23515561e-01 2.24836141e-01 -5.24718575e-02
-4.77853924e-01 -2.66043335e-01 -2.70255566e-01 -6.56615257e-01
-1.25530526e-01 -2.21333116e-01 -1.40558851e+00 -1.40045977e+00
-3.97356153e-02 -2.84739703e-01 1.10541165e+00 6.00442231e-01
7.72895157e-01 4.82760996e-01 -1.19300865e-01 -8.93311083e-01
-5.49405694e-01 -1.01264846e+00 -3.51267815e-01 2.40654156e-01
3.02027673e-01 -7.21796080e-02 -9.55756530e-02 1.64045572e-01] | [8.580625534057617, 3.069040536880493] |
29a39a38-6550-41d8-981e-9864fbd88d46 | safe-collaborative-filtering | 2306.05292 | null | https://arxiv.org/abs/2306.05292v1 | https://arxiv.org/pdf/2306.05292v1.pdf | Safe Collaborative Filtering | Excellent tail performance is crucial for modern machine learning tasks, such as algorithmic fairness, class imbalance, and risk-sensitive decision making, as it ensures the effective handling of challenging samples within a dataset. Tail performance is also a vital determinant of success for personalised recommender systems to reduce the risk of losing users with low satisfaction. This study introduces a "safe" collaborative filtering method that prioritises recommendation quality for less-satisfied users rather than focusing on the average performance. Our approach minimises the conditional value at risk (CVaR), which represents the average risk over the tails of users' loss. To overcome computational challenges for web-scale recommender systems, we develop a robust yet practical algorithm that extends the most scalable method, implicit alternating least squares (iALS). Empirical evaluation on real-world datasets demonstrates the excellent tail performance of our approach while maintaining competitive computational efficiency. | ['Tetsuro Morimura', 'Naoto Ohsaka', 'Tatsushi Oka', 'Riku Togashi'] | 2023-06-08 | null | null | null | null | ['collaborative-filtering'] | ['miscellaneous'] | [ 9.48046520e-02 -1.66004390e-01 -5.14046550e-01 -6.88777506e-01
-7.11532533e-01 -2.58573145e-01 9.16585997e-02 4.48431283e-01
-6.37111723e-01 7.10288882e-01 1.80930436e-01 -6.40218139e-01
-6.02250278e-01 -6.50627136e-01 -9.01575536e-02 -4.97517407e-01
-3.05422433e-02 3.65987092e-01 -1.27850771e-01 -1.27948180e-01
6.11448526e-01 2.02559903e-01 -1.25543857e+00 9.73543599e-02
1.24877656e+00 1.20524395e+00 -5.57891488e-01 3.30127895e-01
1.09442323e-01 6.48950517e-01 -2.10832462e-01 -1.06084514e+00
4.11295503e-01 -2.84415781e-01 -2.22506866e-01 -4.32148427e-01
4.88789171e-01 -2.29010716e-01 2.83577889e-01 7.98151016e-01
6.51398063e-01 5.05180776e-01 7.97260106e-01 -1.30728006e+00
-3.89201850e-01 4.94136840e-01 -6.13278389e-01 2.49541044e-01
1.47735417e-01 -2.22937584e-01 1.46679866e+00 -5.77814102e-01
-2.41483822e-02 7.82377958e-01 8.22928846e-01 4.42468941e-01
-1.60758579e+00 -9.14884388e-01 2.07517147e-01 -7.57399723e-02
-1.27354944e+00 -6.33936465e-01 2.20141366e-01 -4.05993551e-01
3.64475936e-01 7.49113023e-01 3.38794559e-01 3.45984280e-01
4.72402215e-01 4.64917362e-01 9.30042803e-01 -7.58570433e-02
5.29271483e-01 5.67880213e-01 2.02906236e-01 1.91925004e-01
4.39290076e-01 1.27927586e-01 -5.37197232e-01 -7.71031439e-01
2.27474749e-01 3.14443469e-01 5.94543060e-03 -5.51739693e-01
-4.15454656e-01 1.03691304e+00 2.15038154e-02 -2.77478814e-01
-4.70773965e-01 -2.30284601e-01 6.08622804e-02 3.72044533e-01
6.65269315e-01 2.28168786e-01 -5.77164114e-01 -9.40747485e-02
-1.30595064e+00 3.42600137e-01 1.11043012e+00 4.93106097e-01
2.29173243e-01 -2.41963491e-01 -5.04285276e-01 9.41732168e-01
4.75509822e-01 5.11955798e-01 2.40187034e-01 -1.12383831e+00
1.20179027e-01 4.66944903e-01 4.55722123e-01 -8.26650977e-01
-1.78564444e-01 -9.72910941e-01 -7.82811224e-01 2.71749228e-01
7.02477515e-01 -3.04938346e-01 -3.08571942e-02 1.60017741e+00
4.35764194e-01 -1.67688608e-01 -4.98550326e-01 6.50476158e-01
1.47671357e-01 1.21387608e-01 4.37977314e-01 -6.98650897e-01
9.23311234e-01 -3.55052590e-01 -3.42256665e-01 -8.20513442e-02
3.24618369e-01 -7.43303299e-01 8.34519982e-01 8.26666892e-01
-1.09483898e+00 5.18186539e-02 -5.02418339e-01 2.72705972e-01
1.26654625e-01 -1.98574781e-01 8.79360557e-01 1.24667585e+00
-5.63880563e-01 9.26108003e-01 -2.07078934e-01 1.60764113e-01
8.04910243e-01 6.56079352e-01 5.11503499e-03 1.46881759e-01
-8.07276964e-01 7.02711523e-01 -4.08585548e-01 -1.47650421e-01
-3.07059467e-01 -1.39647627e+00 -1.79783687e-01 4.66035932e-01
4.54009742e-01 -6.12098873e-01 1.16964650e+00 -1.08635533e+00
-1.39567375e+00 4.24849689e-01 1.01797149e-01 -5.63738823e-01
7.70355761e-01 -5.36255777e-01 -1.65180728e-01 -7.16045558e-01
-1.72097772e-01 -1.19092181e-01 8.74960959e-01 -7.97440290e-01
-7.86338389e-01 -4.47401434e-01 -3.09414804e-01 1.40575901e-01
-5.94297171e-01 2.36048952e-01 1.34681836e-01 -4.69698399e-01
-3.05983722e-01 -8.06414545e-01 -7.00302899e-01 -2.93529164e-02
2.27366775e-01 -2.64288932e-01 -4.56130765e-02 -4.43036944e-01
1.63880336e+00 -2.13647461e+00 -4.24886256e-01 6.93284929e-01
1.16929457e-01 3.21139723e-01 1.41883954e-01 3.42241019e-01
2.13623151e-01 3.36142741e-02 3.33125591e-02 -1.27469346e-01
2.29043096e-01 -1.87180698e-01 -2.90844649e-01 5.07424891e-01
-3.51673543e-01 4.91288394e-01 -8.01023245e-01 -2.58965760e-01
-6.55527636e-02 2.57866859e-01 -9.72753108e-01 2.19702289e-01
2.85037979e-02 -3.11889481e-02 -2.07848310e-01 2.08841100e-01
5.85854888e-01 -9.63890254e-02 2.48361200e-01 1.07283562e-01
-1.40289053e-01 1.86101422e-01 -1.43187606e+00 8.03179324e-01
-5.51698148e-01 -1.70068890e-01 1.52740777e-01 -6.57046854e-01
8.58721793e-01 -7.30660409e-02 5.29623985e-01 -7.91887879e-01
2.33923092e-01 8.67010355e-02 9.80928820e-03 5.22506498e-02
4.60062504e-01 -3.83748829e-01 -1.26417339e-01 6.29145265e-01
-4.47357446e-01 3.76617879e-01 -1.38054997e-01 5.37989855e-01
8.23098600e-01 -1.79081663e-01 5.19378603e-01 -5.24425566e-01
2.14879468e-01 -4.83027488e-01 8.65816116e-01 9.49739277e-01
-4.58521664e-01 2.89213777e-01 5.48501253e-01 -1.38718501e-01
-6.62827551e-01 -1.10563040e+00 -1.71359405e-01 1.61323547e+00
-1.99555010e-01 -3.61135870e-01 -2.95127749e-01 -8.09806645e-01
6.08392358e-01 1.04977226e+00 -4.94292229e-01 -1.81396216e-01
3.26950513e-02 -1.01876068e+00 3.11889430e-03 2.08636895e-01
-3.11669856e-01 -4.56185251e-01 -3.41915607e-01 2.02383369e-01
1.07131459e-01 -4.08432096e-01 -9.65665638e-01 -1.41273081e-01
-7.87177861e-01 -8.70397925e-01 -4.35217947e-01 2.35851333e-02
4.87477213e-01 3.09406728e-01 1.23355496e+00 1.27827898e-01
-1.21498257e-01 1.97642267e-01 -8.25483128e-02 -5.32770157e-01
1.14642337e-01 -2.38631770e-01 2.97579974e-01 3.59297305e-01
4.13356900e-01 -4.76821452e-01 -9.56589460e-01 5.75969696e-01
-3.91512483e-01 -4.83235627e-01 3.72044832e-01 6.23719156e-01
5.42239130e-01 2.57917166e-01 1.02823126e+00 -1.43924642e+00
9.14412141e-01 -8.18590522e-01 -4.65112329e-01 1.34280697e-01
-1.55320323e+00 -3.22129160e-01 6.33335769e-01 -4.37570482e-01
-1.16254997e+00 -1.83969006e-01 -3.15773115e-02 2.12208450e-01
3.41875434e-01 2.74671078e-01 1.30385399e-01 -6.99812621e-02
6.97757006e-01 -2.33695284e-01 1.18822582e-01 -5.94339311e-01
3.77180457e-01 8.52143705e-01 1.77717969e-01 -3.76162499e-01
5.06550550e-01 1.99074313e-01 9.67130885e-02 -5.00636995e-01
-1.11473513e+00 -7.34906018e-01 -8.34239423e-02 -1.95455506e-01
-9.07473639e-03 -7.96791971e-01 -1.26251590e+00 -5.81602715e-02
-2.25176592e-03 -3.39731038e-01 -3.58157396e-01 4.52578574e-01
-3.21972638e-01 3.29346180e-01 -2.33609334e-01 -1.35895038e+00
-7.97885358e-01 -3.16422582e-01 2.89715767e-01 4.54401702e-01
-5.20262301e-01 -8.83003950e-01 1.93587482e-01 6.59050345e-01
6.47866607e-01 -3.84484045e-02 5.96756577e-01 -9.39136684e-01
-9.64475647e-02 -5.47559917e-01 -1.69669762e-01 3.84251773e-01
-1.49823681e-01 7.64480159e-02 -4.97212857e-01 -5.84953845e-01
-2.38434941e-01 -6.17575869e-02 6.05325341e-01 4.51208681e-01
1.12471116e+00 -4.94076908e-01 8.56081198e-04 3.05695623e-01
1.35648930e+00 -1.00062542e-01 5.03864288e-01 -2.33450998e-02
2.58328766e-01 6.99221432e-01 8.94210637e-01 1.00929594e+00
5.45741975e-01 4.80060279e-01 2.39066646e-01 9.23949406e-02
4.89269614e-01 -7.25175813e-02 4.47660804e-01 3.53411674e-01
4.37990092e-02 -1.26288638e-01 -4.44232941e-01 1.98421910e-01
-1.90998304e+00 -1.08949649e+00 -2.67746955e-01 3.02814078e+00
6.93502069e-01 3.07278752e-01 6.71015263e-01 1.49847463e-01
3.99514198e-01 -4.66907680e-01 -5.73347509e-01 -9.45857644e-01
4.11060154e-01 3.22962373e-01 6.47406161e-01 3.66514236e-01
-8.76694739e-01 4.25922364e-01 6.38853407e+00 1.01486409e+00
-6.19699895e-01 2.53626388e-02 8.61953914e-01 -6.56302989e-01
-3.49392295e-01 -4.31792848e-02 -8.18685532e-01 7.55820870e-01
1.29548943e+00 -5.26659012e-01 4.43134516e-01 8.97094190e-01
5.78223228e-01 -3.48022163e-01 -8.03251088e-01 5.83460629e-01
-1.60116311e-02 -9.21703517e-01 -3.48393559e-01 2.27648959e-01
7.33059466e-01 -1.44559458e-01 2.84935206e-01 3.14890116e-01
6.71454787e-01 -8.15130055e-01 4.37893867e-01 9.71075475e-01
6.57315016e-01 -1.12250018e+00 6.25138342e-01 4.96610761e-01
-5.46897829e-01 -4.31354314e-01 -3.95472169e-01 -2.07189143e-01
2.44834796e-02 1.23328447e+00 -3.88445765e-01 8.34382921e-02
5.86826265e-01 2.19758600e-01 -2.62177914e-01 1.34180605e+00
1.93575442e-01 7.76120245e-01 -3.59755546e-01 -4.06425148e-02
-2.86837846e-01 -4.73199636e-01 1.45108894e-01 1.03559816e+00
1.40193164e-01 3.12253177e-01 1.58410873e-02 4.12819088e-01
-1.28170192e-01 7.72225738e-01 -5.04737981e-02 1.21758871e-01
4.80234146e-01 1.55299449e+00 -4.91630048e-01 6.86091110e-02
-4.19456005e-01 5.03800154e-01 1.45741045e-01 -1.05970018e-01
-4.46424991e-01 -1.63784280e-01 8.24157536e-01 3.67611438e-01
1.93488508e-01 6.31147921e-02 -5.80619752e-01 -9.67439115e-01
-3.46628696e-01 -8.99024189e-01 8.53318810e-01 -2.28926167e-02
-1.72261918e+00 -5.01767360e-02 -6.20549619e-01 -1.09633362e+00
8.59710500e-02 -7.96650425e-02 -8.50683033e-01 8.58706474e-01
-1.37935305e+00 -6.03192925e-01 1.07424952e-01 4.56798851e-01
1.26946056e-02 4.53431644e-02 6.75403178e-01 6.41070783e-01
-6.60233617e-01 1.18015146e+00 5.91993392e-01 -5.11598825e-01
1.05767286e+00 -1.25669205e+00 -1.97195396e-01 5.69384336e-01
-2.74691224e-01 7.82571435e-01 7.39970386e-01 -6.40972197e-01
-1.19653475e+00 -9.73266542e-01 1.09071648e+00 -6.30515754e-01
5.24535120e-01 -1.06015250e-01 -6.99168026e-01 1.45915613e-01
-4.19172585e-01 -2.74479836e-01 1.63335729e+00 6.35034561e-01
-4.27887052e-01 -5.55196345e-01 -1.48368537e+00 4.73119527e-01
7.91795552e-01 -1.53110802e-01 -1.63835399e-02 1.80079907e-01
8.36869478e-02 1.61600098e-01 -1.34176624e+00 1.65200800e-01
1.23882210e+00 -1.16069591e+00 8.76445949e-01 -7.28559554e-01
1.76523969e-01 1.01861455e-01 -5.03222980e-02 -1.15843976e+00
-7.18591332e-01 -8.43539655e-01 -1.31961524e-01 1.37111676e+00
4.96786892e-01 -5.69846928e-01 1.07747543e+00 1.46257985e+00
5.93435824e-01 -9.47222173e-01 -6.37334049e-01 -4.85562325e-01
2.09179148e-01 -2.84038335e-01 4.69394237e-01 8.01033735e-01
6.91271648e-02 1.51503369e-01 -7.05729485e-01 -1.65699348e-01
1.17053211e+00 2.42847338e-01 8.00351381e-01 -1.68275678e+00
-3.36123616e-01 -5.74929476e-01 -7.06570074e-02 -4.94800210e-01
-2.06737176e-01 -8.08653414e-01 -2.40199909e-01 -1.08657146e+00
4.53447938e-01 -8.57048213e-01 -7.70303369e-01 3.00402850e-01
-4.25270051e-01 4.72648114e-01 7.59213269e-02 3.82063761e-02
-9.73939896e-01 3.03512752e-01 7.27165043e-01 4.15770710e-01
-3.82415563e-01 7.97800004e-01 -1.46451366e+00 4.58976537e-01
7.78970897e-01 -7.50773251e-01 -4.79931146e-01 3.70444804e-01
7.74427235e-01 9.64039490e-02 -1.66373834e-01 -5.57754934e-01
1.32970750e-01 -6.89188898e-01 4.47975993e-01 -1.76686987e-01
-1.96814716e-01 -7.73060977e-01 3.26719403e-01 3.76047313e-01
-6.74735010e-01 -3.48783553e-01 -3.31072271e-01 6.38925254e-01
5.44163287e-01 -7.27348253e-02 9.53272879e-01 1.63155690e-01
1.10470779e-01 4.75753427e-01 -1.85956866e-01 1.96732566e-01
1.05849314e+00 1.56792060e-01 -1.74883425e-01 -6.68719590e-01
-6.45119607e-01 5.75570166e-01 3.19118470e-01 2.56263286e-01
2.08635673e-01 -9.57039416e-01 -9.01859462e-01 7.12537020e-02
-6.33646026e-02 -7.56158412e-01 3.31742018e-01 9.34375405e-01
1.24462441e-01 6.00195266e-02 2.75460910e-02 1.81265011e-01
-1.41527438e+00 2.94715136e-01 2.49728903e-01 -1.15408726e-01
-1.00272708e-01 9.15195704e-01 -1.31688133e-01 -1.24725431e-01
2.87820995e-01 4.92582321e-01 -2.90882617e-01 3.62270087e-01
8.09506655e-01 9.65799570e-01 2.43166596e-01 -3.64767671e-01
-3.73954296e-01 -1.11385593e-02 -3.29805970e-01 1.34537175e-01
1.40716898e+00 -4.43379313e-01 -1.19619735e-01 1.77217484e-01
7.30494082e-01 4.41655606e-01 -1.19506669e+00 -3.40647191e-01
7.76381837e-03 -9.04759645e-01 4.93211627e-01 -1.05425823e+00
-8.64574432e-01 3.56561869e-01 5.35775602e-01 3.58144939e-01
1.01529062e+00 -6.05147302e-01 8.55112016e-01 -2.56455708e-02
2.79825091e-01 -1.23826265e+00 -3.28026175e-01 -2.29076147e-02
3.64860058e-01 -1.14595556e+00 4.55358207e-01 -1.14354350e-01
-7.74439096e-01 5.91294467e-01 3.64659816e-01 -1.43154368e-01
1.02233517e+00 -1.26419216e-03 -5.46768727e-03 2.35496432e-01
-1.07964396e+00 -6.62907260e-03 4.44804639e-01 1.82343230e-01
5.18497348e-01 4.51079726e-01 -8.42545927e-01 1.11260402e+00
-3.47190015e-02 2.00517043e-01 4.99837905e-01 4.69607264e-01
-3.88718426e-01 -1.32772171e+00 -1.80280637e-02 1.32215416e+00
-1.04799962e+00 -1.25505105e-01 -1.86767876e-01 -1.44699156e-01
1.34365124e-04 1.24057603e+00 -9.85214603e-04 -2.88870066e-01
4.44564134e-01 3.86109401e-04 1.98058605e-01 -5.00883043e-01
-9.40099180e-01 6.37455881e-02 1.71356007e-01 -6.48116171e-01
-1.89354010e-02 -9.83546019e-01 -7.48700082e-01 -8.42068434e-01
-6.55029058e-01 5.53953350e-01 6.96825624e-01 7.19522536e-01
5.42610228e-01 1.70667976e-01 1.17659581e+00 -3.23563963e-01
-1.17891002e+00 -3.14772040e-01 -9.12120104e-01 4.82832998e-01
8.00550878e-02 -5.47985911e-01 -4.98423070e-01 -5.33589363e-01] | [9.512308120727539, 5.673984050750732] |
08d7efcd-d8a8-439a-8e8f-9b6df3286a56 | leveraging-off-the-shelf-diffusion-model-for | 2210.05872 | null | https://arxiv.org/abs/2210.05872v1 | https://arxiv.org/pdf/2210.05872v1.pdf | Leveraging Off-the-shelf Diffusion Model for Multi-attribute Fashion Image Manipulation | Fashion attribute editing is a task that aims to convert the semantic attributes of a given fashion image while preserving the irrelevant regions. Previous works typically employ conditional GANs where the generator explicitly learns the target attributes and directly execute the conversion. These approaches, however, are neither scalable nor generic as they operate only with few limited attributes and a separate generator is required for each dataset or attribute set. Inspired by the recent advancement of diffusion models, we explore the classifier-guided diffusion that leverages the off-the-shelf diffusion model pretrained on general visual semantics such as Imagenet. In order to achieve a generic editing pipeline, we pose this as multi-attribute image manipulation task, where the attribute ranges from item category, fabric, pattern to collar and neckline. We empirically show that conventional methods fail in our challenging setting, and study efficient adaptation scheme that involves recently introduced attention-pooling technique to obtain a multi-attribute classifier guidance. Based on this, we present a mask-free fashion attribute editing framework that leverages the classifier logits and the cross-attention map for manipulation. We empirically demonstrate that our framework achieves convincing sample quality and attribute alignments. | ['Nojun Kwak', 'Ohjoon Kwon', 'Donghyeon Jeon', 'Chaerin Kong'] | 2022-10-12 | null | null | null | null | ['image-manipulation'] | ['computer-vision'] | [ 6.10055447e-01 9.02561098e-02 -2.27223992e-01 -6.71662927e-01
-7.15960026e-01 -8.73553038e-01 8.33121717e-01 -5.97957820e-02
-7.63416588e-02 6.04777753e-01 1.92960724e-01 -1.49223721e-03
1.03557691e-01 -9.50123489e-01 -1.00493670e+00 -5.65983951e-01
3.65222335e-01 3.91857922e-01 -6.21155761e-02 -3.51970345e-01
1.04875915e-01 2.09798977e-01 -1.39979863e+00 2.73406088e-01
9.09082770e-01 1.26878095e+00 6.43942058e-02 6.06183827e-01
-8.15294683e-02 6.42182291e-01 -4.90372479e-01 -7.97697008e-01
4.60781068e-01 -7.31162131e-01 -5.19588947e-01 1.82415530e-01
5.84910691e-01 -4.65713382e-01 -1.97850034e-01 1.01814818e+00
4.14401114e-01 2.01097857e-02 1.01343095e+00 -1.52770364e+00
-1.59672308e+00 5.20343244e-01 -5.68564534e-01 -2.78076380e-01
2.41825461e-01 4.81035441e-01 9.93093133e-01 -6.09053671e-01
6.73673570e-01 1.11507583e+00 6.01209641e-01 7.55736411e-01
-1.36743057e+00 -5.87041020e-01 5.58319628e-01 -2.58106943e-02
-1.09531868e+00 -1.91972420e-01 1.03606963e+00 -4.06613261e-01
4.81468260e-01 2.64804155e-01 8.72625470e-01 1.68857491e+00
1.75160781e-01 8.27203810e-01 1.59772599e+00 -3.12014222e-01
3.08230639e-01 1.52272686e-01 -2.62208283e-01 7.94685602e-01
-6.49392381e-02 3.87831517e-02 -3.72077018e-01 1.35991216e-01
1.03443491e+00 3.84290293e-02 9.26470906e-02 -5.99588752e-01
-1.31876004e+00 9.22936141e-01 5.74339271e-01 -3.08591008e-01
-5.03023863e-01 3.14452887e-01 2.30304375e-01 5.00084877e-01
6.39933765e-01 4.75730032e-01 -4.19938594e-01 1.77164733e-01
-8.45376074e-01 3.78028214e-01 6.26597106e-01 1.63133216e+00
8.50519836e-01 -5.48342578e-02 -6.38938487e-01 7.25219309e-01
8.28650221e-02 3.48714471e-01 4.41521257e-02 -1.06561065e+00
1.69521555e-01 7.54476607e-01 -1.93247661e-01 -7.14829385e-01
1.40327543e-01 -3.42180938e-01 -9.51040208e-01 3.83130819e-01
2.19478890e-01 4.13928516e-02 -1.32360709e+00 2.01847339e+00
4.19261605e-01 -3.88358869e-02 -1.89010024e-01 8.03901434e-01
5.71192622e-01 1.74900427e-01 4.79632437e-01 2.22387016e-01
1.54029000e+00 -1.28031874e+00 -6.36665881e-01 -1.94800541e-01
6.89730868e-02 -7.36224234e-01 1.59556627e+00 1.53837413e-01
-1.15386307e+00 -5.64914048e-01 -1.31256497e+00 -4.07955825e-01
-7.01060772e-01 2.41252288e-01 9.20351624e-01 6.15648448e-01
-9.99772191e-01 4.11564887e-01 -5.59847713e-01 -5.68838596e-01
7.32788920e-01 3.56550634e-01 -3.36063981e-01 -9.24627855e-02
-9.43940938e-01 7.63782084e-01 -1.14840128e-01 -3.06853093e-02
-1.17651892e+00 -8.44922364e-01 -7.91789830e-01 -1.56351447e-01
4.26513135e-01 -1.41589069e+00 1.00122476e+00 -1.32426167e+00
-1.75809848e+00 1.02072608e+00 1.12847559e-01 -2.62834579e-01
7.45465398e-01 -3.12806785e-01 -5.67902476e-02 -6.46716282e-02
2.35605404e-01 1.05017269e+00 1.15860641e+00 -1.36546052e+00
-4.42403078e-01 -3.28624666e-01 5.26574671e-01 1.44474968e-01
-4.04314578e-01 -4.20821533e-02 -4.71183509e-01 -1.17908943e+00
-3.08488786e-01 -1.06591749e+00 -2.51681417e-01 2.83499897e-01
-7.79674113e-01 -2.70154215e-02 7.39109159e-01 -6.05323613e-01
8.33061457e-01 -2.02395797e+00 3.88624787e-01 1.54715180e-01
2.16744542e-01 -2.81824976e-01 -1.31519273e-01 2.73135632e-01
2.44624734e-01 8.31581950e-02 -4.79554087e-01 -5.12671173e-01
3.33018810e-01 2.19330326e-01 -2.74143755e-01 2.86673337e-01
5.97478688e-01 1.16130543e+00 -7.80356884e-01 -5.13410270e-01
1.12831086e-01 5.94879210e-01 -9.23629403e-01 5.52948236e-01
-5.76892257e-01 7.45617747e-01 -4.90913689e-01 9.20598507e-01
6.20169461e-01 -8.04227144e-02 -2.14856446e-01 -4.96058673e-01
1.87773868e-01 -1.85952291e-01 -9.40454662e-01 2.29835963e+00
-5.54753661e-01 1.42279655e-01 1.74060479e-01 -7.20617771e-01
9.73277390e-01 -1.10216796e-01 2.04331189e-01 -3.82130235e-01
1.22063607e-01 7.21305534e-02 -3.46013486e-01 -4.20558751e-01
5.66916168e-01 3.01669519e-02 -4.64629769e-01 4.84209448e-01
2.43387327e-01 -2.36789733e-01 -8.51819739e-02 2.49157935e-01
1.00240850e+00 8.31790745e-01 1.96627170e-01 -2.26095095e-01
1.74352735e-01 5.19356430e-02 3.58661324e-01 8.58206391e-01
-1.78863164e-02 9.42102253e-01 5.06722689e-01 -3.99562180e-01
-1.36418617e+00 -1.28899264e+00 3.27384830e-01 1.41421485e+00
2.24102020e-01 -2.54826576e-01 -1.14702082e+00 -1.00051701e+00
-8.62517208e-02 5.75916409e-01 -1.00630736e+00 -3.24634552e-01
-5.13802648e-01 -6.29643917e-01 3.16620141e-01 8.54832232e-01
7.26544857e-01 -1.03305364e+00 -2.82386452e-01 8.77401382e-02
1.64378330e-01 -1.07890308e+00 -8.27138841e-01 1.22648142e-01
-4.69119489e-01 -8.05361390e-01 -7.18419909e-01 -8.91350329e-01
8.14475477e-01 -2.80700892e-01 1.09739947e+00 -1.15436718e-01
-2.89711237e-01 5.50775409e-01 -2.54173905e-01 -2.75412887e-01
-4.58594233e-01 3.98730844e-01 -1.75938278e-01 3.70422244e-01
3.45383018e-01 -8.44857514e-01 -1.03425145e+00 1.57405257e-01
-8.75026345e-01 2.89248109e-01 1.00816381e+00 7.89340734e-01
7.93422282e-01 -4.59313512e-01 7.35386431e-01 -1.14896977e+00
7.14349329e-01 -5.33008397e-01 -1.08499244e-01 3.11340928e-01
-6.31581128e-01 8.53085443e-02 6.98258877e-01 -6.25636280e-01
-1.10087073e+00 3.92790169e-01 -1.12690464e-01 -4.75224495e-01
-4.56840783e-01 -1.49756437e-02 -6.02104843e-01 7.08319545e-02
4.07736361e-01 2.05349177e-01 -8.68804529e-02 -5.40659964e-01
1.02307713e+00 5.73130667e-01 7.69415796e-01 -8.17634523e-01
7.65890718e-01 7.30954945e-01 -1.12719551e-01 -2.25616843e-01
-8.69049668e-01 9.70568955e-02 -8.27677965e-01 -9.80532989e-02
1.21929038e+00 -9.08065259e-01 -4.83239919e-01 4.00903255e-01
-1.00184882e+00 -4.03358042e-01 -5.98127306e-01 1.04758248e-01
-1.02327073e+00 -1.19077258e-01 -7.05502450e-01 -3.60785782e-01
-3.30228597e-01 -1.21734500e+00 1.32426667e+00 5.28371260e-02
-3.40825766e-01 -6.90537870e-01 -2.52270788e-01 1.47069946e-01
8.24675918e-01 4.94969726e-01 1.29945421e+00 -5.70523620e-01
-7.94024527e-01 -5.17338961e-02 -2.76340067e-01 3.69608521e-01
2.60593474e-01 -1.51067078e-01 -1.01099312e+00 -1.08484186e-01
-2.01441303e-01 -4.29784656e-01 8.88983965e-01 2.19656020e-01
1.56476450e+00 -3.05938005e-01 -2.54183471e-01 1.00931108e+00
1.30586219e+00 -1.52733833e-01 5.91576934e-01 4.18084115e-01
9.96051729e-01 6.30207479e-01 4.58135843e-01 1.16818145e-01
6.44687593e-01 6.66601300e-01 4.82353270e-01 -4.95668441e-01
-6.22512698e-01 -5.71287572e-01 1.25948697e-01 5.63888967e-01
-2.61566818e-01 -8.24792981e-02 -1.23816013e-01 5.20923674e-01
-1.61894989e+00 -6.16615653e-01 3.14301193e-01 1.92361820e+00
1.07487905e+00 1.86381359e-02 2.35069051e-01 -2.91525275e-01
6.14198625e-01 5.73098585e-02 -8.17699075e-01 -4.52892274e-01
-1.81768328e-01 2.82655269e-01 3.79022539e-01 2.52583086e-01
-1.24205899e+00 1.06946075e+00 6.22540903e+00 8.35062087e-01
-8.28938782e-01 2.84502447e-01 7.77630925e-01 -1.23617142e-01
-6.12422287e-01 -1.98002141e-02 -7.29561925e-01 4.30905551e-01
4.79744643e-01 2.14973673e-01 6.19274259e-01 8.74583185e-01
-1.48598805e-01 3.17555219e-01 -1.43865705e+00 6.62697196e-01
2.03811854e-01 -1.09025705e+00 4.12493289e-01 1.24952672e-02
8.31844866e-01 -5.97444713e-01 4.52688843e-01 4.62498337e-01
6.70470774e-01 -1.13232386e+00 9.15310562e-01 6.93944871e-01
1.24244988e+00 -5.64349711e-01 1.91299453e-01 -1.92292318e-01
-1.08937883e+00 -4.89263386e-02 -1.38408497e-01 2.14526042e-01
1.83410615e-01 2.42193714e-01 -3.05920303e-01 4.42779958e-01
6.49018168e-01 5.77810109e-01 -6.17731631e-01 5.74927747e-01
-4.17459697e-01 3.21042389e-01 -1.54343754e-01 2.41911665e-01
1.08745791e-01 -3.34434092e-01 3.00377876e-01 1.20234287e+00
3.87260437e-01 -1.63734272e-01 1.80341214e-01 1.23401380e+00
-4.10005063e-01 3.68169136e-02 -6.61987543e-01 1.81796595e-01
4.11883354e-01 1.48120880e+00 -7.23737836e-01 -3.22534502e-01
-4.33499038e-01 1.59151733e+00 4.82214421e-01 4.88937050e-01
-1.12788916e+00 -3.20734143e-01 7.76881933e-01 1.74761474e-01
6.64061308e-01 -9.55260023e-02 -4.10416514e-01 -9.69776690e-01
1.53114095e-01 -9.51685548e-01 1.99097738e-01 -6.92121148e-01
-1.67817235e+00 6.32313967e-01 3.60665806e-02 -9.78081107e-01
-6.61374032e-02 -4.39450681e-01 -5.14849305e-01 7.63947427e-01
-1.51358151e+00 -2.10934401e+00 -5.21708727e-01 7.66039610e-01
7.54259825e-01 -1.36221468e-01 8.52787554e-01 2.23451912e-01
-4.01535809e-01 9.93947387e-01 -3.22358996e-01 -4.42161970e-02
9.31765735e-01 -1.49660778e+00 4.98288333e-01 4.96800154e-01
-2.30497092e-01 7.95702040e-01 6.22802556e-01 -6.49962902e-01
-1.44062805e+00 -1.43792129e+00 3.77850950e-01 -6.77405000e-01
5.78001261e-01 -8.01405132e-01 -5.45900941e-01 9.99422669e-01
5.20677030e-01 3.95631641e-02 6.07442737e-01 -1.99706867e-01
-4.12152827e-01 -2.06721261e-01 -1.16822195e+00 7.12889194e-01
1.51756787e+00 -5.03155172e-01 -2.96810746e-01 2.02354267e-01
8.41904998e-01 -3.28421831e-01 -1.08506405e+00 3.63448501e-01
5.46689928e-01 -6.90604627e-01 1.02837253e+00 -7.39942133e-01
8.26680601e-01 -2.69207776e-01 -3.19120616e-01 -1.39851558e+00
-3.83201957e-01 -6.88058913e-01 -1.06736660e-01 1.68074763e+00
3.92666727e-01 -3.45059574e-01 6.99082136e-01 5.14421642e-01
-3.03942353e-01 -8.81080389e-01 -6.17829680e-01 -6.51989520e-01
1.00109175e-01 -1.08937047e-01 1.08496547e+00 9.39791083e-01
-7.39364445e-01 4.19900358e-01 -6.54534280e-01 -8.18304867e-02
7.11330891e-01 2.31785670e-01 8.11107993e-01 -8.00424755e-01
-5.49244821e-01 -4.06235665e-01 -2.65784919e-01 -1.19561124e+00
1.41989246e-01 -9.06717777e-01 -8.07422623e-02 -1.46663320e+00
2.91454405e-01 -7.41982818e-01 -1.72959790e-01 4.40093935e-01
-2.42143631e-01 6.53219283e-01 1.46898866e-01 1.16740376e-01
-5.04022479e-01 7.04734147e-01 1.39251971e+00 -4.12558228e-01
1.26015526e-04 -2.20276073e-01 -1.12978148e+00 5.68090320e-01
6.86906278e-01 -4.41527992e-01 -6.21751547e-01 -5.93140721e-01
1.87351003e-01 -5.66634059e-01 6.41441882e-01 -6.94096625e-01
1.38051257e-01 -1.12326525e-01 6.63606822e-01 -6.06460199e-02
4.97835606e-01 -8.35078537e-01 1.02692701e-01 -7.90054128e-02
-6.51287079e-01 2.47659355e-01 -1.51609778e-01 8.86369348e-01
8.64504278e-02 2.59061038e-01 5.24170280e-01 -1.91716984e-01
-7.51910627e-01 5.48104405e-01 -4.62904908e-02 -8.39661062e-02
1.31825507e+00 -2.08390176e-01 -3.64741385e-01 -2.73197114e-01
-8.46742809e-01 -4.50977869e-02 8.29836965e-01 5.49028635e-01
4.72726822e-01 -1.63957083e+00 -7.20638692e-01 4.07868743e-01
2.48724595e-01 -9.54864547e-02 6.19180463e-02 7.17920184e-01
-1.69738039e-01 -2.05008537e-01 -5.96648097e-01 -4.41009790e-01
-8.28847170e-01 8.75198245e-01 9.80505273e-02 -5.50284050e-02
-5.67636311e-01 9.20330286e-01 5.39585531e-01 -4.72337395e-01
-2.54768450e-02 -2.67761230e-01 2.87290756e-02 -1.71375901e-01
1.45321816e-01 -1.54262846e-02 -1.38439313e-01 -3.16324741e-01
-8.63310546e-02 7.16239393e-01 -1.93108708e-01 -1.77006889e-03
1.27802885e+00 -2.96279907e-01 -6.59099296e-02 2.48679921e-01
1.03160083e+00 3.97256985e-02 -1.89606500e+00 -5.84074333e-02
-4.20185179e-01 -4.23592061e-01 -2.42643490e-01 -9.36935246e-01
-1.20528889e+00 7.56156027e-01 5.47598302e-01 1.05639398e-01
1.40226567e+00 1.99508488e-01 8.02768290e-01 -2.18441531e-01
1.95241228e-01 -1.03279674e+00 1.26346901e-01 -4.92428243e-02
1.07009614e+00 -1.03714347e+00 -1.44667119e-01 -6.70386910e-01
-8.55745375e-01 7.42054462e-01 8.84559751e-01 -3.42912495e-01
3.82613599e-01 3.09802920e-01 -8.87545384e-03 -2.45204017e-01
-5.69198370e-01 -7.60638788e-02 1.84816852e-01 8.96122456e-01
2.17897445e-01 3.65951695e-02 -6.03464544e-02 7.58069158e-01
-3.31152022e-01 -1.43023934e-02 2.02199578e-01 8.68912399e-01
5.09573743e-02 -1.16349483e+00 -3.24431770e-02 5.51658511e-01
-3.11889231e-01 -2.00908199e-01 -5.16627371e-01 7.86451280e-01
3.08624059e-01 5.97094595e-01 7.17198402e-02 -3.62495393e-01
5.03640950e-01 2.24867482e-02 7.79102087e-01 -3.99608523e-01
-7.31697977e-01 -5.27804978e-02 -3.57357375e-02 -6.41721845e-01
-4.78680700e-01 -4.50428337e-01 -5.74794650e-01 -3.50748003e-01
-2.40235589e-02 -3.86732459e-01 6.19428694e-01 7.16617346e-01
5.75622201e-01 6.16336167e-01 3.99813205e-01 -6.55863523e-01
-5.22719979e-01 -9.38580096e-01 -4.39843833e-01 7.80757964e-01
1.98877931e-01 -6.81449354e-01 -4.06457409e-02 4.84330267e-01] | [11.607929229736328, -0.40011367201805115] |
3f65ecd4-8fe0-4b66-92fb-a4fe09b23801 | representing-input-transformations-by-low | 2305.13536 | null | https://arxiv.org/abs/2305.13536v1 | https://arxiv.org/pdf/2305.13536v1.pdf | Representing Input Transformations by Low-Dimensional Parameter Subspaces | Deep models lack robustness to simple input transformations such as rotation, scaling, and translation, unless they feature a particular invariant architecture or undergo specific training, e.g., learning the desired robustness from data augmentations. Alternatively, input transformations can be treated as a domain shift problem, and solved by post-deployment model adaptation. Although a large number of methods deal with transformed inputs, the fundamental relation between input transformations and optimal model weights is unknown. In this paper, we put forward the configuration subspace hypothesis that model weights optimal for parameterized continuous transformations can reside in low-dimensional linear subspaces. We introduce subspace-configurable networks to learn these subspaces and observe their structure and surprisingly low dimensionality on all tested transformations, datasets and architectures from computer vision and audio signal processing domains. Our findings enable efficient model reconfiguration, especially when limited storage and computing resources are at stake. | ['Lothar Thiele', 'Xiaoxi He', 'Dong Wang', 'Olga Saukh'] | 2023-05-22 | null | null | null | null | ['audio-signal-processing'] | ['audio'] | [ 3.61990511e-01 7.05452040e-02 -2.53608227e-01 -1.54019922e-01
-2.35178351e-01 -1.04889703e+00 7.00243056e-01 -2.34857947e-01
-3.52383822e-01 4.45025146e-01 2.38818944e-01 -3.07687432e-01
-2.76767939e-01 -3.27450484e-01 -8.47063124e-01 -6.90390885e-01
-5.22409156e-02 5.98010004e-01 2.11520563e-03 -2.27467582e-01
-6.76041692e-02 9.35518503e-01 -1.15504181e+00 -6.08671568e-02
5.60916722e-01 7.26194143e-01 -2.62356043e-01 6.40371144e-01
1.75480053e-01 7.36396853e-03 -4.09143716e-01 5.21024987e-02
7.80057967e-01 -2.88576126e-01 -7.15693355e-01 9.60389748e-02
4.67022806e-01 -1.35103971e-01 -5.40523827e-01 9.11915362e-01
5.78855276e-01 2.87267357e-01 7.46162176e-01 -1.37366116e+00
-6.11633122e-01 4.65740621e-01 -8.20457470e-03 8.86189044e-02
-1.16272457e-02 5.39603055e-01 6.88058555e-01 -8.63775551e-01
6.56950891e-01 1.12083280e+00 5.70501804e-01 8.06043267e-01
-1.87657702e+00 -6.04067087e-01 2.85804063e-01 1.02095954e-01
-1.23930478e+00 -7.80072927e-01 8.70958090e-01 -6.73141718e-01
1.21120763e+00 3.13829392e-01 5.45408368e-01 1.43924201e+00
3.29850032e-03 2.68117070e-01 6.54412806e-01 -3.21779788e-01
4.92708981e-01 1.85837239e-01 -7.62412176e-02 1.88382939e-01
3.29941988e-01 9.51172933e-02 -4.47353482e-01 -2.06923097e-01
8.67334902e-01 -1.76407337e-01 -3.25381249e-01 -9.91604865e-01
-1.44838512e+00 4.65942204e-01 3.19235086e-01 2.11521924e-01
-2.41124019e-01 1.20928638e-01 3.99599761e-01 7.85163760e-01
6.94110096e-02 1.01552129e+00 -9.24899042e-01 1.12737767e-01
-7.10701227e-01 9.99107659e-02 7.05377638e-01 1.06778848e+00
7.60519207e-01 6.22810721e-01 3.41950580e-02 5.75487673e-01
-7.12308707e-03 5.91201186e-01 6.52446389e-01 -8.47063601e-01
4.87008542e-01 5.29124260e-01 -3.86889912e-02 -8.78672540e-01
-8.37575734e-01 -7.25989640e-01 -1.14836073e+00 5.63557036e-02
4.21422273e-01 -9.15557966e-02 -9.82397914e-01 2.11646390e+00
3.00282419e-01 1.99242085e-01 1.40194997e-01 7.98145115e-01
1.77454323e-01 3.33389103e-01 -1.14684001e-01 -1.75795078e-01
1.13384604e+00 -6.32450104e-01 -3.77724141e-01 -4.04506207e-01
4.92732584e-01 -5.77989280e-01 1.37252116e+00 3.99964839e-01
-9.88441706e-01 -5.43461859e-01 -1.08053732e+00 6.75080046e-02
-3.11357588e-01 4.74984050e-02 3.21065605e-01 4.91577953e-01
-1.01993489e+00 6.71856761e-01 -1.10249174e+00 -6.56019449e-01
1.96833014e-01 7.44252086e-01 -4.94073093e-01 3.14872384e-01
-1.01818252e+00 9.20094550e-01 5.83793700e-01 -1.28925174e-01
-1.08577418e+00 -7.22271144e-01 -5.59743822e-01 1.66356042e-01
1.62431806e-01 -1.01283991e+00 1.06433117e+00 -8.90549123e-01
-1.63918662e+00 5.99378109e-01 5.69519624e-02 -6.66654527e-01
4.53363061e-01 -2.41888762e-01 -4.12233293e-01 -1.05813183e-01
-5.16941428e-01 7.08238661e-01 1.47087586e+00 -1.07311857e+00
-1.16709657e-02 -2.31552482e-01 -4.78135087e-02 1.18031211e-01
-9.92843688e-01 3.29654627e-02 -1.51022449e-01 -6.06646001e-01
4.85944718e-01 -1.28229165e+00 -3.29347789e-01 -2.26673633e-02
-5.61891615e-01 2.82323897e-01 8.01802993e-01 -5.03084064e-01
1.12881458e+00 -2.30562186e+00 5.88374734e-01 4.76973355e-01
4.40615080e-02 2.74879903e-01 -4.61396933e-01 4.55074072e-01
-4.71374154e-01 1.94106102e-01 -1.52063206e-01 3.69016430e-03
2.03024670e-01 3.29131544e-01 -7.97533751e-01 5.91486275e-01
4.02925700e-01 7.08339095e-01 -5.06501555e-01 -7.94547051e-02
1.75245121e-01 4.02854800e-01 -8.64850700e-01 6.20769002e-02
-2.48795897e-01 7.28267014e-01 -2.63740927e-01 2.74233341e-01
4.96075094e-01 2.43481603e-02 2.22277075e-01 -4.56666142e-01
2.19815135e-01 2.73176551e-01 -1.36883509e+00 1.66995609e+00
-3.93415511e-01 7.06372201e-01 4.13502678e-02 -1.22349465e+00
9.66063857e-01 2.40719035e-01 4.31788802e-01 -3.21214408e-01
9.86450389e-02 5.98512329e-02 3.31424564e-01 -1.25483975e-01
2.85734147e-01 1.01497553e-01 -4.49537672e-03 3.27628165e-01
2.74690211e-01 -2.64600486e-01 4.30544950e-02 -1.25683531e-01
1.19892538e+00 5.53961024e-02 2.41077706e-01 -2.65490860e-01
3.42726856e-01 -1.23217851e-01 5.36195338e-01 6.67458475e-01
4.56415229e-02 7.09526539e-01 5.17628551e-01 -5.58783710e-01
-1.56327164e+00 -1.12827575e+00 -2.06441671e-01 1.16217220e+00
-2.50457793e-01 -3.09156835e-01 -6.71503603e-01 -2.72412241e-01
-1.14330150e-01 6.41821682e-01 -4.36185449e-01 -8.17042351e-01
-9.27664876e-01 -4.95965958e-01 5.64385712e-01 4.90029782e-01
1.97485685e-01 -7.70804882e-01 -6.54053032e-01 2.45365739e-01
3.03447694e-01 -1.14028203e+00 -4.71116662e-01 6.19525790e-01
-1.22931838e+00 -8.01719427e-01 -3.76348108e-01 -6.68322980e-01
8.01524937e-01 -3.45141254e-02 9.74379718e-01 -3.11576813e-01
-1.39306039e-01 4.62750286e-01 5.00803702e-02 -2.64257222e-01
-5.12430966e-01 4.51665968e-01 8.03465188e-01 -1.41742066e-01
-1.97488479e-02 -1.07898033e+00 -4.98628289e-01 5.31193316e-01
-9.52077091e-01 8.90860856e-02 5.94673812e-01 9.39924121e-01
5.61664045e-01 -1.09438516e-01 4.53405559e-01 -4.12643701e-01
7.88724065e-01 -9.26686749e-02 -5.62920451e-01 2.77715921e-01
-5.98827541e-01 5.32554924e-01 1.04602039e+00 -9.55697834e-01
-5.48144639e-01 4.80470598e-01 2.04837844e-01 -9.08824265e-01
-4.41470712e-01 2.20245317e-01 -4.93877351e-01 7.65168741e-02
1.22097385e+00 2.18952775e-01 -9.03429464e-02 -5.32252908e-01
6.16468728e-01 1.59805357e-01 6.16790056e-01 -7.70899057e-01
1.52205288e+00 4.03526932e-01 1.61672056e-01 -7.85466909e-01
-3.09879184e-01 6.19231649e-02 -1.03033566e+00 1.44778505e-01
4.24374104e-01 -6.49461627e-01 -3.81626934e-01 1.69499561e-01
-1.07093918e+00 -4.18636471e-01 -7.09552288e-01 3.73122066e-01
-7.08552182e-01 1.35740981e-01 -1.75409764e-01 -3.11152726e-01
-2.70985991e-01 -9.40768838e-01 6.73449755e-01 -5.25791794e-02
-5.56111932e-01 -7.71543741e-01 3.86288390e-02 -3.70639771e-01
6.29537344e-01 2.18381137e-01 1.22848117e+00 -9.82077479e-01
-4.03406143e-01 -2.25595683e-01 2.46402591e-01 3.93897206e-01
1.52477577e-01 1.86570048e-01 -9.41798747e-01 -7.38723695e-01
-8.09372813e-02 -1.46212339e-01 5.74026883e-01 3.38644564e-01
1.12769246e+00 -6.78761363e-01 -2.32838124e-01 1.07953382e+00
8.47955704e-01 5.63929714e-02 1.81219891e-01 4.74988192e-01
6.04594409e-01 4.55167741e-01 4.87390868e-02 2.32416049e-01
-2.40259662e-01 7.64760435e-01 3.80674958e-01 -1.54561643e-02
-7.59429857e-03 -2.87933022e-01 2.52540469e-01 7.25883424e-01
1.19343176e-01 1.52573496e-01 -1.11867011e+00 3.63248348e-01
-1.66128981e+00 -6.83834374e-01 4.30409461e-01 2.12509227e+00
7.20735908e-01 3.73541683e-01 1.55254439e-01 2.67912596e-01
4.54096258e-01 1.04734097e-02 -1.13442707e+00 -2.76354313e-01
-3.90940726e-01 8.79167244e-02 4.32852626e-01 1.92651957e-01
-1.03359258e+00 1.10507536e+00 6.55934191e+00 4.16034311e-01
-1.48346055e+00 -6.09640814e-02 3.21713090e-01 -3.81109208e-01
-2.66176134e-01 8.32707807e-02 -4.23402727e-01 5.06940484e-02
9.34332728e-01 -3.35099667e-01 7.99444914e-01 9.79261458e-01
1.51310071e-01 6.73342943e-01 -1.49349248e+00 9.41570222e-01
-2.70247579e-01 -1.17771554e+00 2.02741697e-01 1.38778672e-01
7.36701012e-01 -3.72548513e-02 4.64912891e-01 3.47708911e-01
1.74622044e-01 -1.03183305e+00 6.40707314e-01 4.54510063e-01
9.22207713e-01 -5.84161401e-01 9.75767598e-02 2.99742132e-01
-8.08434010e-01 -4.17674512e-01 -6.09476328e-01 8.21577385e-02
-1.85536623e-01 1.36567548e-01 -1.04157770e+00 6.02353476e-02
4.87447917e-01 5.26399672e-01 -6.20410919e-01 7.85874724e-01
-8.84156227e-02 8.09645116e-01 -5.33842027e-01 2.71359652e-01
-6.41042367e-02 -1.45447731e-01 8.10832620e-01 1.02647495e+00
4.95323062e-01 -2.23800942e-01 -1.03777319e-01 7.56981194e-01
-8.30498412e-02 -1.07908495e-01 -6.94859385e-01 -1.00902483e-01
6.45640254e-01 1.17370379e+00 -6.14264667e-01 -5.26075214e-02
-1.06004946e-01 7.85530031e-01 2.16198832e-01 8.04938316e-01
-6.50864661e-01 -1.39118180e-01 1.21042430e+00 6.21267371e-02
1.41711727e-01 -6.32331014e-01 -4.98713940e-01 -1.38318634e+00
-1.09787202e-02 -1.18905318e+00 1.60842389e-01 -4.82201189e-01
-8.46008897e-01 6.91694081e-01 9.73071530e-02 -1.56210625e+00
-5.73460102e-01 -7.37598598e-01 -5.84535658e-01 6.05213165e-01
-1.02657127e+00 -1.10320199e+00 -1.63068712e-01 8.47956657e-01
2.54842699e-01 -5.64064980e-01 8.70990813e-01 -4.42323163e-02
-6.95725560e-01 8.53102326e-01 3.78397584e-01 -7.69983605e-02
7.67341197e-01 -1.12351453e+00 6.76218867e-01 9.27128255e-01
4.35294300e-01 8.04004669e-01 1.12399852e+00 -2.00388312e-01
-1.70910645e+00 -1.17217350e+00 3.17304939e-01 -5.33173501e-01
8.07451725e-01 -7.74038255e-01 -1.10658956e+00 7.25378752e-01
2.26441808e-02 2.45184153e-01 4.68006909e-01 9.63740721e-02
-6.81071758e-01 -2.99831480e-01 -9.40885007e-01 9.36068594e-01
1.37266731e+00 -6.46811783e-01 -4.78934497e-01 2.92015284e-01
9.13743675e-01 -4.79887992e-01 -8.31581235e-01 4.49385941e-01
4.28006738e-01 -4.44027930e-01 1.21292770e+00 -1.40788734e+00
-3.13177444e-02 -3.24330121e-01 -3.13475788e-01 -1.69238424e+00
-5.42351186e-01 -8.58174562e-01 -3.70312124e-01 1.08252549e+00
4.92677271e-01 -6.49627924e-01 5.82452595e-01 7.93899119e-01
-2.98189580e-01 -3.79980981e-01 -1.13324189e+00 -1.02492046e+00
3.58169794e-01 -4.00799841e-01 9.50058579e-01 1.18486273e+00
-1.75151080e-01 3.06993395e-01 -3.68143797e-01 4.02571648e-01
3.00166786e-01 -8.38694274e-02 1.08932543e+00 -1.15144265e+00
-3.75491410e-01 -6.57815278e-01 -5.11575520e-01 -7.50178754e-01
3.65072429e-01 -8.93602133e-01 -4.05247688e-01 -1.05972290e+00
-3.06771129e-01 -1.79385483e-01 -3.17562819e-01 7.37812698e-01
3.20402473e-01 -1.93747863e-01 2.25784764e-01 4.26941216e-01
-2.89167076e-01 7.15539277e-01 8.92300546e-01 -2.40895480e-01
-5.23783505e-01 -9.02125835e-02 -6.12793982e-01 7.38940477e-01
1.06986880e+00 -4.20809001e-01 -5.71716070e-01 -5.99841714e-01
3.97661418e-01 -1.95535257e-01 4.11436707e-01 -1.31474650e+00
3.12561482e-01 -4.02666122e-01 5.42264342e-01 -1.52203768e-01
3.55251223e-01 -1.18717873e+00 3.71910393e-01 5.73027074e-01
-4.32397127e-01 7.20544308e-02 4.82532561e-01 4.43010598e-01
-6.05437811e-03 1.72912225e-01 8.76934409e-01 3.01294237e-01
-3.84930730e-01 2.25121379e-01 -2.20964119e-01 3.23373377e-02
9.48634148e-01 -3.79209399e-01 -2.27682188e-01 -3.02883446e-01
-1.06942093e+00 -8.38174224e-02 5.35373449e-01 6.98322237e-01
4.40364450e-01 -1.43523407e+00 -5.73408961e-01 6.81658804e-01
9.38939527e-02 5.11672013e-02 7.77710676e-02 6.06233478e-01
-8.45109597e-02 4.80485648e-01 -4.39303368e-01 -7.61019826e-01
-9.59803879e-01 7.38538921e-01 5.53837061e-01 7.61726405e-03
-5.02703607e-01 5.64539313e-01 2.19080091e-01 -7.37435520e-01
2.18734562e-01 -6.31459475e-01 2.06756502e-01 -6.54547811e-02
1.78569257e-01 1.61282122e-01 2.57544488e-01 -4.99439269e-01
-3.62606347e-01 6.20275497e-01 1.05600350e-01 1.11340650e-03
1.44501185e+00 7.05657676e-02 2.14644238e-01 2.09755734e-01
1.13104999e+00 -4.94584233e-01 -1.43744934e+00 -5.97861588e-01
4.88289148e-02 -2.26369068e-01 -2.12157607e-01 -4.77334827e-01
-8.88725221e-01 1.03509605e+00 7.36396015e-01 1.20511040e-01
1.25528479e+00 -1.24587871e-01 2.11177871e-01 9.72152472e-01
3.84799570e-01 -1.21365142e+00 2.08181217e-01 7.99655795e-01
1.27656841e+00 -8.15396547e-01 -1.94939762e-01 1.93338737e-01
-4.49573934e-01 1.18001509e+00 9.04434979e-01 -1.98633268e-01
6.53509796e-01 2.95128763e-01 -1.30694970e-01 2.77981102e-01
-9.83208179e-01 1.60977527e-01 5.33266902e-01 8.30254018e-01
1.63113728e-01 -1.27376989e-01 2.23703653e-01 4.73207325e-01
-7.08287120e-01 -5.63655019e-01 3.15678626e-01 5.63176632e-01
-3.47739041e-01 -9.59866047e-01 -4.92402017e-01 3.70634526e-01
3.20807509e-02 2.87718549e-02 -3.75293970e-01 8.06531429e-01
-2.20578358e-01 5.42926252e-01 6.48277253e-02 -6.21512949e-01
6.23752117e-01 4.27537203e-01 3.42617989e-01 -4.67345715e-01
-4.98602271e-01 6.05483726e-02 -2.44324788e-01 -4.57725137e-01
2.00349689e-01 -7.53777146e-01 -9.43197370e-01 -1.15399301e-01
-9.25486907e-02 -2.87634164e-01 7.06327498e-01 7.14076400e-01
6.23525918e-01 5.96764088e-01 4.89251196e-01 -9.33538020e-01
-1.10498166e+00 -9.84207094e-01 -2.87737817e-01 4.97661084e-01
5.18806398e-01 -4.78231132e-01 -4.34810668e-01 1.58634961e-01] | [9.028987884521484, 2.6753180027008057] |
2aac02fa-c467-4499-a68c-9f36ed204439 | efficient-explorative-key-term-selection | 2303.00315 | null | https://arxiv.org/abs/2303.00315v1 | https://arxiv.org/pdf/2303.00315v1.pdf | Efficient Explorative Key-term Selection Strategies for Conversational Contextual Bandits | Conversational contextual bandits elicit user preferences by occasionally querying for explicit feedback on key-terms to accelerate learning. However, there are aspects of existing approaches which limit their performance. First, information gained from key-term-level conversations and arm-level recommendations is not appropriately incorporated to speed up learning. Second, it is important to ask explorative key-terms to quickly elicit the user's potential interests in various domains to accelerate the convergence of user preference estimation, which has never been considered in existing works. To tackle these issues, we first propose ``ConLinUCB", a general framework for conversational bandits with better information incorporation, combining arm-level and key-term-level feedback to estimate user preference in one step at each time. Based on this framework, we further design two bandit algorithms with explorative key-term selection strategies, ConLinUCB-BS and ConLinUCB-MCR. We prove tighter regret upper bounds of our proposed algorithms. Particularly, ConLinUCB-BS achieves a regret bound of $O(\sqrt{dT\log T})$, better than the previous result $O(d\sqrt{T}\log T)$. Extensive experiments on synthetic and real-world data show significant advantages of our algorithms in learning accuracy (up to 54\% improvement) and computational efficiency (up to 72\% improvement), compared to the classic ConUCB algorithm, showing the potential benefit to recommender systems. | ['John C. S. Lui', 'Shuai Li', 'Xutong Liu', 'Zhiyong Wang'] | 2023-03-01 | null | null | null | null | ['multi-armed-bandits'] | ['miscellaneous'] | [ 5.35714719e-03 -1.88204199e-01 -7.92841077e-01 -4.35304433e-01
-1.27177751e+00 -8.05275500e-01 6.92477729e-03 7.37820119e-02
-4.43813056e-01 1.24990129e+00 3.06973010e-01 -7.02972531e-01
-7.77336001e-01 -6.23734236e-01 -8.01694989e-01 -8.26783836e-01
-2.36577570e-01 5.74303508e-01 6.88926652e-02 -2.02496678e-01
2.05448970e-01 7.97726139e-02 -1.27095020e+00 2.02958047e-01
1.19980276e+00 1.44377160e+00 2.31982410e-01 5.21699846e-01
-1.19362161e-01 5.25141656e-01 -2.13619635e-01 -5.12488961e-01
3.37659389e-01 -3.60120684e-01 -8.61358225e-01 -2.42298678e-01
1.01719774e-01 -4.25016403e-01 6.32859580e-03 8.69621933e-01
4.74031925e-01 6.23144865e-01 1.84746832e-01 -7.91895926e-01
-2.31204912e-01 9.75530863e-01 -7.15043366e-01 2.38008156e-01
4.05600131e-01 -2.02443480e-01 1.53203952e+00 -3.77441555e-01
1.14274919e-01 1.24485815e+00 5.06785274e-01 4.86143917e-01
-1.17321885e+00 -8.82812917e-01 9.35740232e-01 2.62484372e-01
-8.85672569e-01 -1.20003596e-01 7.16054380e-01 9.64848474e-02
3.53990614e-01 1.11785579e+00 9.04402375e-01 8.80627096e-01
-7.23690033e-01 1.38043439e+00 1.27903116e+00 -3.59008491e-01
4.50472564e-01 3.27861607e-01 5.26159286e-01 4.21135932e-01
1.10740922e-01 4.30111795e-05 -6.98950350e-01 -4.70506907e-01
5.49605727e-01 2.81641990e-01 -5.16000867e-01 -2.21324652e-01
-8.78416181e-01 8.52659941e-01 4.76746410e-01 -1.49300648e-02
-4.97044176e-01 1.18222803e-01 1.18849173e-01 5.46399057e-01
7.62593210e-01 3.87329638e-01 -6.68518007e-01 -5.98589301e-01
-7.61294961e-01 3.61716509e-01 8.64233494e-01 1.08797002e+00
8.17108452e-01 -4.54940349e-01 -3.53726268e-01 1.33880770e+00
2.56987035e-01 4.92091060e-01 3.91391754e-01 -9.08247113e-01
8.09560180e-01 4.40081835e-01 7.01251566e-01 -6.43473864e-01
-1.29708663e-01 -7.66805828e-01 -7.73631930e-01 -5.76435983e-01
4.49420601e-01 -4.52508807e-01 -3.60572904e-01 1.70643282e+00
4.93624806e-01 3.34189273e-02 -9.63712111e-02 1.06174612e+00
3.67309779e-01 6.46927595e-01 -3.46297234e-01 -7.89833009e-01
1.29114020e+00 -1.11173272e+00 -5.10086834e-01 -1.87094942e-01
6.05591536e-01 -6.21239245e-01 1.15943420e+00 5.98229527e-01
-1.21686423e+00 -4.17114086e-02 -7.22319067e-01 4.45769072e-01
-6.46695644e-02 -6.78384006e-02 1.16767907e+00 1.14237761e+00
-6.81755960e-01 3.46481770e-01 -2.69836426e-01 2.55911145e-02
2.91639835e-01 5.44168234e-01 3.67962718e-01 -1.59996510e-01
-1.31046236e+00 1.95916548e-01 8.61398056e-02 1.97859094e-01
-4.74854499e-01 -9.06620979e-01 -2.63791949e-01 1.41036212e-01
8.37489128e-01 -5.71290255e-01 1.68897128e+00 -8.97367001e-01
-1.78064167e+00 1.37565836e-01 -5.03001809e-01 -4.72716779e-01
7.40852714e-01 -4.77131963e-01 -1.76436707e-01 -3.17287058e-01
-3.29223394e-01 1.16104975e-01 3.89284223e-01 -1.02947474e+00
-1.17666042e+00 -3.56054664e-01 6.71291471e-01 4.97827381e-01
-6.85386300e-01 -2.66786903e-01 -6.50667071e-01 -6.29440963e-01
1.84638992e-01 -1.14793789e+00 -4.53220695e-01 -5.30666411e-01
-1.74355626e-01 -4.50157613e-01 2.72493273e-01 -2.51449257e-01
1.45509648e+00 -1.85917115e+00 -2.67235160e-01 5.46457231e-01
-2.74377108e-01 2.98941761e-01 7.78958711e-05 4.52947170e-01
2.97001302e-01 1.83776900e-01 2.48682410e-01 -2.15054601e-01
1.83795825e-01 1.58861399e-01 -5.10531187e-01 1.93415210e-01
-7.88793206e-01 6.68966830e-01 -9.76565480e-01 -1.69740289e-01
-1.71989053e-02 -2.99642812e-02 -9.40917253e-01 1.59970671e-01
-4.46965098e-01 2.62096167e-01 -9.01303947e-01 7.29913652e-01
6.66657865e-01 -2.89312214e-01 4.31628764e-01 1.89085752e-02
-1.99410990e-02 6.68091118e-01 -1.34393537e+00 1.33456349e+00
-8.51056874e-01 9.78708714e-02 2.52626270e-01 -1.13958979e+00
5.01149774e-01 3.38327080e-01 5.48192918e-01 -7.04454184e-01
-1.28411045e-02 2.73315370e-01 -3.71070653e-01 -1.81366354e-01
2.67767131e-01 9.17256903e-03 9.07051377e-03 7.62627184e-01
-4.47907299e-01 3.29157561e-01 1.44950211e-01 -2.46896548e-03
6.80548847e-01 -1.62504002e-01 2.14724466e-01 -2.77867287e-01
5.94473124e-01 -3.73501420e-01 6.26015306e-01 1.21416461e+00
3.14521492e-02 -4.86904662e-03 4.69980568e-01 -4.31403905e-01
-2.42005199e-01 -5.07949650e-01 -8.12072214e-03 1.91187978e+00
4.42235023e-01 -1.81960031e-01 -5.95076442e-01 -8.54945183e-01
1.25164792e-01 7.07424164e-01 -6.46101117e-01 3.03237557e-01
-3.77249271e-01 -1.10345256e+00 -7.69163519e-02 3.53759915e-01
3.46335679e-01 -6.79305553e-01 -3.05505544e-01 2.51968801e-01
-4.98840392e-01 -6.67897642e-01 -9.27643001e-01 2.49386504e-01
-1.06157899e+00 -7.09600091e-01 -9.07125235e-01 -3.98455530e-01
6.14339769e-01 6.77451193e-01 8.21430087e-01 -1.43868446e-01
2.60011971e-01 2.53594249e-01 -6.20300293e-01 -3.01675498e-01
3.13961565e-01 2.83583760e-01 -4.54014577e-02 2.55442619e-01
1.28032953e-01 -8.52063239e-01 -1.18233478e+00 7.64501631e-01
-5.50123036e-01 3.96814719e-02 8.37031662e-01 9.27090824e-01
3.02939862e-01 -2.57677525e-01 6.11775875e-01 -1.20927155e+00
8.86197567e-01 -4.87660229e-01 -7.30055273e-01 3.50309193e-01
-8.85878325e-01 3.76672372e-02 4.35849369e-01 -7.10578680e-01
-1.16351867e+00 -2.72467524e-01 -1.65074930e-01 2.06600204e-02
3.42522204e-01 5.70967853e-01 6.68833330e-02 -4.41637114e-02
4.84762996e-01 3.96335095e-01 -4.65307206e-01 -6.59539700e-01
3.73457938e-01 9.78279948e-01 -1.03934057e-01 -1.11697888e+00
3.72124702e-01 2.42481306e-01 -4.67670470e-01 -5.69860816e-01
-1.19105399e+00 -7.66346097e-01 1.28871426e-01 -3.94156277e-02
-4.48550731e-02 -8.59997034e-01 -1.59527612e+00 -1.64877594e-01
-6.02976561e-01 -2.97093660e-01 4.97590899e-02 7.24954486e-01
-5.69588661e-01 2.59213954e-01 -3.12330693e-01 -1.61318874e+00
-6.26366198e-01 -9.44813669e-01 6.10048950e-01 3.73333812e-01
-6.27200007e-02 -7.87413239e-01 -1.41462088e-01 8.90366077e-01
2.84702063e-01 -3.46578062e-01 8.23303580e-01 -7.24359453e-01
-7.59386122e-01 -1.02313221e-01 -1.86448544e-01 -1.20440843e-02
1.59000590e-01 -7.74269879e-01 -7.80483842e-01 -5.77740371e-01
-2.01960534e-01 -3.97097319e-01 7.13592172e-01 4.06567544e-01
1.31859136e+00 -8.93961906e-01 -4.71602082e-01 4.70099509e-01
1.12544453e+00 6.32450402e-01 1.64720699e-01 2.58582115e-01
1.28624037e-01 3.00554454e-01 9.14607525e-01 9.05040264e-01
2.80504376e-01 7.08434045e-01 3.35402519e-01 5.36458753e-02
5.50948679e-01 -1.20363981e-01 2.90060610e-01 6.57417357e-01
-3.67085636e-01 -2.91823000e-01 -2.82475263e-01 7.50043511e-01
-2.22817421e+00 -6.37027740e-01 2.28981018e-01 2.50120091e+00
1.10955250e+00 2.52543151e-01 5.00255704e-01 7.58790225e-02
5.15537322e-01 8.82253945e-02 -7.94547200e-01 -4.47812408e-01
2.79389828e-01 -2.34930236e-02 5.66632628e-01 6.31178439e-01
-7.68666863e-01 6.47087455e-01 5.30253983e+00 1.02950549e+00
-1.03095782e+00 1.22419074e-01 9.06940520e-01 -6.67895079e-01
-5.27055204e-01 -7.03010634e-02 -8.64865661e-01 5.92047870e-01
5.67734659e-01 -1.93990618e-01 1.03449583e+00 1.02361965e+00
2.23055974e-01 -2.26223096e-01 -1.03270268e+00 1.03242517e+00
-3.40330690e-01 -1.18282413e+00 -2.90403426e-01 1.16064161e-01
9.05049443e-01 -1.13993026e-01 1.10953525e-01 5.45317829e-01
8.68389070e-01 -4.52534884e-01 3.92420888e-01 1.62228659e-01
4.71490443e-01 -1.11765993e+00 6.17166758e-01 6.61607742e-01
-9.50295269e-01 -4.26107377e-01 -4.18762118e-01 -1.35726452e-01
-8.83446708e-02 6.87697589e-01 -9.00325775e-01 7.86099195e-01
8.93972754e-01 2.45480001e-01 1.58074260e-01 1.07187390e+00
7.37004280e-02 9.15172696e-01 -5.40920317e-01 -8.35204482e-01
5.87011397e-01 -4.18894053e-01 3.80852282e-01 1.07926023e+00
4.32793498e-01 4.06043738e-01 2.52269775e-01 2.46967047e-01
-1.42486140e-01 4.88995135e-01 9.74114835e-02 1.71502069e-01
7.14014590e-01 1.11293960e+00 -2.78991610e-01 -2.42034957e-01
-2.33598247e-01 6.57687902e-01 2.98022509e-01 5.27208924e-01
-8.64486992e-01 -7.47237504e-02 8.23591053e-01 6.73496723e-03
6.04121983e-01 2.75351584e-01 -1.42105654e-01 -9.01162446e-01
1.06083840e-01 -9.58866715e-01 7.02357292e-01 -3.10701340e-01
-1.12254357e+00 2.63763070e-01 8.84570368e-03 -1.05157292e+00
-9.24302489e-02 -1.46110907e-01 -2.86282659e-01 8.12906027e-01
-1.51171708e+00 -9.40508962e-01 -2.12158877e-02 5.60435534e-01
6.60436034e-01 1.74887121e-01 7.82095671e-01 2.65582204e-01
-3.98372173e-01 1.04167998e+00 7.51148820e-01 -4.10048991e-01
6.03201330e-01 -1.17219353e+00 -6.11640327e-02 2.04268962e-01
6.84543625e-02 9.14857924e-01 7.59497821e-01 -1.76830560e-01
-1.57822728e+00 -7.79591084e-01 4.96899962e-01 3.23094577e-02
4.36071813e-01 -3.95317107e-01 -2.94733822e-01 5.87104857e-01
-2.75295954e-02 -2.80122757e-01 1.09307897e+00 1.01901901e+00
-2.13458613e-01 -8.07351530e-01 -1.01183987e+00 8.53106737e-01
1.20681572e+00 -1.64192513e-01 -2.82534838e-01 4.45840627e-01
7.50349760e-01 -5.12416363e-01 -7.77218103e-01 3.64328623e-01
1.04555142e+00 -1.06734586e+00 9.58221018e-01 -4.76641387e-01
-2.63460994e-01 2.62071639e-01 -1.79192632e-01 -1.32720828e+00
-2.04038978e-01 -1.38752389e+00 -3.14340025e-01 1.08647633e+00
6.64268970e-01 -9.43835974e-01 1.04328835e+00 6.86379373e-01
2.72856712e-01 -1.21164799e+00 -9.88977790e-01 -5.06149709e-01
-1.75199345e-01 -5.75997233e-01 9.39019024e-01 6.69313073e-01
3.43809545e-01 1.75176427e-01 -9.17245507e-01 -6.89532934e-03
4.15569693e-01 8.28418672e-01 7.74492502e-01 -1.13970029e+00
-6.29597902e-01 -5.60544252e-01 5.92868030e-01 -2.02230263e+00
-3.27387542e-01 -4.34031665e-01 -1.04078770e-01 -1.23739326e+00
3.16863477e-01 -9.14438486e-01 -9.03501034e-01 3.86015415e-01
-3.10514241e-01 5.05970651e-03 5.51084951e-02 4.76081967e-02
-8.63818288e-01 4.54478681e-01 1.38388503e+00 -9.45216939e-02
-4.85859007e-01 8.15087795e-01 -1.17856944e+00 4.53775585e-01
5.71949959e-01 -3.56531322e-01 -7.45498180e-01 -1.19600385e-01
5.18369615e-01 5.48621297e-01 -2.55943090e-01 -5.30825675e-01
1.75717399e-01 -6.38381481e-01 3.19147855e-02 -7.86130607e-01
5.84581494e-01 -7.64735401e-01 -1.62506655e-01 3.04412752e-01
-6.04059339e-01 -3.53747576e-01 3.88305187e-02 9.51240003e-01
1.17114909e-01 -2.34556764e-01 2.95423657e-01 -2.32008547e-01
-9.48367864e-02 3.92049402e-01 -3.10190052e-01 -9.58297253e-02
8.10268104e-01 -1.01913735e-01 -5.56357689e-02 -8.47303987e-01
-5.36147833e-01 6.10142350e-01 -1.69243321e-01 1.50402471e-01
9.78865251e-02 -1.28362572e+00 -3.33335102e-01 -4.90052290e-02
-6.39347285e-02 -2.67260429e-02 3.96618873e-01 1.02165890e+00
1.60119593e-01 9.37290967e-01 4.52526480e-01 -3.83119881e-01
-1.36633480e+00 4.00775999e-01 9.27089080e-02 -5.92131197e-01
-6.40650392e-02 1.14216924e+00 2.82045424e-01 -4.20997381e-01
6.95441008e-01 -5.38477004e-01 1.00026820e-02 3.04089159e-01
4.92566913e-01 5.29624164e-01 -2.62422282e-02 2.94935316e-01
-2.77002960e-01 2.75985807e-01 -3.81725699e-01 -1.92853987e-01
1.44837189e+00 -3.77450347e-01 3.60022709e-02 3.92891318e-01
8.77671003e-01 2.92649299e-01 -1.12227690e+00 -6.43627584e-01
7.97078535e-02 -7.16893494e-01 1.02510124e-01 -1.13912022e+00
-8.42235506e-01 4.37895328e-01 4.79287654e-01 5.35005212e-01
1.22750199e+00 -2.87818849e-01 1.04138505e+00 7.83886373e-01
5.68262756e-01 -1.19926071e+00 -1.67091191e-01 3.37211221e-01
6.89473689e-01 -1.13741922e+00 9.87490192e-02 -2.91036218e-01
-5.03454864e-01 6.49062574e-01 3.91891330e-01 2.82455802e-01
6.62851393e-01 -3.55276078e-01 -3.45846638e-02 4.02365744e-01
-9.94498730e-01 -2.17097342e-01 3.78753394e-01 -9.91800576e-02
4.43896621e-01 2.50870109e-01 -6.35572791e-01 1.02958047e+00
-1.91257045e-01 5.08008245e-03 -1.54265419e-01 7.28694916e-01
-5.39985716e-01 -1.41950262e+00 -3.39711308e-01 5.88475943e-01
-6.80273652e-01 -2.29415625e-01 -1.71108499e-01 4.15261209e-01
-2.58995771e-01 1.16250491e+00 -3.91342819e-01 -2.82600999e-01
2.21003816e-01 4.37740870e-02 3.33267599e-01 -1.09565623e-01
-6.46112978e-01 5.20179987e-01 3.72055590e-01 -4.86718267e-01
-3.89361560e-01 -5.80630720e-01 -7.15397358e-01 -3.04780751e-01
-6.56266034e-01 1.00387228e+00 4.19573277e-01 9.06656027e-01
3.24038327e-01 2.31123731e-01 1.02340031e+00 -5.99118412e-01
-7.93261290e-01 -1.04570866e+00 -4.79343176e-01 2.08834186e-01
4.64922786e-01 -5.63776731e-01 -2.42852956e-01 -4.60536391e-01] | [4.639928817749023, 3.337019205093384] |
9ccad644-1023-43c8-85f3-7a0e6e566058 | positive-unlabeled-classification-under-class | 1809.07011 | null | https://arxiv.org/abs/1809.07011v4 | https://arxiv.org/pdf/1809.07011v4.pdf | Positive-Unlabeled Classification under Class Prior Shift and Asymmetric Error | Bottlenecks of binary classification from positive and unlabeled data (PU classification) are the requirements that given unlabeled patterns are drawn from the test marginal distribution, and the penalty of the false positive error is identical to the false negative error. However, such requirements are often not fulfilled in practice. In this paper, we generalize PU classification to the class prior shift and asymmetric error scenarios. Based on the analysis of the Bayes optimal classifier, we show that given a test class prior, PU classification under class prior shift is equivalent to PU classification with asymmetric error. Then, we propose two different frameworks to handle these problems, namely, a risk minimization framework and density ratio estimation framework. Finally, we demonstrate the effectiveness of the proposed frameworks and compare both frameworks through experiments using benchmark datasets. | ['Masashi Sugiyama', 'Nontawat Charoenphakdee'] | 2018-09-19 | null | null | null | null | ['density-ratio-estimation'] | ['methodology'] | [ 4.28087056e-01 2.37431094e-01 -5.23420632e-01 -6.21230304e-01
-6.33493483e-01 -5.03451645e-01 1.39628425e-01 6.97395578e-02
-2.29306206e-01 1.15907860e+00 -6.19539440e-01 -5.27787328e-01
-1.61769301e-01 -8.13414037e-01 -7.21584618e-01 -8.04107070e-01
5.20623326e-01 4.77349132e-01 1.77817255e-01 6.52370393e-01
3.27201486e-01 3.12265456e-01 -1.57487905e+00 -1.45614818e-01
9.78480041e-01 1.46470904e+00 -1.97120234e-01 4.27948743e-01
5.72293773e-02 5.85295916e-01 -5.62817752e-01 -4.88238990e-01
4.63680714e-01 -5.84267676e-01 -8.59421492e-01 3.70130569e-01
2.23357260e-01 -3.45681965e-01 3.62110585e-01 1.43685853e+00
2.36667886e-01 1.74107686e-01 1.24142420e+00 -1.77004671e+00
-5.21089733e-01 2.36430928e-01 -6.88929498e-01 1.37411296e-01
8.78252089e-02 -2.67674297e-01 9.77945507e-01 -8.95743012e-01
2.44304910e-01 9.49665487e-01 5.64123809e-01 6.20838284e-01
-1.20980322e+00 -4.95486170e-01 1.27636835e-01 6.11364022e-02
-1.35857129e+00 -1.08097278e-01 2.03729957e-01 -4.61806446e-01
5.32613218e-01 1.34548202e-01 2.70745426e-01 1.06268561e+00
2.18922257e-01 8.83189201e-01 1.18591702e+00 -5.71247041e-01
6.52008533e-01 5.81831634e-01 7.58774102e-01 3.73930484e-01
7.08473802e-01 8.91175270e-02 -1.80609882e-01 -3.08502913e-01
2.71963567e-01 5.06289080e-02 -3.34101230e-01 -6.60115302e-01
-6.10357881e-01 7.69477606e-01 -1.09970443e-01 -6.20115958e-02
-7.20676109e-02 -4.59204376e-01 1.42207250e-01 2.66459078e-01
3.51436257e-01 -1.75565094e-01 -4.88727272e-01 1.96368441e-01
-5.30273974e-01 -7.70937800e-02 1.09000087e+00 1.19915235e+00
6.37920260e-01 -1.80034265e-01 -1.96662650e-01 6.62445784e-01
5.60695887e-01 6.61977828e-01 4.19297695e-01 -4.69311804e-01
2.31083855e-01 5.08671820e-01 3.81416827e-01 -5.11064827e-01
1.76400021e-02 -5.29383183e-01 -5.92268407e-01 -2.43593957e-02
6.39009476e-01 -2.14977592e-01 -9.07638431e-01 1.66980970e+00
3.32705170e-01 3.57957184e-01 3.43410462e-01 6.24884188e-01
1.59622639e-01 4.19098139e-01 8.32228065e-02 -5.69222510e-01
1.00907922e+00 -6.29117250e-01 -7.41178393e-01 -3.03972550e-02
7.86518455e-01 -5.24284184e-01 1.04675972e+00 4.76203561e-01
-8.20791185e-01 -2.82178849e-01 -1.26119089e+00 4.25629169e-01
-1.23109892e-01 1.78085297e-01 2.35827416e-01 1.29392231e+00
-4.45391953e-01 3.50076884e-01 -7.50383735e-01 -4.51120317e-01
2.59938985e-01 3.27550769e-01 -7.82110542e-02 -2.83024536e-04
-7.91294992e-01 6.37846529e-01 5.86898386e-01 1.08870991e-01
-7.14573860e-01 -6.22741282e-01 -7.63477445e-01 3.39372039e-01
4.27375555e-01 -2.08798662e-01 1.35693812e+00 -9.70202267e-01
-1.36729705e+00 5.98036587e-01 -1.86623856e-01 -4.01505172e-01
7.00829625e-01 -1.53161973e-01 -2.14747936e-01 -1.32025883e-01
1.42899543e-01 4.92273197e-02 5.93488395e-01 -1.07422173e+00
-1.04587376e+00 -4.93135244e-01 -2.39189819e-01 -2.76524927e-02
-2.64000297e-01 -1.81925416e-01 -9.13676545e-02 -2.65327692e-01
3.48642856e-01 -6.84079647e-01 -1.12537807e-02 -6.84320554e-02
-4.53379244e-01 -3.45648855e-01 1.11547410e+00 -3.27368677e-01
9.61902559e-01 -2.41974545e+00 -4.40626621e-01 6.00623488e-01
-7.94865116e-02 6.29516095e-02 3.16952586e-01 -1.88437238e-01
-5.71873710e-02 3.08655441e-01 -4.27132934e-01 -6.22280650e-02
1.43982008e-01 3.99416327e-01 -4.55383003e-01 6.16567612e-01
4.97482300e-01 3.49240154e-01 -8.87425125e-01 -4.97945219e-01
1.14493212e-02 -5.78994267e-02 -4.73779023e-01 1.68262377e-01
9.88303348e-02 1.53653726e-01 -3.71161669e-01 8.01261604e-01
9.96605814e-01 -4.20180887e-01 3.94421309e-01 2.45995186e-02
2.54317462e-01 4.48280238e-02 -1.45477474e+00 6.19114161e-01
-7.82319754e-02 8.20962787e-02 1.52501033e-03 -1.39267707e+00
9.66716528e-01 3.32615346e-01 4.50025611e-02 -3.19750756e-01
4.04164702e-01 3.46234530e-01 4.63550240e-02 -1.89754948e-01
6.13555647e-02 -4.80258405e-01 1.99612513e-01 5.60064435e-01
2.43655711e-01 2.45908812e-01 2.10725263e-01 -1.33853942e-01
7.39770651e-01 4.27258201e-02 6.08041346e-01 -3.68332952e-01
5.35206497e-01 -4.34891999e-01 1.02608109e+00 1.03939080e+00
-6.43583417e-01 4.95724857e-01 8.80800009e-01 -9.78863686e-02
-6.54192865e-01 -1.43564045e+00 -6.19431913e-01 5.76357424e-01
3.81718814e-01 2.00155675e-02 -6.52578533e-01 -1.34404707e+00
-3.83848064e-02 6.03243053e-01 -5.12663484e-01 -1.43748075e-01
-4.69891727e-03 -1.19711220e+00 2.04398766e-01 6.77986026e-01
4.14049864e-01 -4.46663857e-01 -4.92582440e-01 -1.77838966e-01
-2.65364125e-02 -9.28269923e-01 -9.96186659e-02 4.76083130e-01
-7.75966406e-01 -1.31387651e+00 -5.60863137e-01 -8.59439194e-01
8.23152423e-01 1.52207702e-01 7.53849089e-01 -1.70881659e-01
-5.09515069e-02 5.05039632e-01 -4.21047807e-01 -4.20089811e-01
-3.77289236e-01 -2.05243751e-01 2.83516824e-01 2.50579000e-01
6.20270729e-01 -1.91066027e-01 -3.04964125e-01 6.92631125e-01
-9.45834458e-01 -3.20978582e-01 5.75044334e-01 1.00545120e+00
8.33148718e-01 7.68030807e-02 7.82391608e-01 -8.49670589e-01
2.02588052e-01 -7.05286145e-01 -8.34840238e-01 7.40740716e-01
-8.85130048e-01 6.30610064e-02 4.17122483e-01 -6.11721277e-01
-1.07160592e+00 1.69533882e-02 2.02249303e-01 -3.31842065e-01
-1.39060646e-01 3.20859104e-01 -5.71598291e-01 1.00067966e-02
4.80435461e-01 4.52978015e-02 -1.02186263e-01 -3.49305838e-01
-2.01930881e-01 9.06305194e-01 2.76879251e-01 -7.84539282e-01
4.27793026e-01 3.80909771e-01 9.81483608e-02 -6.53246880e-01
-1.11709607e+00 -3.98617446e-01 -5.27848065e-01 -1.06563643e-01
6.73985600e-01 -3.54413569e-01 -6.11042440e-01 1.75863251e-01
-7.78562069e-01 1.50590134e-03 -3.29475671e-01 7.71372080e-01
-5.67914605e-01 6.78383887e-01 -3.42819899e-01 -1.26619530e+00
6.77131210e-03 -1.23113143e+00 6.60780728e-01 4.29263622e-01
8.35704058e-02 -8.86340439e-01 -2.40729988e-01 3.70234326e-02
-1.18790716e-01 2.79499255e-02 1.06870437e+00 -1.15826631e+00
-2.18181491e-01 -4.75826710e-01 -3.46390575e-01 7.54622877e-01
1.65744245e-01 1.62447561e-02 -9.59326506e-01 -2.12953046e-01
1.60061821e-01 -4.15382743e-01 5.21331251e-01 2.95000583e-01
1.27647257e+00 -1.47311389e-01 -4.37821120e-01 3.54803294e-01
1.35284150e+00 4.13026661e-01 5.14595330e-01 -1.23882294e-01
1.11404344e-01 6.60839915e-01 9.93614674e-01 6.37015820e-01
-1.08975321e-01 4.11178350e-01 1.93794876e-01 3.52887392e-01
4.80016202e-01 -1.86563909e-01 1.93865940e-01 3.97637308e-01
3.49876016e-01 -5.03673971e-01 -7.88780093e-01 3.33388448e-01
-1.79230249e+00 -7.94108808e-01 -2.71221966e-01 2.75375295e+00
6.28964722e-01 2.17411458e-01 -5.32015339e-02 4.92471486e-01
1.10547590e+00 -6.74430728e-01 -5.11602819e-01 -2.24160671e-01
-3.44508439e-02 1.36862367e-01 4.87123281e-01 5.02866745e-01
-1.16881871e+00 3.69662791e-01 6.84615183e+00 8.68755996e-01
-8.72845709e-01 8.92269835e-02 1.09170318e+00 3.35103422e-01
1.09934462e-02 1.22435361e-01 -1.09304428e+00 4.52319115e-01
7.75486827e-01 -2.63343096e-01 -9.15895179e-02 1.06431794e+00
-2.87626505e-01 -3.59516501e-01 -1.40692520e+00 9.64919746e-01
4.70741140e-03 -4.68069494e-01 -1.52439684e-01 1.09155163e-01
5.39801478e-01 -4.73675907e-01 1.26202613e-01 5.66784680e-01
3.47433597e-01 -6.66977346e-01 7.74219871e-01 7.56057054e-02
5.65330327e-01 -6.80149913e-01 1.04469025e+00 5.00447214e-01
-9.39454496e-01 -1.43427745e-01 -4.65393126e-01 -7.30013996e-02
-3.89179550e-02 6.94931209e-01 -8.78627062e-01 5.86016476e-01
6.08165026e-01 5.59439361e-01 -2.75199950e-01 1.05076158e+00
-1.69505879e-01 7.25986838e-01 -4.38810557e-01 -4.88490798e-02
-1.31678611e-01 -3.37977707e-01 2.32762575e-01 9.67634320e-01
3.64797205e-01 1.64194833e-02 2.66271591e-01 8.25135767e-01
1.14715971e-01 2.27744758e-01 -4.63972449e-01 2.13549927e-01
4.40467983e-01 1.01290846e+00 -1.13858211e+00 -3.84086400e-01
-6.66893005e-01 6.91728473e-01 2.22781613e-01 4.27289248e-01
-9.55101609e-01 -1.83671191e-01 1.91527739e-01 -3.32922339e-01
1.86945349e-01 4.73660469e-01 -4.20460284e-01 -1.22432876e+00
3.54330033e-01 -4.98940319e-01 6.65954173e-01 -2.26029634e-01
-1.69884956e+00 1.15463167e-01 1.12218805e-01 -1.43892431e+00
-5.49477153e-02 -9.85472023e-01 -4.75048363e-01 8.24544132e-01
-1.31407738e+00 -4.61765736e-01 5.32405227e-02 3.27826738e-01
1.64302796e-01 1.59764037e-01 6.56816185e-01 2.44771406e-01
-8.29195619e-01 8.36138964e-01 1.72438145e-01 1.17288217e-01
5.51229417e-01 -1.28165817e+00 -3.62277776e-01 9.82644737e-01
-3.28698128e-01 3.48021924e-01 4.04006898e-01 -6.02599919e-01
-9.65975404e-01 -1.00348723e+00 6.09767199e-01 -3.42035621e-01
5.16309679e-01 -1.13021575e-01 -7.81210244e-01 7.73481488e-01
-4.57515508e-01 3.58550996e-01 1.07647943e+00 9.44048390e-02
-3.47962618e-01 -3.23168300e-02 -1.60367024e+00 3.30027044e-01
6.46670043e-01 -1.92704827e-01 -4.53807473e-01 2.81019866e-01
2.77496845e-01 -1.79072823e-02 -6.86768949e-01 8.50783885e-01
7.85680830e-01 -8.04737508e-01 5.30814350e-01 -7.70236611e-01
5.13236970e-02 -3.26312214e-01 -4.69962627e-01 -8.76266479e-01
6.13799021e-02 -2.97835674e-02 -2.91915629e-02 1.21648896e+00
6.20026588e-01 -9.55500901e-01 7.88835406e-01 8.05188417e-01
1.64473251e-01 -5.43128014e-01 -1.12101471e+00 -1.00336635e+00
-2.79589724e-02 -3.80576581e-01 1.78534508e-01 9.00632560e-01
1.20062940e-01 2.22778350e-01 -2.41942018e-01 4.65459466e-01
5.04853606e-01 2.63878517e-03 4.34725970e-01 -1.36136949e+00
-6.66954815e-01 -3.24383378e-01 -7.13987231e-01 -9.10267413e-01
2.73788452e-01 -6.10262692e-01 2.91917443e-01 -1.00597274e+00
5.22164762e-01 -3.92446518e-01 -5.37777960e-01 3.47763419e-01
-4.30596888e-01 5.06470054e-02 -1.83879822e-01 1.20362468e-01
-4.83812690e-01 4.91643429e-01 7.38965094e-01 3.34649906e-02
3.14347334e-02 4.79528755e-01 -3.88582170e-01 7.21732557e-01
8.29024851e-01 -4.91391659e-01 -5.69563508e-01 1.29582584e-02
1.23941138e-01 1.65443897e-01 1.40843868e-01 -8.89319003e-01
-2.82007784e-01 -2.58375287e-01 3.77021492e-01 -5.92117727e-01
-3.19217518e-02 -1.00731862e+00 -2.53901303e-01 6.02647364e-01
-2.70045877e-01 -3.48165452e-01 -1.09956391e-01 9.41696167e-01
-9.27944034e-02 -7.14142859e-01 9.76964355e-01 2.46985719e-01
-1.99531794e-01 8.49444345e-02 -4.03514653e-01 -2.24310923e-02
1.43631661e+00 -3.25514048e-01 -3.23216170e-01 -1.24999627e-01
-8.57090056e-01 2.01410264e-01 2.22781539e-01 -8.32877532e-02
4.60936964e-01 -1.18528152e+00 -3.69757503e-01 2.92125791e-01
2.60793000e-01 -2.04906195e-01 -9.99238249e-03 1.10698211e+00
-1.88005924e-01 3.51146519e-01 6.21531978e-02 -6.84128404e-01
-1.15630531e+00 6.56696022e-01 4.11006957e-01 -1.23848386e-01
-1.33145586e-01 5.27509987e-01 5.56590676e-01 -4.64047611e-01
3.93024087e-01 -3.98939878e-01 -9.70208570e-02 -2.83264667e-01
6.94616437e-01 5.35855055e-01 2.27857471e-01 -2.06782952e-01
-2.06588253e-01 1.62247971e-01 -1.22428909e-01 -2.20526218e-01
7.10091054e-01 -5.71744591e-02 9.30658057e-02 5.83109438e-01
9.26607490e-01 -1.32965118e-01 -1.14920962e+00 -1.59506783e-01
2.58867592e-01 -5.39685190e-01 -2.53038287e-01 -6.26867592e-01
-7.59993494e-01 9.25652742e-01 8.88014555e-01 4.38144833e-01
1.00599468e+00 -1.31812036e-01 -3.84763926e-02 3.69239569e-01
3.31213176e-01 -1.03478551e+00 -3.00436080e-01 4.10159737e-01
1.54680550e-01 -1.30628753e+00 -3.08614522e-01 -8.59808207e-01
-5.15447140e-01 9.35770214e-01 9.20874655e-01 6.70001805e-02
1.00207078e+00 2.77856618e-01 -2.43310213e-01 4.53706801e-01
-5.92192233e-01 -1.21006951e-01 1.90881640e-01 7.74904311e-01
4.28618193e-01 1.97789222e-01 -6.74318790e-01 9.10664856e-01
1.96867332e-01 6.08353689e-02 4.86175328e-01 1.24774373e+00
-3.73116463e-01 -9.81103003e-01 -4.77306515e-01 6.78868532e-01
-5.44464946e-01 1.20598383e-01 -2.16590852e-01 6.39025629e-01
-4.52529341e-02 1.11316311e+00 2.51757383e-01 -1.78874850e-01
-5.79625480e-02 4.39913750e-01 5.31703591e-01 -7.77781427e-01
2.30818421e-01 -6.11152165e-02 -2.78481215e-01 -1.05940271e-02
-2.99270570e-01 -6.40405059e-01 -1.01773965e+00 -1.19455755e-02
-7.87655711e-01 2.96930283e-01 3.97575140e-01 1.16792595e+00
-4.15849994e-04 6.82713389e-02 6.86778426e-01 -1.79619133e-01
-1.16450346e+00 -1.09022403e+00 -9.94945645e-01 -3.85066913e-03
-2.86160018e-02 -9.56117928e-01 -7.03792512e-01 3.51200365e-02] | [9.0558443069458, 4.049476623535156] |
b40b644c-5d99-43e0-87ca-d8175735ee61 | to-root-artificial-intelligence-deeply-in | 2009.05678 | null | https://arxiv.org/abs/2009.05678v1 | https://arxiv.org/pdf/2009.05678v1.pdf | To Root Artificial Intelligence Deeply in Basic Science for a New Generation of AI | One of the ambitions of artificial intelligence is to root artificial intelligence deeply in basic science while developing brain-inspired artificial intelligence platforms that will promote new scientific discoveries. The challenges are essential to push artificial intelligence theory and applied technologies research forward. This paper presents the grand challenges of artificial intelligence research for the next 20 years which include:~(i) to explore the working mechanism of the human brain on the basis of understanding brain science, neuroscience, cognitive science, psychology and data science; (ii) how is the electrical signal transmitted by the human brain? What is the coordination mechanism between brain neural electrical signals and human activities? (iii)~to root brain-computer interface~(BCI) and brain-muscle interface~(BMI) technologies deeply in science on human behaviour; (iv)~making research on knowledge-driven visual commonsense reasoning~(VCR), develop a new inference engine for cognitive network recognition~(CNR); (v)~to develop high-precision, multi-modal intelligent perceptrons; (vi)~investigating intelligent reasoning and fast decision-making systems based on knowledge graph~(KG). We believe that the frontier theory innovation of AI, knowledge-driven modeling methodologies for commonsense reasoning, revolutionary innovation and breakthroughs of the novel algorithms and new technologies in AI, and developing responsible AI should be the main research strategies of AI scientists in the future. | ['Jingan Yang', 'Yang Peng'] | 2020-09-11 | null | null | null | null | ['visual-commonsense-reasoning'] | ['reasoning'] | [ 2.60108620e-01 6.89368546e-02 1.29708841e-01 -2.55238593e-01
9.50731874e-01 -2.54714698e-01 3.60664636e-01 -4.08087462e-01
-2.17129469e-01 5.51756084e-01 -6.68460801e-02 -5.51870286e-01
-5.38562357e-01 -1.08277249e+00 -4.43054557e-01 -4.11783904e-01
-1.65487945e-01 3.26468766e-01 -5.74807748e-02 -6.78336918e-01
6.10854447e-01 7.24300444e-01 -1.68825543e+00 1.53519377e-01
1.04580009e+00 1.13334322e+00 2.72235632e-01 6.79349601e-01
-2.83771247e-01 1.37055206e+00 -4.90058541e-01 -2.05835730e-01
-2.72743046e-01 -5.89842558e-01 -1.16757107e+00 -3.84766817e-01
-5.79245031e-01 3.00043464e-01 -6.26775324e-01 1.55420625e+00
2.94481546e-01 1.03595041e-01 7.82049894e-01 -1.46983516e+00
-1.39048874e+00 7.14603364e-01 -3.65032822e-01 7.29465187e-01
4.19627309e-01 3.15200418e-01 3.05931211e-01 -3.86274695e-01
5.24964929e-01 1.28976989e+00 3.06863725e-01 7.73051322e-01
-7.76251912e-01 -9.65146244e-01 -2.48827748e-02 1.27323723e+00
-1.41780376e+00 -2.16571510e-01 1.00948596e+00 -6.18893743e-01
1.12228358e+00 2.68346459e-01 1.13902712e+00 8.92973006e-01
6.36388183e-01 7.80187249e-01 1.48177361e+00 -5.12133121e-01
3.74043077e-01 2.39036933e-01 7.31174707e-01 9.88586426e-01
3.09098244e-01 1.49638265e-01 -6.94096625e-01 4.88731772e-01
9.00476515e-01 -2.75882185e-02 -3.40875924e-01 5.32172143e-01
-1.46210408e+00 6.12608612e-01 6.97923124e-01 7.63450623e-01
-7.01202214e-01 2.76636988e-01 2.50564784e-01 5.20543337e-01
-1.86391547e-01 6.14749014e-01 -2.53212899e-01 -1.08759597e-01
-5.53912699e-01 -1.99603196e-03 8.88100982e-01 8.01109850e-01
6.43464148e-01 6.51458442e-01 2.53703117e-01 6.70939863e-01
4.17481691e-01 8.77972305e-01 7.48167992e-01 -1.21818078e+00
-1.64817184e-01 8.98961782e-01 -6.43320382e-01 -1.19376993e+00
-6.24013186e-01 -4.37105834e-01 -1.24130070e+00 3.53402525e-01
-2.14164317e-01 -2.69963324e-01 -8.46627235e-01 1.51036310e+00
-2.03756958e-01 -2.14386526e-02 2.28305608e-02 6.98823333e-01
1.15300179e+00 5.51114142e-01 3.94149777e-03 -2.91584998e-01
1.45826423e+00 -4.27710414e-01 -1.16530693e+00 -2.90144473e-01
-1.51096727e-03 -2.66654715e-02 6.96847498e-01 5.36188662e-01
-1.14188254e+00 -6.20450258e-01 -1.06476629e+00 4.79501560e-02
-9.87222254e-01 -2.74891794e-01 1.16231370e+00 5.61531782e-01
-1.13636792e+00 2.37320751e-01 -4.94369894e-01 -4.63828415e-01
6.42533243e-01 3.85968387e-01 -2.39850394e-02 1.57636292e-02
-1.64830303e+00 1.55682850e+00 6.66885018e-01 2.74403747e-02
-5.37239671e-01 -4.26074117e-01 -5.80145061e-01 -7.94000179e-02
7.12602288e-02 -1.16124928e+00 5.97642004e-01 -1.12763476e+00
-1.50578415e+00 9.16941345e-01 -1.69490874e-01 -4.41911817e-01
-2.57269830e-01 3.32751453e-01 -1.00553191e+00 3.62113476e-01
-2.68126786e-01 8.40774894e-01 4.62547600e-01 -1.07175374e+00
-4.54068691e-01 -6.02298975e-01 -4.02110517e-01 5.74735738e-02
-2.02825770e-01 2.71360893e-02 -7.82953203e-02 -4.59992200e-01
1.80224776e-01 -6.36072338e-01 1.91201195e-01 -1.39618978e-01
-1.92294225e-01 -4.87792701e-01 6.58106029e-01 -7.48526692e-01
1.16621506e+00 -2.04789352e+00 4.56521600e-01 5.33796668e-01
4.45525467e-01 3.62411320e-01 1.72339410e-01 5.61354123e-02
-1.43735215e-01 6.55538291e-02 1.88496321e-01 9.74183857e-01
6.83689956e-04 2.24394470e-01 -2.44742408e-01 3.08018737e-02
5.80639672e-03 1.61147785e+00 -8.06175709e-01 -3.23602021e-01
4.05742854e-01 3.60464305e-01 -4.93346602e-02 -1.16462894e-01
1.14891723e-01 1.40441224e-01 -6.20103359e-01 8.70346963e-01
1.80757895e-01 -1.50542483e-01 -1.77938387e-01 -3.02791238e-01
-3.13469744e-03 -1.56427324e-01 -8.79023612e-01 1.34027815e+00
2.02836283e-02 1.19652903e+00 -8.93564075e-02 -1.44610155e+00
9.74473178e-01 3.78564894e-01 1.33787245e-01 -1.09615636e+00
5.87576270e-01 -8.28080326e-02 5.07314444e-01 -9.59997535e-01
-4.41151351e-01 -2.17575252e-01 4.07820314e-01 2.86978096e-01
-4.08316180e-02 -3.21489990e-01 -7.65446052e-02 1.20337635e-01
1.30373955e+00 -3.32562625e-01 3.55817080e-01 -3.98441225e-01
5.39991617e-01 2.42986411e-01 2.12065995e-01 5.34528971e-01
-5.20920336e-01 -3.91181737e-01 6.56296313e-02 -6.19803965e-01
-4.69750911e-01 -1.19067335e+00 4.29984629e-02 1.12631142e+00
2.94696093e-01 3.77117604e-01 -8.50642145e-01 4.01225835e-01
-1.85467631e-01 1.23699605e+00 -4.38161969e-01 -5.51715612e-01
-3.20601650e-02 -7.90993929e-01 8.45297933e-01 5.83070755e-01
1.21386504e+00 -1.63761270e+00 -7.97032118e-01 2.11682856e-01
-2.00917751e-01 -1.00972641e+00 6.60555482e-01 3.58449847e-01
-9.32047844e-01 -1.04817963e+00 -1.35049596e-01 -9.03057516e-01
4.96340245e-01 1.44987758e-02 1.00420654e+00 1.43204719e-01
-6.31454349e-01 5.48220217e-01 -1.14255458e-01 -1.09556043e+00
-9.34292823e-02 -5.08230329e-01 2.41326988e-01 -5.06555200e-01
7.70982504e-01 -8.78512859e-01 -5.97918689e-01 2.54207812e-02
-7.19769955e-01 3.07801038e-01 8.09070766e-01 4.32663709e-01
4.85070683e-02 9.00878370e-01 9.00006115e-01 -3.97192985e-01
9.95197713e-01 -7.48472750e-01 8.82308185e-02 4.01470244e-01
-7.55940616e-01 1.00309290e-02 4.47110623e-01 -4.38415498e-01
-1.14031541e+00 -6.39477491e-01 7.23248497e-02 -1.73323646e-01
-3.70953530e-01 5.02470016e-01 -5.29930666e-02 -2.17196569e-01
8.92925262e-01 7.57300198e-01 4.52320911e-02 5.42654991e-02
4.78623569e-01 9.26172316e-01 1.03770673e+00 -3.97794008e-01
4.52272922e-01 4.20991957e-01 5.38513530e-03 -1.04372978e+00
-1.08370960e-01 -4.86902632e-02 -3.23528945e-01 -6.81374311e-01
1.15098321e+00 -5.64243376e-01 -1.27915907e+00 5.48133910e-01
-1.23821080e+00 1.63878053e-01 -1.12616181e-01 5.05316556e-01
-4.76891756e-01 1.91666791e-03 -6.90410733e-01 -1.02909410e+00
-6.87056363e-01 -9.68737602e-01 1.34721741e-01 5.56200981e-01
-3.84909600e-01 -1.09802401e+00 -2.04321742e-01 8.10955405e-01
5.02085686e-01 1.20927297e-01 1.43800592e+00 -4.67024535e-01
-4.53098238e-01 -3.16608883e-02 -4.78041410e-01 4.26774263e-01
-2.28969321e-01 1.16108246e-02 -8.84672046e-01 4.88061517e-01
1.25208378e-01 -2.09866360e-01 5.84357977e-01 4.17452782e-01
1.19419241e+00 1.14310868e-01 -4.09327567e-01 3.64170402e-01
1.17071402e+00 1.07374442e+00 1.01985335e+00 2.83134073e-01
4.56574023e-01 5.96893311e-01 -1.27078459e-01 -2.29920652e-02
6.49837673e-01 -3.25842351e-02 3.31732780e-01 2.45971888e-01
-1.88531503e-01 3.40760350e-01 4.25345600e-01 1.30448461e+00
-9.12279010e-01 2.16702163e-01 -1.20961809e+00 4.97222304e-01
-1.74127519e+00 -1.39181828e+00 -1.85556144e-01 1.42271733e+00
6.67086422e-01 5.20587936e-02 -3.55009288e-01 5.37031710e-01
6.83015466e-01 -5.39880037e-01 -7.67941952e-01 -8.62757742e-01
-1.80790097e-01 4.60186929e-01 -3.29105929e-02 -3.57969431e-03
-4.22568768e-01 9.22347665e-01 6.98306274e+00 6.61574841e-01
-1.20291293e+00 2.56975032e-02 2.18686000e-01 2.41631702e-01
-3.22994977e-01 -3.77213389e-01 -2.91247070e-01 4.92551744e-01
1.08119452e+00 -3.36596698e-01 1.30170214e+00 6.62983358e-01
3.07626557e-02 -2.68794358e-01 -8.75372767e-01 1.35738444e+00
1.77427500e-01 -1.53093171e+00 2.11352542e-01 -2.93012913e-02
4.40347284e-01 1.82847723e-01 -3.20253670e-01 4.21575516e-01
2.90797561e-01 -1.24230373e+00 4.65759933e-01 1.17593265e+00
3.50402772e-01 -7.82957792e-01 5.48786283e-01 5.50771892e-01
-8.15497160e-01 -2.69632429e-01 -4.96324390e-01 -5.25503933e-01
-3.81009936e-01 6.80172801e-01 -3.57202619e-01 2.09307522e-01
9.61579144e-01 5.77295005e-01 -4.24428254e-01 6.59764647e-01
-2.36761630e-01 5.13958573e-01 2.68054418e-02 -5.83092213e-01
-3.18434507e-01 2.53384057e-02 5.08898199e-01 1.13694918e+00
-1.44390076e-01 7.23923266e-01 -5.97495854e-01 1.49085844e+00
1.58815458e-01 -4.61190045e-01 -7.49899030e-01 -5.52327693e-01
5.25708616e-01 9.59584534e-01 -8.58931601e-01 -3.80252182e-01
-1.66261673e-01 8.25140357e-01 2.61172485e-02 6.41289055e-01
-6.98854148e-01 -8.04088116e-01 5.86406529e-01 -2.70353764e-01
-2.17396349e-01 -5.04042320e-02 -7.37284005e-01 -1.03155661e+00
-2.43877262e-01 -7.74486959e-01 1.53743267e-01 -1.35194969e+00
-1.34043777e+00 6.00329697e-01 8.95343572e-02 -3.58459473e-01
-2.66451478e-01 -9.99551356e-01 -7.63273120e-01 9.38967407e-01
-1.43049014e+00 -8.95706832e-01 -3.51266116e-01 1.10676324e+00
3.12685043e-01 -4.94140506e-01 1.02092838e+00 -6.03603199e-03
-5.37524819e-01 -4.57230769e-02 -1.40190706e-01 2.18991861e-01
-1.70748860e-01 -9.77784395e-01 2.94923149e-02 4.26953912e-01
-1.10686600e-01 6.74228013e-01 4.26052541e-01 -5.09505332e-01
-1.94018769e+00 -5.23728609e-01 5.79151332e-01 -2.85302609e-01
7.94579625e-01 8.65286887e-02 -9.57749546e-01 3.68225664e-01
2.70936400e-01 -2.84340858e-01 7.84382939e-01 -9.92316157e-02
-1.36759253e-02 -1.79390639e-01 -1.19470048e+00 4.97336626e-01
1.08292651e+00 -6.48036897e-01 -1.27384245e+00 1.91945761e-01
5.65628290e-01 2.47354046e-01 -8.88129294e-01 3.19740325e-01
7.20135808e-01 -6.87526047e-01 1.08245957e+00 -5.81933618e-01
2.68940240e-01 -4.59609121e-01 2.22837590e-02 -1.21609104e+00
-7.99454629e-01 -2.73033679e-01 -3.33327234e-01 6.42263532e-01
9.25065950e-02 -9.66504574e-01 1.94156662e-01 8.28135192e-01
-6.78821206e-02 -6.40293717e-01 -8.31195116e-01 -4.60328221e-01
-1.47564128e-01 -8.46782267e-01 3.60094130e-01 1.08082867e+00
7.42700696e-01 3.91791761e-01 1.81968451e-01 -1.79152451e-02
5.14224529e-01 -2.29630485e-01 7.39258379e-02 -1.54846835e+00
-2.40501929e-02 -9.31810677e-01 -1.06983733e+00 -3.26395482e-01
2.32382044e-01 -9.48307157e-01 -4.40851361e-01 -1.86877739e+00
3.25237751e-01 2.84928560e-01 -6.31863117e-01 6.60770237e-01
2.70737372e-02 1.24989852e-01 4.51284386e-02 1.88083962e-01
-4.07705516e-01 6.89740628e-02 1.48712814e+00 -2.78088957e-01
-8.72227829e-03 -5.64102173e-01 -9.75096464e-01 1.02703154e+00
6.83525264e-01 1.03457302e-01 -7.06699431e-01 -2.63751328e-01
5.93675077e-01 1.29536927e-01 6.40051723e-01 -1.24684548e+00
7.60842204e-01 -4.30179238e-01 8.51234496e-01 -3.26611519e-01
4.89772409e-02 -9.92489576e-01 2.15076283e-01 7.73450553e-01
7.11554587e-02 -1.16902135e-01 2.94593632e-01 2.88339972e-01
-1.47338763e-01 -8.80433172e-02 7.89723396e-01 -3.85800302e-01
-1.39337933e+00 -4.03319091e-01 -8.88882279e-01 -3.30710448e-02
1.18273544e+00 -6.75375402e-01 -6.21405005e-01 -1.99824467e-01
-7.23336220e-01 1.02093592e-01 -1.65868059e-01 4.04280603e-01
1.00421655e+00 -1.07045805e+00 -2.99208730e-01 2.02496529e-01
-1.56193674e-01 -4.26252216e-01 5.99843599e-02 6.06270790e-01
-7.02880204e-01 7.04005480e-01 -7.69106925e-01 -1.69679374e-01
-9.01471674e-01 4.64408249e-01 2.74440229e-01 4.64280456e-01
-4.34490979e-01 8.75637114e-01 -9.18321460e-02 -7.23748282e-02
3.04601401e-01 -2.21178293e-01 -5.37157357e-01 -3.54324132e-01
7.28216529e-01 8.87710154e-01 -2.39945337e-01 -3.73923957e-01
-6.81501985e-01 4.89444762e-01 3.13829750e-01 -1.11617595e-01
1.35316479e+00 4.38217819e-02 -7.08747327e-01 7.00407565e-01
5.99570155e-01 -9.42688227e-01 -1.33063167e-01 1.32434264e-01
-2.35843167e-01 8.59270841e-02 4.84189332e-01 -1.53822279e+00
-9.80093539e-01 9.96767938e-01 6.96204245e-01 2.44829521e-01
1.60293734e+00 -1.19554594e-01 5.32799125e-01 8.67009640e-01
7.25217104e-01 -1.44959486e+00 2.72681713e-02 3.65371853e-01
1.01614797e+00 -9.18929815e-01 1.47950679e-01 -1.07356511e-01
-6.78391814e-01 1.40465915e+00 4.21255797e-01 8.85785073e-02
1.04587340e+00 2.15598285e-01 -8.71148780e-02 -9.28202093e-01
-7.59039164e-01 -2.59783059e-01 5.37320197e-01 8.84590149e-01
3.01600009e-01 2.99034327e-01 -2.75100112e-01 7.13322878e-01
-4.45583761e-01 6.76219165e-01 1.56704837e-03 8.88633192e-01
-8.74042332e-01 -4.87641871e-01 -5.70632517e-01 6.42509937e-01
3.74493096e-03 -1.42734662e-01 -5.77074707e-01 5.86683214e-01
7.07440197e-01 1.30854225e+00 -1.28118485e-01 -7.66603708e-01
2.61759698e-01 3.57973337e-01 6.47533059e-01 -2.00348511e-01
-3.01886201e-01 -6.35982156e-01 -2.08364919e-01 -3.60260993e-01
-5.36890209e-01 -1.23462327e-01 -1.85392618e+00 -8.70703638e-01
-2.38577381e-01 -1.53880820e-01 1.15912735e+00 1.24086642e+00
3.99434060e-01 9.90707636e-01 1.58197537e-01 -4.43874478e-01
2.08645344e-01 -7.76837349e-01 -5.57907701e-01 2.07833171e-01
-4.31540489e-01 -6.59477472e-01 -1.67850852e-01 3.48618001e-01] | [9.157012939453125, 6.414112567901611] |
e5d379b1-79aa-425a-ad07-6c90d3089727 | clirmatrix-a-massively-large-collection-of | null | null | https://aclanthology.org/2020.emnlp-main.340 | https://aclanthology.org/2020.emnlp-main.340.pdf | CLIRMatrix: A massively large collection of bilingual and multilingual datasets for Cross-Lingual Information Retrieval | We present CLIRMatrix, a massively large collection of bilingual and multilingual datasets for Cross-Lingual Information Retrieval extracted automatically from Wikipedia. CLIRMatrix comprises (1) BI-139, a bilingual dataset of queries in one language matched with relevant documents in another language for 139x138=19,182 language pairs, and (2) MULTI-8, a multilingual dataset of queries and documents jointly aligned in 8 different languages. In total, we mined 49 million unique queries and 34 billion (query, document, label) triplets, making it the largest and most comprehensive CLIR dataset to date. This collection is intended to support research in end-to-end neural information retrieval and is publicly available at [url]. We provide baseline neural model results on BI-139, and evaluate MULTI-8 in both single-language retrieval and mix-language retrieval settings. | ['Kevin Duh', 'Shuo Sun'] | null | null | null | null | emnlp-2020-11 | ['cross-lingual-information-retrieval'] | ['natural-language-processing'] | [-4.27031428e-01 -5.29116035e-01 -7.34114766e-01 -2.46889666e-01
-1.96581817e+00 -1.03296423e+00 8.99443507e-01 2.00901255e-01
-1.00749660e+00 7.78811932e-01 5.48674643e-01 -7.49828741e-02
-2.15921327e-01 -3.22138578e-01 -6.99660003e-01 -1.69311147e-02
3.24310482e-01 1.05925035e+00 -1.32905051e-01 -3.32870513e-01
-1.11733703e-02 1.93560660e-01 -9.54747021e-01 5.65696955e-01
5.38390517e-01 9.18828070e-01 1.40109286e-01 3.94808233e-01
5.49510540e-03 5.69648504e-01 -2.84700483e-01 -7.96415567e-01
1.78237483e-01 1.51713407e-02 -9.39316273e-01 -7.76339650e-01
1.06841421e+00 -2.96991080e-01 -6.56846106e-01 7.21859097e-01
1.03715205e+00 -1.60850838e-01 8.59147549e-01 -9.38053131e-01
-1.15497077e+00 1.06582880e+00 -6.46047175e-01 4.49819505e-01
2.80340135e-01 -1.85264796e-01 1.57713056e+00 -1.38299012e+00
1.21266162e+00 1.38388824e+00 3.68796974e-01 2.38473877e-01
-1.09280515e+00 -9.61534321e-01 -3.25349033e-01 -4.11208086e-02
-1.90817297e+00 -8.66688192e-01 1.24388270e-01 -3.24943841e-01
1.36302114e+00 2.05706313e-01 7.05012307e-02 1.32849693e+00
3.96012608e-03 1.21087682e+00 7.80894279e-01 -6.01749957e-01
-6.70930862e-01 2.82902837e-01 3.12481105e-01 4.19950962e-01
1.12448461e-01 6.91710785e-02 -5.91207862e-01 -3.96276176e-01
1.46146163e-01 -4.68254119e-01 -1.53073922e-01 -4.46536429e-02
-1.49528694e+00 7.24892020e-01 4.50404882e-01 4.49018687e-01
-2.76780009e-01 1.21660501e-01 7.01911151e-01 6.65362179e-01
5.42263567e-01 7.28040457e-01 -7.26373613e-01 1.92877799e-01
-8.11274290e-01 6.41502678e-01 8.76278222e-01 1.43232167e+00
8.24985206e-01 -4.55231637e-01 -6.32448137e-01 1.57832778e+00
4.82158959e-01 1.24855554e+00 5.62328815e-01 -8.45564961e-01
9.07751024e-01 3.37637037e-01 -2.83438843e-02 -9.30075884e-01
-3.48859787e-01 -3.47217888e-01 -4.87822413e-01 -1.07014012e+00
9.63459536e-02 -7.50248507e-02 -3.62425417e-01 1.81432199e+00
-2.12689862e-01 -8.13282132e-01 2.42548078e-01 7.93349445e-01
1.29965222e+00 5.99460542e-01 5.12156344e-04 1.54376343e-01
1.53399563e+00 -1.21130002e+00 -5.81656814e-01 -3.67865801e-01
9.48577702e-01 -1.35243785e+00 9.24114108e-01 -1.01278260e-01
-1.04512835e+00 -2.88467586e-01 -7.37305880e-01 -8.46286237e-01
-9.66323376e-01 4.94196504e-01 4.71076101e-01 6.28441432e-03
-1.52640426e+00 -4.52653468e-01 -1.92922249e-01 -5.24156928e-01
-1.91736817e-01 6.76441416e-02 -5.26589990e-01 -5.43141901e-01
-1.59864926e+00 9.07321095e-01 5.48938692e-01 -2.38159075e-02
-7.75469363e-01 -6.90834522e-01 -7.91634202e-01 -1.61303222e-01
2.95141697e-01 -4.90023792e-01 1.38673246e+00 -2.74651885e-01
-5.26164711e-01 1.56094944e+00 -1.40543491e-01 -2.35115275e-01
2.70779520e-01 -3.57992172e-01 -6.55698419e-01 -1.87945552e-02
7.11919069e-01 1.16562974e+00 8.33743438e-02 -9.39407289e-01
-4.99701321e-01 -5.47521591e-01 -2.29825303e-01 5.03214717e-01
-3.97063822e-01 6.28360033e-01 -1.38854492e+00 -8.38162601e-01
-2.00830117e-01 -1.11313736e+00 3.53500247e-01 -5.20658374e-01
-9.38485920e-01 -5.94362378e-01 2.87343204e-01 -1.00789618e+00
1.28419387e+00 -1.80509245e+00 -1.29245380e-02 6.06426708e-02
6.34174868e-02 2.35537756e-02 -7.11650074e-01 1.03690839e+00
8.40528384e-02 1.71856403e-01 2.12670714e-01 -5.49735129e-01
3.64097089e-01 -2.65445203e-01 -5.54714084e-01 2.09341034e-01
-1.28396437e-01 1.38778329e+00 -7.73597240e-01 -9.45271671e-01
-4.76677060e-01 4.14138615e-01 -3.22945654e-01 -2.77426802e-02
-1.52717546e-01 -2.37181000e-02 -5.74179828e-01 8.64179790e-01
3.99513960e-01 -1.17034681e-01 -7.59708658e-02 -3.13819200e-01
6.00032657e-02 5.17446280e-01 -2.99931437e-01 2.15292501e+00
-6.08780801e-01 9.12693739e-01 1.85174406e-01 -2.37194985e-01
5.16651988e-01 4.77877915e-01 6.59253597e-01 -1.48555231e+00
-7.61984661e-02 6.72914326e-01 -6.42352760e-01 -4.83222842e-01
1.16283691e+00 5.46565473e-01 -8.16024840e-01 8.67072999e-01
1.95854440e-01 1.58053637e-01 7.12005854e-01 5.72828591e-01
8.43712628e-01 -1.38132662e-01 -8.69173184e-02 -4.29933906e-01
4.05170739e-01 1.05103821e-01 2.03281775e-01 1.15966928e+00
1.10015057e-01 4.07684565e-01 -6.64818138e-02 -1.03132948e-01
-1.11374390e+00 -9.69541788e-01 -3.94546092e-01 1.60735500e+00
-2.43817776e-01 -3.79596740e-01 -4.01553035e-01 -3.61165702e-01
5.15132129e-01 3.66926610e-01 -2.38421857e-01 2.28120834e-02
-6.22219682e-01 -7.06131577e-01 1.25777209e+00 8.15242380e-02
2.58786559e-01 -1.26343513e+00 3.11592162e-01 9.46528986e-02
-7.09947109e-01 -1.45171332e+00 -1.32028127e+00 1.11017779e-01
-2.53917933e-01 -1.03440535e+00 -9.76449490e-01 -1.21863782e+00
1.84166566e-01 4.19155568e-01 2.00216746e+00 -5.16782105e-02
-4.93428081e-01 7.69779205e-01 -1.73953742e-01 -1.12953477e-01
-3.30117464e-01 8.29077065e-01 1.89431474e-01 -4.71672297e-01
9.75260139e-01 8.68868455e-02 -4.01050925e-01 1.58408210e-01
-8.70380521e-01 -4.00235772e-01 7.10686862e-01 9.45025861e-01
6.61396444e-01 -6.28492653e-01 6.71586990e-01 -6.47416174e-01
1.15932214e+00 -7.46764958e-01 -8.85936379e-01 9.38286781e-01
-6.50858879e-01 7.17217997e-02 1.00543216e-01 -3.71894613e-02
-7.96489775e-01 -4.28099930e-01 1.91352010e-01 -2.09920168e-01
1.23184077e-01 9.78181362e-01 3.35757107e-01 2.60047764e-01
7.91697502e-01 2.41787732e-01 -4.00204837e-01 -9.78026509e-01
7.56913841e-01 1.20309913e+00 6.79231167e-01 -1.01269877e+00
4.05843437e-01 -2.00789273e-01 -7.57012963e-01 -6.57098532e-01
-7.35161424e-01 -9.41520333e-01 -6.70461535e-01 1.57557756e-01
7.08718479e-01 -1.47379005e+00 -6.57031178e-01 5.30177891e-01
-1.18380606e+00 -1.39690563e-01 3.80632490e-01 4.62876290e-01
-3.79288681e-02 -9.81067196e-02 -1.11848032e+00 -4.03718382e-01
-8.22003961e-01 -1.17148566e+00 1.54993248e+00 -2.18352586e-01
-1.35216229e-02 -9.85456526e-01 4.21649128e-01 6.44375086e-01
4.99875844e-01 -4.00436282e-01 9.62646544e-01 -9.07450855e-01
-5.49311042e-01 -3.21920753e-01 -6.64183557e-01 6.13602176e-02
-5.27009033e-02 -1.44881845e-01 -8.27352583e-01 -7.66328216e-01
-6.10351145e-01 -1.30182338e+00 1.02941835e+00 5.16024679e-02
6.43872201e-01 -3.06222886e-01 -4.19625968e-01 1.99327722e-01
1.44623470e+00 5.34099825e-02 -1.25095854e-02 3.56309831e-01
6.38760388e-01 6.77131832e-01 2.54130602e-01 -7.05220550e-02
9.23592985e-01 1.22235119e+00 -3.13447088e-01 -1.53938130e-01
-2.80532002e-01 -3.77965033e-01 2.31910408e-01 1.55080593e+00
8.66206944e-01 -3.71285617e-01 -1.22887528e+00 8.88845325e-01
-1.49044263e+00 -7.28706837e-01 3.90313119e-01 2.05759811e+00
1.42064524e+00 -4.00657773e-01 -1.02205597e-01 -8.65540206e-01
4.99992341e-01 5.78568056e-02 -6.52375460e-01 6.36152700e-02
-7.65797257e-01 -3.92840877e-02 6.78469479e-01 7.27099538e-01
-1.27505088e+00 1.31223619e+00 6.59022999e+00 1.06804180e+00
-9.65027273e-01 1.63366750e-01 4.88860786e-01 -2.76564807e-01
-5.60869157e-01 -2.62124687e-01 -1.29612887e+00 3.10665846e-01
8.64318669e-01 -5.13404071e-01 7.44003952e-01 4.21573788e-01
-4.41205382e-01 1.02470092e-01 -1.28766155e+00 1.00007594e+00
3.99017721e-01 -1.00458097e+00 3.65567178e-01 3.11945565e-03
6.43714607e-01 1.17908680e+00 3.21542710e-01 8.26431274e-01
4.88803804e-01 -9.18608844e-01 6.55992627e-01 5.14181972e-01
1.11646950e+00 -5.37735879e-01 4.56899673e-01 3.69574130e-01
-8.95751119e-01 2.45457649e-01 -1.66972995e-01 9.86668169e-01
-1.12435773e-01 1.11783221e-01 -9.17137742e-01 5.66215992e-01
9.81930733e-01 7.96073735e-01 -1.12860572e+00 7.77648270e-01
3.63030285e-01 1.06494926e-01 -4.85855758e-01 -4.56447806e-03
5.55766761e-01 -1.35507479e-01 4.89744842e-01 1.63772368e+00
-2.86441129e-02 -5.97010016e-01 2.40712732e-01 7.30318069e-01
-8.18630278e-01 5.60786486e-01 -1.11476707e+00 -4.03392494e-01
1.04248965e+00 1.14955962e+00 -1.41279489e-01 -4.03269291e-01
-5.57989836e-01 7.38991261e-01 6.43774927e-01 7.97142148e-01
-3.03134829e-01 -6.34565234e-01 3.86549026e-01 -5.23866415e-01
-2.94339210e-01 2.19739857e-03 5.05939126e-01 -1.44424224e+00
3.55024457e-01 -1.23947227e+00 8.02929878e-01 -8.83181274e-01
-1.83903229e+00 8.78956974e-01 2.41637394e-01 -8.99015009e-01
-8.96214902e-01 -4.88934129e-01 6.93012536e-01 1.18880069e+00
-1.68992305e+00 -1.46691072e+00 2.21257240e-01 7.27873921e-01
4.24620241e-01 -7.66713798e-01 9.91307557e-01 1.09699225e+00
-4.76926208e-01 1.14538646e+00 7.84562230e-01 6.49224043e-01
1.42672431e+00 -8.57762396e-01 3.72218817e-01 3.53577405e-01
6.06618941e-01 1.15825760e+00 -6.82183579e-02 -4.87890780e-01
-1.64878035e+00 -9.21915293e-01 1.55329192e+00 -9.58293319e-01
7.78430104e-01 -6.26997590e-01 -7.16797054e-01 9.09832418e-01
6.66279256e-01 -3.62090886e-01 6.59820557e-01 2.58250743e-01
-8.07127595e-01 -3.29798102e-01 -5.93935370e-01 6.79403722e-01
7.01615572e-01 -1.48170543e+00 -4.55458969e-01 8.28103960e-01
1.04707491e+00 -5.06925464e-01 -1.27856314e+00 3.66439849e-01
7.09185183e-01 -3.04741085e-01 1.29693151e+00 -6.92820132e-01
1.29899144e-01 1.83055878e-01 -5.75868487e-01 -1.07087064e+00
-6.48811609e-02 -1.96777716e-01 3.41101974e-01 1.17052233e+00
1.02514350e+00 -5.96404970e-01 7.56613538e-02 3.95382643e-01
2.68628448e-02 -6.58742547e-01 -9.85535085e-01 -6.35852695e-01
7.23564625e-01 -2.95234740e-01 5.15010238e-01 1.14384449e+00
-2.38183677e-01 7.58713603e-01 -1.93445921e-01 -3.58571798e-01
4.74703401e-01 2.23241791e-01 7.11211145e-01 -9.20620084e-01
-9.14976597e-02 -7.77359188e-01 3.95715237e-01 -1.28523946e+00
6.01627767e-01 -1.63557291e+00 -4.84751835e-02 -1.26067543e+00
6.14807487e-01 -7.46706605e-01 -3.66516292e-01 8.47800016e-01
8.81671626e-03 5.17054975e-01 -1.73003078e-01 6.98047757e-01
-9.40091491e-01 6.39486670e-01 8.36451471e-01 -6.20065153e-01
1.46362737e-01 -6.67786121e-01 -7.79370010e-01 -1.40633062e-01
1.97578385e-03 -5.58207631e-01 -1.85562298e-01 -1.38655806e+00
5.59065223e-01 2.29000613e-01 -1.98147148e-02 -3.08118522e-01
5.65615714e-01 4.38324898e-01 2.40376770e-01 -8.72136712e-01
4.15594220e-01 -6.00969076e-01 -2.58370876e-01 -9.91183072e-02
-1.10201931e+00 7.44669437e-01 3.36645961e-01 3.12047124e-01
-6.00998998e-01 3.32855619e-02 7.86463544e-02 -1.14738069e-01
-6.00787163e-01 4.13848698e-01 -1.29791379e-01 8.34827542e-01
1.52993977e-01 8.72598767e-01 -7.95985460e-01 -4.47552115e-01
-2.42523134e-01 8.52498293e-01 2.26787686e-01 1.27117956e+00
2.71309495e-01 -1.78589487e+00 -1.20630884e+00 -3.76048312e-02
9.98121440e-01 -4.23498362e-01 -9.30650234e-02 6.25182271e-01
-3.22953433e-01 1.55706191e+00 1.72669411e-01 -4.97301102e-01
-7.87068665e-01 4.34067398e-01 1.91949487e-01 -4.45875496e-01
-1.08419947e-01 7.47241557e-01 -1.63597688e-01 -1.22420514e+00
4.53257740e-01 2.64711559e-01 -2.19874442e-01 3.62172365e-01
4.54977810e-01 -2.11655181e-02 3.24125230e-01 -1.19792318e+00
-3.27977866e-01 5.11817634e-01 -6.23960555e-01 -6.77201152e-01
8.28777730e-01 -3.37072998e-01 -8.31822455e-01 5.09951472e-01
2.03335214e+00 5.13683893e-02 -7.67993554e-02 -1.05334389e+00
5.31755149e-01 -1.67292263e-02 1.28217742e-01 -1.08108056e+00
-1.11657536e+00 5.03336966e-01 4.09004837e-01 -2.34601051e-01
8.75158846e-01 1.53079420e-01 9.63435948e-01 1.23920727e+00
5.83171725e-01 -1.18619215e+00 -1.63129121e-01 8.75791609e-01
1.22622299e+00 -1.50641668e+00 -1.73398599e-01 4.25206155e-01
-3.16482097e-01 5.93436778e-01 5.28452814e-01 4.43719238e-01
6.81842804e-01 -1.35677114e-01 5.65460801e-01 -5.59077084e-01
-9.45810020e-01 8.00403729e-02 8.54628026e-01 -2.74960883e-02
8.32126737e-01 -1.08564995e-01 -1.50511339e-01 2.35395253e-01
-3.22702825e-01 -1.94732144e-01 -2.64586955e-01 8.90483677e-01
3.20963264e-01 -1.17139268e+00 -7.69681633e-02 3.92573506e-01
-6.80916846e-01 -7.39249468e-01 -6.52468264e-01 7.88752317e-01
-5.73839724e-01 9.43074644e-01 -1.42790675e-02 -1.12912789e-01
1.36082992e-01 9.93739292e-02 1.50143415e-01 -3.80943507e-01
-6.63772225e-01 1.01638921e-01 4.02396023e-01 -6.17791593e-01
-1.30620509e-01 -5.06353438e-01 -4.93368149e-01 5.76514564e-02
-1.36878863e-01 3.84008050e-01 8.11118543e-01 5.51843703e-01
6.69375956e-01 -1.30482256e-01 4.62604284e-01 -3.60326618e-01
-5.13999522e-01 -1.36882031e+00 -2.60005713e-01 4.00610328e-01
3.96962225e-01 -3.19475740e-01 -1.10686578e-01 -2.37909093e-01] | [11.391020774841309, 9.770442008972168] |
78f499f4-3d26-4b45-840b-ea271e08af36 | retrosynthesis-prediction-with-local-template | 2306.04123 | null | https://arxiv.org/abs/2306.04123v1 | https://arxiv.org/pdf/2306.04123v1.pdf | Retrosynthesis Prediction with Local Template Retrieval | Retrosynthesis, which predicts the reactants of a given target molecule, is an essential task for drug discovery. In recent years, the machine learing based retrosynthesis methods have achieved promising results. In this work, we introduce RetroKNN, a local reaction template retrieval method to further boost the performance of template-based systems with non-parametric retrieval. We first build an atom-template store and a bond-template store that contain the local templates in the training data, then retrieve from these templates with a k-nearest-neighbor (KNN) search during inference. The retrieved templates are combined with neural network predictions as the final output. Furthermore, we propose a lightweight adapter to adjust the weights when combing neural network and KNN predictions conditioned on the hidden representation and the retrieved templates. We conduct comprehensive experiments on two widely used benchmarks, the USPTO-50K and USPTO-MIT. Especially for the top-1 accuracy, we improved 7.1% on the USPTO-50K dataset and 12.0% on the USPTO-MIT dataset. These results demonstrate the effectiveness of our method. | ['Tao Qin', 'Lijun Wu', 'Yingce Xia', 'Junliang Guo', 'Rui Yan', 'Shufang Xie'] | 2023-06-07 | null | null | null | null | ['drug-discovery', 'retrosynthesis'] | ['medical', 'medical'] | [ 4.00512099e-01 -2.47907177e-01 -8.67593229e-01 -1.56485066e-01
-8.24148417e-01 -6.01756752e-01 4.91347194e-01 9.41223279e-02
-2.21439481e-01 9.53649938e-01 2.81385899e-01 -3.33862901e-01
-8.17749277e-03 -7.85226345e-01 -7.99799860e-01 -1.08659267e+00
2.76488572e-01 2.20870405e-01 3.17074001e-01 -8.29064995e-02
5.24633825e-01 6.27371013e-01 -1.23622262e+00 6.01300478e-01
5.97453773e-01 1.33886313e+00 2.11743146e-01 2.10152730e-01
1.71890050e-01 8.28573883e-01 -1.72267705e-01 -3.15935910e-01
5.71772456e-02 -1.63686693e-01 -6.35216415e-01 -8.83192420e-01
1.01559557e-01 -1.36636361e-01 -4.70941067e-01 5.80433190e-01
8.26911151e-01 3.78329575e-01 1.08234978e+00 -7.20068991e-01
-3.88739794e-01 5.93671679e-01 -1.66029911e-02 -1.74400751e-02
3.39781046e-01 6.46168068e-02 1.05166125e+00 -1.43810856e+00
7.07380533e-01 1.04357576e+00 4.42812413e-01 5.01836419e-01
-9.75100458e-01 -1.17891610e+00 -1.97555274e-02 4.97657239e-01
-1.49939597e+00 -6.75675988e-01 7.65214503e-01 -3.41413856e-01
1.45409083e+00 1.90164894e-01 4.59215045e-01 1.14542449e+00
7.93327451e-01 5.44224322e-01 6.39711201e-01 -2.29340307e-02
1.89965263e-01 -3.01013589e-01 -2.10992455e-01 7.98083603e-01
-6.07899614e-02 4.03165579e-01 -7.61639953e-01 -4.05535549e-01
5.91053784e-01 3.50215316e-01 -3.57918233e-01 2.50300951e-03
-1.05137086e+00 9.34312522e-01 7.95306206e-01 7.82907680e-02
-4.07839447e-01 2.27099657e-01 5.22967994e-01 -1.98745430e-01
3.61198872e-01 8.12152386e-01 -7.62323856e-01 3.59874934e-01
-7.87783623e-01 1.19371489e-01 7.26231098e-01 1.06588042e+00
5.87345183e-01 -6.77003935e-02 -4.72056925e-01 7.83853352e-01
1.84100717e-01 1.55340999e-01 7.38430381e-01 -4.25860345e-01
4.99515563e-01 5.12365460e-01 6.26309365e-02 -9.04425085e-01
-5.60785055e-01 -3.09177309e-01 -9.90801394e-01 -5.99698901e-01
-5.33144139e-02 1.97159857e-01 -1.10243249e+00 1.39737904e+00
5.71330070e-01 3.59766662e-01 1.49016634e-01 5.48817992e-01
1.10752797e+00 1.34459436e+00 3.85132849e-01 -6.59698844e-01
9.95036125e-01 -1.36197317e+00 -6.84950233e-01 2.10141838e-01
7.18077421e-01 -1.03283095e+00 3.55739713e-01 7.08045304e-01
-8.38056922e-01 -4.42824453e-01 -1.20091641e+00 -3.63557249e-01
-7.96253562e-01 1.74248159e-01 7.09270060e-01 1.59461305e-01
-2.97580868e-01 1.21026337e+00 -5.70435524e-01 3.61409992e-01
3.50333869e-01 8.19388151e-01 -3.58287662e-01 -3.60444598e-02
-1.35552764e+00 7.16345310e-01 1.08590996e+00 3.43642026e-01
-1.28704858e+00 -9.25222099e-01 -6.23006999e-01 7.41991997e-02
7.90130615e-01 -3.61053437e-01 1.13497806e+00 -1.28472388e-01
-1.77310789e+00 4.30460274e-02 -1.73090085e-01 -1.53509706e-01
-7.50565156e-02 8.26405585e-02 -5.15251458e-01 -1.31885037e-01
-1.41688734e-01 6.95024371e-01 6.87510431e-01 -6.97794259e-01
-3.24010491e-01 -2.47108683e-01 -3.06374282e-01 2.26572510e-02
2.61639133e-02 -1.50923714e-01 -4.10174042e-01 -9.00563896e-01
2.43188694e-01 -9.89627182e-01 -4.40896660e-01 -3.18215221e-01
-7.92369545e-01 -7.01416433e-01 3.70627999e-01 -3.54597330e-01
1.46609914e+00 -1.79400098e+00 3.97099137e-01 4.09945935e-01
5.68507537e-02 3.55705976e-01 -1.73259497e-01 7.05861509e-01
-4.72540826e-01 8.40779468e-02 6.49621785e-02 1.61893532e-01
-1.51803389e-01 -2.02822283e-01 -4.99982387e-01 1.94770038e-01
1.12394951e-02 8.44998538e-01 -7.47875571e-01 -4.29754734e-01
4.78433780e-02 3.89470100e-01 -5.77534795e-01 2.75750011e-01
-6.70294702e-01 -1.09648975e-02 -7.15965450e-01 7.89884448e-01
1.64137006e-01 -3.42771083e-01 3.58496845e-01 -7.03650475e-01
-1.86976522e-01 6.24602079e-01 -7.97500730e-01 1.83274055e+00
-2.18093202e-01 -3.76700521e-01 -9.26145971e-01 -5.62646985e-01
1.06238461e+00 5.73333561e-01 3.30149919e-01 -7.51756489e-01
1.23873390e-01 4.72402900e-01 -1.46271408e-01 -1.69619575e-01
5.38706958e-01 -1.18555225e-01 6.38000388e-03 6.60902262e-02
2.23382622e-01 1.42292492e-02 8.40161219e-02 1.81111891e-03
7.24142969e-01 3.92339557e-01 8.24701965e-01 -1.56656697e-01
7.46317208e-01 -1.30688995e-01 7.22674787e-01 4.67068702e-01
3.95138204e-01 2.60390788e-01 2.52778649e-01 -8.77120137e-01
-8.98091614e-01 -5.93827307e-01 -6.22837059e-02 1.18810892e+00
-1.66538998e-01 -1.03049004e+00 -5.01342058e-01 -8.09011340e-01
-1.77829355e-01 5.16175210e-01 -5.76960206e-01 -6.41238391e-01
-5.15268266e-01 -7.08369195e-01 4.84887213e-01 5.34562111e-01
2.65381426e-01 -1.15136230e+00 2.95672715e-01 7.06117868e-01
1.18536569e-01 -5.95223844e-01 -8.93815756e-01 5.39396822e-01
-8.21323097e-01 -1.13667369e+00 -4.66009885e-01 -8.15401912e-01
3.49718511e-01 1.27180085e-01 4.90572870e-01 -7.39648640e-02
-2.51357168e-01 -6.04172885e-01 -2.48422418e-02 -2.96486735e-01
-1.98953196e-01 4.23774153e-01 4.63651828e-02 -2.85305411e-01
8.09826925e-02 -5.76737046e-01 -7.94247329e-01 4.85521853e-01
-6.98925018e-01 1.71450645e-01 6.46031260e-01 9.87458110e-01
1.24250829e+00 -8.70041177e-02 6.53792620e-01 -8.57868254e-01
1.27806067e-01 -4.95832413e-01 -8.76671493e-01 4.41042453e-01
-8.72090816e-01 5.25754333e-01 1.07956612e+00 -6.64461672e-01
-7.70545602e-01 5.50956905e-01 -1.06894292e-01 -4.59035218e-01
4.29817557e-01 8.24749947e-01 -4.34955150e-01 -3.38354707e-02
6.70601130e-01 3.19526970e-01 -5.00928700e-01 -5.83300173e-01
3.10930967e-01 5.57192802e-01 3.32711488e-01 -7.46440113e-01
5.62344432e-01 -2.43144676e-01 4.64545369e-01 -3.27908814e-01
-9.98268425e-01 -4.74861652e-01 -2.40733713e-01 3.06687742e-01
7.36559212e-01 -9.15291786e-01 -1.28088462e+00 1.78753734e-01
-1.23987317e+00 -2.31800437e-01 3.96306187e-01 4.82128084e-01
-5.24384677e-01 3.72272909e-01 -8.45169067e-01 -3.51119131e-01
-1.09004843e+00 -1.70134664e+00 9.51851904e-01 1.36348233e-01
3.31919678e-02 -5.09206891e-01 1.71674505e-01 2.51913965e-01
2.65556909e-02 5.86037003e-02 1.27134299e+00 -9.36576247e-01
-7.51646876e-01 -1.34780988e-01 -1.18793659e-01 -3.73813286e-02
2.94809997e-01 -1.64924726e-01 -9.39814270e-01 -9.99049544e-02
-4.09854025e-01 -3.39339107e-01 1.10637248e+00 1.63871229e-01
1.58769977e+00 -5.75753927e-01 -7.39614427e-01 4.34263706e-01
1.32457638e+00 7.60113537e-01 6.54981434e-01 -3.45116779e-02
8.91629934e-01 1.89658120e-01 8.57247353e-01 3.97201121e-01
-1.96889136e-02 8.10922623e-01 3.30652624e-01 3.12674224e-01
1.56721294e-01 -7.52284944e-01 2.84136593e-01 8.91165853e-01
-2.51079708e-01 -3.42676848e-01 -6.57062709e-01 -2.17719719e-01
-1.98279798e+00 -8.47349048e-01 2.06248462e-01 2.22387075e+00
1.46068180e+00 1.12425134e-01 -7.66081735e-02 -3.87389064e-02
4.34367448e-01 1.15749463e-01 -8.31174314e-01 -1.89411506e-01
2.64516085e-01 6.97230279e-01 6.31913662e-01 3.62177849e-01
-1.27933180e+00 1.47426200e+00 6.32913399e+00 1.58753908e+00
-1.27773070e+00 -3.54788095e-01 8.00270796e-01 -2.99879964e-02
-4.66169901e-02 -4.24907878e-02 -1.36962461e+00 5.09667456e-01
1.25435436e+00 -1.42045349e-01 5.41886210e-01 9.37084556e-01
3.15294951e-01 3.36765707e-01 -1.36352003e+00 9.02209044e-01
-1.29088433e-02 -2.07300949e+00 5.16754448e-01 1.58485416e-02
5.90775549e-01 -1.98788121e-01 7.80967623e-02 3.31908971e-01
-7.16287130e-03 -1.36835682e+00 4.65456039e-01 6.41557574e-01
8.39367986e-01 -1.07384002e+00 4.89770561e-01 7.91177154e-02
-1.41626501e+00 7.49620721e-02 -5.42814195e-01 4.20541644e-01
-3.74036551e-01 4.88595963e-01 -1.27612829e+00 9.55703795e-01
1.68237001e-01 1.10481560e+00 -2.80348361e-01 1.05120862e+00
-2.70194203e-01 3.42787564e-01 -3.47907126e-01 -4.00234163e-01
1.56311035e-01 -1.83499351e-01 -2.02964917e-01 1.04371703e+00
2.13401452e-01 3.40860039e-01 3.59202176e-01 5.58748066e-01
-5.06416082e-01 4.92332637e-01 -4.48876381e-01 -1.23517700e-01
7.08560228e-01 1.15765679e+00 -4.14147407e-01 -3.17093641e-01
5.75778708e-02 5.45101047e-01 2.17864424e-01 2.76082069e-01
-9.74945605e-01 -4.01551634e-01 2.49541730e-01 -3.79421890e-01
3.59461576e-01 2.50610989e-02 3.75106663e-01 -8.60832572e-01
-2.81495601e-01 -1.10354149e+00 5.60522795e-01 -8.31760108e-01
-9.93572295e-01 2.86872417e-01 -1.80305377e-01 -1.08577549e+00
5.31252883e-02 -8.98586571e-01 -3.30326021e-01 5.50550938e-01
-1.26098454e+00 -9.05062854e-01 2.81780571e-01 2.62272656e-01
6.08926117e-01 -8.13036859e-02 1.10805023e+00 3.51664871e-01
-1.05121982e+00 6.06493533e-01 3.27799499e-01 -2.43735626e-01
9.76953387e-01 -8.38085413e-01 1.86336786e-01 1.38081908e-01
-1.84867289e-02 1.04164922e+00 3.01107287e-01 -7.37266958e-01
-1.71503711e+00 -1.48472023e+00 7.00059175e-01 -5.89102507e-02
4.36179221e-01 -3.70858014e-01 -6.99513733e-01 4.87149268e-01
-2.72243083e-01 1.30628492e-03 9.11019146e-01 -2.03793332e-01
-7.35919476e-01 -3.16553384e-01 -8.57490301e-01 7.56544888e-01
1.03964162e+00 -5.22315145e-01 -2.85695225e-01 7.43199527e-01
1.20935512e+00 -6.60144448e-01 -1.39326346e+00 3.52923870e-01
8.08293700e-01 -1.97876260e-01 1.28597069e+00 -9.55271602e-01
5.37647784e-01 -4.04069901e-01 -2.44791359e-01 -9.78228748e-01
-4.72256184e-01 -7.19651818e-01 -1.20776780e-01 8.43807161e-01
9.24416006e-01 -3.47031444e-01 8.48041534e-01 4.65565294e-01
-4.93434608e-01 -1.17233050e+00 -9.42201734e-01 -6.45258427e-01
7.01222643e-02 -1.22541569e-01 8.36165309e-01 6.97867215e-01
7.52527937e-02 7.54110336e-01 -6.53888345e-01 -8.84626135e-02
8.44716132e-02 1.17801666e-01 4.66057450e-01 -1.05825901e+00
-2.93664783e-01 -3.11310679e-01 -9.38609689e-02 -1.09077072e+00
1.05597556e-01 -1.25011647e+00 1.30843639e-01 -1.04884017e+00
4.07585114e-01 -1.79801404e-01 -6.71294332e-01 8.73721063e-01
-2.02900305e-01 -1.23443499e-01 -1.66407317e-01 2.71571636e-01
-7.59044647e-01 7.50567019e-01 1.29702771e+00 -4.05375570e-01
-2.93886900e-01 -2.60536641e-01 -3.78733724e-01 2.98404038e-01
7.99574435e-01 -8.53834987e-01 -3.22947562e-01 2.97329664e-01
3.86901647e-01 2.38929048e-01 -1.66585222e-01 -6.89153433e-01
4.99495327e-01 -4.90978897e-01 6.79129601e-01 -1.10712361e+00
6.54604733e-01 -7.15533137e-01 5.34925699e-01 6.73506021e-01
-5.06975889e-01 -2.85066307e-01 1.35652974e-01 7.93228328e-01
2.27987587e-01 -2.65617013e-01 4.47296113e-01 -8.75413883e-03
-2.21279532e-01 9.60760415e-01 -1.07567728e-01 -5.30901909e-01
7.23558068e-01 1.92111716e-01 -4.77885962e-01 2.76116550e-01
-5.31681597e-01 1.41769141e-01 -8.88895690e-02 2.14125618e-01
7.76281476e-01 -1.40547669e+00 -3.13558616e-02 -1.68701008e-01
3.09480906e-01 2.14102119e-01 -8.09839889e-02 6.03033960e-01
-3.63260686e-01 1.10497701e+00 2.34831721e-01 -1.13364771e-01
-1.14368725e+00 1.04660928e+00 4.54371393e-01 -4.20029998e-01
-4.46038321e-02 5.99287868e-01 2.39286378e-01 -3.44156682e-01
3.58791053e-01 -5.31006217e-01 -2.37355098e-01 -5.72441034e-02
6.07084095e-01 2.43377760e-01 2.63502389e-01 -2.15058014e-01
-3.35578203e-01 5.05741715e-01 -7.27733612e-01 5.20395160e-01
1.35454619e+00 8.32960308e-01 -1.48450702e-01 7.60777071e-02
1.31304705e+00 -3.29122096e-01 -8.27449799e-01 -3.52131754e-01
8.35788548e-02 6.03366941e-02 1.79668248e-01 -8.53792250e-01
-8.74549448e-01 5.89419603e-01 3.44996661e-01 -5.48858464e-01
7.58880079e-01 -1.73310935e-01 8.95038068e-01 1.17802370e+00
1.89389542e-01 -9.36328053e-01 1.60139516e-01 3.63922209e-01
9.96263325e-01 -8.22894752e-01 4.55908000e-01 -5.48754930e-01
-1.09017193e-01 1.40792775e+00 2.91388631e-01 2.09215879e-01
5.27280450e-01 -2.94313163e-01 -6.59232616e-01 -1.45494640e-01
-9.28161740e-01 2.86089301e-01 4.85969394e-01 4.98618037e-02
5.59101999e-01 6.29932731e-02 -3.06208730e-01 8.47121656e-01
4.48999088e-03 1.77035540e-01 -3.94080505e-02 8.52122784e-01
-6.01434529e-01 -1.46825123e+00 -6.01564161e-02 4.31743294e-01
-4.68053132e-01 -6.27103806e-01 -6.76354587e-01 5.48979580e-01
-5.98938763e-02 7.62563765e-01 -7.06733882e-01 -5.07527351e-01
2.52338707e-01 2.87890285e-01 6.35670543e-01 -7.03517437e-01
-7.20992446e-01 2.79584795e-01 2.21515477e-01 -6.64666653e-01
-1.57402590e-01 -1.29444793e-01 -1.39695060e+00 -1.43359855e-01
-7.62503743e-01 4.91516382e-01 5.34825206e-01 9.13904011e-01
5.37072837e-01 3.01165462e-01 8.88020515e-01 -6.72855675e-01
-6.88937426e-01 -7.96474159e-01 -9.92486030e-02 -2.39368349e-01
-1.00590326e-01 -6.90065622e-01 2.43060309e-02 -1.23297580e-01] | [4.499020099639893, 6.106736660003662] |
1a283830-67f7-4534-bb47-04ec1800d851 | old-is-mathbf-mathcal-g-old-redefining-the | 2004.07657 | null | https://arxiv.org/abs/2004.07657v4 | https://arxiv.org/pdf/2004.07657v4.pdf | Old is Gold: Redefining the Adversarially Learned One-Class Classifier Training Paradigm | A popular method for anomaly detection is to use the generator of an adversarial network to formulate anomaly scores over reconstruction loss of input. Due to the rare occurrence of anomalies, optimizing such networks can be a cumbersome task. Another possible approach is to use both generator and discriminator for anomaly detection. However, attributed to the involvement of adversarial training, this model is often unstable in a way that the performance fluctuates drastically with each training step. In this study, we propose a framework that effectively generates stable results across a wide range of training steps and allows us to use both the generator and the discriminator of an adversarial model for efficient and robust anomaly detection. Our approach transforms the fundamental role of a discriminator from identifying real and fake data to distinguishing between good and bad quality reconstructions. To this end, we prepare training examples for the good quality reconstruction by employing the current generator, whereas poor quality examples are obtained by utilizing an old state of the same generator. This way, the discriminator learns to detect subtle distortions that often appear in reconstructions of the anomaly inputs. Extensive experiments performed on Caltech-256 and MNIST image datasets for novelty detection show superior results. Furthermore, on UCSD Ped2 video dataset for anomaly detection, our model achieves a frame-level AUC of 98.1%, surpassing recent state-of-the-art methods. | ['Seung-Ik Lee', 'Jin-ha Lee', 'Muhammad Zaigham Zaheer', 'Marcella Astrid'] | 2020-04-16 | old-is-gold-redefining-the-adversarially | http://openaccess.thecvf.com/content_CVPR_2020/html/Zaheer_Old_Is_Gold_Redefining_the_Adversarially_Learned_One-Class_Classifier_Training_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Zaheer_Old_Is_Gold_Redefining_the_Adversarially_Learned_One-Class_Classifier_Training_CVPR_2020_paper.pdf | cvpr-2020-6 | ['one-class-classifier'] | ['methodology'] | [ 2.87395626e-01 -1.57531440e-01 3.83871496e-01 -4.02175151e-02
-7.31056392e-01 -6.78520024e-01 6.27789497e-01 2.25855425e-01
-3.98381948e-01 5.66581786e-01 -4.33821648e-01 -1.78501919e-01
2.88704634e-01 -8.66902113e-01 -9.68884051e-01 -8.48955631e-01
-8.50344002e-02 1.38800204e-01 2.93043286e-01 -1.01055078e-01
1.96534306e-01 5.47178507e-01 -1.48404765e+00 2.67100483e-01
9.79816914e-01 1.23235321e+00 -3.30852926e-01 5.99680483e-01
4.06555869e-02 5.58798730e-01 -1.01217687e+00 -6.98021650e-01
6.08816206e-01 -7.08091676e-01 -3.36431116e-01 4.17747423e-02
4.86410469e-01 -4.01576132e-01 -3.29676837e-01 1.40739751e+00
4.50243890e-01 6.29866347e-02 6.29133463e-01 -1.30323732e+00
-4.23937678e-01 2.68211663e-01 -3.35529596e-01 5.50127745e-01
3.58129114e-01 4.02565658e-01 7.62004316e-01 -8.56756389e-01
3.97102356e-01 8.99546564e-01 5.12520671e-01 6.31875157e-01
-1.38013744e+00 -7.22623408e-01 9.00212005e-02 1.25959933e-01
-1.36965597e+00 -4.62069273e-01 8.31186414e-01 -4.06946480e-01
4.55982208e-01 1.76862106e-01 4.86720562e-01 1.44175816e+00
3.82885545e-01 5.31284153e-01 8.48765373e-01 -1.49694383e-01
3.08041632e-01 1.07501298e-01 -3.81823272e-01 7.04313159e-01
3.61988842e-01 2.11728826e-01 -2.99972147e-01 -3.17974508e-01
6.22901797e-01 1.13627017e-01 -4.64025855e-01 -2.96187818e-01
-8.89022052e-01 6.20316625e-01 4.81033534e-01 2.67466992e-01
-5.91984332e-01 1.97713319e-02 4.75914448e-01 7.56161690e-01
3.56479794e-01 7.41652429e-01 -2.92728413e-02 -5.65662757e-02
-8.97845447e-01 1.46779865e-01 6.34040654e-01 3.17974180e-01
4.55770433e-01 4.03756231e-01 -2.85307676e-01 7.79724479e-01
1.08194295e-02 2.69450635e-01 6.96091592e-01 -6.23935819e-01
3.93944174e-01 5.11095405e-01 -1.58732548e-01 -1.11783600e+00
1.69902638e-01 -7.00889111e-01 -1.01009262e+00 7.45014310e-01
6.99652791e-01 -1.77768767e-02 -1.03742421e+00 1.80738485e+00
2.17689112e-01 4.30530369e-01 2.52431571e-01 7.77232409e-01
3.41140568e-01 5.19409239e-01 -6.89232945e-02 -9.70239714e-02
9.61892009e-01 -7.04487145e-01 -4.38280880e-01 -1.51369691e-01
4.55140084e-01 -5.97533107e-01 1.06475270e+00 6.87238038e-01
-1.01349640e+00 -5.22460818e-01 -1.14059329e+00 4.86181468e-01
-2.98161030e-01 -1.04392052e-01 1.58802658e-01 6.03904188e-01
-6.75404549e-01 9.45316017e-01 -7.12079287e-01 -8.13175142e-02
5.62532544e-01 1.22105196e-01 -5.78458488e-01 -6.33254722e-02
-1.03621471e+00 7.17461944e-01 4.58253562e-01 -8.93675014e-02
-1.12170172e+00 -5.74104428e-01 -6.86232984e-01 -2.21166690e-03
2.64663517e-01 -4.14853007e-01 8.69458973e-01 -1.40446389e+00
-1.21408808e+00 7.58021474e-01 3.45119834e-01 -7.50923455e-01
9.80511129e-01 -1.07246503e-01 -6.35348141e-01 1.54656410e-01
1.09637268e-02 2.17547819e-01 1.38101244e+00 -1.07168317e+00
-5.19001782e-01 -1.43794373e-01 -3.10462974e-02 -1.01235755e-01
-3.18701148e-01 -3.50353539e-01 -2.25808457e-01 -1.04581153e+00
2.42820323e-01 -7.34796226e-01 -1.42528698e-01 3.03300112e-01
-4.06484723e-01 1.67858198e-01 8.60619307e-01 -5.82466483e-01
1.17993271e+00 -2.46041846e+00 -1.07934758e-01 4.87293810e-01
9.43225920e-02 5.21850109e-01 -2.05385506e-01 4.84437123e-02
-3.98527801e-01 6.31191880e-02 -5.76792955e-01 -2.57840276e-01
-3.18457514e-01 2.85892844e-01 -5.14913321e-01 6.13558769e-01
5.85696101e-01 6.54464185e-01 -9.57537770e-01 -1.05357498e-01
4.82906513e-02 2.21892178e-01 -6.78007305e-01 4.58080918e-01
-6.27638549e-02 6.25970423e-01 -2.69234419e-01 7.08336592e-01
5.73114693e-01 2.80464813e-03 -2.76468515e-01 1.07118599e-01
3.33162457e-01 8.58315006e-02 -1.37275708e+00 1.47751355e+00
-2.24073842e-01 5.04247189e-01 -1.91425428e-01 -1.24972999e+00
9.68806982e-01 3.91369373e-01 2.93442994e-01 -7.07113028e-01
1.19235896e-01 4.28689778e-01 2.45190516e-01 -3.49357098e-01
1.96871370e-01 -1.65666297e-01 6.66258037e-02 4.59389031e-01
2.48525828e-01 5.98644279e-02 1.51860818e-01 1.38370350e-01
1.52263594e+00 -1.92023486e-01 2.06846118e-01 2.00497404e-01
7.86751211e-01 -5.30670166e-01 7.17827618e-01 8.36486459e-01
-2.82068849e-01 8.10308993e-01 7.03724623e-01 -5.01418650e-01
-1.18818724e+00 -1.37401509e+00 -1.41565755e-01 4.84973222e-01
-1.06919579e-01 -1.12857133e-01 -6.73003137e-01 -1.11433339e+00
3.39072905e-02 5.50301313e-01 -5.52519441e-01 -6.49303913e-01
-5.67117035e-01 -6.95780277e-01 8.66979063e-01 3.51007432e-01
5.92651129e-01 -1.03528583e+00 -3.50950807e-01 2.19604447e-01
-1.53697982e-01 -1.13224208e+00 -2.22082898e-01 2.28857547e-02
-8.45952451e-01 -1.11077952e+00 -5.53088486e-01 -3.33984196e-01
9.08110023e-01 -1.60145253e-01 1.03469646e+00 4.07626569e-01
-2.08234444e-01 2.67627761e-02 -3.71246725e-01 -3.16853762e-01
-9.15536880e-01 -3.78033906e-01 2.71777630e-01 4.46095675e-01
-6.80188909e-02 -7.44026005e-01 -4.93554860e-01 3.07205737e-01
-1.24901462e+00 -4.97946769e-01 6.84339941e-01 9.99754131e-01
6.71724737e-01 2.73629874e-01 7.24733770e-01 -8.15599442e-01
5.39334297e-01 -6.66148007e-01 -6.10302448e-01 -1.14912309e-01
-6.74041867e-01 1.91116288e-01 1.11707306e+00 -7.24197149e-01
-6.39313102e-01 -1.12543311e-02 -5.15778363e-01 -8.33220363e-01
-2.55450696e-01 1.60632119e-01 -1.81421965e-01 -1.10553652e-01
9.86573994e-01 3.74292016e-01 2.76323885e-01 -3.27301085e-01
3.02788913e-02 2.73032725e-01 8.43366385e-01 -4.13394779e-01
1.27150619e+00 3.65412593e-01 1.35566130e-01 -6.24173045e-01
-6.17446184e-01 -2.56248005e-03 -2.76067019e-01 -1.98641822e-01
4.33079451e-01 -6.31879628e-01 -2.44633630e-01 7.90514708e-01
-9.07170832e-01 1.49182370e-03 -6.14961386e-01 2.57128745e-01
-2.86714137e-01 6.00239217e-01 -2.55400062e-01 -6.99127972e-01
-2.90624171e-01 -1.16778767e+00 5.90555251e-01 -2.37465557e-02
3.95749994e-02 -8.30660701e-01 3.11808251e-02 -8.81341100e-02
4.10246044e-01 8.37949753e-01 7.16458380e-01 -1.00098705e+00
-5.23691297e-01 -7.67628312e-01 7.47446939e-02 1.03544843e+00
1.75190136e-01 8.42713285e-03 -9.53877866e-01 -4.21914250e-01
2.53697395e-01 -1.74871877e-01 8.54177296e-01 -1.17073223e-01
1.52204132e+00 -5.09270012e-01 4.82875071e-02 7.63949275e-01
1.26658702e+00 1.57363728e-01 9.48050499e-01 4.07797545e-01
5.68655908e-01 2.35546827e-01 4.26994264e-01 2.41160467e-01
-2.31189370e-01 4.82227564e-01 7.18366444e-01 7.29153380e-02
8.21301267e-02 -2.88526416e-01 6.28352523e-01 2.67829627e-01
-1.83561407e-02 -3.78078252e-01 -7.25619078e-01 4.02216643e-01
-1.57225168e+00 -1.22023094e+00 1.88713595e-01 2.47477484e+00
6.32409394e-01 5.61763465e-01 1.07616119e-01 7.10603595e-01
6.31465733e-01 1.88625768e-01 -5.32432199e-01 -3.72658670e-01
-1.79408550e-01 2.84646332e-01 1.83480665e-01 -4.57342118e-02
-1.16889417e+00 5.87803304e-01 5.65866661e+00 7.93380618e-01
-1.27838469e+00 -2.45866571e-02 6.12077057e-01 -6.77951872e-02
-1.82824675e-02 -2.42449775e-01 -1.47568688e-01 9.59362209e-01
9.01137412e-01 -7.53923357e-02 3.37670326e-01 8.52600217e-01
-4.57438231e-02 1.06109167e-02 -1.19119656e+00 8.62701654e-01
1.36074856e-01 -9.95018363e-01 2.74016470e-01 8.50433111e-03
5.99834561e-01 -2.21624538e-01 1.97444543e-01 2.33410925e-01
-1.35140970e-01 -1.01286328e+00 6.81677639e-01 5.22976100e-01
7.18369126e-01 -7.98294246e-01 9.48409617e-01 4.61507440e-01
-8.97041619e-01 -1.82454363e-01 -2.69608647e-01 9.94437486e-02
-2.46062130e-03 7.34146476e-01 -7.56563723e-01 5.38901269e-01
5.73703706e-01 5.55862069e-01 -6.31436348e-01 1.12672806e+00
-3.77275616e-01 6.11166716e-01 -3.06398898e-01 5.32896876e-01
1.48344845e-01 -1.64685786e-01 9.94926810e-01 9.40183699e-01
4.02608216e-01 -2.25133300e-01 -4.23514843e-02 8.39624405e-01
-3.70619953e-01 -6.67286068e-02 -1.00614166e+00 -1.26152635e-02
2.52992988e-01 1.02064037e+00 -5.15422881e-01 -1.85433775e-01
-1.59892246e-01 1.25074410e+00 2.49623820e-01 2.35474020e-01
-9.47806835e-01 -2.99301863e-01 7.87228942e-01 1.32675335e-01
3.07549208e-01 1.45588875e-01 -4.21312414e-02 -1.29585207e+00
4.66432869e-01 -1.13901174e+00 5.40424347e-01 -2.38322303e-01
-1.43303239e+00 7.60054350e-01 -3.85847032e-01 -1.81230783e+00
-3.89530599e-01 -4.03094679e-01 -1.00337982e+00 6.97007716e-01
-1.36221755e+00 -6.47619545e-01 -4.11768049e-01 6.76475942e-01
4.57351029e-01 -4.54895139e-01 7.03756750e-01 5.25588453e-01
-6.12952828e-01 1.04342020e+00 -1.43581375e-01 4.54385310e-01
6.20906830e-01 -1.22255516e+00 4.44806695e-01 1.52896810e+00
3.47419888e-01 2.13117808e-01 7.43216872e-01 -4.92403001e-01
-9.02591884e-01 -1.23448265e+00 3.58301997e-01 -3.90370071e-01
5.95781684e-01 -1.42698660e-01 -1.46799457e+00 4.07871097e-01
-2.92975277e-01 4.49052453e-01 3.98867339e-01 -4.94332850e-01
-5.16202331e-01 -1.07092269e-01 -1.52004611e+00 4.81221378e-01
8.58214676e-01 -4.07937616e-01 -5.03280640e-01 1.99859999e-02
3.86889815e-01 -5.79597950e-01 -5.89371681e-01 4.71948862e-01
4.21703756e-01 -1.16752684e+00 1.03986919e+00 -5.08277237e-01
6.27171814e-01 -3.96079808e-01 -4.88765016e-02 -1.47097468e+00
2.95271948e-02 -3.87054414e-01 -4.83811468e-01 1.17104614e+00
2.98008293e-01 -9.36198711e-01 5.70879400e-01 2.31633052e-01
-8.79593492e-02 -8.58962774e-01 -1.07045233e+00 -1.03050876e+00
-2.16175225e-02 -4.73210722e-01 5.36434710e-01 8.68254066e-01
-6.02146626e-01 -2.06613421e-01 -2.88509458e-01 3.67251843e-01
6.51713908e-01 -2.66001016e-01 8.50278556e-01 -1.21298194e+00
-3.79585862e-01 -4.37948883e-01 -9.40024197e-01 -6.38549030e-01
2.10071564e-01 -9.26071525e-01 -1.42940432e-02 -8.08488905e-01
-3.22643518e-01 -4.76386160e-01 -5.07290900e-01 2.77446777e-01
-3.25163245e-01 5.44944286e-01 3.88594642e-02 2.33694509e-01
-2.64837593e-01 5.62027991e-01 9.62508619e-01 -8.34485143e-02
-4.39208895e-02 3.10511351e-01 -5.06299436e-01 8.30744386e-01
8.26417565e-01 -7.35128164e-01 -2.25673974e-01 -1.90327972e-01
7.99731985e-02 -2.14905411e-01 6.89917564e-01 -1.39051199e+00
-3.71699780e-02 1.17711812e-01 6.10251069e-01 -1.75331369e-01
2.24582613e-01 -8.99957359e-01 9.77343321e-02 5.90494692e-01
-9.37545970e-02 3.04264694e-01 9.23766121e-02 8.17011118e-01
-4.07151550e-01 -3.46557826e-01 1.16630125e+00 -2.04828978e-02
-4.73253101e-01 4.11067158e-01 -2.06932947e-01 2.07129002e-01
1.06689608e+00 -2.75181323e-01 -2.03739524e-01 -2.93959111e-01
-5.78806400e-01 -3.52372229e-02 7.13024199e-01 3.93718302e-01
7.39157498e-01 -1.37506986e+00 -6.19033575e-01 6.57574177e-01
2.11376101e-01 6.61912141e-03 3.68383192e-02 6.97390974e-01
-3.70070577e-01 -4.16397423e-01 -4.67843145e-01 -6.80334985e-01
-9.87805128e-01 5.01210093e-01 5.42830586e-01 -2.44505182e-01
-7.89369404e-01 6.44170165e-01 -2.40182355e-02 -1.15740662e-02
2.55358964e-01 -5.41126616e-02 2.54400354e-03 -7.06664324e-02
5.23409426e-01 2.67443806e-01 3.34083736e-01 -5.86971045e-01
-2.73452669e-01 1.49820775e-01 1.40112294e-02 6.96606562e-02
1.09100807e+00 1.82149455e-01 1.03802636e-01 3.18800300e-01
1.13714671e+00 -4.16249372e-02 -1.17089164e+00 -2.52141982e-01
-1.64727196e-01 -7.89255977e-01 -1.36809021e-01 -5.10705769e-01
-1.28638899e+00 6.89130068e-01 8.41368735e-01 3.98153335e-01
1.34418559e+00 -2.84441859e-01 7.37356067e-01 3.05721104e-01
7.79299587e-02 -8.34975421e-01 4.42051142e-01 3.17332268e-01
8.23779762e-01 -1.46204662e+00 -3.09100121e-01 2.27591936e-02
-5.84042907e-01 9.85372663e-01 6.59612298e-01 -4.34125781e-01
3.27414244e-01 6.88885227e-02 5.84014021e-02 -9.36110765e-02
-5.57903707e-01 1.26584858e-01 4.02216315e-01 4.27029222e-01
-4.66048382e-02 -1.45344645e-01 -1.59814760e-01 3.20047170e-01
-1.87742338e-01 -3.79575223e-01 6.95586801e-01 8.88410807e-01
-2.16900527e-01 -1.09866786e+00 -4.54667568e-01 6.75671697e-01
-8.91980290e-01 1.13458872e-01 -1.46124646e-01 5.66436231e-01
1.18246600e-01 7.34422088e-01 1.99283510e-01 -3.80134225e-01
5.42374969e-01 2.83360660e-01 8.98339078e-02 -4.07231599e-01
-6.29429698e-01 -3.40105742e-01 -2.38663629e-01 -7.94338048e-01
-1.08730895e-02 -6.01019263e-01 -1.05907655e+00 -2.47313157e-01
-2.31581166e-01 6.93926364e-02 5.54716468e-01 8.79813671e-01
2.37314045e-01 5.74072659e-01 1.01928425e+00 -6.35191977e-01
-8.37919831e-01 -7.44243503e-01 -5.03916979e-01 8.99635553e-01
7.35401869e-01 -8.07949722e-01 -7.51344085e-01 -9.48331282e-02] | [7.703464508056641, 2.2436957359313965] |
a9227d1d-a04b-4708-85ed-85bede0dc496 | improving-language-identification-of-accented | 2203.16972 | null | https://arxiv.org/abs/2203.16972v3 | https://arxiv.org/pdf/2203.16972v3.pdf | Improving Language Identification of Accented Speech | Language identification from speech is a common preprocessing step in many spoken language processing systems. In recent years, this field has seen fast progress, mostly due to the use of self-supervised models pretrained on multilingual data and the use of large training corpora. This paper shows that for speech with a non-native or regional accent, the accuracy of spoken language identification systems drops dramatically, and that the accuracy of identifying the language is inversely correlated with the strength of the accent. We also show that using the output of a lexicon-free speech recognition system of the particular language helps to improve language identification performance on accented speech by a large margin, without sacrificing accuracy on native speech. We obtain relative error rate reductions ranging from to 35 to 63% over the state-of-the-art model across several non-native speech datasets. | ['Tanel Alumäe', 'Kunnar Kukk'] | 2022-03-31 | null | null | null | null | ['spoken-language-identification'] | ['speech'] | [ 4.34897803e-02 6.27800301e-02 -2.39651158e-01 -7.34559000e-01
-1.15626812e+00 -8.24851036e-01 6.07842505e-01 5.01459278e-02
-7.54767060e-01 5.89312494e-01 4.47023302e-01 -4.27735448e-01
3.37040424e-01 -2.21137285e-01 -3.15345943e-01 -5.02551556e-01
1.57075912e-01 7.73507833e-01 1.01308666e-01 -4.18080270e-01
1.05786063e-02 3.43293846e-01 -1.27474451e+00 -1.13028012e-01
7.13565886e-01 6.86955333e-01 8.49376321e-02 7.60424376e-01
-3.98581594e-01 6.11833930e-01 -7.46477902e-01 -2.95560896e-01
7.07239658e-02 -4.03897017e-01 -1.01418412e+00 1.03255726e-01
5.18887103e-01 4.14990596e-02 -1.26740023e-01 1.12617171e+00
4.98943180e-01 -9.61507857e-02 3.41831177e-01 -7.51677215e-01
-2.66832322e-01 9.52258110e-01 -1.10257253e-01 3.07841331e-01
4.45030272e-01 -1.49226218e-01 8.75278115e-01 -8.81572723e-01
4.01908547e-01 1.40257001e+00 6.10025465e-01 4.45468843e-01
-1.29564619e+00 -7.59858072e-01 1.07196040e-01 -2.88114637e-01
-1.74091148e+00 -1.10704851e+00 4.22803968e-01 -3.45920295e-01
1.24369526e+00 1.71977073e-01 6.13552406e-02 7.15237498e-01
-2.67508924e-01 5.07276475e-01 1.20293021e+00 -7.79174089e-01
1.66652277e-01 4.54795122e-01 3.09554815e-01 6.58764064e-01
-2.80397713e-01 -1.79916844e-02 -6.34258091e-01 -5.58189526e-02
2.84542501e-01 -7.57808983e-01 -2.54518092e-01 1.01295404e-01
-1.15926266e+00 8.29767704e-01 -1.14140458e-01 6.60599530e-01
-2.63338953e-01 -4.96679813e-01 5.29460371e-01 4.75771219e-01
7.19730675e-01 3.24770898e-01 -5.74917316e-01 -3.97688776e-01
-1.04535091e+00 -1.76336408e-01 1.24113357e+00 6.09478354e-01
7.07610011e-01 3.91408622e-01 5.18939853e-01 1.37725413e+00
9.29760486e-02 7.57250726e-01 7.74661362e-01 -5.33654332e-01
3.99203330e-01 3.94948661e-01 -1.71592012e-01 -3.26379985e-01
-4.33374763e-01 -2.86816061e-01 -5.74021757e-01 -7.54763559e-02
6.84921145e-01 -4.08588290e-01 -8.79962444e-01 1.83244872e+00
1.77221701e-01 -2.82416731e-01 4.80351299e-01 5.44513941e-01
3.93301696e-01 6.44859076e-01 1.35315850e-01 -3.52187872e-01
1.29612899e+00 -7.60745823e-01 -6.44199729e-01 -5.83614290e-01
5.50593019e-01 -1.07031751e+00 1.08030772e+00 2.48562261e-01
-7.91024089e-01 -5.79995871e-01 -8.35421562e-01 2.23980129e-01
-3.82470399e-01 -4.14034650e-02 2.01717466e-01 1.18004799e+00
-1.30210567e+00 -9.94086862e-02 -6.22988701e-01 -7.34902084e-01
-2.36210823e-01 6.52421832e-01 -6.72886968e-01 1.37135997e-01
-1.11580765e+00 1.02245331e+00 3.85254800e-01 -2.91701645e-01
-3.93441319e-01 -5.73194385e-01 -9.13572431e-01 -1.76627710e-01
1.74705744e-01 9.86251608e-03 1.37779999e+00 -1.30036223e+00
-1.73622215e+00 1.09714437e+00 -4.87914234e-01 -5.34583032e-01
3.18297178e-01 -1.60921723e-01 -8.74425769e-01 -1.07775718e-01
1.60143211e-01 4.44160640e-01 6.62999094e-01 -9.41755652e-01
-7.98672259e-01 -3.60269725e-01 -3.48481864e-01 2.60983080e-01
-5.32946169e-01 6.56018674e-01 -2.65847266e-01 -5.75595140e-01
-4.27786149e-02 -1.09102654e+00 6.27738088e-02 -6.48495078e-01
-3.09067100e-01 -3.99287760e-01 6.59106255e-01 -9.79468524e-01
1.14547801e+00 -2.35725904e+00 2.98195519e-03 2.39320919e-01
-3.29228610e-01 4.45948124e-01 -6.30030483e-02 3.26447278e-01
-8.11570883e-02 1.46827862e-01 -1.73005193e-01 -4.43959981e-01
-5.67070395e-02 2.78636038e-01 -2.72614568e-01 3.20057482e-01
1.19741619e-01 3.86663079e-01 -7.45836973e-01 -2.39527389e-01
2.14465976e-01 4.26111519e-01 -5.24108261e-02 2.04034567e-01
2.28267893e-01 3.07976782e-01 1.30403459e-01 4.20944840e-01
3.55387837e-01 1.48627222e-01 3.99455518e-01 2.87605226e-01
-3.73888671e-01 9.16960418e-01 -1.18550229e+00 1.29126704e+00
-7.07861543e-01 4.81055796e-01 5.59986889e-01 -6.85113132e-01
1.03880537e+00 6.13088191e-01 6.80505112e-02 -4.88402605e-01
6.96397200e-02 5.77568710e-01 3.68597120e-01 -1.35837197e-02
3.57567370e-01 -1.72792226e-01 -1.58017367e-01 5.22440374e-01
2.68895566e-01 9.89987282e-04 3.34942453e-02 -1.51847452e-01
6.43629193e-01 -6.33367598e-01 4.41358566e-01 -6.64132178e-01
9.17042911e-01 4.95662801e-02 3.38222831e-01 6.20414078e-01
-2.31369421e-01 3.32502097e-01 -1.01984665e-01 -2.69837230e-01
-9.68931496e-01 -1.05901897e+00 -3.09588939e-01 1.57286549e+00
-4.70221788e-01 -1.10865407e-01 -1.05005801e+00 -5.03624618e-01
-7.46294856e-02 8.30464363e-01 -3.85163799e-02 1.22496700e-02
-7.77319908e-01 -6.95220590e-01 9.62085068e-01 4.28374469e-01
4.90469038e-01 -9.62492168e-01 3.80452663e-01 3.13710362e-01
-1.23666957e-01 -1.44666255e+00 -8.39392245e-01 3.40519369e-01
-4.09161359e-01 -6.36439025e-01 -6.61564469e-01 -1.33692193e+00
4.99617189e-01 -1.03496395e-01 9.60045457e-01 -1.23258948e-01
1.96292311e-01 3.89313906e-01 -1.85940266e-01 -3.59140217e-01
-1.18650460e+00 6.85167313e-01 6.42020047e-01 2.33161703e-01
6.51013434e-01 -1.17544435e-01 1.47674024e-01 3.33479643e-01
-4.58647281e-01 -4.87874627e-01 2.54228532e-01 7.65419602e-01
2.13521838e-01 -1.21171668e-01 8.67098451e-01 -7.14351118e-01
5.55356085e-01 -2.01890096e-01 -7.03748047e-01 3.00821394e-01
-7.75124252e-01 8.80350620e-02 7.01988816e-01 -4.83855844e-01
-1.00231874e+00 5.06820858e-01 -5.36024153e-01 3.84821407e-02
-5.41204453e-01 3.79371583e-01 -2.07064569e-01 -2.18282491e-01
4.80260551e-01 6.35686934e-01 2.73127347e-01 -4.79788840e-01
1.83332697e-01 1.14797187e+00 7.38153458e-01 -3.23249310e-01
6.60883367e-01 -1.08430035e-01 -5.78190386e-01 -1.42374182e+00
-5.62946320e-01 -6.95043087e-01 -8.16297174e-01 7.28472620e-02
7.00505316e-01 -1.11250877e+00 -3.48126888e-01 8.30139875e-01
-9.45559442e-01 -3.39090079e-01 7.84269571e-02 6.17001772e-01
-9.87709239e-02 3.70597392e-01 -6.39742076e-01 -1.03750074e+00
-4.71761525e-01 -1.12767231e+00 8.35854411e-01 -2.36052219e-02
-5.77574968e-01 -1.13700032e+00 6.00027554e-02 4.22924697e-01
5.43212175e-01 -6.31693900e-01 8.07208478e-01 -1.23709977e+00
-7.78581053e-02 -2.96712875e-01 2.45551705e-01 7.38805115e-01
5.22743642e-01 -1.81354865e-01 -1.16776180e+00 -3.09877694e-01
-2.57586479e-01 -3.27486783e-01 4.82099265e-01 8.56501758e-02
1.69608250e-01 -2.57417202e-01 4.36943956e-02 3.43860149e-01
9.32289600e-01 2.83358365e-01 1.42322704e-01 2.04784602e-01
6.32310569e-01 8.23098183e-01 2.59156879e-02 -1.20164908e-01
6.31344140e-01 5.42868733e-01 -4.20814186e-01 -4.14257050e-02
-1.97517082e-01 -1.35532424e-01 5.78360200e-01 1.29429865e+00
3.99871826e-01 -4.76960465e-02 -1.32643318e+00 8.38382721e-01
-1.32580245e+00 -6.87163770e-01 1.11501053e-01 2.48495197e+00
1.10952330e+00 1.37007892e-01 5.42625248e-01 2.13587657e-01
8.72980773e-01 7.30408281e-02 -1.80475220e-01 -6.22197449e-01
-4.23179775e-01 2.66423970e-01 5.86952209e-01 1.17733455e+00
-1.18694568e+00 1.39093709e+00 7.73382759e+00 6.14756048e-01
-1.50885570e+00 -1.17546916e-01 6.15908086e-01 3.50960582e-01
2.19459891e-01 -2.93786496e-01 -1.09067500e+00 3.25585306e-01
1.46564686e+00 -5.06494582e-01 6.76292598e-01 7.41530120e-01
1.12545982e-01 -5.31636477e-02 -9.58796859e-01 8.33168507e-01
3.02475512e-01 -6.21918082e-01 -5.39180525e-02 7.32254330e-03
5.12014925e-01 4.90969926e-01 -4.91736494e-02 1.66624144e-01
4.26184207e-01 -9.40048516e-01 6.38173223e-01 -1.72970831e-01
7.43888080e-01 -8.85096967e-01 7.13585496e-01 5.64381957e-01
-1.09599912e+00 1.70060888e-01 6.44558147e-02 -8.16575810e-02
1.25361562e-01 1.92468584e-01 -1.32640028e+00 9.08173472e-02
5.28827906e-01 6.62954524e-02 -4.30436403e-01 5.76861024e-01
-1.35875732e-01 1.17499948e+00 -6.38332367e-01 -9.15083140e-02
1.65012598e-01 -3.95309702e-02 4.16352123e-01 1.49598336e+00
1.29733220e-01 -4.84590419e-02 4.62579638e-01 2.02533156e-01
-4.25868444e-02 5.92165530e-01 -5.47609150e-01 -2.88007021e-01
6.13968492e-01 1.02382731e+00 -4.67860878e-01 -4.68735248e-01
-5.70741117e-01 9.51616287e-01 2.98059464e-01 1.10699199e-01
-1.33814476e-02 -3.85597825e-01 9.34091985e-01 -2.66256109e-02
1.79103136e-01 -3.77455384e-01 -2.47304663e-01 -8.61481071e-01
-1.46639675e-01 -1.14917147e+00 1.77358791e-01 -1.86144263e-01
-1.32143962e+00 1.01728177e+00 -2.25978240e-01 -5.64930737e-01
-9.81840074e-01 -5.67394078e-01 -3.16641629e-01 1.37829721e+00
-1.26814592e+00 -9.90152419e-01 2.22444475e-01 4.53282565e-01
6.43586934e-01 -6.66465104e-01 1.33956730e+00 3.41857731e-01
-5.60347259e-01 8.87651086e-01 1.72971651e-01 7.29190230e-01
9.16350663e-01 -1.29743755e+00 7.43558764e-01 8.03807020e-01
4.01671290e-01 6.23765707e-01 6.06298625e-01 -4.17475581e-01
-1.12051547e+00 -8.10334682e-01 1.62339282e+00 -5.65059900e-01
9.67394948e-01 -5.27237058e-01 -1.10158098e+00 7.20102847e-01
4.99219269e-01 -2.78303474e-01 8.46002698e-01 4.90143508e-01
-5.42952180e-01 -2.15177238e-01 -9.95242238e-01 3.13858926e-01
6.23087704e-01 -1.16697180e+00 -6.99198663e-01 2.00518504e-01
4.83985573e-01 -1.06140442e-01 -8.19416285e-01 5.37042171e-02
5.10610878e-01 -5.60377955e-01 7.51848817e-01 -4.12121117e-01
-3.84587318e-01 -2.41537809e-01 -2.95032710e-01 -1.53877091e+00
-1.73818827e-01 -5.68471253e-01 5.98566830e-01 1.63517404e+00
6.30417466e-01 -8.49206150e-01 5.42541087e-01 5.11590362e-01
4.48924676e-02 8.80173147e-02 -1.06883144e+00 -8.92515719e-01
3.04781169e-01 -2.80945361e-01 3.98868978e-01 1.04912639e+00
1.57309532e-01 7.85376191e-01 -8.21241140e-02 3.85725021e-01
4.28956211e-01 -3.32871467e-01 7.55814433e-01 -1.15120232e+00
-7.55478665e-02 -5.33002138e-01 -4.87807423e-01 -8.74605060e-01
6.89160883e-01 -8.53304684e-01 2.81605661e-01 -9.34990346e-01
-2.06199735e-01 -2.45192572e-01 -2.07362518e-01 5.71129441e-01
-2.02199742e-01 2.36715510e-01 8.54627416e-02 8.62190798e-02
-2.41769642e-01 4.74140793e-02 3.41351539e-01 -1.71349347e-01
-3.37524146e-01 -4.31163125e-02 -6.06539547e-01 8.06798339e-01
8.23593259e-01 -3.10413182e-01 -1.72243193e-01 -4.60471332e-01
-4.82450306e-01 -4.01555449e-02 -1.32029161e-01 -1.03442633e+00
3.08032840e-01 2.33977661e-01 2.43748003e-03 -1.70584366e-01
2.84014791e-01 -4.79199558e-01 -1.78260386e-01 3.39200616e-01
-3.62324148e-01 3.44618857e-01 3.83257657e-01 1.69462673e-02
-6.15440071e-01 -2.22079065e-02 9.18841302e-01 -4.80555296e-02
-8.91896725e-01 -2.63441086e-01 -7.06856966e-01 1.20752245e-01
4.62210447e-01 1.40174866e-01 -1.25829548e-01 -4.92658705e-01
-5.45815051e-01 -6.10740632e-02 4.24591422e-01 7.11888671e-01
-7.39201680e-02 -9.24454451e-01 -8.50180209e-01 6.05085313e-01
1.45197198e-01 -5.92326581e-01 -2.26842016e-01 4.72413093e-01
-3.44370037e-01 6.75842583e-01 1.04990341e-01 -5.82452774e-01
-1.57623601e+00 1.81326628e-01 5.88228524e-01 7.25639611e-02
-7.05021396e-02 9.07965958e-01 -2.37859190e-02 -8.50414395e-01
4.15480882e-01 -7.16106668e-02 -1.18713249e-02 -3.60804163e-02
6.69572830e-01 9.60359648e-02 2.30462343e-01 -1.39266515e+00
-8.00224185e-01 5.23205340e-01 -2.93827862e-01 -4.28679466e-01
8.10966372e-01 -2.15447873e-01 -5.57321273e-02 7.72009969e-01
1.17331731e+00 4.70531762e-01 -6.46553159e-01 -4.28379953e-01
3.29461217e-01 7.05149546e-02 2.84302115e-01 -9.68389571e-01
-7.29895949e-01 5.91690898e-01 6.95235014e-01 2.82052666e-01
8.47180426e-01 5.53617552e-02 6.08069420e-01 4.69959050e-01
4.89135653e-01 -1.13222229e+00 -6.33102298e-01 1.00072849e+00
5.79545021e-01 -1.51292765e+00 -4.49638605e-01 -5.01076937e-01
-6.35395408e-01 9.46862578e-01 3.02957177e-01 9.72868130e-02
9.40186918e-01 5.93485415e-01 9.23582375e-01 3.17147642e-01
-3.60062808e-01 -2.79904038e-01 3.24221134e-01 5.05145907e-01
7.65552759e-01 3.38683933e-01 -8.00108686e-02 3.86957109e-01
-6.45984828e-01 -4.46628362e-01 8.33327770e-02 6.77239597e-01
-5.40272117e-01 -1.56012523e+00 -5.81272364e-01 1.33122012e-01
-5.68801820e-01 -4.17980313e-01 -8.22918117e-01 7.22632527e-01
-1.74927652e-01 1.30331588e+00 2.27689296e-01 -3.64497900e-01
3.40814680e-01 7.41173267e-01 1.22470990e-01 -6.69287980e-01
-8.88412178e-01 1.93675250e-01 4.37621921e-01 4.80693169e-02
-3.62561166e-01 -9.18480694e-01 -1.23606992e+00 -3.27098697e-01
-1.87436849e-01 3.33376139e-01 7.22806334e-01 9.84471440e-01
1.10453360e-01 2.18068082e-02 7.01815546e-01 -4.82000411e-01
-7.24350810e-01 -1.11935079e+00 -7.27508247e-01 1.43601418e-01
5.49359083e-01 -1.87375903e-01 -5.34140348e-01 8.60035270e-02] | [14.164530754089355, 6.629953861236572] |
73f02050-9b4d-4dd0-8baa-f93d50fa4824 | segregated-temporal-assembly-recurrent | 1811.07460 | null | http://arxiv.org/abs/1811.07460v1 | http://arxiv.org/pdf/1811.07460v1.pdf | Segregated Temporal Assembly Recurrent Networks for Weakly Supervised Multiple Action Detection | This paper proposes a segregated temporal assembly recurrent (STAR) network
for weakly-supervised multiple action detection. The model learns from
untrimmed videos with only supervision of video-level labels and makes
prediction of intervals of multiple actions. Specifically, we first assemble
video clips according to class labels by an attention mechanism that learns
class-variable attention weights and thus helps the noise relieving from
background or other actions. Secondly, we build temporal relationship between
actions by feeding the assembled features into an enhanced recurrent neural
network. Finally, we transform the output of recurrent neural network into the
corresponding action distribution. In order to generate more precise temporal
proposals, we design a score term called segregated temporal gradient-weighted
class activation mapping (ST-GradCAM) fused with attention weights. Experiments
on THUMOS'14 and ActivityNet1.3 datasets show that our approach outperforms the
state-of-the-art weakly-supervised method, and performs at par with the
fully-supervised counterparts. | ['ShiLiang Pu', 'Zhanzhan Cheng', 'Yunlu Xu', 'Yi Niu', 'Jianwen Xie', 'Fei Wu', 'Chengwei Zhang'] | 2018-11-19 | null | null | null | null | ['multiple-action-detection'] | ['computer-vision'] | [ 3.86517286e-01 6.59520328e-02 -4.98467356e-01 -4.05663341e-01
-8.76554012e-01 -2.75091588e-01 5.68693519e-01 -5.20390570e-01
-4.47546870e-01 5.13676167e-01 5.49526989e-01 1.70092002e-01
1.86724171e-01 -3.48334610e-01 -8.31589341e-01 -8.31784606e-01
-3.89298499e-01 1.57964617e-01 6.06114507e-01 1.38735354e-01
-5.60814179e-02 9.65374112e-02 -1.61693144e+00 8.67307544e-01
3.33902746e-01 1.12774980e+00 1.58624038e-01 7.30885923e-01
1.30726576e-01 1.72144735e+00 -4.99504119e-01 2.80736070e-02
1.46100312e-01 -8.97668958e-01 -7.28449941e-01 5.09054482e-01
2.37508684e-01 -3.73778909e-01 -5.77868938e-01 7.50927627e-01
1.67960927e-01 5.20194948e-01 5.12702703e-01 -9.78969812e-01
-4.35192287e-01 6.30015552e-01 -6.15195572e-01 5.41462481e-01
3.60588342e-01 1.11786187e-01 1.13993692e+00 -9.89092886e-01
7.16512918e-01 1.18305099e+00 4.08754289e-01 6.44565940e-01
-1.09007955e+00 -3.88013273e-01 7.56879687e-01 5.23982346e-01
-1.09611344e+00 -2.73538291e-01 7.17870414e-01 -4.37757462e-01
1.18358803e+00 9.80222248e-04 7.48989344e-01 1.46704137e+00
5.96915418e-03 1.28274989e+00 6.60711586e-01 -2.01196000e-01
7.68931285e-02 -2.83483982e-01 -1.10996507e-01 9.09070790e-01
-6.39599264e-01 -1.06184803e-01 -6.63694799e-01 1.94410339e-01
8.77517700e-01 2.84743905e-01 -1.35016292e-01 -2.08868265e-01
-1.51492846e+00 4.57203060e-01 3.59831363e-01 4.95384485e-01
-5.72692454e-01 5.15826643e-01 4.92698550e-01 1.55118123e-01
7.14981139e-01 1.14405558e-01 -4.64567304e-01 -2.35775664e-01
-9.61651385e-01 -1.40304357e-01 1.76736131e-01 7.97335386e-01
6.01948559e-01 2.05071464e-01 -6.78731084e-01 6.85132504e-01
1.51306599e-01 1.34180859e-01 6.66849613e-01 -1.19180882e+00
5.11365175e-01 4.37804043e-01 1.10361941e-01 -7.45254755e-01
-4.24577057e-01 -3.33866566e-01 -5.84909976e-01 1.81431994e-01
2.91919857e-01 -9.31392312e-02 -9.88926291e-01 1.70152044e+00
1.98103264e-02 8.31599176e-01 5.44486335e-03 1.05067647e+00
4.76557761e-01 8.57010365e-01 1.88304082e-01 -4.31969583e-01
1.04702389e+00 -1.59829175e+00 -8.61900747e-01 -1.58239126e-01
6.03937447e-01 -2.07595557e-01 8.78384948e-01 4.38944310e-01
-1.07457829e+00 -8.32009733e-01 -8.66940081e-01 8.33264142e-02
-1.03866294e-01 5.77370524e-01 4.73297268e-01 -1.09191254e-01
-9.86579955e-01 7.13049352e-01 -1.19580805e+00 -2.61890709e-01
5.47003627e-01 1.06537707e-01 -3.38169664e-01 2.39712045e-01
-1.06880641e+00 6.91618145e-01 2.57202655e-01 2.03768849e-01
-1.40371752e+00 -3.23192686e-01 -1.07259130e+00 -1.45203657e-02
5.49395382e-01 -3.08976859e-01 1.37944591e+00 -1.44397044e+00
-1.70802736e+00 6.84632301e-01 -1.76532045e-01 -7.51958132e-01
4.00571525e-01 -4.28800017e-01 -4.01989222e-01 5.13620198e-01
1.66852906e-01 8.78193796e-01 1.12257445e+00 -6.75948858e-01
-8.66312385e-01 5.90137253e-03 9.53482017e-02 4.57614660e-01
-2.53005207e-01 2.71380574e-01 -7.35558510e-01 -8.24894786e-01
-1.50396377e-02 -8.40214133e-01 -4.65241015e-01 -8.30126405e-02
-1.48602933e-01 -4.83853400e-01 9.05679286e-01 -6.99474037e-01
1.29666722e+00 -2.33975029e+00 5.86799145e-01 -1.49864852e-01
-1.24806846e-02 1.76026970e-02 -4.10093188e-01 -4.10667062e-02
-3.61206293e-01 -2.33290777e-01 9.76608042e-03 -6.36916816e-01
-2.35171497e-01 3.61704856e-01 -2.96269745e-01 5.52058578e-01
4.50523645e-01 9.35696840e-01 -1.22783887e+00 -5.38304329e-01
3.57697845e-01 4.25637573e-01 -4.48645771e-01 3.83573592e-01
-5.04596293e-01 5.03070414e-01 -4.31255430e-01 7.25567639e-01
-1.78773329e-01 -3.14025402e-01 1.52744785e-01 -1.45537227e-01
-1.48137853e-01 3.48630786e-01 -8.34839582e-01 2.06769228e+00
-3.31845820e-01 6.41982198e-01 -4.47446495e-01 -1.26384878e+00
7.02051282e-01 5.93728244e-01 7.37590194e-01 -6.45531356e-01
-3.21942777e-03 -2.98524648e-01 -2.67444640e-01 -7.23981619e-01
2.27726623e-01 2.47403726e-01 -2.02267140e-01 5.05830705e-01
5.38837612e-01 4.08315748e-01 3.39475423e-01 2.56629467e-01
1.32339013e+00 9.32720423e-01 1.68452322e-01 1.90643489e-01
5.73023140e-01 -3.79565090e-01 8.95106673e-01 6.30106509e-01
-3.08071733e-01 8.04751456e-01 7.45345771e-01 -5.87160587e-01
-7.27618694e-01 -9.01852071e-01 4.68684524e-01 1.67442656e+00
-6.31120056e-02 -4.29343283e-01 -6.07354522e-01 -1.20313334e+00
-5.33644199e-01 3.82399648e-01 -9.47300196e-01 -7.88654760e-02
-7.28970885e-01 -4.54285741e-01 3.46950203e-01 1.02032685e+00
4.37045097e-01 -1.54703116e+00 -7.29668498e-01 5.46350956e-01
-3.53736311e-01 -1.17262018e+00 -6.75726712e-01 4.17611629e-01
-8.09319913e-01 -9.97107744e-01 -7.18605101e-01 -7.73574769e-01
6.67354226e-01 1.88993365e-01 9.45019901e-01 -3.32304090e-01
-8.63456056e-02 3.22412699e-01 -5.80208719e-01 1.53454706e-01
-2.65446100e-02 -1.78966060e-01 2.78276261e-02 5.18407285e-01
3.09118271e-01 -6.87153161e-01 -5.26640236e-01 5.00418782e-01
-7.13150620e-01 1.06483713e-01 7.85386264e-01 8.00031245e-01
7.87005961e-01 -6.30697384e-02 4.25130248e-01 -6.32606864e-01
6.42543584e-02 -4.07154113e-01 -5.09461582e-01 3.10977876e-01
4.66409735e-02 1.21397860e-01 5.02832294e-01 -8.00950110e-01
-1.30055535e+00 5.37119925e-01 1.30720124e-01 -9.48015213e-01
-2.14172557e-01 3.35470527e-01 6.08406365e-02 4.00849164e-01
5.93484223e-01 3.30184013e-01 -3.64029884e-01 -3.23075742e-01
4.32856053e-01 2.58505851e-01 8.07840645e-01 -2.51708031e-01
3.46771479e-01 5.76695204e-01 -3.87496024e-01 -5.67838788e-01
-1.31427157e+00 -6.05532646e-01 -8.36339712e-01 -7.37346351e-01
1.33374703e+00 -1.09246385e+00 -4.20959979e-01 4.76792634e-01
-1.13979101e+00 -6.97659850e-01 -4.68710572e-01 7.12589145e-01
-8.44746411e-01 1.86739847e-01 -9.38960791e-01 -7.96553850e-01
8.96543078e-03 -8.76058340e-01 1.18437076e+00 4.33415920e-02
-2.90720910e-01 -6.86868012e-01 1.79470211e-01 3.84893805e-01
1.38636632e-02 2.99860299e-01 3.04298878e-01 -5.36356449e-01
-6.07394934e-01 -1.86704725e-01 1.32540604e-02 3.96049380e-01
1.62093434e-02 -9.08600725e-03 -9.83559728e-01 2.99990773e-02
-1.18837066e-01 -5.87201834e-01 1.26723552e+00 5.81365466e-01
1.45553851e+00 -2.74165392e-01 -2.05507740e-01 4.00287151e-01
8.94125223e-01 2.27588892e-01 7.88638413e-01 8.06338340e-02
7.37892270e-01 5.16770482e-01 8.55878651e-01 5.54507613e-01
9.05649960e-02 8.42267513e-01 4.35168386e-01 -1.32998660e-01
-1.07384466e-01 -2.32498676e-01 9.44480598e-01 6.73528373e-01
-5.27303994e-01 -9.67917293e-02 -3.00550252e-01 5.97648263e-01
-2.35057783e+00 -1.54270697e+00 2.69063026e-01 1.80601645e+00
6.26144111e-01 3.69227499e-01 4.22211975e-01 2.82281395e-02
7.61235416e-01 5.48549592e-01 -6.30575895e-01 -6.70110658e-02
-1.37259766e-01 2.96933223e-02 2.31314987e-01 2.71569043e-01
-1.53974044e+00 1.05208373e+00 6.16495466e+00 7.74730802e-01
-7.44676471e-01 3.35118175e-01 6.52724922e-01 -3.97476465e-01
1.99501276e-01 -1.58300743e-01 -6.94638073e-01 3.70722324e-01
8.88962328e-01 2.63488084e-01 3.13898414e-01 8.76593947e-01
5.15236974e-01 -6.62231073e-02 -1.12765765e+00 7.50342190e-01
3.25666875e-01 -1.16292310e+00 -1.36497065e-01 -2.09158480e-01
7.92939305e-01 1.07247099e-01 -1.53897591e-02 5.91790855e-01
4.36653823e-01 -7.59276748e-01 8.75563860e-01 8.16130996e-01
5.13974071e-01 -5.61439693e-01 5.80750883e-01 2.13896468e-01
-1.56361639e+00 -2.94556171e-01 -1.10564768e-01 -2.11316526e-01
3.13665003e-01 1.96483091e-01 -4.18573320e-01 2.81454682e-01
8.18635523e-01 1.60813999e+00 -4.13154960e-01 7.03800559e-01
-6.28287077e-01 7.20398307e-01 -5.65199926e-02 2.01937735e-01
5.64427972e-01 -8.98718834e-02 3.29714268e-01 1.25638556e+00
1.75553873e-01 2.41705447e-01 4.07793194e-01 4.59737509e-01
-3.86645645e-02 -2.21354768e-01 -5.05551338e-01 -9.01290774e-02
-4.30407152e-02 1.32655394e+00 -7.72041917e-01 -6.25943482e-01
-7.43018389e-01 1.37372589e+00 4.64615613e-01 5.83267987e-01
-1.09617746e+00 3.24444100e-02 4.68087524e-01 -1.64368838e-01
6.37659192e-01 6.77967966e-02 2.47174293e-01 -1.37315655e+00
-2.92612822e-03 -5.49131572e-01 8.04191411e-01 -9.64091659e-01
-9.40961838e-01 6.91998482e-01 -1.71634480e-01 -1.75028896e+00
-4.58898008e-01 -2.60501921e-01 -7.08957553e-01 2.74645239e-01
-1.17504656e+00 -1.05071485e+00 -4.18531783e-02 6.22789204e-01
1.17126191e+00 -1.85116291e-01 7.03500807e-01 2.42802069e-01
-7.20362425e-01 2.85857946e-01 -4.32634383e-01 2.97903419e-01
5.76652229e-01 -1.12934864e+00 1.43149048e-01 1.02027357e+00
4.77194071e-01 1.34409331e-02 4.04121578e-01 -5.32879949e-01
-6.78067803e-01 -1.47563422e+00 6.77493930e-01 -4.62487996e-01
8.77939284e-01 -2.85948157e-01 -8.44384491e-01 9.33304369e-01
2.89627373e-01 4.38459337e-01 5.03518879e-01 -2.49580845e-01
-3.52522403e-01 -7.97849745e-02 -6.63634956e-01 5.14098227e-01
1.31756055e+00 -6.54445410e-01 -6.71008170e-01 4.57822859e-01
8.29363823e-01 -3.05525661e-01 -6.33406460e-01 3.42911541e-01
5.57337523e-01 -8.98421764e-01 8.37442458e-01 -7.76145339e-01
4.67555076e-01 -4.46106255e-01 -2.11647246e-02 -1.02638507e+00
-5.27743101e-01 -6.74677491e-01 -4.13539529e-01 8.29841733e-01
5.56577742e-01 1.74506549e-02 8.93459976e-01 1.39957428e-01
-4.25306916e-01 -8.01372647e-01 -8.53180289e-01 -5.91164351e-01
-5.75869083e-01 -4.63936627e-01 -5.76809160e-02 7.62019277e-01
2.26880223e-01 3.08052361e-01 -9.38457489e-01 1.05943322e-01
4.01144803e-01 5.26004378e-03 4.04662549e-01 -6.56154990e-01
-6.01218879e-01 -4.29471254e-01 -5.47192931e-01 -1.31797862e+00
4.18319970e-01 -4.86571103e-01 2.95419395e-01 -1.27772307e+00
1.92818701e-01 1.80230349e-01 -8.35578859e-01 9.21725333e-01
-5.54709136e-02 5.36352932e-01 6.53824350e-03 5.64721487e-02
-1.48218954e+00 7.24967122e-01 1.08039474e+00 -1.68485999e-01
-3.47035021e-01 1.49635032e-01 -4.74936403e-02 8.74312222e-01
6.14303648e-01 -4.24956501e-01 -3.99036944e-01 -2.81925678e-01
-4.35828865e-02 4.28891182e-01 3.88978988e-01 -1.18581080e+00
2.63265282e-01 -1.75134167e-01 4.16460693e-01 -8.34373832e-01
4.40889537e-01 -6.51359975e-01 -9.57822055e-02 3.60654026e-01
-8.30939412e-01 -1.64046586e-01 -2.05740079e-01 8.65901947e-01
-3.77936363e-01 1.81675151e-01 6.46062553e-01 -1.99184552e-01
-1.02838635e+00 3.64492357e-01 -9.38234150e-01 -1.98295325e-01
1.10883391e+00 -1.48875073e-01 -5.86662926e-02 -4.02304143e-01
-1.25724399e+00 3.50929171e-01 -5.14245778e-02 6.40256763e-01
7.62391508e-01 -1.56475830e+00 -4.86504465e-01 1.21647440e-01
1.89955428e-01 -2.38725066e-01 3.32604498e-01 9.13480163e-01
2.46543307e-02 2.49988019e-01 -1.11146420e-01 -6.32252991e-01
-1.18845081e+00 5.81161857e-01 4.12138224e-01 -3.91399950e-01
-7.41608262e-01 9.57775295e-01 2.83795238e-01 -9.86389257e-03
6.28043950e-01 -4.67339218e-01 -5.70646405e-01 2.35958889e-01
7.44446099e-01 2.19577000e-01 -3.15530568e-01 -7.46737123e-01
-3.09249192e-01 4.13495302e-01 -3.86933424e-02 -2.16864586e-01
1.44770384e+00 1.99068282e-02 1.35886356e-01 6.84830248e-01
1.11956048e+00 -5.53576291e-01 -1.96171999e+00 -3.87116134e-01
5.19577935e-02 -1.51582792e-01 -1.20349981e-01 -5.50142288e-01
-1.24701762e+00 7.43522167e-01 4.84802932e-01 -6.00729175e-02
1.21992731e+00 2.26161823e-01 5.29127657e-01 4.53117490e-01
1.20170534e-01 -1.23494017e+00 8.12882483e-01 4.92794514e-01
7.72929728e-01 -1.11220539e+00 -2.70964295e-01 -3.32372598e-02
-9.22166944e-01 9.85201597e-01 9.10377443e-01 -2.36416727e-01
5.13616383e-01 1.76637992e-01 5.09744063e-02 -2.05848768e-01
-1.09324598e+00 -4.41869527e-01 3.84164542e-01 2.87805170e-01
2.86631912e-01 -1.86347425e-01 -2.89324317e-02 4.84573454e-01
5.55351138e-01 6.73559532e-02 2.83537686e-01 8.76853585e-01
-4.81548160e-01 -7.50505805e-01 -5.15566804e-02 3.83974016e-01
-4.36586142e-01 1.38955191e-01 -1.21172011e-01 3.58770102e-01
2.02177793e-01 7.45051324e-01 3.58066499e-01 -5.59355438e-01
3.15126032e-01 9.89589021e-02 3.09881717e-01 -8.07216406e-01
-4.66721743e-01 5.44653594e-01 1.80312052e-01 -1.12945211e+00
-1.01453340e+00 -8.09354782e-01 -1.25523126e+00 4.90129352e-01
-9.66075212e-02 1.28791779e-01 4.06884700e-02 1.09870696e+00
2.13029817e-01 9.26908374e-01 7.63790369e-01 -1.31871557e+00
-7.36670047e-02 -1.11073911e+00 -5.78890979e-01 3.86434764e-01
3.73202652e-01 -6.96531415e-01 -4.28567767e-01 4.82430667e-01] | [8.411823272705078, 0.5530698895454407] |
b3986740-85dd-4cc3-acaf-3ee41df69ce1 | s3m-scalable-statistical-shape-modeling | 2304.07515 | null | https://arxiv.org/abs/2304.07515v1 | https://arxiv.org/pdf/2304.07515v1.pdf | S3M: Scalable Statistical Shape Modeling through Unsupervised Correspondences | Statistical shape models (SSMs) are an established way to geometrically represent the anatomy of a population with various clinically relevant applications. However, they typically require domain expertise and labor-intensive manual segmentations or landmark annotations to generate. Methods to estimate correspondences for SSMs typically learn with such labels as supervision signals. We address these shortcomings by proposing an unsupervised method that leverages deep geometric features and functional correspondences to learn local and global shape structures across complex anatomies simultaneously. Our pipeline significantly improves unsupervised correspondence estimation for SSMs compared to baseline methods, even on highly irregular surface topologies. We demonstrate this for two different anatomical structures: the thyroid and a multi-chamber heart dataset. Furthermore, our method is robust enough to learn from noisy neural network predictions, enabling scaling SSMs to larger patient populations without manual annotation. | ['Nassir Navab', 'Benjamin Busam', 'Mahdi Saleh', 'Ha Young Kim', 'Vincent Bürgin', 'Emily Hoppe', 'Alexander Bauman', 'Lennart Bastian'] | 2023-04-15 | null | null | null | null | ['anatomy'] | ['miscellaneous'] | [ 2.8788078e-01 5.6183410e-01 -1.1540575e-01 -6.1800206e-01
-1.1292760e+00 -7.0797008e-01 2.7805391e-01 4.8800325e-01
3.0066535e-02 4.5005742e-01 1.7465553e-01 -3.0468491e-01
-1.8011231e-02 -4.4840014e-01 -5.9235871e-01 -2.2288935e-01
-2.3398812e-01 9.6514368e-01 1.3834345e-01 4.0104073e-02
1.5039930e-01 5.9000307e-01 -6.1992496e-01 7.2661310e-02
9.4719809e-01 9.0203238e-01 5.1697161e-02 5.2510983e-01
-1.4088628e-01 2.5054678e-01 -2.6205468e-01 -3.1667483e-01
4.6070147e-01 -4.6206129e-01 -8.4663409e-01 2.5053617e-01
7.3568046e-01 -1.3226990e-01 1.9839537e-01 5.7708585e-01
6.2987399e-01 -2.8149092e-01 7.9762477e-01 -6.7059422e-01
-1.2206786e-01 4.3667230e-01 -4.0180653e-01 8.5411876e-02
-4.2172562e-02 3.0687392e-01 1.0114509e+00 -7.8061628e-01
7.0921797e-01 9.3577462e-01 1.2533567e+00 3.7086308e-01
-1.6515383e+00 -2.4456905e-01 -2.0039555e-01 -6.0069680e-01
-1.1636750e+00 -4.9310905e-01 6.3500577e-01 -6.5714073e-01
5.3994262e-01 -2.1810461e-02 8.2433516e-01 6.4414358e-01
1.7426421e-01 5.8097935e-01 1.0506883e+00 -5.6904268e-02
7.6714471e-02 -2.5214815e-01 -2.9565141e-01 1.0560766e+00
2.7150467e-01 -1.3381526e-01 -1.4946613e-01 -4.6879870e-01
1.4090605e+00 -1.5778148e-01 -4.6319041e-02 -7.9019296e-01
-1.3772926e+00 6.3323438e-01 4.4947132e-01 7.4837044e-02
-3.5476983e-01 4.2466736e-01 2.2833228e-01 -1.7970097e-01
2.7747807e-01 8.7760401e-01 -6.1015481e-01 1.5310582e-02
-1.1655661e+00 1.6308645e-02 7.3420596e-01 9.6359169e-01
5.5609703e-01 -2.1659095e-02 -9.5684282e-02 7.9443258e-01
4.7817132e-01 3.6257237e-01 1.6034044e-01 -1.1604025e+00
2.8753701e-01 7.0359284e-01 -7.8039557e-02 -9.7426832e-01
-8.8523114e-01 -6.8221229e-01 -7.2959000e-01 -4.2981807e-02
7.7203256e-01 -2.4017297e-01 -1.1611571e+00 1.5597050e+00
3.7302122e-01 2.9141235e-01 -3.9714032e-01 7.1594071e-01
8.7962800e-01 -8.9028329e-02 1.7591833e-01 5.0333917e-02
1.1098490e+00 -5.8648050e-01 -2.2789717e-01 -3.7031424e-01
8.4614825e-01 -5.9935749e-01 1.1813476e+00 8.4165253e-02
-1.3815806e+00 -6.9826737e-02 -6.8732309e-01 6.8491869e-02
2.7452475e-01 4.1457602e-01 6.9279289e-01 5.5461395e-01
-1.1164986e+00 8.3680063e-01 -1.4130298e+00 -1.7410818e-01
1.0751337e+00 6.0304356e-01 -2.1886133e-01 3.7255284e-01
-4.3806666e-01 8.7525195e-01 -4.0905397e-02 1.0750318e-01
-9.2145264e-01 -1.1817669e+00 -9.5919454e-01 -3.7889250e-02
1.8448980e-01 -8.9958549e-01 1.2749047e+00 -7.0945799e-01
-1.4740866e+00 1.2296315e+00 -9.8036006e-02 -4.6731877e-01
7.6576865e-01 1.1345887e-01 1.8151456e-01 3.0906010e-01
9.3206488e-02 6.5683246e-01 8.3783108e-01 -1.1846859e+00
6.0232520e-02 -2.7662402e-01 -5.2872276e-01 -4.4877059e-03
1.7188905e-01 -9.5026396e-02 -1.5249415e-01 -8.7387931e-01
7.4671048e-01 -9.9276567e-01 -8.0344939e-01 4.7565556e-01
-5.9907812e-01 2.7088806e-01 2.4637879e-01 -6.2222141e-01
7.8964555e-01 -1.8831851e+00 1.8577093e-01 6.6995007e-01
6.0117000e-01 4.6160959e-02 3.5040062e-02 -1.3474479e-01
1.6897665e-01 3.0692020e-01 -7.3747247e-01 -3.6332306e-01
-3.2269165e-01 1.8306026e-01 8.6555466e-02 4.7007599e-01
3.1224391e-01 1.3838570e+00 -8.4884173e-01 -8.1992352e-01
1.1180921e-01 2.7747443e-01 -8.3660662e-01 8.0915608e-02
-3.7542331e-01 1.1771338e+00 -6.7579198e-01 6.6672301e-01
3.6966681e-01 -8.2042986e-01 5.8152932e-01 -4.4615200e-01
2.9519808e-01 2.7940363e-01 -8.7929863e-01 2.1847315e+00
-4.8034281e-01 1.2461981e-01 3.8758215e-01 -1.1299212e+00
9.1661268e-01 3.5453486e-01 1.1281414e+00 -1.0463365e-02
4.2603600e-01 6.6126269e-01 1.3633506e-01 -2.6757714e-01
-3.1847519e-01 -4.9940047e-01 2.7179612e-02 3.4249011e-01
2.4733551e-01 -7.2377479e-01 -2.3481210e-01 2.8673897e-03
1.0423193e+00 3.3700746e-01 2.9765615e-01 -5.8384722e-01
1.9789270e-01 6.3953847e-02 8.2496023e-01 5.3200877e-01
-2.4772258e-01 1.0454035e+00 3.9847031e-01 -6.7205441e-01
-1.0819271e+00 -1.3303804e+00 -4.4596475e-01 4.7383493e-01
4.9477228e-04 -4.5364267e-01 -7.5680059e-01 -8.2132286e-01
1.6853800e-01 1.2962395e-01 -4.3313637e-01 1.2806961e-01
-8.2821131e-01 -6.8679178e-01 7.2802132e-01 6.5749979e-01
1.0062451e-02 -9.3724227e-01 -5.8758146e-01 4.2960018e-01
-1.9461019e-02 -1.2940304e+00 -6.2746686e-01 -1.5574564e-01
-1.3661451e+00 -1.2934847e+00 -6.9156700e-01 -8.6485064e-01
1.0861528e+00 -4.9076608e-01 1.4112227e+00 3.6873078e-01
-9.3244559e-01 4.5891619e-01 2.2508574e-01 -4.4552019e-01
-5.5830216e-01 1.3400742e-01 -5.4633088e-02 -4.9642541e-02
-3.7524346e-01 -8.8691050e-01 -8.3647251e-01 4.8965380e-01
-3.8741341e-01 1.1538232e-01 6.8421739e-01 7.4483085e-01
1.0727564e+00 -6.6069716e-01 6.9646579e-01 -1.2482767e+00
3.1853747e-01 -2.8662038e-01 -6.6410267e-01 1.4398079e-01
-5.1724505e-01 1.9526023e-01 3.9605671e-01 -1.4154695e-01
-8.1725019e-01 5.3222042e-01 -1.2226484e-01 -4.6939808e-01
-3.6245194e-01 3.7849689e-01 2.5598356e-01 -4.3455052e-01
8.3133572e-01 6.1224937e-02 4.7033376e-01 -5.1529133e-01
3.6335093e-01 1.6256852e-02 7.5094116e-01 -8.4372157e-01
8.6116791e-01 6.7827660e-01 5.2181816e-01 -5.5941319e-01
-9.1286063e-01 -1.8656380e-01 -1.2132849e+00 1.6564955e-01
7.4134988e-01 -5.3334594e-01 -6.0410148e-01 9.7959183e-02
-8.9246708e-01 -5.7506800e-01 -3.9667690e-01 4.6391797e-01
-9.1222787e-01 2.8030691e-01 -6.8042779e-01 -2.8583696e-01
-5.1992244e-01 -1.1987563e+00 1.4326543e+00 -1.2425291e-01
-5.4130828e-01 -1.3407736e+00 -1.1133539e-01 3.5120711e-01
5.7355434e-01 6.9661117e-01 9.4947326e-01 -5.8653182e-01
-7.1705103e-01 -8.3825327e-03 1.3608444e-02 4.2240437e-02
5.2018899e-01 -3.6904667e-02 -5.9880406e-01 -2.0350873e-01
-2.6346150e-01 -2.6426768e-01 5.0652301e-01 8.8394880e-01
1.5429327e+00 -1.7357214e-01 -5.7304704e-01 9.4462639e-01
9.5972431e-01 -3.2570055e-01 2.1869245e-01 -2.4193649e-01
9.5187378e-01 6.0627395e-01 3.0906036e-01 2.3198214e-01
3.8390848e-01 4.5058414e-01 2.6766950e-01 -6.0029298e-01
-2.3431121e-01 -3.0258745e-01 -2.4043132e-01 7.4654943e-01
-5.4694705e-02 4.3421653e-01 -1.2966387e+00 6.3426942e-01
-1.5088109e+00 -3.8020974e-01 -1.7922817e-01 2.2539427e+00
1.1318545e+00 1.1301860e-01 1.6182058e-01 -3.8508034e-01
4.4345391e-01 -1.9921263e-01 -6.7619199e-01 1.2701537e-01
1.5856485e-01 4.8446369e-01 4.6063742e-01 3.9565569e-01
-1.1080552e+00 9.2429024e-01 7.5784459e+00 3.1653684e-01
-1.2623556e+00 3.3177558e-02 9.1989946e-01 3.3816442e-02
-3.5878760e-01 1.6262992e-01 -4.2799598e-01 1.1303928e-01
6.8215728e-01 2.5271073e-01 1.6832590e-01 4.8477861e-01
3.3322760e-01 1.9183749e-01 -1.1237104e+00 9.3151122e-01
-1.7339560e-01 -1.7348853e+00 -2.5981870e-01 -3.3331271e-02
7.7409887e-01 2.2376332e-01 -5.0933804e-02 -2.2062594e-01
3.7843031e-01 -1.3877292e+00 3.4493458e-01 3.7751195e-01
1.1424297e+00 -1.4030732e-01 3.6190373e-01 2.1413666e-01
-1.1825869e+00 2.7712750e-01 3.5470940e-02 5.3001904e-01
3.9408377e-01 3.9456037e-01 -1.3394861e+00 2.1628481e-01
3.3137399e-01 8.0217385e-01 -5.7666898e-01 1.2545354e+00
-4.5978069e-02 7.9228002e-01 -7.1295363e-01 6.3717526e-01
-5.3831112e-02 -2.0672014e-01 6.3663560e-01 9.8516262e-01
1.8641533e-01 -2.0714784e-02 4.4586691e-01 1.2108953e+00
-7.9242334e-02 3.0005372e-01 -4.8302883e-01 -3.6611941e-02
4.1105708e-01 1.5232936e+00 -1.1326522e+00 -8.4067658e-02
-1.4846799e-01 4.9114019e-01 2.3662007e-01 2.3504600e-02
-5.6534141e-01 1.6675989e-01 3.4708160e-01 7.1640223e-01
-6.1513152e-02 -4.2221513e-01 -8.0413967e-01 -1.0645548e+00
-8.2551122e-02 -4.8161915e-01 4.7262782e-01 -4.1791496e-01
-1.3862944e+00 2.8987932e-01 -8.7559737e-02 -1.3531500e+00
-9.2760414e-02 -4.3937600e-01 -7.5788879e-01 6.4081359e-01
-1.4322619e+00 -1.3920389e+00 -3.8646886e-01 1.5908550e-01
2.2183524e-01 -5.5625336e-03 9.0500855e-01 2.0542639e-01
-2.8834978e-01 6.1231822e-01 -3.7935567e-01 3.1994712e-01
6.8665141e-01 -1.4985693e+00 4.8570874e-01 2.3954786e-01
2.8223974e-01 6.6395825e-01 2.8442407e-01 -8.4421664e-01
-1.2696069e+00 -1.1249800e+00 2.8921428e-01 -8.0047119e-01
4.4278386e-01 -2.2812666e-01 -8.6405170e-01 9.5374018e-01
-3.3339918e-01 7.8960395e-01 7.3220628e-01 8.5795902e-02
-4.4187240e-02 6.7967542e-02 -1.3191701e+00 4.5908996e-01
1.2181858e+00 -2.0455807e-01 -3.2664981e-01 5.5748659e-01
2.7073216e-01 -9.6187526e-01 -1.6085024e+00 6.9401866e-01
4.4619128e-01 -5.6454515e-01 1.1898559e+00 -6.3226265e-01
2.6880899e-01 -3.7410074e-01 3.0570537e-01 -1.2680659e+00
-2.6988141e-02 -1.0223200e+00 4.1562494e-02 7.1130568e-01
6.6853362e-01 -5.1740134e-01 1.0805446e+00 7.5208318e-01
-5.0460303e-01 -9.4938952e-01 -8.9157236e-01 -3.7619698e-01
3.6294913e-01 -2.6288727e-01 4.2875060e-01 1.0163275e+00
-1.3870122e-01 1.3232990e-01 1.3442244e-01 1.0361507e-01
7.7095050e-01 1.8505421e-01 7.0789194e-01 -1.6261640e+00
-2.7863988e-01 -5.6244659e-01 -4.4541752e-01 -7.6773310e-01
1.0949265e-01 -1.4431949e+00 1.4171749e-01 -1.5896399e+00
-3.0680856e-02 -1.1978102e+00 -7.6165885e-02 7.4808794e-01
-1.2203930e-01 4.2592564e-01 -1.9525783e-01 3.7697795e-01
-3.4845310e-01 3.4125108e-01 1.7644740e+00 1.5094745e-01
-1.9729331e-01 2.2051108e-01 -6.2717551e-01 1.1466687e+00
8.4558380e-01 -3.7936816e-01 -4.2297259e-01 -3.6082458e-01
-1.8411880e-02 2.6964673e-01 5.0496435e-01 -7.0925516e-01
-6.4979466e-03 -1.5468122e-01 5.6865954e-01 -3.0920100e-01
1.6103593e-01 -7.6145953e-01 1.5485462e-02 3.8335121e-01
-3.7626517e-01 -1.9057408e-01 3.2216248e-01 5.0557059e-01
1.1985308e-01 5.5025365e-02 1.0716829e+00 -4.1271299e-01
3.8637303e-02 6.2765723e-01 -8.1122808e-02 5.3645635e-01
9.1340441e-01 -2.4704288e-01 1.9923174e-01 -3.3480680e-01
-1.2578751e+00 3.0994052e-01 4.9436238e-01 -1.8327427e-03
5.5903834e-01 -1.1677989e+00 -8.1939018e-01 2.7159375e-01
-9.3615286e-02 7.6858485e-01 -7.9113215e-02 1.2913529e+00
-8.3161354e-01 1.4841545e-01 4.6521965e-02 -1.1678762e+00
-9.4449842e-01 -7.4148603e-02 9.1802609e-01 -1.9401169e-01
-1.0321671e+00 7.1282524e-01 2.2348830e-01 -9.9430454e-01
-2.2860441e-01 -6.8082535e-01 2.2720974e-02 -4.8760968e-01
-2.2827671e-01 6.9144227e-02 8.5414357e-02 -7.3697114e-01
-3.0274564e-01 8.7724161e-01 2.7360484e-01 1.7504273e-01
1.4692867e+00 4.3724559e-02 -7.1141630e-02 8.3239935e-02
9.2760080e-01 6.1669920e-02 -1.4807122e+00 -3.3433437e-01
3.8022894e-01 -3.4059352e-01 -9.6672989e-02 -7.4114960e-01
-1.2603110e+00 8.5034996e-01 3.8113752e-01 -2.4613072e-01
7.0436782e-01 2.7617827e-01 8.3888358e-01 1.9101648e-01
3.3438101e-01 -8.1225097e-01 1.6537231e-01 1.4602779e-01
8.3422250e-01 -1.3502597e+00 1.7841174e-01 -9.1498482e-01
-6.6653788e-01 1.1163324e+00 4.8016188e-01 -2.6012936e-01
8.6439031e-01 5.9654921e-01 2.9722595e-01 -5.8639538e-01
-4.0553874e-01 1.5690868e-01 7.3710090e-01 5.4761183e-01
7.1623671e-01 7.8998856e-02 -7.0634790e-02 4.7244254e-01
-2.6019681e-01 -1.8360317e-01 3.6200112e-01 8.2485831e-01
-2.1564318e-01 -1.3044261e+00 -1.3411663e-01 8.9710283e-01
-7.2859794e-01 -1.0575110e-01 -2.7132750e-01 3.9579555e-01
-2.5504199e-01 2.8194994e-01 8.4890164e-02 2.2529598e-01
1.4151481e-01 3.9609019e-02 7.5456858e-01 -1.1363963e+00
-6.3488269e-01 4.8880419e-01 -9.0707451e-02 -7.2638714e-01
-1.1371059e-01 -9.2809486e-01 -1.8201849e+00 3.2945687e-01
-1.8443820e-01 -2.4284203e-01 6.8083268e-01 9.5726758e-01
6.4097327e-01 5.1446342e-01 3.1243488e-01 -9.2120099e-01
-5.1901489e-01 -8.0953318e-01 -2.7237740e-01 7.4637413e-01
1.9907407e-01 -7.3721039e-01 2.8883340e-02 2.6333109e-01] | [14.199416160583496, -2.4564080238342285] |
d02092e3-3e2a-4a68-ada7-1bd34c8524f5 | unsupervised-image-to-image-translation-with-3 | 2204.03641 | null | https://arxiv.org/abs/2204.03641v1 | https://arxiv.org/pdf/2204.03641v1.pdf | Unsupervised Image-to-Image Translation with Generative Prior | Unsupervised image-to-image translation aims to learn the translation between two visual domains without paired data. Despite the recent progress in image translation models, it remains challenging to build mappings between complex domains with drastic visual discrepancies. In this work, we present a novel framework, Generative Prior-guided UNsupervised Image-to-image Translation (GP-UNIT), to improve the overall quality and applicability of the translation algorithm. Our key insight is to leverage the generative prior from pre-trained class-conditional GANs (e.g., BigGAN) to learn rich content correspondences across various domains. We propose a novel coarse-to-fine scheme: we first distill the generative prior to capture a robust coarse-level content representation that can link objects at an abstract semantic level, based on which fine-level content features are adaptively learned for more accurate multi-level content correspondences. Extensive experiments demonstrate the superiority of our versatile framework over state-of-the-art methods in robust, high-quality and diversified translations, even for challenging and distant domains. | ['Chen Change Loy', 'Ziwei Liu', 'Liming Jiang', 'Shuai Yang'] | 2022-04-07 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Yang_Unsupervised_Image-to-Image_Translation_With_Generative_Prior_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Yang_Unsupervised_Image-to-Image_Translation_With_Generative_Prior_CVPR_2022_paper.pdf | cvpr-2022-1 | ['unsupervised-image-to-image-translation'] | ['computer-vision'] | [ 5.85389733e-01 -4.02650833e-02 -2.60838479e-01 -3.61238301e-01
-1.00330460e+00 -6.35773122e-01 8.91927600e-01 -5.24151325e-01
1.37435257e-01 8.40574026e-01 3.05288941e-01 1.96608886e-01
2.00006023e-01 -8.04566920e-01 -1.01619232e+00 -7.14714706e-01
7.40165234e-01 5.06419063e-01 1.03007160e-01 -2.10062698e-01
-1.04597144e-01 2.65449047e-01 -1.40316725e+00 4.48296607e-01
1.07543421e+00 8.29766631e-01 3.84875029e-01 3.13288301e-01
-3.09009869e-02 5.40812016e-01 -2.15235665e-01 -5.67316413e-01
4.31939125e-01 -1.04040873e+00 -5.86563051e-01 4.77944463e-01
6.10623419e-01 -4.49157119e-01 -1.71664208e-01 1.18177235e+00
3.29911470e-01 -1.39230147e-01 8.33048344e-01 -1.28887105e+00
-1.16624832e+00 2.65925884e-01 -6.85198069e-01 -2.91813999e-01
4.37660396e-01 4.14588392e-01 9.78788435e-01 -1.05590403e+00
1.07588232e+00 1.33779085e+00 3.32766145e-01 4.48552012e-01
-1.64355779e+00 -6.81331575e-01 1.05450444e-01 -1.63543841e-03
-1.22753501e+00 -4.35312033e-01 9.19654727e-01 -5.94496608e-01
4.13191497e-01 -1.47795305e-01 7.26371527e-01 1.44092858e+00
1.62343338e-01 5.64407766e-01 1.50493515e+00 -5.10496318e-01
6.07460225e-03 6.80704042e-02 -9.12261605e-01 7.26380229e-01
2.34878868e-01 2.77518362e-01 -7.90015519e-01 1.83439508e-01
1.29114842e+00 1.79368034e-01 -2.07717150e-01 -7.10285306e-01
-1.52621675e+00 7.87357688e-01 5.70872366e-01 2.95543015e-01
-4.96259600e-01 2.10899174e-01 -7.54449889e-02 2.50642478e-01
4.68879879e-01 3.11626017e-01 -1.94318846e-01 6.02424741e-02
-1.09962511e+00 2.74564862e-01 3.98266286e-01 1.22056735e+00
1.13397133e+00 2.84888297e-01 -5.95858872e-01 7.12448955e-01
2.77701378e-01 7.77495205e-01 1.41218051e-01 -9.47833002e-01
3.50295365e-01 4.47292298e-01 7.64164776e-02 -9.42205667e-01
3.40225637e-01 -6.19939089e-01 -9.93640006e-01 1.88890800e-01
1.91458672e-01 2.05045551e-01 -1.07542586e+00 1.86169291e+00
4.13575083e-01 1.36525467e-01 -7.69774243e-02 8.86073887e-01
6.11331105e-01 4.52059209e-01 8.02067891e-02 8.35260302e-02
1.29099095e+00 -1.07716727e+00 -6.61477447e-01 -3.08098733e-01
3.39276418e-02 -9.56371963e-01 1.30966985e+00 -1.25410020e-01
-1.33531940e+00 -8.06487978e-01 -8.77358854e-01 -2.54960656e-01
-1.58503830e-01 5.36508411e-02 3.51753980e-01 2.55986154e-01
-1.13943827e+00 2.00611636e-01 -6.57786191e-01 -4.46658283e-01
6.33828044e-01 5.56115359e-02 -4.14791852e-01 -3.88096869e-01
-1.05015647e+00 7.32778490e-01 3.66927326e-01 -3.45402002e-01
-1.27395856e+00 -8.55986774e-01 -8.94301593e-01 -7.54243284e-02
2.05754355e-01 -1.40118694e+00 9.27304268e-01 -1.32620573e+00
-1.80826712e+00 1.21518743e+00 -7.62342438e-02 -1.17909715e-01
7.23408580e-01 5.56710549e-02 2.94208303e-02 2.31024638e-01
4.35813785e-01 1.13284683e+00 1.26759422e+00 -1.67918694e+00
-4.84667361e-01 -1.66934416e-01 -1.09383091e-01 4.00387585e-01
-2.15415061e-01 -1.65197596e-01 -6.55026913e-01 -9.85514879e-01
-9.40198973e-02 -9.23807919e-01 5.12917386e-03 3.31175625e-01
-2.71945566e-01 3.08125705e-01 6.90076351e-01 -6.74090803e-01
6.02202535e-01 -1.96467757e+00 5.65320134e-01 -6.73688501e-02
1.99623257e-01 5.18405102e-02 -3.00982773e-01 5.31561732e-01
7.89536685e-02 -2.05759242e-01 -3.79222095e-01 -5.18811345e-01
1.37097597e-01 3.32303852e-01 -3.71694356e-01 2.95938492e-01
4.43506449e-01 1.49974751e+00 -1.01380444e+00 -5.60459375e-01
2.74936467e-01 7.12765276e-01 -8.06987941e-01 5.75086296e-01
-3.56621116e-01 1.25631988e+00 -5.17518163e-01 7.95910478e-01
6.84117258e-01 -4.04136777e-01 -4.03506272e-02 -4.49769080e-01
2.60897219e-01 5.11068776e-02 -6.53146207e-01 2.25583744e+00
-4.74506617e-01 4.92718399e-01 -1.36846704e-02 -9.94643390e-01
9.84595418e-01 5.10953255e-02 3.90368849e-01 -9.23605323e-01
1.04020305e-01 3.09682131e-01 -3.97599161e-01 -1.78229541e-01
2.72641629e-01 -3.50205600e-01 -2.33420298e-01 2.43436038e-01
3.30927819e-01 -5.49297035e-01 -1.09162405e-01 1.29323944e-01
5.92073917e-01 6.54297113e-01 3.43006641e-01 -1.49265051e-01
4.15293336e-01 -9.37041342e-02 4.21321213e-01 5.17352819e-01
5.43274134e-02 1.08346272e+00 1.67437986e-01 -8.89287442e-02
-1.35154104e+00 -1.34717655e+00 9.94434804e-02 1.08182967e+00
2.77651072e-01 2.71959492e-04 -9.62516844e-01 -5.56799293e-01
-1.28426760e-01 4.69516277e-01 -5.96571565e-01 -2.70686477e-01
-3.91722918e-01 -2.54844785e-01 3.00869793e-01 4.14798945e-01
8.90634596e-01 -9.68742549e-01 -1.94159761e-01 1.54864460e-01
-4.66074765e-01 -1.49676585e+00 -7.16877997e-01 -1.95167243e-01
-8.01883996e-01 -5.90805590e-01 -1.08135712e+00 -9.60192978e-01
8.15221488e-01 2.34754279e-01 1.49248230e+00 -1.66599870e-01
-1.14414409e-01 4.45613950e-01 -4.29992139e-01 -1.18009493e-01
-6.38353646e-01 -6.20223954e-02 -2.23658085e-01 2.25720361e-01
-3.80311720e-02 -8.89675021e-01 -8.39177191e-01 3.49356920e-01
-1.15890777e+00 5.54497659e-01 9.35910106e-01 1.14771163e+00
9.63691294e-01 -3.94816041e-01 5.00353932e-01 -8.56593013e-01
4.45189655e-01 -2.63523698e-01 -5.00766039e-01 3.87966752e-01
-5.23355424e-01 1.27124906e-01 5.09021282e-01 -4.66051966e-01
-1.10653591e+00 5.95466122e-02 6.10386692e-02 -6.34897292e-01
-1.10732630e-01 2.39598021e-01 -5.34180462e-01 -2.27202579e-01
5.82298934e-01 6.80757046e-01 -5.11502810e-02 -2.34434783e-01
9.40640390e-01 3.07124406e-01 9.23625410e-01 -9.36200023e-01
1.12411630e+00 6.22481465e-01 -1.28477752e-01 -2.65351981e-01
-8.51104975e-01 -3.54099646e-02 -9.10562158e-01 -2.95690671e-02
1.20156860e+00 -1.33102798e+00 -3.87145020e-02 3.90791953e-01
-1.08968878e+00 -6.03014469e-01 -4.98073727e-01 1.16772637e-01
-1.06070530e+00 1.04303233e-01 -4.32868779e-01 -3.94242592e-02
-3.34176868e-01 -1.19923282e+00 1.61869371e+00 1.58402845e-01
3.58105190e-02 -1.13761497e+00 1.08974338e-01 4.55138475e-01
4.85759169e-01 5.39368272e-01 6.26505673e-01 5.98450191e-02
-9.67681885e-01 2.95657963e-01 -4.81150091e-01 3.22624683e-01
3.16950619e-01 -1.53853253e-01 -6.72456741e-01 -4.78819132e-01
-1.57870740e-01 -4.22145188e-01 7.51060367e-01 2.52119869e-01
1.04629850e+00 -3.08867007e-01 -3.11499298e-01 1.01264405e+00
1.51333022e+00 -1.64559722e-01 9.11303997e-01 1.23624146e-01
1.03541517e+00 3.20515364e-01 5.61449349e-01 1.93098500e-01
5.54994464e-01 9.36452508e-01 2.50754088e-01 -5.70244253e-01
-7.34967113e-01 -7.55076826e-01 4.79088873e-01 7.32917011e-01
-5.11280112e-02 -8.12383816e-02 -4.98655140e-01 6.95195735e-01
-1.68301094e+00 -7.79663861e-01 4.64530110e-01 1.88060486e+00
1.11105371e+00 -1.34079710e-01 -4.86588441e-02 -4.40752715e-01
6.18373692e-01 1.52484059e-01 -5.41445673e-01 7.18346238e-03
-1.28841966e-01 3.13282996e-01 2.94024825e-01 3.75380576e-01
-8.15805376e-01 1.32975292e+00 6.05726242e+00 1.08464932e+00
-1.11300099e+00 4.01350290e-01 6.69761240e-01 2.22650692e-01
-7.36594915e-01 3.78589556e-02 -5.96905708e-01 4.85525131e-01
4.11130190e-01 -5.59424795e-03 4.82777417e-01 6.21993303e-01
-6.67976541e-03 2.49263242e-01 -1.18435645e+00 1.11679053e+00
2.25830704e-01 -1.59528041e+00 5.05292356e-01 1.45204037e-01
1.51241934e+00 -1.89209342e-01 4.72160578e-01 2.48802695e-02
5.65843940e-01 -9.58020091e-01 1.00592256e+00 5.66155493e-01
1.38949823e+00 -5.68980157e-01 2.84539074e-01 2.28470359e-02
-1.16212940e+00 4.59288120e-01 -3.82195830e-01 3.92366290e-01
2.72816718e-01 5.16538322e-01 -5.13592005e-01 8.83659542e-01
5.98800600e-01 1.08771026e+00 -3.41040105e-01 4.65596437e-01
-5.15636086e-01 2.69010305e-01 6.57453686e-02 5.15099347e-01
2.34830543e-01 -3.99509698e-01 3.29041064e-01 1.07228637e+00
5.28007329e-01 -8.73192549e-02 3.16224426e-01 1.49249244e+00
-2.24281311e-01 -1.50102854e-01 -7.88976133e-01 -4.03323770e-02
3.77133787e-01 1.20853353e+00 -6.23485863e-01 -4.38177228e-01
-3.85437548e-01 1.51378989e+00 3.35973918e-01 6.14351749e-01
-9.86092448e-01 1.22524746e-01 7.12124705e-01 3.00929636e-01
6.35522842e-01 -2.62806267e-01 -2.69367069e-01 -1.44893885e+00
-7.76111782e-02 -1.17038167e+00 8.54496732e-02 -1.03032827e+00
-1.52695262e+00 6.19847596e-01 -2.22509238e-03 -1.54486322e+00
-4.27092224e-01 -2.70082325e-01 -3.82536411e-01 9.69645739e-01
-1.65921438e+00 -2.02716327e+00 -5.85069776e-01 9.84397650e-01
5.55363119e-01 -2.00895250e-01 7.07694530e-01 2.18425393e-01
-6.07620832e-03 8.35481763e-01 2.09129766e-01 7.73161054e-02
1.02347779e+00 -9.16679502e-01 2.95319378e-01 9.66461658e-01
3.22950631e-01 4.04084474e-01 5.12901604e-01 -6.30725384e-01
-1.31891811e+00 -1.38146102e+00 4.10501689e-01 -5.83035946e-01
3.83688420e-01 -4.61826384e-01 -6.42824471e-01 7.41135597e-01
5.07273316e-01 2.12407544e-01 3.82390410e-01 -4.93693322e-01
-6.04338944e-01 -1.77886128e-01 -1.16981304e+00 6.53048575e-01
1.22477388e+00 -7.59738386e-01 -3.48184645e-01 1.07383125e-01
8.58813226e-01 -5.22345483e-01 -9.07335997e-01 2.63548225e-01
5.02680361e-01 -1.02821255e+00 1.24494612e+00 -2.44179472e-01
8.65256011e-01 -4.33647633e-01 -3.39724451e-01 -1.48828042e+00
-4.45069283e-01 -7.10986197e-01 1.02815107e-01 1.25282848e+00
2.56395154e-02 -5.64276636e-01 5.72055638e-01 1.36830434e-01
-9.75764245e-02 -4.87397730e-01 -6.62785590e-01 -6.41576469e-01
1.79767698e-01 6.01011813e-02 7.14672804e-01 9.96496558e-01
-4.60331947e-01 4.25658971e-01 -7.18700290e-01 -7.61125088e-02
8.80007446e-01 4.92376417e-01 1.13824177e+00 -8.72067869e-01
-5.12875199e-01 -3.77909333e-01 -5.42977154e-01 -1.32953024e+00
1.62987903e-01 -9.22912300e-01 3.98004688e-02 -1.60873067e+00
4.78905767e-01 -2.41172135e-01 -1.19673751e-01 4.14514542e-01
-2.19961762e-01 7.28683054e-01 2.14953676e-01 4.12386477e-01
-6.29086196e-01 9.83577311e-01 2.03259301e+00 -2.52518356e-01
1.13398626e-01 -5.57354808e-01 -8.65824044e-01 3.37598711e-01
4.12989974e-01 -5.33676207e-01 -5.70629537e-01 -5.97235620e-01
-3.60113978e-02 -3.63199823e-02 5.35935163e-01 -8.51120651e-01
-6.15156330e-02 -3.93531978e-01 6.01738214e-01 -3.83673698e-01
2.95211494e-01 -7.42135048e-01 5.07353663e-01 2.48675391e-01
-3.52239043e-01 -2.18490139e-01 -3.87182087e-02 7.18858838e-01
-5.18008947e-01 4.21139002e-01 1.05196381e+00 -1.51116863e-01
-6.13814831e-01 6.20238245e-01 2.81240284e-01 2.36787066e-01
9.87820506e-01 -2.51840293e-01 -1.31785825e-01 -6.38005495e-01
-5.91166437e-01 -1.49507955e-01 1.02532816e+00 4.66793507e-01
6.29099607e-01 -1.79621422e+00 -9.14073944e-01 4.37773585e-01
5.10268688e-01 1.02080494e-01 1.58790886e-01 8.22187424e-01
-3.70710164e-01 2.39989594e-01 -7.13457406e-01 -9.89998698e-01
-9.64292288e-01 4.60424006e-01 1.69306323e-01 -3.76552373e-01
-6.68169498e-01 7.76680231e-01 7.78054953e-01 -3.58801901e-01
-2.05359057e-01 -2.53613114e-01 3.94262046e-01 -2.99999952e-01
2.71984696e-01 -1.98476911e-01 -1.82395965e-01 -8.67593706e-01
-1.29993930e-01 1.00280070e+00 5.09407520e-02 -3.12424630e-01
1.28238571e+00 -4.32688445e-01 -9.00237784e-02 1.15886241e-01
1.13634777e+00 -1.63010150e-01 -1.84680748e+00 -5.65252483e-01
-5.77309966e-01 -8.45494390e-01 -1.23803690e-01 -7.70392597e-01
-1.16938126e+00 9.37234461e-01 5.99584401e-01 -3.58700722e-01
1.35640395e+00 1.86174706e-01 9.00411665e-01 -1.46122307e-01
6.38186932e-01 -7.74547279e-01 5.90467870e-01 2.10140243e-01
1.05906630e+00 -1.27414906e+00 -7.31589124e-02 -4.70291436e-01
-8.22463214e-01 7.68647790e-01 5.08599043e-01 -2.69769937e-01
2.91858017e-01 1.23903574e-02 8.74702446e-03 -5.56589663e-02
-5.06488323e-01 -3.37134093e-01 6.45545065e-01 9.98947740e-01
2.82776117e-01 8.29676241e-02 -2.92210560e-02 2.53979146e-01
-2.39621960e-02 1.70957983e-01 1.88185558e-01 6.06705368e-01
-1.05154119e-01 -1.33425426e+00 -2.94587553e-01 -7.30184540e-02
-2.00861320e-01 -1.71863213e-01 -3.81988317e-01 6.93451703e-01
2.16348797e-01 6.64197922e-01 -2.98535656e-02 -2.90459245e-01
2.17996001e-01 -2.60615796e-01 9.01165605e-01 -6.49148583e-01
-2.34123766e-01 3.94614995e-01 -3.52588147e-01 -7.73442805e-01
-7.37989485e-01 -5.27308345e-01 -8.32662523e-01 -1.36443377e-01
1.56804651e-01 -3.32361281e-01 4.07485723e-01 8.36861789e-01
6.41790450e-01 5.68571866e-01 4.88448948e-01 -1.19245899e+00
-2.78357983e-01 -7.39858091e-01 -4.06431109e-01 8.63218904e-01
1.84491545e-01 -8.89829397e-01 -1.03240356e-01 6.44841313e-01] | [11.625657081604004, -0.4575669765472412] |
800e9b20-7bd3-4cb4-b5c4-62ce0eed74dd | is-mapping-necessary-for-realistic-pointgoal | 2206.00997 | null | https://arxiv.org/abs/2206.00997v2 | https://arxiv.org/pdf/2206.00997v2.pdf | Is Mapping Necessary for Realistic PointGoal Navigation? | Can an autonomous agent navigate in a new environment without building an explicit map? For the task of PointGoal navigation ('Go to $\Delta x$, $\Delta y$') under idealized settings (no RGB-D and actuation noise, perfect GPS+Compass), the answer is a clear 'yes' - map-less neural models composed of task-agnostic components (CNNs and RNNs) trained with large-scale reinforcement learning achieve 100% Success on a standard dataset (Gibson). However, for PointNav in a realistic setting (RGB-D and actuation noise, no GPS+Compass), this is an open question; one we tackle in this paper. The strongest published result for this task is 71.7% Success. First, we identify the main (perhaps, only) cause of the drop in performance: the absence of GPS+Compass. An agent with perfect GPS+Compass faced with RGB-D sensing and actuation noise achieves 99.8% Success (Gibson-v2 val). This suggests that (to paraphrase a meme) robust visual odometry is all we need for realistic PointNav; if we can achieve that, we can ignore the sensing and actuation noise. With that as our operating hypothesis, we scale the dataset and model size, and develop human-annotation-free data-augmentation techniques to train models for visual odometry. We advance the state of art on the Habitat Realistic PointNav Challenge from 71% to 94% Success (+23, 31% relative) and 53% to 74% SPL (+21, 40% relative). While our approach does not saturate or 'solve' this dataset, this strong improvement combined with promising zero-shot sim2real transfer (to a LoCoBot) provides evidence consistent with the hypothesis that explicit mapping may not be necessary for navigation, even in a realistic setting. | ['Oleksandr Maksymets', 'Dhruv Batra', 'Oles Dobosevych', 'Naoki Yokoyama', 'Erik Wijmans', 'Ruslan Partsey'] | 2022-06-02 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Partsey_Is_Mapping_Necessary_for_Realistic_PointGoal_Navigation_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Partsey_Is_Mapping_Necessary_for_Realistic_PointGoal_Navigation_CVPR_2022_paper.pdf | cvpr-2022-1 | ['pointgoal-navigation'] | ['robots'] | [-3.08288485e-02 2.12874532e-01 1.52716100e-01 -2.07670286e-01
-6.65552855e-01 -7.56696343e-01 6.07813597e-01 -7.11655170e-02
-1.00553715e+00 9.09111440e-01 -1.23217948e-01 -6.33804083e-01
-6.60650432e-02 -7.95401216e-01 -1.18763983e+00 -5.00174165e-01
-4.53491569e-01 6.21332705e-01 3.96608472e-01 -1.05809236e+00
1.41812235e-01 3.52764904e-01 -1.66342330e+00 -4.04720783e-01
5.99300504e-01 8.10301542e-01 3.07661206e-01 1.02290535e+00
3.87490541e-01 8.29434216e-01 -6.79184020e-01 2.93832570e-01
7.26246059e-01 -1.25700161e-01 -7.82369733e-01 -5.05427122e-01
5.03850698e-01 -5.07749021e-01 -3.81190062e-01 7.53001571e-01
8.01521838e-01 3.02834094e-01 3.69773507e-01 -1.35645723e+00
-4.36229646e-01 1.14426889e-01 -3.00311744e-01 6.04896545e-02
4.94579732e-01 9.96114075e-01 6.68929994e-01 -3.82400274e-01
7.32599795e-01 1.03481710e+00 1.13199413e+00 4.81141657e-01
-1.31296170e+00 -4.29830998e-01 8.49215686e-02 -1.78282320e-01
-1.44276845e+00 -4.08814251e-01 1.25916693e-02 -3.17232847e-01
1.63389206e+00 8.69739801e-02 9.17655647e-01 1.33292043e+00
2.40473270e-01 3.63849282e-01 1.07138014e+00 -6.33592978e-02
5.34576833e-01 -2.16107547e-01 -4.24609035e-01 6.64033890e-01
2.58838266e-01 7.63106585e-01 -2.80363828e-01 9.13191587e-02
8.55688214e-01 -3.03628802e-01 -3.24327588e-01 -7.70039558e-01
-1.28186929e+00 6.44568861e-01 1.06917369e+00 -1.15902290e-01
-3.09038341e-01 6.97915018e-01 1.87552303e-01 6.14126086e-01
-2.25321084e-01 9.02767301e-01 -5.23825705e-01 -6.11937761e-01
-5.08442700e-01 8.48125458e-01 9.17386949e-01 1.34112537e+00
8.63610387e-01 4.21626240e-01 6.27975762e-01 3.97182405e-01
2.26227164e-01 8.36368978e-01 4.55375761e-01 -1.40379500e+00
4.83474940e-01 2.38044187e-01 6.45688295e-01 -7.44716704e-01
-1.07169330e+00 -4.30631340e-01 -3.80539030e-01 1.04142702e+00
7.07501054e-01 -4.91488725e-01 -1.17459118e+00 1.93394685e+00
9.17905942e-02 -2.14669377e-01 3.55786920e-01 1.19837093e+00
5.75112760e-01 5.15653312e-01 4.36228178e-02 4.83964980e-01
1.05937052e+00 -6.80556178e-01 -1.61892056e-01 -1.06338394e+00
9.06802118e-01 -2.85911798e-01 1.41118181e+00 2.49815881e-01
-6.06273234e-01 -5.13738275e-01 -1.53245926e+00 -1.66805878e-01
-5.22080898e-01 -3.18931609e-01 7.61706173e-01 3.89170885e-01
-1.46705735e+00 5.48920333e-01 -1.05483496e+00 -8.84965539e-01
-3.40491645e-02 5.36268055e-01 -5.71726024e-01 -1.08124159e-01
-1.12423551e+00 1.51560915e+00 2.63111532e-01 -5.22571355e-02
-1.31397069e+00 -2.97655970e-01 -1.03230858e+00 -3.59751463e-01
3.18693340e-01 -9.61481154e-01 1.48078454e+00 -5.64033449e-01
-1.49040496e+00 6.09037340e-01 2.43760988e-01 -8.05252790e-01
7.82349467e-01 -3.18396121e-01 -1.33194346e-02 -3.19701254e-01
3.98338914e-01 1.16686058e+00 3.52821559e-01 -1.55686092e+00
-8.08475375e-01 -4.26393241e-01 4.12205279e-01 5.67133188e-01
6.21201932e-01 -7.38261163e-01 -2.92767197e-01 -2.28019636e-02
3.51638705e-01 -1.29752278e+00 -5.14629245e-01 1.53105378e-01
-3.82838547e-02 2.34477326e-01 3.39469820e-01 -4.01425689e-01
2.59828001e-01 -2.09202838e+00 -3.70778851e-02 3.00277080e-02
1.64128821e-02 -7.97942504e-02 -9.32151303e-02 4.38189536e-01
1.94274232e-01 -6.03932189e-03 -3.58092934e-01 -4.60922658e-01
2.60783255e-01 5.94150424e-01 -2.69134790e-01 5.24792194e-01
3.30965817e-02 9.81952548e-01 -9.91154909e-01 2.22248398e-02
4.22375083e-01 3.37315172e-01 -5.98209918e-01 -1.40814736e-01
-2.42693275e-01 4.73686725e-01 3.67117533e-03 5.21948695e-01
4.38000053e-01 1.18179925e-01 1.52008161e-01 3.05147141e-01
-5.15743792e-01 3.79717827e-01 -1.20317721e+00 2.28024769e+00
-4.74934280e-01 7.07134724e-01 3.25220317e-01 -6.43566489e-01
8.97719920e-01 -2.54260093e-01 2.33689055e-01 -1.30555642e+00
1.24886632e-01 4.70159709e-01 2.39988551e-01 -4.93019223e-01
8.22087526e-01 -2.59322703e-01 -3.65947366e-01 -1.79308549e-01
2.62237823e-04 -8.55020225e-01 -1.81779996e-01 -1.71710011e-02
1.53474283e+00 6.05047464e-01 8.05020481e-02 -3.63893330e-01
-3.19058031e-01 6.98769212e-01 2.87131399e-01 1.11964774e+00
-5.16989946e-01 6.34908438e-01 2.59193748e-01 -5.03120303e-01
-1.35800278e+00 -1.08642054e+00 1.02135822e-01 9.89383936e-01
6.21890008e-01 -2.04861745e-01 -3.55942816e-01 -1.75546810e-01
2.64948189e-01 6.50865197e-01 -6.87617898e-01 -2.61591792e-01
-6.56217515e-01 -6.82567120e-01 8.77556264e-01 6.37238026e-01
5.67767143e-01 -7.27618098e-01 -1.27672732e+00 4.19995040e-01
5.17851226e-02 -9.73925769e-01 3.40449005e-01 7.74478555e-01
-6.97271645e-01 -8.73302042e-01 -5.14133394e-01 -6.89910114e-01
1.80498987e-01 3.30299288e-01 9.95787323e-01 -9.26574320e-02
1.40238971e-01 3.25073540e-01 -3.52080941e-01 -3.12281042e-01
-4.07220088e-02 7.38391578e-02 2.13459700e-01 -1.11737776e+00
-3.00131571e-02 -6.51509702e-01 -6.97544813e-01 2.96784967e-01
-4.66143399e-01 -3.22794057e-02 6.86025441e-01 7.52729893e-01
3.61849308e-01 -4.69920695e-01 2.70901680e-01 -6.32540882e-02
4.98266220e-01 -4.46790606e-01 -7.96852171e-01 -5.98564029e-01
-5.75693309e-01 1.21435739e-01 5.76603830e-01 -2.00638548e-01
-3.72918695e-01 1.54656753e-01 -5.42622566e-01 -7.19158649e-02
-2.66933888e-01 4.42831099e-01 1.91844374e-01 -4.82566923e-01
1.42308605e+00 2.05837548e-01 1.88648611e-01 -2.47771353e-01
4.27548915e-01 2.63118476e-01 7.41207421e-01 -5.19531488e-01
7.13926494e-01 7.91182280e-01 4.35224399e-02 -6.73014641e-01
-5.76817058e-02 -8.58445689e-02 -1.47101626e-01 1.68236837e-01
8.30763757e-01 -1.24660993e+00 -1.28633928e+00 1.85105786e-01
-7.53775418e-01 -1.24261737e+00 -5.23855388e-01 5.20553827e-01
-1.00788796e+00 3.13419737e-02 -2.19571739e-01 -8.28764200e-01
1.25863358e-01 -1.26870394e+00 1.05505526e+00 2.57223248e-01
-3.31353366e-01 -4.70179915e-01 3.93608421e-01 -1.14737280e-01
6.87873483e-01 4.66602296e-01 3.87388289e-01 -2.64131695e-01
-6.48003817e-01 -1.21457435e-01 -2.95555532e-01 -1.59286112e-01
-1.97407290e-01 -5.12632370e-01 -8.54099274e-01 -4.82825637e-01
-3.35735619e-01 -4.72203344e-01 7.19386876e-01 1.15480505e-01
2.30902210e-01 -2.85062995e-02 -3.40502381e-01 7.21855104e-01
1.58535254e+00 3.89847338e-01 5.86981297e-01 1.05102146e+00
6.37369454e-01 2.27574468e-01 6.22855484e-01 2.93667257e-01
1.09149837e+00 8.39372754e-01 1.19225287e+00 -2.76366137e-02
-4.78860028e-02 -3.83862525e-01 4.70943719e-01 -3.90338302e-02
-2.28637636e-01 -2.28349581e-01 -1.30746770e+00 7.37605035e-01
-1.84175217e+00 -6.84472501e-01 -7.73778707e-02 2.21060848e+00
3.40346962e-01 5.48213303e-01 1.76441714e-01 -8.51734355e-02
-1.82311069e-02 1.44061506e-01 -7.48951554e-01 -3.71233553e-01
-1.08130902e-01 8.29876065e-02 1.10379744e+00 8.25277448e-01
-9.36052978e-01 1.08728778e+00 6.27523041e+00 1.82039753e-01
-1.26789522e+00 -8.99449065e-02 -6.82178140e-02 -2.71728724e-01
7.04051694e-03 2.23009005e-01 -5.17844617e-01 3.34708005e-01
1.05123401e+00 4.99112338e-01 7.35909820e-01 1.20430160e+00
8.18552747e-02 -6.14759266e-01 -9.95039582e-01 1.03781521e+00
-1.34224862e-01 -1.01914859e+00 -6.77194774e-01 1.20513976e-01
4.07278389e-01 9.29848611e-01 9.71872825e-03 9.04511750e-01
9.88674402e-01 -1.20562625e+00 1.10182118e+00 2.84256816e-01
7.80351400e-01 -4.85034853e-01 7.76271224e-01 6.95703983e-01
-9.88804817e-01 -1.23797484e-01 -3.83031875e-01 -7.47656405e-01
2.26712152e-01 -3.60710055e-01 -1.12974489e+00 4.42030370e-01
1.04628074e+00 4.05258328e-01 -5.22684276e-01 1.00378823e+00
-3.83106023e-01 -1.84211712e-02 -9.17653501e-01 -2.28304356e-01
6.31034315e-01 1.74348533e-01 4.91713315e-01 8.53004813e-01
3.82624298e-01 2.54151933e-02 1.52598441e-01 6.03842735e-01
3.59005034e-01 -4.79019314e-01 -9.75168407e-01 4.89782453e-01
4.11320776e-01 6.79250300e-01 -3.68975371e-01 -1.78434089e-01
-2.27808971e-02 8.48037481e-01 4.70610619e-01 4.41483021e-01
-7.94926405e-01 -3.74171257e-01 9.39530730e-01 5.94361089e-02
2.07173392e-01 -7.95793355e-01 -4.62379247e-01 -8.03680241e-01
2.52688434e-02 -8.00887227e-01 -2.04530191e-02 -1.16611302e+00
-6.61919594e-01 4.54526931e-01 -2.52149403e-01 -1.26475918e+00
-4.91587520e-01 -6.76466644e-01 -1.31184727e-01 8.86546910e-01
-1.47409952e+00 -1.08653903e+00 -6.90993428e-01 4.18858677e-01
1.98947519e-01 2.96766579e-01 9.71383214e-01 1.59910679e-01
1.86308682e-01 2.36479208e-01 4.64668572e-02 -9.80009437e-02
6.02506816e-01 -1.49643397e+00 9.68291461e-01 7.25388408e-01
-2.87899166e-01 5.98968923e-01 1.19788730e+00 -6.59165740e-01
-1.75780392e+00 -7.85003185e-01 3.68603587e-01 -9.24381495e-01
4.68621343e-01 -4.54687297e-01 -6.48743808e-01 1.01787066e+00
-5.59567586e-02 -6.19831271e-02 -3.05277426e-02 -5.36743086e-03
-1.60211578e-01 7.66181052e-02 -1.25961065e+00 9.33811128e-01
1.42605186e+00 -2.56354004e-01 -5.61845183e-01 1.14829458e-01
9.05416965e-01 -9.10784543e-01 -7.17377782e-01 5.46329856e-01
7.44729042e-01 -1.06935632e+00 1.09545743e+00 -3.86182606e-01
8.56546219e-03 -7.29455709e-01 -5.84259272e-01 -1.57160521e+00
-2.10294709e-01 -4.19413120e-01 3.10195744e-01 3.95683080e-01
3.48870218e-01 -7.89913237e-01 8.75589013e-01 5.82423925e-01
-6.21911347e-01 -2.65453935e-01 -1.20293581e+00 -9.01671410e-01
1.80234492e-01 -8.47767651e-01 4.99683022e-01 7.16588259e-01
-9.68310796e-03 1.76232323e-01 -3.40634227e-01 3.98272753e-01
2.52384692e-01 -3.07114214e-01 1.42922688e+00 -8.80793869e-01
-2.85876006e-01 -3.90317798e-01 -6.68881774e-01 -1.36903763e+00
-3.13089371e-01 -5.04044831e-01 6.09849989e-01 -1.91490293e+00
-7.27508962e-01 -8.36149096e-01 2.10321397e-01 6.85345292e-01
3.27808321e-01 2.31621608e-01 2.82155395e-01 2.17178091e-01
-5.87442100e-01 5.23362219e-01 9.84780908e-01 -4.10635732e-02
-2.84666479e-01 -3.24595749e-01 -7.13079453e-01 5.28346419e-01
7.42672443e-01 -2.54213214e-01 -2.95347512e-01 -8.28775167e-01
5.22753775e-01 -1.55587886e-02 8.29042196e-01 -1.53092921e+00
3.31298113e-01 -3.31360735e-02 4.37974036e-01 -1.04162425e-01
8.53041887e-01 -8.61123979e-01 2.29014724e-01 8.43692541e-01
1.99526906e-01 3.78730267e-01 6.75274134e-01 4.89632398e-01
2.36886948e-01 1.88647330e-01 5.92861056e-01 -4.20828938e-01
-1.28972518e+00 -5.96526265e-02 -5.93016505e-01 -4.57155402e-04
7.37805903e-01 -4.88341331e-01 -7.54211605e-01 -6.28768981e-01
-7.48304605e-01 4.16235209e-01 9.67791855e-01 4.51271892e-01
4.07081127e-01 -1.02709401e+00 -3.58120650e-01 2.47388974e-01
3.34087104e-01 4.77220505e-01 1.21657014e-01 7.03599989e-01
-9.71674860e-01 2.76240766e-01 -4.77288544e-01 -7.09529638e-01
-3.17411959e-01 3.64714950e-01 6.22249663e-01 2.82691538e-01
-4.87115175e-01 9.49848354e-01 -2.22173601e-01 -9.13796067e-01
9.52016786e-02 -6.91933990e-01 2.80350626e-01 -3.52322787e-01
1.17569715e-01 1.06496476e-01 1.04970746e-01 -5.56691349e-01
-5.44944346e-01 5.33999383e-01 4.52683508e-01 -5.87325513e-01
1.21352303e+00 -2.14601845e-01 5.74825764e-01 4.16166097e-01
8.57096732e-01 -2.67895192e-01 -1.75175107e+00 2.91163623e-01
-2.58446217e-01 -2.68949270e-01 -2.95650065e-01 -1.03657854e+00
-3.21207523e-01 8.81148219e-01 9.88547742e-01 8.28713477e-02
4.66135651e-01 -2.00471103e-01 3.55966121e-01 8.31062496e-01
8.95719707e-01 -9.55224454e-01 -2.26610854e-01 1.15420711e+00
9.16273475e-01 -1.62891316e+00 -1.75989926e-01 4.29684132e-01
-6.17693484e-01 5.37774980e-01 9.41895187e-01 -4.76070464e-01
3.51024866e-02 3.33297074e-01 3.21501166e-01 -1.31268486e-01
-5.34458578e-01 -5.71365476e-01 -4.56640065e-01 1.15640235e+00
-1.04066484e-01 -6.59539644e-03 2.82989711e-01 2.10410282e-01
-8.87559414e-01 -2.66033709e-01 3.96082729e-01 1.38971710e+00
-8.23329031e-01 -4.71353382e-01 -2.97542602e-01 6.28782734e-02
3.07128578e-01 -4.66003716e-02 -3.40703160e-01 1.41531289e+00
1.63500607e-01 8.54298532e-01 1.02484874e-01 -6.83586419e-01
7.77212918e-01 -2.41465308e-02 4.25212443e-01 -4.37406927e-01
-5.90330482e-01 -1.58981919e-01 2.83428043e-01 -8.38655472e-01
1.67246386e-02 -3.66084576e-01 -1.67904615e+00 -5.93133807e-01
5.46882637e-02 -2.35500172e-01 1.06317747e+00 7.81700075e-01
5.34681916e-01 6.03610456e-01 -2.59403646e-01 -1.35352945e+00
-5.12353778e-01 -9.21551824e-01 -1.46111906e-01 1.86833367e-01
7.13547289e-01 -8.14415872e-01 -3.73108119e-01 -4.29166675e-01] | [4.7040629386901855, 0.6989579200744629] |
31f452ad-3a63-4396-87c8-aedc40ae020a | compm-context-modeling-with-speaker-s-pre | 2108.11626 | null | https://arxiv.org/abs/2108.11626v3 | https://arxiv.org/pdf/2108.11626v3.pdf | CoMPM: Context Modeling with Speaker's Pre-trained Memory Tracking for Emotion Recognition in Conversation | As the use of interactive machines grow, the task of Emotion Recognition in Conversation (ERC) became more important. If the machine-generated sentences reflect emotion, more human-like sympathetic conversations are possible. Since emotion recognition in conversation is inaccurate if the previous utterances are not taken into account, many studies reflect the dialogue context to improve the performances. Many recent approaches show performance improvement by combining knowledge into modules learned from external structured data. However, structured data is difficult to access in non-English languages, making it difficult to extend to other languages. Therefore, we extract the pre-trained memory using the pre-trained language model as an extractor of external knowledge. We introduce CoMPM, which combines the speaker's pre-trained memory with the context model, and find that the pre-trained memory significantly improves the performance of the context model. CoMPM achieves the first or second performance on all data and is state-of-the-art among systems that do not leverage structured data. In addition, our method shows that it can be extended to other languages because structured knowledge is not required, unlike previous methods. Our code is available on github (https://github.com/rungjoo/CoMPM). | ['Wooin Lee', 'Joosung Lee'] | 2021-08-26 | null | null | null | null | ['emotion-recognition-in-conversation'] | ['natural-language-processing'] | [-1.86536491e-01 2.66001821e-01 -2.21907511e-01 -6.24940097e-01
-6.68114603e-01 -5.86854160e-01 5.50218403e-01 -1.42459497e-01
-4.43313122e-01 6.56544030e-01 6.86909318e-01 -1.12011224e-01
6.42112017e-01 -5.40236294e-01 -3.67634684e-01 -2.77319759e-01
9.92993191e-02 3.33669811e-01 -6.87642395e-02 -5.37360251e-01
4.40622941e-02 -5.88442609e-02 -1.22020853e+00 7.95669854e-01
7.44069338e-01 7.52345622e-01 -1.61667960e-03 7.62811005e-01
-5.10842621e-01 1.13161242e+00 -6.94148600e-01 -4.36451793e-01
-1.29052952e-01 -7.11731672e-01 -1.28729796e+00 -8.43402520e-02
-1.74149707e-01 -1.85589150e-01 -8.91157761e-02 7.07286060e-01
5.24658203e-01 2.16276601e-01 3.56477618e-01 -1.05548680e+00
-3.11340094e-01 8.16144705e-01 -1.47805050e-01 -2.01340839e-01
6.27456665e-01 -5.39085791e-02 8.49318743e-01 -9.35201108e-01
7.54340470e-01 1.28350437e+00 4.38849956e-01 1.00245106e+00
-1.01728320e+00 -6.06888711e-01 3.57933342e-01 1.14617944e-01
-1.10061204e+00 -6.87772155e-01 8.60497415e-01 -1.86817661e-01
1.37833476e+00 4.35067058e-01 6.63755417e-01 1.56026983e+00
-1.35281771e-01 1.04162288e+00 1.35134089e+00 -5.71805060e-01
3.39884013e-02 6.65206611e-01 1.82272449e-01 4.63091642e-01
-5.54827631e-01 -2.89726585e-01 -9.22258437e-01 -2.79765069e-01
1.34359688e-01 -2.83484280e-01 -4.36742872e-01 2.15725780e-01
-9.25858438e-01 9.17915523e-01 2.47549787e-02 5.15313625e-01
-2.45543748e-01 -2.68375039e-01 6.21291935e-01 5.78981221e-01
7.10607111e-01 7.03397274e-01 -6.60444438e-01 -8.50393176e-01
-8.86838317e-01 -1.68870240e-01 1.63797832e+00 7.34590411e-01
5.87308109e-01 -1.75879285e-01 1.64639294e-01 1.23100519e+00
1.43176079e-01 3.47777367e-01 6.37236118e-01 -1.13958108e+00
4.08767462e-01 7.21886396e-01 -1.50517046e-01 -8.99690092e-01
-5.31073630e-01 -2.99417794e-01 -6.98484063e-01 -2.78811425e-01
3.98170620e-01 -5.96920431e-01 -4.01791662e-01 1.90732121e+00
2.00360611e-01 -6.82955980e-02 5.43997049e-01 7.38701165e-01
1.14876640e+00 9.48113322e-01 2.38273595e-03 -4.73034114e-01
1.20592439e+00 -1.20260799e+00 -9.20498073e-01 -6.45853400e-01
8.41311872e-01 -9.01014209e-01 1.02530456e+00 4.68204200e-01
-8.69434476e-01 -1.72340602e-01 -7.56162286e-01 -1.50271086e-02
-5.12317181e-01 -1.59141451e-01 7.55676270e-01 4.32296097e-01
-1.05107629e+00 2.18484089e-01 -5.43909371e-01 -6.03825510e-01
1.02865547e-02 6.22011088e-02 -3.64926606e-01 2.36994982e-01
-1.53097034e+00 1.12218702e+00 1.73507318e-01 1.48695782e-01
-4.60204780e-01 -9.48762819e-02 -8.42445314e-01 -2.52865523e-01
5.07127047e-01 -3.10971648e-01 1.50372875e+00 -1.54768682e+00
-2.09561753e+00 9.45103467e-01 -6.89381003e-01 -2.15881437e-01
3.41899008e-01 -3.63666385e-01 -5.81877589e-01 8.28870237e-02
-3.56247127e-01 5.99712372e-01 4.73581374e-01 -1.18208528e+00
-3.50450337e-01 2.69071292e-03 2.22906172e-01 2.75224388e-01
-4.60497051e-01 1.72084019e-01 -7.72246242e-01 -2.02133298e-01
-1.99999541e-01 -1.06204677e+00 -9.39039513e-02 -6.67720854e-01
-4.77540016e-01 -2.50140458e-01 6.06544077e-01 -7.99734712e-01
1.59159207e+00 -2.19546247e+00 1.43923923e-01 2.46093601e-01
2.17731530e-03 1.29848361e-01 -2.50948697e-01 7.55321562e-01
1.10091604e-01 3.35530698e-01 -1.04194932e-01 -4.55833524e-01
4.60295714e-02 2.10141286e-01 -2.26308763e-01 -1.36710122e-01
9.85869467e-02 8.44246030e-01 -8.11416090e-01 -4.92082506e-01
-9.40478966e-02 3.77619058e-01 -6.06190503e-01 5.10801136e-01
-2.26690784e-01 5.22395551e-01 -3.51707429e-01 3.38396490e-01
2.89901674e-01 -3.23259205e-01 5.22623241e-01 1.17144540e-01
-5.25718974e-03 7.91496038e-01 -9.33724880e-01 1.50318456e+00
-8.70154202e-01 8.03507686e-01 2.53203005e-01 -6.96532369e-01
9.40169036e-01 6.93760395e-01 2.14025348e-01 -4.89365935e-01
5.21316528e-02 1.94096953e-01 1.06467836e-01 -7.06564009e-01
5.99694550e-01 -2.10290059e-01 -4.30996627e-01 6.44080818e-01
1.00117870e-01 -1.85744181e-01 6.78159744e-02 5.32445729e-01
9.63462114e-01 -6.44849390e-02 4.59122300e-01 9.87746269e-02
4.57399219e-01 -4.12667505e-02 7.27170467e-01 6.74387395e-01
-1.79175764e-01 2.60755688e-01 6.40740871e-01 -1.22274391e-01
-4.75072563e-01 -5.01573324e-01 -5.49266627e-03 1.42610574e+00
-1.96433634e-01 -8.75445783e-01 -8.65970671e-01 -6.38124824e-01
-4.02061969e-01 8.37027550e-01 -5.63087702e-01 -1.64644003e-01
-3.42403740e-01 -5.33444345e-01 6.99376523e-01 3.94736081e-01
5.38573861e-01 -1.37577510e+00 -3.52314055e-01 2.92672604e-01
-8.46108437e-01 -1.02802873e+00 -3.64335716e-01 1.85416266e-01
-6.33204043e-01 -6.83891177e-01 -2.24961236e-01 -5.17175138e-01
2.94787556e-01 -1.69473737e-01 1.33857822e+00 2.54209548e-01
1.33462951e-01 5.64806700e-01 -5.07262945e-01 -5.10094941e-01
-6.23247981e-01 4.34061110e-01 -7.73459747e-02 -1.83992088e-02
8.34345698e-01 -4.77344692e-01 -2.27901503e-01 2.53567845e-01
-4.45393533e-01 3.78231078e-01 3.13853383e-01 9.04355168e-01
1.51619278e-02 -2.63419628e-01 8.44557226e-01 -1.15795374e+00
7.72502124e-01 -5.12859225e-01 8.69262367e-02 5.01088202e-02
-2.65513539e-01 5.64865442e-03 4.05089110e-01 -4.49597448e-01
-1.38394499e+00 -1.43367186e-01 -4.24770653e-01 1.38221219e-01
-2.93415308e-01 7.46647954e-01 -2.48286709e-01 5.40384054e-01
4.13226157e-01 3.18607353e-02 -5.68046868e-02 -4.09576386e-01
3.44220668e-01 1.21877897e+00 2.08463773e-01 -5.52240551e-01
1.85427472e-01 1.38266340e-01 -8.09530258e-01 -9.47423577e-01
-1.00769544e+00 -4.87397403e-01 -4.13839519e-01 -5.14078557e-01
7.30672240e-01 -9.37148809e-01 -7.77156293e-01 3.57686341e-01
-1.26744056e+00 -7.20918775e-01 7.05068856e-02 5.65330744e-01
-3.15081626e-01 1.25342667e-01 -1.01531959e+00 -1.12795532e+00
-4.62346673e-01 -9.13323224e-01 6.32327080e-01 2.82909155e-01
-1.01975727e+00 -1.23085272e+00 1.25492826e-01 5.11974394e-01
5.63851237e-01 -1.40270725e-01 7.22099364e-01 -9.85325456e-01
-4.19119224e-02 -1.28127426e-01 2.92626590e-01 4.39043075e-01
-6.73541129e-02 -6.46122321e-02 -1.41066384e+00 8.32428485e-02
2.85491735e-01 -7.73113012e-01 7.41566956e-01 -1.88826382e-01
8.22613060e-01 -4.16570663e-01 -1.85646802e-01 6.70885593e-02
8.09736013e-01 1.36657178e-01 6.20420754e-01 2.27394253e-01
3.49254727e-01 9.95534778e-01 3.61369133e-01 4.41472083e-01
8.35911572e-01 5.08014858e-01 -8.37463140e-02 -1.41404226e-01
2.62554079e-01 -1.93864509e-01 8.76648664e-01 1.45639288e+00
-1.81342345e-02 -2.83743054e-01 -8.80056143e-01 4.64330643e-01
-1.94561386e+00 -1.12924397e+00 3.37631628e-02 1.88824725e+00
1.48782945e+00 1.90346390e-01 -1.17004745e-01 -2.31691718e-01
3.88918430e-01 2.68094093e-01 -3.87261063e-01 -1.04257762e+00
-1.58527851e-01 2.00268179e-02 -1.98510602e-01 9.48781252e-01
-8.75926971e-01 1.01559842e+00 6.10849524e+00 5.59089124e-01
-1.36417532e+00 3.26186597e-01 7.10894704e-01 -3.10370654e-01
-3.55246395e-01 4.46452312e-02 -6.76095784e-01 2.73150802e-01
1.28574562e+00 -6.62038997e-02 4.47222263e-01 8.79166663e-01
2.61983067e-01 -3.08151305e-01 -1.21128607e+00 9.22605336e-01
2.97104001e-01 -8.06257904e-01 -3.42140824e-01 -1.17654078e-01
4.96966928e-01 1.05770536e-01 -3.02014649e-01 8.70723903e-01
2.91436195e-01 -9.30290759e-01 3.92852902e-01 7.28407562e-01
3.49532604e-01 -7.00857997e-01 9.70378935e-01 5.89025915e-01
-6.97230518e-01 2.78655350e-01 5.50606921e-02 -3.51521850e-01
2.49043480e-01 5.28490543e-01 -8.62083495e-01 2.73324162e-01
5.42210042e-01 5.65715373e-01 -4.17569727e-01 2.48627216e-01
-4.64503527e-01 1.06203437e+00 -3.96249086e-01 -2.62724489e-01
1.76356614e-01 -1.31110726e-02 5.31197608e-01 1.71950603e+00
-1.13534540e-01 2.86413908e-01 4.86649245e-01 5.28812706e-01
-3.07592362e-01 5.14587104e-01 -7.01061368e-01 -3.81615132e-01
3.67199510e-01 1.44524181e+00 -2.74525404e-01 -5.56320071e-01
-6.54917598e-01 1.10345292e+00 5.65960884e-01 5.17633259e-01
-5.16945422e-01 -3.48376781e-01 5.67516565e-01 -2.32745171e-01
-1.94160700e-01 -6.77635223e-02 -1.57225266e-01 -1.36971450e+00
-1.50802523e-01 -1.32518971e+00 4.00214463e-01 -7.30788410e-01
-1.33444154e+00 9.70625222e-01 -3.63091081e-01 -8.46483409e-01
-8.16607952e-01 -3.83047909e-01 -5.71233690e-01 8.05190861e-01
-1.05511785e+00 -8.42532516e-01 -6.47261590e-02 4.48385656e-01
6.65306032e-01 -4.91090752e-02 1.39576542e+00 1.08083144e-01
-5.74283838e-01 6.47465646e-01 -1.44577920e-01 5.28543353e-01
1.26593983e+00 -1.05607581e+00 -1.21012010e-01 6.31527007e-01
7.31699467e-02 8.39721501e-01 6.49504840e-01 -5.78487098e-01
-1.37151790e+00 -6.32323503e-01 1.29513133e+00 -7.43078411e-01
6.55184031e-01 -7.17788577e-01 -1.04400432e+00 8.14641058e-01
8.85139763e-01 -5.97820997e-01 1.26094151e+00 7.18059599e-01
-4.62819368e-01 9.72969308e-02 -7.66777813e-01 6.42050862e-01
7.83861458e-01 -8.95813823e-01 -7.37978458e-01 -1.25175878e-01
5.52970648e-01 -4.70336050e-01 -8.66853476e-01 1.93125695e-01
6.24134541e-01 -7.49666095e-01 2.71817684e-01 -5.88161647e-01
6.16457522e-01 1.59378484e-01 -2.44383946e-01 -1.48364615e+00
1.37520418e-01 -7.70761669e-01 -2.35931695e-01 1.51312542e+00
1.07600081e+00 -8.12917948e-01 3.00194591e-01 1.11887503e+00
1.27447560e-01 -7.91172802e-01 -5.58070719e-01 -3.86073530e-01
-4.35523465e-02 -7.39574909e-01 3.19618851e-01 1.28538465e+00
8.48791182e-01 9.29753363e-01 -5.87088585e-01 -2.65666813e-01
-1.49467543e-01 3.57719690e-01 1.02417183e+00 -9.61912155e-01
-5.22550344e-01 -3.51657361e-01 9.59980711e-02 -1.10649180e+00
4.78709072e-01 -9.45195377e-01 2.10561156e-01 -1.40999234e+00
3.10453922e-01 -2.09107563e-01 -5.17862439e-02 7.47079968e-01
-2.36722648e-01 1.27465039e-01 3.20218205e-01 -2.91544460e-02
-8.13523471e-01 5.45451045e-01 1.03735530e+00 3.63434218e-02
-5.49526036e-01 -9.97501165e-02 -9.71427739e-01 9.90989566e-01
1.09857488e+00 -3.89738977e-01 -2.34361172e-01 -2.05238029e-01
3.19310635e-01 -7.06533417e-02 -5.07959686e-02 -5.74601293e-01
2.41667911e-01 -1.39714062e-01 1.89459875e-01 -3.14460099e-01
6.01607323e-01 -4.35847044e-01 -9.21236649e-02 2.91677425e-03
-5.95714390e-01 -1.37083203e-01 4.35001165e-01 1.06211156e-01
-5.57279587e-01 -2.33940944e-01 4.60869700e-01 -3.24213266e-01
-6.34342849e-01 -2.92372197e-01 -8.70455980e-01 2.83989340e-01
4.08243060e-01 1.66937187e-01 -2.31621906e-01 -1.03621376e+00
-8.67248297e-01 4.23642099e-01 2.96885163e-01 7.45918036e-01
3.80606562e-01 -1.06037402e+00 -6.46492839e-01 -3.14108506e-02
1.69873625e-01 -3.67587805e-01 2.43651003e-01 1.01577330e+00
1.05620749e-01 2.71848708e-01 2.58431286e-01 -4.51241344e-01
-1.36341131e+00 1.75875232e-01 3.94780636e-01 -4.10101920e-01
-3.56305838e-01 6.66863143e-01 1.31646663e-01 -8.83213639e-01
3.09775621e-01 -7.44756013e-02 -2.42131680e-01 3.13688070e-01
6.41604364e-01 -1.07293747e-01 -2.11630538e-01 -6.17480040e-01
-4.07855004e-01 8.09681863e-02 -2.44298816e-01 -5.56259692e-01
1.28641284e+00 -3.43991250e-01 -4.09023225e-01 1.14357972e+00
1.26070130e+00 3.84451419e-01 -6.86758280e-01 -5.07289648e-01
8.06429163e-02 -5.73948734e-02 -1.15498133e-01 -1.10056186e+00
-7.87175417e-01 8.95963550e-01 1.70971960e-01 9.82333571e-02
1.10684168e+00 1.51884809e-01 6.29837692e-01 7.04915106e-01
3.08948785e-01 -1.56524420e+00 1.39501825e-01 1.11726153e+00
1.05260992e+00 -1.38676405e+00 -4.02829021e-01 -3.38722140e-01
-1.17719257e+00 1.08017230e+00 9.20321882e-01 4.17236626e-01
7.12583005e-01 4.67735916e-01 7.60079980e-01 -1.22596271e-01
-1.28669620e+00 -1.13369934e-01 6.07497618e-02 2.55469941e-02
1.00866222e+00 2.76717871e-01 -3.68472099e-01 1.16889894e+00
-4.64032769e-01 -4.00758833e-02 5.60973048e-01 1.04183316e+00
-2.01943979e-01 -1.31545281e+00 -1.36442229e-01 3.99428397e-01
-5.64250350e-01 -3.07393670e-01 -9.86427724e-01 5.99349201e-01
-1.15246452e-01 1.36737394e+00 -1.76124632e-01 -4.37263936e-01
2.97190130e-01 8.53720605e-01 3.76187414e-02 -7.07662821e-01
-8.63205910e-01 1.68700635e-01 8.87964547e-01 -7.76518583e-01
-5.74538469e-01 -4.32056457e-01 -1.35269082e+00 -2.47782215e-01
-3.16560298e-01 4.84273225e-01 5.46016037e-01 1.05293798e+00
5.08230329e-01 1.91842034e-01 6.11119807e-01 -5.45102537e-01
-5.24300598e-02 -1.36513412e+00 -2.14084551e-01 5.07199764e-01
9.46651474e-02 -2.23796546e-01 -4.13996339e-01 6.02116808e-02] | [12.990486145019531, 6.260193347930908] |
31814033-61fd-4c3b-8aa5-82221844285e | weakly-supervised-headline-dependency-parsing | 2301.10371 | null | https://arxiv.org/abs/2301.10371v1 | https://arxiv.org/pdf/2301.10371v1.pdf | Weakly Supervised Headline Dependency Parsing | English news headlines form a register with unique syntactic properties that have been documented in linguistics literature since the 1930s. However, headlines have received surprisingly little attention from the NLP syntactic parsing community. We aim to bridge this gap by providing the first news headline corpus of Universal Dependencies annotated syntactic dependency trees, which enables us to evaluate existing state-of-the-art dependency parsers on news headlines. To improve English news headline parsing accuracies, we develop a projection method to bootstrap silver training data from unlabeled news headline-article lead sentence pairs. Models trained on silver headline parses demonstrate significant improvements in performance over models trained solely on gold-annotated long-form texts. Ultimately, we find that, although projected silver training data improves parser performance across different news outlets, the improvement is moderated by constructions idiosyncratic to outlet. | ['Igor Malioutov', 'Ozan İrsoy', 'Tianze Shi', 'Adrian Benton'] | 2023-01-25 | null | null | null | null | ['dependency-parsing'] | ['natural-language-processing'] | [-2.53514230e-01 5.52843750e-01 -6.17438674e-01 -6.28166795e-01
-1.49967790e+00 -9.15596128e-01 2.46704757e-01 5.26587844e-01
-6.01680219e-01 7.81346142e-01 1.12340367e+00 -6.62059009e-01
3.21802229e-01 -5.96406221e-01 -8.12803566e-01 -2.84942966e-02
1.95692509e-01 3.78021985e-01 2.01222628e-01 -3.50211024e-01
-6.08354295e-03 -1.25320956e-01 -6.88319623e-01 5.82578957e-01
5.90738893e-01 1.83105230e-01 4.25794214e-01 4.96274650e-01
-3.33220959e-01 7.83459246e-01 -4.39242184e-01 -9.34924364e-01
-2.75921468e-02 -5.05809128e-01 -1.01770449e+00 -1.81160703e-01
2.83283263e-01 4.44794595e-02 -9.83673055e-03 8.55091691e-01
1.94534287e-01 -3.95115227e-01 2.77145088e-01 -2.96953529e-01
-7.25132227e-01 1.61990917e+00 -4.95334089e-01 7.15795755e-01
4.46828753e-01 -3.36827308e-01 1.95062006e+00 -5.38510799e-01
1.30665874e+00 1.09683681e+00 9.02613044e-01 4.52717692e-01
-1.33280563e+00 -3.70087028e-01 3.03977072e-01 -3.10505033e-01
-5.90570569e-01 -5.32393634e-01 7.11786568e-01 -4.04493928e-01
1.21020758e+00 -1.35438759e-02 4.30077434e-01 1.25782359e+00
3.24494392e-01 9.90945816e-01 1.12586617e+00 -6.94166243e-01
-2.74382085e-01 9.38977376e-02 8.76076579e-01 5.45279443e-01
3.33667129e-01 -3.76537859e-01 -5.26905298e-01 -8.34596157e-02
1.64537996e-01 -8.11209440e-01 -1.77441001e-01 5.47345638e-01
-7.35985219e-01 1.19617665e+00 3.68230753e-02 5.81598639e-01
-2.36542150e-01 -1.87313914e-01 7.48630881e-01 3.57465237e-01
8.48300755e-01 5.40879071e-01 -1.05371761e+00 -3.14241171e-01
-5.79242170e-01 1.24616295e-01 1.31805730e+00 9.38568294e-01
2.55504072e-01 -4.81190890e-01 7.49033689e-02 9.20166790e-01
2.21639335e-01 5.76001048e-01 3.53715360e-01 -4.85306680e-01
1.05519056e+00 3.66053671e-01 -4.83261719e-02 -7.93599129e-01
-8.20723176e-01 -3.98949653e-01 2.71619707e-01 -7.73583651e-01
5.24844408e-01 -6.89015925e-01 -4.34042782e-01 1.74762809e+00
1.77582607e-01 -9.17953908e-01 4.04080063e-01 5.17305613e-01
9.40781355e-01 7.67852604e-01 5.80979705e-01 -4.44845200e-01
1.79545712e+00 -8.48812222e-01 -7.03293145e-01 -5.40433645e-01
1.12645674e+00 -1.13345146e+00 1.34600139e+00 -1.57434031e-01
-1.24007344e+00 -1.64421573e-01 -7.68265724e-01 -3.84926409e-01
7.60801360e-02 2.01183915e-01 6.45309389e-01 7.57125139e-01
-4.77145612e-01 2.31775269e-01 -1.11322272e+00 -4.14719015e-01
2.39783049e-01 -1.09712005e-01 -3.99809480e-01 5.92701845e-02
-1.09683657e+00 9.41111684e-01 2.20450759e-01 -2.42752731e-01
-1.40395701e-01 -7.79244721e-01 -1.08781195e+00 -1.40965655e-01
3.06959927e-01 -2.02583179e-01 1.78841019e+00 -8.27501595e-01
-1.13583326e+00 1.24109006e+00 -3.73532832e-01 -3.89608383e-01
8.89160559e-02 -6.06453538e-01 -4.44422781e-01 -2.80702189e-02
6.56383514e-01 -4.00941037e-02 1.63997054e-01 -9.12163854e-01
-9.77666259e-01 -4.46096957e-01 -5.92973344e-02 3.97955738e-02
-3.50911140e-01 1.09057248e+00 -3.23900670e-01 -5.72025716e-01
1.19744152e-01 -1.02388084e+00 -1.12679519e-01 -8.76295924e-01
-6.24954164e-01 -4.65142578e-01 3.35475355e-01 -1.01877320e+00
1.39485967e+00 -2.08586860e+00 -1.07030131e-01 -2.93035120e-01
-3.25147539e-01 -2.11218089e-01 7.80157745e-02 5.56952596e-01
1.30632237e-01 4.10587549e-01 4.92246449e-03 -5.97513258e-01
-5.08217551e-02 5.74041009e-01 -3.70268941e-01 3.57898653e-01
2.84379989e-01 1.15616357e+00 -6.27333879e-01 -5.30219436e-01
-6.43105924e-01 1.05385840e-01 -5.98380208e-01 6.76860213e-02
-3.59200358e-01 3.54034126e-01 -5.04748642e-01 5.02528548e-01
2.19510287e-01 -4.88574617e-02 6.22780919e-01 -3.58127393e-02
-5.70289373e-01 1.35370159e+00 -3.86884004e-01 1.50617552e+00
-5.64619541e-01 4.98054415e-01 2.16772482e-01 -6.19172037e-01
6.22877955e-01 1.87675580e-01 2.16500815e-02 -7.19253719e-01
2.79412210e-01 2.62946159e-01 2.52711266e-01 -6.96548700e-01
6.94240928e-01 -4.52829301e-01 -8.92750800e-01 7.13126957e-01
2.03718081e-01 -7.61671141e-02 4.07227218e-01 6.15744293e-02
1.22595429e+00 1.97384834e-01 3.60844433e-01 -4.39359248e-01
5.17020747e-02 6.44203663e-01 8.40761542e-01 4.94594157e-01
7.78250247e-02 4.06532466e-01 7.61982322e-01 -3.62521619e-01
-1.05803847e+00 -6.74202979e-01 -6.86726391e-01 1.69166172e+00
-3.69461030e-01 -7.97706664e-01 -8.08971405e-01 -1.03420055e+00
-3.40163738e-01 8.91646802e-01 -5.85404396e-01 7.43695021e-01
-1.43725908e+00 -1.08883202e+00 6.33212507e-01 6.54606879e-01
1.09357677e-01 -1.05618000e+00 -4.89014298e-01 7.31291950e-01
-3.09943348e-01 -1.80344284e+00 -3.16068709e-01 5.57602167e-01
-4.97298479e-01 -1.06569278e+00 -3.84202123e-01 -1.28145194e+00
3.80190820e-01 -2.06289217e-01 1.33561897e+00 -1.84490606e-01
1.37849823e-01 4.75273915e-02 -7.55057633e-01 -6.13122106e-01
-9.74077523e-01 6.22772157e-01 -2.72993684e-01 -6.91498578e-01
6.52845681e-01 -1.70323789e-01 1.37090236e-01 -1.21507160e-01
-3.18804741e-01 -4.57226597e-02 3.58024031e-01 8.43135297e-01
3.80747437e-01 -5.93587101e-01 5.56183279e-01 -1.74278677e+00
6.89299941e-01 -5.31763136e-01 -5.37363648e-01 6.09301180e-02
-2.09632635e-01 1.40422866e-01 8.15848112e-01 7.90924728e-02
-1.51085544e+00 -1.29411519e-01 -7.37391591e-01 8.44553709e-01
-1.62373036e-02 1.10420036e+00 7.69399405e-02 5.32158136e-01
8.25251043e-01 -3.59887779e-01 -4.04156029e-01 -7.40863085e-01
4.14958984e-01 6.88873529e-01 6.43309593e-01 -8.13422918e-01
4.62308496e-01 6.49577193e-03 -4.92704660e-01 -8.47767532e-01
-1.75104749e+00 -3.92536372e-01 -6.27909422e-01 3.57792526e-01
1.24795258e+00 -9.89022136e-01 1.99004095e-02 -1.66901797e-01
-1.39052093e+00 -3.44002068e-01 -2.80794203e-01 5.13098240e-01
-1.10283129e-01 1.40976608e-01 -1.42239535e+00 -4.65842456e-01
-4.29000854e-01 -7.38785326e-01 8.30522120e-01 3.18127684e-02
-6.33788645e-01 -1.11795032e+00 4.08285767e-01 3.17541659e-01
-3.57851125e-02 6.68513775e-02 1.17905152e+00 -1.03835893e+00
4.51875888e-02 -2.13483915e-01 -2.31581032e-02 2.30648592e-01
5.56666926e-02 -3.77808698e-02 -6.90214097e-01 1.92147806e-01
4.80210371e-02 -3.24224532e-01 7.11645305e-01 4.63272542e-01
1.14574842e-01 -5.50717473e-01 -3.42242032e-01 3.59168291e-01
1.11890686e+00 -2.25367635e-01 1.19939804e-01 7.94735610e-01
3.81912798e-01 7.15019166e-01 5.09353042e-01 1.83078408e-01
7.91074097e-01 1.69041499e-01 -2.45857552e-01 1.65217653e-01
-2.31751993e-01 -4.29789543e-01 5.88562727e-01 1.35905695e+00
2.73228794e-01 -2.24545315e-01 -8.91241074e-01 8.45120668e-01
-1.62048733e+00 -6.33837283e-01 -5.32342792e-01 1.40021682e+00
1.16058624e+00 6.21612191e-01 9.18531045e-02 -2.38038436e-01
6.33991599e-01 2.86173791e-01 2.06890419e-01 -5.48694968e-01
-4.14577514e-01 3.45041186e-01 5.73388100e-01 6.54457211e-01
-1.32250869e+00 1.40591979e+00 6.73284101e+00 2.48815000e-01
-7.72965968e-01 4.38753635e-01 4.34804708e-01 3.49875182e-01
-5.26391685e-01 2.78056324e-01 -1.55383015e+00 2.97976226e-01
1.22534680e+00 -4.60113287e-02 -2.89996326e-01 1.08350599e+00
1.82469577e-01 -3.40694189e-02 -9.52489674e-01 3.00472200e-01
3.86358500e-02 -1.48726237e+00 -5.45211077e-01 -1.95262074e-01
7.84037888e-01 6.01686716e-01 -3.40277523e-01 3.36935371e-01
7.96713054e-01 -6.10280752e-01 8.25778306e-01 -3.15076530e-01
4.39175367e-01 -5.36640406e-01 9.33303058e-01 4.21021640e-01
-8.68053436e-01 -4.71681589e-03 -3.12239200e-01 -2.53676534e-01
1.01568937e+00 5.42109847e-01 -8.98943067e-01 2.97800899e-01
5.39340973e-01 5.90442836e-01 -4.53973681e-01 4.90883082e-01
-8.76043320e-01 1.46895337e+00 -3.51663172e-01 -3.96019012e-01
3.47322971e-01 2.68497076e-02 5.67862511e-01 1.79350471e+00
-1.97591722e-01 2.87834466e-01 3.77802670e-01 3.90531033e-01
-2.80638129e-01 6.86264217e-01 -3.53208035e-01 -3.36260378e-01
3.29070330e-01 1.34265876e+00 -9.51805353e-01 -9.30779278e-02
-9.61832643e-01 5.82562804e-01 8.09884727e-01 -4.86795716e-02
-5.80337763e-01 -1.29585832e-01 2.79232830e-01 2.97714770e-01
2.25484028e-01 -5.64515471e-01 -5.06888092e-01 -1.04447985e+00
9.73924622e-02 -5.28715968e-01 5.68500817e-01 -2.65420645e-01
-1.53499711e+00 8.72236550e-01 -1.64926454e-01 -5.02546787e-01
-1.47121489e-01 -6.54176831e-01 -4.35040146e-01 6.41371787e-01
-1.66336536e+00 -1.41368079e+00 3.10589045e-01 -1.49027500e-02
7.57317185e-01 6.13534339e-02 1.05868840e+00 1.10175855e-01
-6.77190661e-01 5.73199213e-01 -8.10062513e-02 6.40343308e-01
6.65217936e-01 -1.17449272e+00 9.23431039e-01 9.42657530e-01
3.43091875e-01 6.36944652e-01 8.82249415e-01 -8.54389429e-01
-1.19759130e+00 -7.69562185e-01 1.61990404e+00 -7.68229842e-01
1.30987835e+00 -3.28012943e-01 -7.20480919e-01 1.25087011e+00
3.28787863e-01 -1.37995958e-01 1.03717220e+00 8.69043410e-01
-4.87344503e-01 4.31021988e-01 -6.29250348e-01 4.61947799e-01
9.49729979e-01 -6.13175333e-01 -1.24978650e+00 5.35465658e-01
1.00360072e+00 -5.36079884e-01 -1.04238307e+00 2.59362478e-02
3.79019111e-01 -5.27876318e-01 1.53638721e-01 -9.23580825e-01
7.91997850e-01 2.70451069e-01 -1.88677579e-01 -1.22208774e+00
-2.79664487e-01 -6.44567847e-01 5.55081129e-01 1.55270684e+00
1.24712527e+00 -4.89064872e-01 5.99557877e-01 7.83580959e-01
-7.58438826e-01 -3.99948925e-01 -7.56720066e-01 -5.48213005e-01
4.70719874e-01 -6.84370697e-01 3.27251367e-02 8.20442677e-01
7.05933213e-01 1.03813803e+00 1.62760720e-01 -8.14701021e-02
1.72927946e-01 3.18666577e-01 5.62458277e-01 -1.15852380e+00
-6.06066465e-01 -5.94028421e-02 1.04223594e-01 -1.22026038e+00
5.68017006e-01 -1.09342062e+00 3.47173244e-01 -1.57218218e+00
2.52470732e-01 -7.59770989e-01 4.39539999e-01 6.81423187e-01
-3.35280359e-01 1.52669042e-01 1.69723660e-01 2.24146605e-01
-4.60036427e-01 1.46423383e-02 9.93443847e-01 4.19440746e-01
-5.75918496e-01 -1.88879129e-02 -1.13194227e+00 9.98308659e-01
7.29924202e-01 -8.41483474e-01 1.78336889e-01 -7.57869899e-01
7.03499794e-01 6.22753836e-02 -3.59405041e-01 -5.06602645e-01
4.95615462e-03 -9.89985242e-02 7.80297220e-02 -2.53682315e-01
-2.52051234e-01 -2.56951809e-01 -4.33486909e-01 7.94212371e-02
-4.58202749e-01 3.35723221e-01 2.42748812e-01 2.45421633e-01
-1.10835090e-01 -6.40791237e-01 4.81192440e-01 -3.88616741e-01
-3.78074557e-01 -1.47116274e-01 -5.53867221e-01 7.27760375e-01
5.73116124e-01 3.61422151e-01 -7.07897842e-01 2.40415446e-02
-6.69573903e-01 -7.57852644e-02 2.44067416e-01 2.63756275e-01
-2.05263108e-01 -5.92090786e-01 -9.86080110e-01 -2.57631969e-02
-6.12097159e-02 -5.73359020e-02 -2.68694222e-01 7.34254122e-01
-5.54746509e-01 4.12630022e-01 2.60487288e-01 -1.69724151e-01
-1.18024409e+00 3.53731900e-01 -3.69564176e-01 -5.60339928e-01
-1.05045533e+00 1.10109854e+00 -8.44756886e-02 -4.01453793e-01
-2.04825759e-01 -5.03249407e-01 -3.22256267e-01 2.38068044e-01
5.02470911e-01 -1.67169541e-01 4.69217636e-02 -1.01670873e+00
-2.27819845e-01 2.54236817e-01 -3.92740250e-01 -3.44546139e-01
1.68560386e+00 -2.10347787e-01 1.16532510e-02 5.41597724e-01
1.19553542e+00 9.91620302e-01 -8.68507326e-01 -1.40237585e-02
6.28509164e-01 1.89068124e-01 -1.48457229e-01 -6.20601177e-01
-4.26762670e-01 1.99204549e-01 -4.56923544e-01 5.45943856e-01
4.64434773e-01 6.56586587e-01 1.23950040e+00 3.27281415e-01
1.73078060e-01 -1.26802886e+00 -3.99912655e-01 1.00450146e+00
6.89039886e-01 -1.11373305e+00 -1.66249443e-02 -7.12414622e-01
-8.55919480e-01 1.10737562e+00 2.75385469e-01 -3.27886224e-01
8.27216506e-01 8.73516381e-01 2.72323519e-01 -4.50541317e-01
-4.59442943e-01 -3.25708956e-01 -6.85264245e-02 2.99724042e-01
1.07488668e+00 1.16688371e-01 -1.10677207e+00 1.39112866e+00
-9.30237591e-01 -4.84036565e-01 6.05696738e-01 1.01434517e+00
-6.28554583e-01 -1.35841107e+00 -4.09055948e-02 3.87037694e-01
-1.30321276e+00 -7.00954914e-01 -1.01031199e-01 9.50679243e-01
-3.10874999e-01 9.75746810e-01 2.33668715e-01 9.26868096e-02
3.46912265e-01 3.05023968e-01 3.51339340e-01 -1.13077605e+00
-8.66655052e-01 3.96719575e-01 1.20438218e+00 -1.08889423e-01
-5.43704808e-01 -1.01529360e+00 -1.61671102e+00 1.28484750e-02
-4.10227239e-01 5.01467407e-01 7.83560455e-01 1.28661740e+00
7.19147846e-02 2.23268241e-01 2.45338395e-01 -5.76279938e-01
-3.99823010e-01 -1.07858670e+00 -4.23695564e-01 2.93905526e-01
5.65202087e-02 -4.40511517e-02 -3.57764035e-01 2.38800973e-01] | [10.231389999389648, 9.77672290802002] |
5edb0f70-19b2-460b-8514-616923dc541f | casp-net-rethinking-video-saliency-prediction | 2303.06357 | null | https://arxiv.org/abs/2303.06357v1 | https://arxiv.org/pdf/2303.06357v1.pdf | CASP-Net: Rethinking Video Saliency Prediction from an Audio-VisualConsistency Perceptual Perspective | Incorporating the audio stream enables Video Saliency Prediction (VSP) to imitate the selective attention mechanism of human brain. By focusing on the benefits of joint auditory and visual information, most VSP methods are capable of exploiting semantic correlation between vision and audio modalities but ignoring the negative effects due to the temporal inconsistency of audio-visual intrinsics. Inspired by the biological inconsistency-correction within multi-sensory information, in this study, a consistency-aware audio-visual saliency prediction network (CASP-Net) is proposed, which takes a comprehensive consideration of the audio-visual semantic interaction and consistent perception. In addition a two-stream encoder for elegant association between video frames and corresponding sound source, a novel consistency-aware predictive coding is also designed to improve the consistency within audio and visual representations iteratively. To further aggregate the multi-scale audio-visual information, a saliency decoder is introduced for the final saliency map generation. Substantial experiments demonstrate that the proposed CASP-Net outperforms the other state-of-the-art methods on six challenging audio-visual eye-tracking datasets. For a demo of our system please see our project webpage. | ['Guangtao Zhai', 'Yufei zha', 'Wei Huang', 'Peng Zhang', 'Ganglai Wang', 'Junwen Xiong'] | 2023-03-11 | null | null | null | null | ['saliency-prediction'] | ['computer-vision'] | [ 3.72205347e-01 -1.15744323e-01 -7.88846333e-03 -3.09792906e-01
-6.21309102e-01 3.87317240e-02 3.38917851e-01 7.15384707e-02
-1.00603536e-01 4.92039084e-01 5.50844550e-01 3.27769220e-01
3.13338302e-02 -6.10167533e-02 -9.20912206e-01 -4.22377765e-01
2.32080042e-01 -5.24891376e-01 8.47424030e-01 -5.52305579e-02
6.54491603e-01 -1.67530969e-01 -2.31579208e+00 5.48930526e-01
9.32331264e-01 1.22609663e+00 7.73634374e-01 5.18612981e-01
9.56697315e-02 8.70564938e-01 -1.92695469e-01 3.63893472e-02
-1.83920145e-01 -7.77054250e-01 -4.26183999e-01 -2.88457684e-02
5.85678577e-01 -1.95057020e-01 -3.67607892e-01 1.26676834e+00
6.78487957e-01 1.68403462e-01 1.47983864e-01 -1.85700154e+00
-8.59499454e-01 5.05258083e-01 -5.57122529e-01 6.78804219e-01
4.55989927e-01 2.49735951e-01 1.12687147e+00 -1.13314474e+00
4.16354030e-01 1.26715088e+00 2.84122944e-01 4.56915408e-01
-9.33609724e-01 -7.98583329e-01 3.54566127e-01 1.01544714e+00
-1.37594497e+00 -7.43780017e-01 1.14398408e+00 -3.16887379e-01
7.12235034e-01 3.90635908e-01 1.02869093e+00 9.75964546e-01
3.19943070e-01 1.02150404e+00 7.63062954e-01 -1.08385548e-01
1.29526228e-01 1.15628183e-01 -1.68236166e-01 4.04884309e-01
-1.44918174e-01 2.42290139e-01 -1.42369914e+00 1.04964353e-01
6.67796731e-01 2.38208491e-02 -3.46452624e-01 -4.14306462e-01
-1.51293278e+00 3.37122977e-01 6.60843492e-01 2.16298938e-01
-4.49816614e-01 3.19690377e-01 4.39395428e-01 -9.64340940e-02
2.36731648e-01 1.44760564e-01 -1.19651824e-01 6.12371191e-02
-1.18944943e+00 5.76764829e-02 -2.11607009e-01 9.04617310e-01
6.15866005e-01 4.40284103e-01 -6.44876599e-01 6.33776188e-01
7.12027788e-01 3.87209237e-01 6.67782247e-01 -1.06228590e+00
1.07036039e-01 4.34500903e-01 1.79425962e-02 -1.14265072e+00
-3.50895107e-01 -5.07701635e-01 -5.38417876e-01 1.00669481e-01
-1.42858639e-01 4.22395945e-01 -5.90831280e-01 1.92402947e+00
4.24763560e-01 7.83244908e-01 -8.32061768e-02 1.35981619e+00
1.19315553e+00 3.96908879e-01 1.63158819e-01 -4.97262061e-01
1.46704495e+00 -1.05899251e+00 -9.72194552e-01 -2.85661399e-01
-1.99444652e-01 -7.16754735e-01 9.61521864e-01 2.42763385e-02
-1.17627275e+00 -8.55605364e-01 -1.08809710e+00 -2.94748396e-01
4.51487936e-02 6.09659515e-02 4.35014099e-01 -1.40883967e-01
-1.20619154e+00 2.04800963e-01 -6.87266052e-01 -3.71136099e-01
5.26085854e-01 6.55560242e-03 4.49035615e-02 2.96417981e-01
-1.39456737e+00 7.20721245e-01 2.57563472e-01 1.43764496e-01
-1.24075556e+00 -8.27040136e-01 -9.47091341e-01 1.35992780e-01
4.11312580e-01 -8.02738786e-01 1.11298370e+00 -1.58715641e+00
-1.30215001e+00 5.44514358e-01 -7.62537897e-01 -3.94854397e-01
6.30059838e-02 -1.71480179e-01 -4.74756420e-01 5.56328475e-01
2.81582266e-01 1.20129943e+00 1.29952157e+00 -1.25049150e+00
-6.68579757e-01 -1.46489128e-01 -3.52160126e-01 5.24938762e-01
-1.36925429e-01 3.21196765e-01 -3.82752866e-01 -8.07411015e-01
1.71659872e-01 -6.01420105e-01 2.40429699e-01 2.12874502e-01
-3.66368204e-01 -1.27703905e-01 6.72713339e-01 -6.63509965e-01
1.28200841e+00 -2.27740955e+00 3.52987945e-01 -2.76272118e-01
2.09814429e-01 1.07040733e-01 -3.52911592e-01 6.50473982e-02
-1.91975459e-01 -3.50565583e-01 -8.22859779e-02 -2.24619836e-01
-1.17837094e-01 -4.33323234e-01 -4.74679381e-01 3.37774158e-01
3.37797493e-01 1.00556624e+00 -1.13463950e+00 -7.63906240e-01
2.00621963e-01 6.59206867e-01 -6.22855902e-01 7.39609450e-02
-1.67828381e-01 5.61660230e-01 -1.80723786e-01 8.40440392e-01
4.86344010e-01 -2.88603008e-01 -2.58021146e-01 -4.10544574e-01
-2.10565999e-01 2.20272094e-01 -9.61333275e-01 2.08277535e+00
6.79618418e-02 8.82120490e-01 -1.24277242e-01 -6.95439875e-01
7.74902701e-01 3.49408209e-01 4.12906557e-01 -8.85592163e-01
2.04761550e-02 1.40275493e-01 3.14544104e-02 -2.94092953e-01
5.46661437e-01 1.01285473e-01 2.66692936e-01 7.49143139e-02
4.10318941e-01 1.13412410e-01 -1.50945067e-01 3.67035717e-01
6.49950147e-01 1.15767621e-01 1.50518164e-01 -1.46707997e-01
7.09406197e-01 -3.58734071e-01 7.41949737e-01 4.11647350e-01
-7.76059091e-01 8.37538064e-01 2.28121340e-01 -3.26388590e-02
-8.86502743e-01 -1.21125889e+00 2.01614648e-02 1.29425132e+00
7.27676272e-01 -6.14823341e-01 -7.56180167e-01 -2.36597791e-01
-1.94399178e-01 7.54092515e-01 -6.33454025e-01 -4.60023403e-01
8.68876800e-02 -1.93265453e-01 2.03376949e-01 3.77435237e-01
5.32020807e-01 -1.25373995e+00 -8.03839684e-01 8.71965438e-02
-6.33722425e-01 -9.76081073e-01 -8.46325278e-01 -3.80567908e-01
-5.34617186e-01 -9.35967565e-01 -6.12135053e-01 -8.57013524e-01
2.82367647e-01 8.15234125e-01 6.46095395e-01 3.97232063e-02
-2.18045548e-01 4.07983750e-01 -2.80603856e-01 -6.18618906e-01
4.51748446e-02 -3.59387696e-01 1.52882531e-01 3.83680105e-01
3.76550049e-01 -6.74712896e-01 -8.06911945e-01 3.71564567e-01
-7.54700780e-01 6.12408936e-01 4.07164246e-01 7.24935055e-01
7.82069147e-01 -2.58530259e-01 8.72607827e-01 2.32697979e-01
1.80031404e-01 -4.71809655e-01 -3.08576494e-01 2.18129247e-01
-4.01174933e-01 -6.28560930e-02 1.27902314e-01 -4.51854229e-01
-1.08385491e+00 1.97086230e-01 1.58717677e-01 -8.20395648e-01
-6.00254498e-02 2.59624928e-01 -1.86648503e-01 5.48448488e-02
2.99452186e-01 7.37738907e-01 9.60602164e-02 -1.72663003e-01
2.69377828e-01 5.07486701e-01 8.69623005e-01 1.20295875e-01
3.40584427e-01 4.98897910e-01 -8.20892453e-02 -7.67930269e-01
-1.06177175e+00 -4.46646988e-01 -5.14221549e-01 -6.80448592e-01
1.01943707e+00 -1.30042386e+00 -6.42168701e-01 6.79945886e-01
-1.33701897e+00 1.01293072e-01 -4.92883986e-03 5.75639009e-01
-7.85728395e-01 4.75865543e-01 -2.43283153e-01 -8.02964032e-01
-1.46834612e-01 -1.18373084e+00 1.15715933e+00 4.33478117e-01
-5.24897613e-02 -4.03076798e-01 -1.36036634e-01 2.62709409e-01
3.49093914e-01 -1.69593811e-01 3.36658239e-01 -3.84910226e-01
-7.51555085e-01 3.72785926e-01 -5.05253017e-01 9.29119363e-02
8.79031047e-03 -2.78012753e-02 -1.27043664e+00 -1.69730876e-02
-5.81740588e-02 -2.67522216e-01 1.07913363e+00 7.45233238e-01
1.21518159e+00 -1.90163925e-01 -1.68854296e-01 3.71452451e-01
1.15340602e+00 -8.22015032e-02 6.49952948e-01 2.50040144e-01
7.91237831e-01 7.11500466e-01 8.72311354e-01 5.16349018e-01
7.79467881e-01 8.52287769e-01 8.95103633e-01 1.50117338e-01
-4.24243867e-01 -5.88916421e-01 6.19190931e-01 8.41437876e-01
1.27321005e-01 1.47483766e-01 -4.72684830e-01 6.44500315e-01
-2.01872206e+00 -1.29775441e+00 -2.15163186e-01 2.17496562e+00
9.12346900e-01 -9.69648175e-03 2.56510705e-01 8.06504395e-03
1.16249037e+00 3.26861590e-01 -8.00897837e-01 -6.13028742e-03
-4.02010858e-01 -3.04058015e-01 -6.86410218e-02 4.13211614e-01
-1.01280344e+00 8.85779619e-01 6.14197540e+00 9.08270657e-01
-1.08302915e+00 3.49080473e-01 1.80165917e-01 -6.19186640e-01
-5.96545458e-01 8.28327518e-03 -6.38797641e-01 8.30759108e-01
8.71734977e-01 -2.53555655e-01 5.06082356e-01 4.84790146e-01
5.20994663e-01 -4.31071103e-01 -9.40722167e-01 1.11036134e+00
4.29089516e-01 -1.35596561e+00 1.86365157e-01 -3.01737994e-01
4.67349619e-01 7.09489882e-02 2.35562474e-01 -2.40217924e-01
-3.11348319e-01 -7.07576692e-01 1.36454844e+00 9.89952743e-01
6.09764099e-01 -6.85811102e-01 3.99049550e-01 1.03491835e-01
-1.41745663e+00 -2.02324748e-01 -3.17734092e-01 4.74803708e-02
2.98769474e-01 4.59916025e-01 -2.42882505e-01 3.06256860e-01
1.14317226e+00 1.16878223e+00 -7.26250470e-01 1.39199293e+00
-3.45166683e-01 2.41575703e-01 -3.09365969e-02 4.40593772e-02
-1.74676999e-01 2.84736961e-01 1.03287315e+00 9.93454933e-01
3.76920819e-01 -1.52193695e-01 -2.27012590e-01 1.11608124e+00
3.24638605e-01 -3.68223749e-02 -2.58044720e-01 1.21725149e-01
7.77506649e-01 1.08819568e+00 -5.19676447e-01 -3.23306829e-01
-3.48936170e-01 1.09629261e+00 5.86004481e-02 3.06203723e-01
-1.13042378e+00 -3.34672928e-01 8.45905364e-01 -2.29108706e-01
5.31949282e-01 7.96626061e-02 -3.31162035e-01 -1.01360393e+00
2.05537677e-01 -5.63786030e-01 1.91062525e-01 -1.52276349e+00
-7.55986392e-01 4.10013199e-01 -2.34887138e-01 -1.51428759e+00
5.23165651e-02 1.05560370e-01 -5.86427629e-01 7.52299368e-01
-1.80508184e+00 -1.22047341e+00 -3.30996275e-01 8.33894014e-01
7.72469163e-01 -1.14199907e-01 3.29805911e-01 8.02755207e-02
-4.01642948e-01 5.24512351e-01 -2.80080974e-01 -4.67829078e-01
8.97461474e-01 -7.59889185e-01 -3.72174010e-03 1.12371802e+00
2.45492935e-01 3.29165608e-01 9.35795844e-01 -7.83944905e-01
-1.19245958e+00 -1.05117488e+00 1.05871916e+00 -2.95663327e-01
7.55270183e-01 -2.30580360e-01 -1.03793132e+00 3.62001121e-01
4.21734989e-01 4.02110592e-02 5.40624380e-01 -3.58428746e-01
-4.02501822e-01 -2.35046223e-01 -7.09875822e-01 5.07771075e-01
1.12307024e+00 -9.34618056e-01 -8.44293535e-01 -1.15838334e-01
1.19305551e+00 -2.36035794e-01 -2.50288486e-01 2.65790224e-01
6.34660542e-01 -1.23761654e+00 1.08148444e+00 -3.04294676e-01
5.80022097e-01 -8.94956350e-01 -2.00017467e-01 -1.24203610e+00
-4.60584134e-01 -4.86101717e-01 -3.44831467e-01 1.37916994e+00
2.16473676e-02 -1.24586813e-01 1.00281358e-01 3.38077433e-02
-4.02412236e-01 -3.03603858e-01 -1.19477797e+00 -4.73126203e-01
-8.42406988e-01 -6.75596535e-01 2.69334435e-01 6.55828655e-01
3.31559837e-01 3.78138423e-01 -7.18792856e-01 4.12327528e-01
6.77072644e-01 1.31827682e-01 2.52726942e-01 -1.21605563e+00
-1.12144604e-01 -4.45553035e-01 -5.37741899e-01 -8.74327004e-01
2.09757864e-01 -8.54071558e-01 2.83586234e-01 -1.35824835e+00
4.84058112e-01 4.02690291e-01 -7.95090079e-01 3.69881123e-01
-2.77647555e-01 5.09192348e-01 4.10452783e-01 2.96736538e-01
-1.22949827e+00 1.07633030e+00 1.39512932e+00 -4.08158302e-02
-2.73213238e-01 -4.09767240e-01 -9.40845847e-01 6.09339952e-01
3.99511039e-01 -3.25979352e-01 -4.89684075e-01 -2.59411484e-01
1.84164301e-01 1.39252096e-01 1.04210997e+00 -1.15889478e+00
6.06570482e-01 -2.37624422e-01 3.05112064e-01 -8.40967953e-01
4.56878334e-01 -5.50712347e-01 -3.21039483e-02 2.59340554e-01
-6.48101151e-01 -1.78912953e-01 2.75433838e-01 9.66704249e-01
-4.94401813e-01 2.09071338e-01 8.96683097e-01 1.30554035e-01
-1.03927970e+00 1.57807365e-01 -4.03433621e-01 -8.05527419e-02
9.60605681e-01 -2.77830333e-01 -3.72158408e-01 -3.82228732e-01
-6.72420681e-01 3.28730881e-01 3.31775397e-01 7.40030229e-01
1.09280550e+00 -1.75382972e+00 -5.18544078e-01 2.43046314e-01
2.86438376e-01 -3.19936246e-01 8.75915945e-01 1.25010836e+00
2.62441188e-01 4.27919894e-01 -6.59388244e-01 -8.87174487e-01
-1.41057813e+00 7.95411050e-01 2.07114935e-01 6.96716547e-01
-4.72803354e-01 9.53604639e-01 5.55564284e-01 3.92839938e-01
3.26240063e-01 -2.49509513e-01 -4.25257534e-01 1.09058179e-01
8.40525568e-01 2.95869261e-01 -2.99012095e-01 -1.02415502e+00
-6.90424740e-01 4.20007139e-01 2.93711960e-01 -8.83310735e-02
1.07952201e+00 -8.49001229e-01 -1.60764530e-01 8.70427310e-01
7.91295111e-01 -2.31672481e-01 -1.66437733e+00 -4.26047236e-01
-2.97276139e-01 -5.61753392e-01 2.63693780e-01 -7.27835596e-01
-1.05080950e+00 1.16900480e+00 7.17898309e-01 -1.25567660e-01
1.42626286e+00 2.08504677e-01 6.37745261e-01 -2.46022135e-01
1.99986160e-01 -1.15612388e+00 5.70161104e-01 3.29367191e-01
1.24759746e+00 -1.31955862e+00 -1.67076543e-01 -5.08802652e-01
-1.12212217e+00 8.90924811e-01 7.79630959e-01 2.35265456e-02
5.33857763e-01 -3.59246105e-01 -2.42951572e-01 -3.49454395e-02
-1.04729164e+00 -5.35092413e-01 6.82434857e-01 7.82163322e-01
1.76878348e-01 -2.84653574e-01 1.28806999e-03 9.31134224e-01
-5.46285138e-03 8.07733238e-02 4.83739406e-01 4.88490850e-01
-7.69262910e-01 -3.15948486e-01 -2.46867806e-01 1.49115786e-01
-1.95752993e-01 -3.88727009e-01 -3.08236241e-01 6.78952113e-02
4.76424471e-02 1.06894064e+00 3.10251534e-01 -5.91954887e-01
-8.27558264e-02 8.74749199e-02 3.18550974e-01 -3.43912572e-01
-3.05628568e-01 4.79375035e-01 -2.70824879e-01 -9.15617824e-01
-8.56693447e-01 -8.07468474e-01 -1.37951398e+00 1.76579759e-01
-1.84483662e-01 -1.39618441e-01 5.64853072e-01 9.22604203e-01
8.02078903e-01 7.93954253e-01 5.28050244e-01 -1.17736244e+00
-9.44486409e-02 -8.45740855e-01 -7.99443960e-01 2.04428926e-01
7.26469636e-01 -1.05924845e+00 -5.04593670e-01 1.64067656e-01] | [9.765751838684082, -0.24203626811504364] |
663aba92-961a-461e-94f6-1df4ffc096fb | cmcgan-a-uniform-framework-for-cross-modal | 1711.08102 | null | http://arxiv.org/abs/1711.08102v2 | http://arxiv.org/pdf/1711.08102v2.pdf | CMCGAN: A Uniform Framework for Cross-Modal Visual-Audio Mutual Generation | Visual and audio modalities are two symbiotic modalities underlying videos,
which contain both common and complementary information. If they can be mined
and fused sufficiently, performances of related video tasks can be
significantly enhanced. However, due to the environmental interference or
sensor fault, sometimes, only one modality exists while the other is abandoned
or missing. By recovering the missing modality from the existing one based on
the common information shared between them and the prior information of the
specific modality, great bonus will be gained for various vision tasks. In this
paper, we propose a Cross-Modal Cycle Generative Adversarial Network (CMCGAN)
to handle cross-modal visual-audio mutual generation. Specifically, CMCGAN is
composed of four kinds of subnetworks: audio-to-visual, visual-to-audio,
audio-to-audio and visual-to-visual subnetworks respectively, which are
organized in a cycle architecture. CMCGAN has several remarkable advantages.
Firstly, CMCGAN unifies visual-audio mutual generation into a common framework
by a joint corresponding adversarial loss. Secondly, through introducing a
latent vector with Gaussian distribution, CMCGAN can handle dimension and
structure asymmetry over visual and audio modalities effectively. Thirdly,
CMCGAN can be trained end-to-end to achieve better convenience. Benefiting from
CMCGAN, we develop a dynamic multimodal classification network to handle the
modality missing problem. Abundant experiments have been conducted and validate
that CMCGAN obtains the state-of-the-art cross-modal visual-audio generation
results. Furthermore, it is shown that the generated modality achieves
comparable effects with those of original modality, which demonstrates the
effectiveness and advantages of our proposed method. | ['Zhao-Xiang Zhang', 'Wangli Hao', 'He Guan'] | 2017-11-22 | null | null | null | null | ['audio-generation'] | ['audio'] | [ 2.53268987e-01 -5.80588058e-02 -3.49748731e-02 1.63802937e-01
-7.94281483e-01 -4.66942132e-01 5.82485557e-01 -7.67740250e-01
1.65831387e-01 7.95127094e-01 3.75584632e-01 1.04619391e-01
-3.67347617e-03 -6.88336551e-01 -8.26286256e-01 -1.14083207e+00
2.34712139e-01 -2.50571162e-01 -3.76490131e-02 -2.15323776e-01
-2.61614114e-01 1.75981317e-02 -1.69167936e+00 4.24643755e-01
7.95442939e-01 9.80030715e-01 1.25605673e-01 5.56627035e-01
1.35805160e-01 8.89274061e-01 -4.90967810e-01 -3.82130474e-01
2.03474253e-01 -7.36637533e-01 -4.57027316e-01 3.15645427e-01
6.73208199e-03 -2.62177438e-01 -7.81090379e-01 1.06199372e+00
7.36449540e-01 -5.07692434e-02 5.97643971e-01 -1.86725712e+00
-6.02638245e-01 5.54543316e-01 -7.48756111e-01 -4.21043336e-01
5.50694525e-01 4.52139556e-01 6.57490313e-01 -6.44197047e-01
6.00208282e-01 1.39497244e+00 3.88807148e-01 6.59168243e-01
-1.03461611e+00 -9.86193061e-01 1.81509942e-01 1.67967036e-01
-1.28173566e+00 -4.52238649e-01 1.10517144e+00 -3.71034056e-01
3.36074531e-01 2.17349485e-01 7.20623136e-01 1.48412073e+00
5.58636412e-02 1.02396595e+00 1.06506765e+00 -2.17773616e-01
-2.88031131e-01 1.31200522e-01 -7.38071918e-01 5.42040825e-01
-2.83864170e-01 2.70802170e-01 -5.89347780e-01 7.94268996e-02
5.98292530e-01 1.56820625e-01 -7.05654502e-01 -3.62146199e-01
-1.40902328e+00 5.79007804e-01 2.90487707e-01 3.05433571e-01
-2.51015544e-01 1.70403153e-01 5.32301843e-01 4.22349304e-01
5.22908866e-02 -1.67773783e-01 -6.65030479e-02 -1.37060387e-02
-7.12565720e-01 -1.67327058e-02 3.56250018e-01 1.08454359e+00
5.06443679e-01 5.68332314e-01 -2.76433110e-01 7.25275636e-01
5.99063098e-01 9.27868187e-01 5.46042025e-01 -9.98069227e-01
5.63100994e-01 3.44086289e-01 -2.23815963e-01 -1.17762160e+00
-3.71423177e-02 -2.90557802e-01 -1.44452882e+00 1.38544530e-01
-7.60060921e-02 -3.62939268e-01 -8.12098980e-01 2.18442917e+00
2.66875446e-01 4.18479562e-01 3.59126985e-01 8.42789829e-01
1.19724381e+00 9.81557488e-01 8.21640119e-02 -3.77913803e-01
1.19434965e+00 -7.77183294e-01 -8.79751742e-01 5.09620719e-02
-1.90119982e-01 -8.82978380e-01 7.39564896e-01 2.19175637e-01
-1.01876557e+00 -7.71300733e-01 -1.13741136e+00 3.42078149e-01
2.12773439e-02 -7.07954988e-02 4.71047431e-01 3.98258656e-01
-9.00693417e-01 7.24246949e-02 -4.87026274e-01 2.93442141e-02
1.85310155e-01 1.13533236e-01 -6.64650023e-01 -1.86029702e-01
-1.52902067e+00 2.07386047e-01 6.43029869e-01 1.73825681e-01
-1.30165923e+00 -4.86758024e-01 -9.85345900e-01 -1.24870978e-01
3.96062821e-01 -9.75949943e-01 9.04201746e-01 -1.22375524e+00
-1.44449472e+00 4.58257228e-01 5.37114777e-02 7.25339400e-03
5.80075026e-01 2.56013691e-01 -7.41172016e-01 3.82671684e-01
1.29367784e-01 8.19571078e-01 1.31185102e+00 -1.54294789e+00
-7.61677682e-01 -1.11477159e-01 9.98046547e-02 2.96287239e-01
-3.79542589e-01 -2.89546400e-01 -7.71291196e-01 -8.44930649e-01
6.32571578e-02 -8.92047703e-01 3.60828221e-01 -8.31844509e-02
-5.40560961e-01 1.55946434e-01 1.19011569e+00 -7.55538702e-01
9.84660268e-01 -2.35571718e+00 7.38048255e-01 2.93260347e-02
2.37126961e-01 1.60374314e-01 -2.80835479e-01 5.73615074e-01
-2.10990787e-01 3.73690277e-02 -1.83734611e-01 -2.84077376e-01
-3.43475528e-02 2.43672386e-01 -3.11715126e-01 3.07401031e-01
7.81775042e-02 8.57598543e-01 -8.01360428e-01 -5.60584426e-01
2.94130802e-01 8.28196287e-01 -3.35391790e-01 3.78118843e-01
-3.50313261e-02 7.90270567e-01 -3.13696474e-01 9.82700646e-01
7.79218256e-01 -4.54423577e-03 3.37010389e-03 -5.52380741e-01
3.45230848e-01 -5.25493801e-01 -1.17605984e+00 1.79646885e+00
-4.80288714e-01 5.68955898e-01 3.17351937e-01 -1.00898528e+00
6.39207244e-01 7.21972883e-01 5.36565721e-01 -6.74635768e-01
2.29272902e-01 1.49203777e-01 -2.96327144e-01 -7.27427185e-01
1.25689849e-01 -4.15875316e-01 -2.16652125e-01 1.57529533e-01
4.11060423e-01 2.40747072e-03 -3.27614844e-02 2.64651567e-01
6.77702188e-01 2.13138059e-01 -1.41547784e-01 4.29276466e-01
8.09394896e-01 -5.15636027e-01 6.47732019e-01 2.56733030e-01
-1.31832287e-02 6.54945374e-01 4.60369945e-01 2.87736416e-01
-9.68557715e-01 -1.11872578e+00 2.41466403e-01 7.52217054e-01
3.51020157e-01 -8.15928262e-03 -6.04386687e-01 -5.99892676e-01
-1.74832344e-01 2.61786371e-01 -4.80068356e-01 -5.37272334e-01
-1.43891454e-01 -4.70857143e-01 8.30887556e-01 4.58708346e-01
9.12978172e-01 -1.06340313e+00 -2.03966387e-02 5.83167449e-02
-6.42240703e-01 -1.00468314e+00 -5.79214931e-01 -3.75479788e-01
-5.67931712e-01 -1.10594404e+00 -1.12350380e+00 -8.76552820e-01
3.14699739e-01 5.25518358e-01 7.40934253e-01 -2.12325722e-01
3.24294418e-02 7.40795493e-01 -4.02005821e-01 -1.49895832e-01
-5.49925864e-01 -2.22720921e-01 4.79829758e-02 4.40357149e-01
-1.70214862e-01 -8.76049519e-01 -5.75851500e-01 3.22226971e-01
-1.26846564e+00 2.10542351e-01 5.74956477e-01 1.09052432e+00
4.83017832e-01 3.88190091e-01 8.41853976e-01 -3.81178647e-01
4.34393615e-01 -6.52437806e-01 -2.08513111e-01 1.88195020e-01
-2.43935570e-01 -2.00207889e-01 6.12190068e-01 -7.19664812e-01
-1.26640558e+00 3.57046127e-02 -6.09626658e-02 -1.00104284e+00
-3.60770384e-03 4.82285500e-01 -9.54229116e-01 -9.34860029e-04
7.69048482e-02 6.36090517e-01 4.23774242e-01 -1.86262548e-01
5.23603618e-01 7.12113380e-01 8.66443872e-01 -3.37218553e-01
1.00633037e+00 4.70556855e-01 9.13292542e-03 -6.16839707e-01
-3.52966189e-01 -1.81503713e-01 -5.18909954e-02 -5.33639073e-01
1.01226664e+00 -1.34399867e+00 -8.01330447e-01 1.03051877e+00
-1.01454639e+00 1.80167809e-01 -1.97141290e-01 6.20615661e-01
-5.49610019e-01 6.21097982e-01 -6.21454179e-01 -5.87985039e-01
-2.91956723e-01 -1.27427053e+00 1.08115935e+00 4.53088701e-01
4.32543427e-01 -9.93621647e-01 -1.43419117e-01 5.31070352e-01
1.63424939e-01 6.02409184e-01 6.88906014e-01 -8.95193517e-02
-6.37277484e-01 -1.82726368e-01 -1.57760292e-01 6.62596762e-01
8.79536271e-02 1.60638224e-02 -1.02814829e+00 -4.41555262e-01
-7.84182642e-03 -3.67183387e-01 8.35786581e-01 2.00068906e-01
9.06153977e-01 -2.96719879e-01 -2.00469926e-01 7.28369474e-01
1.32604158e+00 3.76357913e-01 9.50142443e-01 1.43506065e-01
9.07954216e-01 4.23788637e-01 5.06963849e-01 2.67137319e-01
3.34003925e-01 4.70869720e-01 8.41681063e-01 -2.45585144e-01
-1.93694293e-01 -5.13012767e-01 7.10703254e-01 9.51676846e-01
-4.93538193e-02 -4.51955974e-01 -2.57486522e-01 5.58719575e-01
-1.73808885e+00 -1.30357885e+00 8.12968686e-02 2.06132174e+00
7.31201231e-01 -9.44570452e-02 1.04885638e-01 3.75841588e-01
9.67835784e-01 2.88279742e-01 -4.95909274e-01 1.26780823e-01
-5.06035864e-01 -2.27449313e-01 7.93128386e-02 9.25078392e-02
-1.11508024e+00 4.18445706e-01 5.41352224e+00 1.30237734e+00
-1.13607788e+00 2.33027667e-01 3.87398809e-01 1.85943618e-02
-5.85181773e-01 -1.13360353e-01 -3.02457303e-01 7.53107369e-01
4.12255734e-01 -6.28738701e-02 5.42458773e-01 5.61952591e-01
-5.85383289e-02 1.55077711e-01 -8.12170267e-01 1.22813892e+00
1.71168908e-01 -1.05522490e+00 3.04064453e-01 7.59077445e-02
7.68611431e-01 -4.54812855e-01 2.86843538e-01 4.20433193e-01
-9.22211111e-02 -1.04747462e+00 7.45259941e-01 7.06629694e-01
1.20491064e+00 -1.11291564e+00 9.31829572e-01 3.63047600e-01
-1.39122486e+00 -7.56056979e-03 -5.87371513e-02 4.48834956e-01
2.50957549e-01 3.42334270e-01 -9.37463120e-02 1.22363675e+00
6.38623655e-01 8.35144341e-01 -2.26152182e-01 7.54083335e-01
-2.83097595e-01 4.51325655e-01 -3.21810739e-03 5.29676437e-01
1.87223941e-01 -1.24491408e-01 9.23920691e-01 9.23756897e-01
5.91712773e-01 -1.27175525e-01 1.02811545e-01 6.74711823e-01
-1.11423701e-01 -1.76133096e-01 -8.79393518e-01 -1.13463506e-01
4.76286918e-01 1.22995412e+00 -1.84080765e-01 -2.67997384e-01
-4.33105111e-01 1.01303637e+00 -3.80195171e-01 4.49161381e-01
-1.20725060e+00 -6.00989640e-01 4.14439619e-01 -2.62311459e-01
2.26984680e-01 3.35697114e-01 2.37077117e-01 -1.31497681e+00
1.32570699e-01 -1.28713942e+00 3.30384433e-01 -1.15692234e+00
-1.27077401e+00 6.35297060e-01 -2.57029142e-02 -1.75178850e+00
-3.98079723e-01 -1.74947560e-01 -5.80140829e-01 7.78663218e-01
-1.36521196e+00 -1.61155498e+00 -5.52564442e-01 1.11823463e+00
3.48922402e-01 -5.96437693e-01 6.39778018e-01 5.53760707e-01
-3.24481040e-01 8.09221148e-01 2.26249903e-01 1.94466069e-01
8.10584009e-01 -8.26313257e-01 -4.47943002e-01 9.33833659e-01
-1.84154034e-01 2.06870183e-01 5.25771558e-01 -4.04447824e-01
-1.46934426e+00 -1.23259437e+00 2.76601702e-01 5.46419658e-02
4.16068316e-01 -2.51525696e-02 -7.11997747e-01 5.42133212e-01
5.84850967e-01 -2.19678879e-01 5.25504887e-01 -5.00184178e-01
-5.14273405e-01 -2.99318701e-01 -1.17210484e+00 5.28400302e-01
8.93082380e-01 -7.93144822e-01 -4.69590425e-01 1.44560710e-01
1.10549366e+00 -3.66241485e-01 -8.93966377e-01 5.49168110e-01
5.51470459e-01 -1.03940201e+00 1.06551826e+00 -3.56014341e-01
8.15148175e-01 -5.47072113e-01 -4.20232266e-01 -1.39457393e+00
8.26617144e-03 -7.04150259e-01 -2.38609746e-01 1.72682464e+00
4.55657952e-02 -5.72202444e-01 2.90812939e-01 -1.39130265e-01
-1.31631404e-01 -4.03400868e-01 -9.56192732e-01 -7.44162738e-01
-1.69502676e-01 -4.34384674e-01 6.88329637e-01 1.03972852e+00
-1.88066587e-01 4.81830686e-01 -1.05330873e+00 1.12935066e-01
7.64614224e-01 1.01882152e-01 8.30428958e-01 -8.69791090e-01
-5.44475377e-01 -2.05394298e-01 -5.42423308e-01 -9.44175899e-01
1.94137022e-01 -7.67752886e-01 -9.40336436e-02 -1.34837413e+00
3.75471383e-01 -8.41381550e-02 -3.74262393e-01 4.32972223e-01
-6.83847517e-02 5.41097403e-01 5.06368041e-01 1.75321922e-01
-5.09756029e-01 8.86798680e-01 1.68307412e+00 -4.97925550e-01
-5.89855872e-02 -6.36735037e-02 -7.62978613e-01 5.94899654e-01
5.01577377e-01 -2.01336443e-01 -6.38381064e-01 -1.53528661e-01
8.81820079e-03 6.88301146e-01 5.93468845e-01 -1.04200113e+00
7.58730546e-02 4.26835231e-02 3.56858760e-01 -5.48284650e-01
5.90716183e-01 -9.57760751e-01 5.82384944e-01 3.50268394e-01
-2.93336366e-03 -3.20702344e-01 1.79002896e-01 9.34233546e-01
-7.49580979e-01 1.06032938e-01 7.01080143e-01 6.73396606e-03
-6.22421980e-01 3.67457479e-01 -3.67840499e-01 -9.68127139e-03
1.18276358e+00 -2.36077115e-01 -2.68379360e-01 -9.38940465e-01
-8.89095664e-01 3.41067791e-01 2.55090266e-01 5.79709113e-01
7.52935171e-01 -1.94821393e+00 -7.31976092e-01 3.06507230e-01
1.85555190e-01 -7.96123743e-02 7.96003640e-01 9.82345343e-01
-1.44694313e-01 1.55060232e-01 -3.98990840e-01 -8.39317143e-01
-1.38196278e+00 8.20450068e-01 2.49384969e-01 -8.24791938e-02
-2.85751194e-01 5.99330783e-01 4.86143887e-01 -4.41242993e-01
2.32277974e-01 2.08024427e-01 -2.58165687e-01 2.67839491e-01
4.19058591e-01 3.96450549e-01 -4.35742617e-01 -1.01067150e+00
-1.24247327e-01 6.44535303e-01 3.87155265e-01 -1.85034513e-01
1.15354073e+00 -3.22558582e-01 -1.55740395e-01 3.27217221e-01
1.29664195e+00 1.41035184e-01 -1.25589216e+00 -2.34939620e-01
-1.01170826e+00 -4.55439925e-01 -5.88340089e-02 -5.69876015e-01
-1.63646102e+00 9.05033946e-01 6.70931816e-01 1.53991878e-01
1.69866669e+00 -1.84710145e-01 8.80322337e-01 -2.05997139e-01
2.65557349e-01 -8.59336972e-01 3.99851531e-01 1.12504244e-01
9.67342496e-01 -1.15829694e+00 -2.52644509e-01 -3.76155585e-01
-8.34279478e-01 8.59434187e-01 5.08989871e-01 2.04022661e-01
5.39081633e-01 6.32396266e-02 -6.69409409e-02 1.58125367e-02
-6.02407455e-01 -1.40980884e-01 4.11256492e-01 9.14602757e-01
1.90282017e-02 -2.41490640e-02 5.55414148e-02 6.12680733e-01
1.39295757e-01 -9.37964991e-02 3.50310981e-01 7.54643977e-01
-9.49800760e-02 -1.01752090e+00 -7.03000903e-01 -2.94918250e-02
-4.71859038e-01 6.16617650e-02 -1.45164743e-01 9.94499862e-01
4.08015937e-01 1.16878641e+00 -2.51407623e-01 -8.60628486e-01
1.26909897e-01 -3.40499058e-02 4.32408065e-01 -5.81866093e-02
-1.89514071e-01 5.18030167e-01 -2.24080130e-01 -4.94290501e-01
-8.61939013e-01 -4.49474812e-01 -1.08228970e+00 -3.63475651e-01
-3.07387859e-01 -1.62783358e-02 3.79934996e-01 7.80119538e-01
3.69231790e-01 9.58349109e-01 8.50472033e-01 -9.64240193e-01
-3.82393718e-01 -8.56493711e-01 -6.58273160e-01 5.96912742e-01
3.61988783e-01 -7.05082357e-01 -5.12945414e-01 2.73703694e-01] | [11.382927894592285, 1.1058692932128906] |
7b0c7184-5baa-4afc-b36f-91763f215a83 | leaf-counting-with-deep-convolutional-and | 1708.07570 | null | http://arxiv.org/abs/1708.07570v2 | http://arxiv.org/pdf/1708.07570v2.pdf | Leaf Counting with Deep Convolutional and Deconvolutional Networks | In this paper, we investigate the problem of counting rosette leaves from an
RGB image, an important task in plant phenotyping. We propose a data-driven
approach for this task generalized over different plant species and imaging
setups. To accomplish this task, we use state-of-the-art deep learning
architectures: a deconvolutional network for initial segmentation and a
convolutional network for leaf counting. Evaluation is performed on the leaf
counting challenge dataset at CVPPP-2017. Despite the small number of training
samples in this dataset, as compared to typical deep learning image sets, we
obtain satisfactory performance on segmenting leaves from the background as a
whole and counting the number of leaves using simple data augmentation
strategies. Comparative analysis is provided against methods evaluated on the
previous competition datasets. Our framework achieves mean and standard
deviation of absolute count difference of 1.62 and 2.30 averaged over all five
test datasets. | ['Ian Stavness', 'Shubhra Aich'] | 2017-08-24 | null | null | null | null | ['plant-phenotyping'] | ['computer-vision'] | [ 3.36018711e-01 -2.76285142e-01 8.54179710e-02 -2.10655257e-01
-5.44918358e-01 -1.11687779e+00 2.84601212e-01 6.33668154e-02
-3.42765450e-01 4.31960762e-01 -8.03293586e-01 -4.90426689e-01
4.64780003e-01 -8.64460468e-01 -6.53515637e-01 -8.29142749e-01
9.74418744e-02 6.89362586e-01 3.45966011e-01 3.20787162e-01
-7.13633001e-02 7.65274704e-01 -1.28217542e+00 1.12769350e-01
7.18757212e-01 1.35300601e+00 5.56516469e-01 1.20897555e+00
-4.14109200e-01 7.79201269e-01 -6.79419339e-01 -7.76848197e-02
3.62179190e-01 -2.12713972e-01 -9.62980330e-01 4.02084738e-01
8.82233500e-01 -6.56270444e-01 1.07092291e-01 9.66380179e-01
5.95409334e-01 -6.77331030e-01 5.40878892e-01 -1.21831107e+00
-8.62964094e-01 6.01754725e-01 -9.93220568e-01 1.86605677e-01
-4.50558156e-01 4.12420005e-01 8.69857788e-01 -6.28692687e-01
2.26762578e-01 9.12347376e-01 9.60175395e-01 4.06752795e-01
-1.62518752e+00 -3.33469570e-01 6.94352761e-03 -2.03922361e-01
-1.29551256e+00 1.99621590e-03 8.52552578e-02 -3.11314434e-01
3.71569544e-01 2.82119989e-01 5.24908900e-01 7.73974061e-01
-3.76600295e-01 1.03887165e+00 8.97592723e-01 -3.73816133e-01
1.67800069e-01 -3.22533041e-01 3.62442955e-02 8.44903827e-01
6.28189385e-01 -2.09146559e-01 3.47834170e-01 1.91902667e-01
8.60591173e-01 -1.01317354e-01 -1.62073389e-01 -3.01493108e-01
-9.16631758e-01 6.27760828e-01 6.45172894e-01 3.34128618e-01
-4.02804106e-01 2.67536402e-01 4.71012533e-01 -5.27058601e-01
3.00703049e-01 2.14449540e-01 -1.04800868e+00 3.65059406e-01
-1.18376040e+00 1.71260521e-01 1.13077009e+00 1.11581779e+00
2.76876688e-01 3.03444434e-02 -7.21855402e-01 6.32569849e-01
5.21332286e-02 6.56704009e-01 -1.78117022e-01 -1.42151833e+00
-1.12190329e-01 7.63269961e-01 3.45741361e-01 -2.66321510e-01
-4.84411627e-01 -3.10285240e-01 -9.01565194e-01 5.10750115e-01
1.14260340e+00 -1.19395792e-01 -1.27784908e+00 1.65472114e+00
1.20719634e-01 1.52050570e-01 -2.74525315e-01 6.20929062e-01
1.19830441e+00 4.55752790e-01 4.55241859e-01 6.63217157e-02
1.64965582e+00 -8.39612126e-01 -2.47040316e-01 -2.85368353e-01
6.57076716e-01 -7.77323306e-01 1.15599978e+00 4.36502814e-01
-1.14039278e+00 -5.65260053e-01 -9.44430351e-01 -2.78773278e-01
-5.43210983e-01 1.28547347e+00 7.84880102e-01 7.80709386e-01
-8.55468273e-01 7.87716210e-01 -7.65129685e-01 -7.53926694e-01
1.16941702e+00 2.19248831e-01 -9.42355171e-02 -6.45171180e-02
-7.40845874e-02 5.26195884e-01 5.79524219e-01 3.03640872e-01
-1.39882207e+00 -1.02053237e+00 -2.13879138e-01 5.91272354e-01
4.46952999e-01 -4.36448783e-01 1.64130962e+00 -5.09074211e-01
-1.41316676e+00 1.51112866e+00 1.95699483e-01 -5.11614501e-01
2.89050251e-01 4.47471589e-02 1.70002207e-01 -1.20602861e-01
2.02526912e-01 1.06681943e+00 4.63332027e-01 -1.29887640e+00
-8.86525750e-01 -5.26498258e-01 1.50467694e-01 -3.68884414e-01
-7.64813870e-02 -2.43800536e-01 -4.14112210e-01 -3.62738632e-02
-5.14505506e-02 -1.00168955e+00 -1.93420783e-01 6.34963751e-01
-5.15038013e-01 -7.71416575e-02 1.26973259e+00 -7.05653846e-01
4.83883709e-01 -1.62630141e+00 -1.21253945e-01 -6.75580680e-01
2.98786730e-01 8.68166924e-01 -3.87064755e-01 -1.81595474e-01
-6.27823994e-02 2.24993527e-01 -5.84573686e-01 -5.33266485e-01
-2.62054950e-01 3.61903250e-01 -1.46406516e-01 3.00540745e-01
6.56268001e-01 1.06096828e+00 -6.49957657e-01 -5.30685484e-01
3.89125019e-01 4.71405774e-01 -7.55255893e-02 3.25952917e-01
-6.13817573e-01 6.25959516e-01 -1.24830872e-01 1.29491889e+00
1.42518115e+00 -4.67393309e-01 1.46058872e-01 -5.34783900e-01
-4.21060592e-01 -1.12455197e-01 -7.53443480e-01 1.49827814e+00
-4.30422515e-01 5.51147044e-01 4.00094837e-01 -1.02607560e+00
7.40478516e-01 3.52463312e-02 6.04192376e-01 -1.27531216e-01
5.44515312e-01 1.01257585e-01 1.77884161e-01 6.52402490e-02
-9.72093418e-02 5.25108159e-01 3.17985147e-01 2.81849634e-02
6.43780470e-01 -5.83001614e-01 1.02958846e+00 1.03444688e-01
1.32538736e+00 3.66996169e-01 2.79307872e-01 -5.21496594e-01
5.56440294e-01 1.97249651e-01 5.29369175e-01 6.72044873e-01
-6.71848059e-01 7.53798842e-01 7.03405201e-01 -6.37911439e-01
-9.80981767e-01 -9.68220949e-01 -4.27580118e-01 1.11814713e+00
-2.33231843e-01 -4.51685131e-01 -1.00586629e+00 -7.70719945e-01
-4.04839814e-02 6.10806704e-01 -6.86570287e-01 2.76877344e-01
-2.51246065e-01 -1.35911477e+00 8.20559144e-01 8.35925102e-01
9.76216495e-01 -1.31773162e+00 -1.00661194e+00 1.69241011e-01
-1.83284819e-01 -1.73453128e+00 1.07992351e-01 8.55591178e-01
-7.23761618e-01 -1.41651130e+00 -6.74430847e-01 -6.76128626e-01
2.54338831e-01 4.40883219e-01 1.76512504e+00 1.43211529e-01
-1.15291619e+00 1.27367854e-01 -1.26808643e-01 -9.89788830e-01
-2.12541685e-01 3.54220301e-01 -7.18163431e-01 -6.77375376e-01
5.26350677e-01 -4.38494712e-01 -7.55736947e-01 -5.42752072e-02
-7.81893194e-01 -3.23116422e-01 5.25416374e-01 7.40110815e-01
5.95356584e-01 -1.67030498e-01 -3.83069739e-02 -8.38957548e-01
5.64934537e-02 -4.61885631e-02 -1.54876292e+00 5.29009104e-01
-1.54160634e-01 -3.06642145e-01 6.18223488e-01 -2.06545860e-01
-7.55319953e-01 6.37564301e-01 9.83715579e-02 -1.51291015e-02
-4.96885628e-01 3.02690510e-02 -3.60329211e-01 -4.14650798e-01
6.40973210e-01 -1.73358098e-02 -2.54643112e-01 -4.32159245e-01
5.38396597e-01 2.38440350e-01 8.58278573e-01 -4.52131629e-01
6.57022834e-01 7.54663765e-01 4.93251503e-01 -8.26719999e-01
-1.45964801e+00 -5.58057427e-01 -1.20198548e+00 1.30835455e-02
9.03573990e-01 -7.74691820e-01 -1.35499418e+00 8.07917178e-01
-1.44175971e+00 -6.81513190e-01 -5.36004364e-01 -1.38173789e-01
-5.02824306e-01 3.77268732e-01 -6.31668150e-01 -7.14674234e-01
-6.89266801e-01 -9.61402655e-01 1.54585147e+00 6.35529101e-01
3.89471829e-01 -8.15715730e-01 -2.12799489e-01 -6.85784668e-02
3.40106666e-01 5.09846509e-01 5.92456698e-01 -2.04542890e-01
-7.03999460e-01 -5.29424489e-01 -1.19142926e+00 5.89735270e-01
1.72607929e-01 7.06283927e-01 -1.35636950e+00 -8.09604228e-02
-2.50122070e-01 -7.62049556e-01 1.17967725e+00 9.70396340e-01
1.99422920e+00 4.14648801e-01 -4.00177747e-01 8.17935646e-01
1.61555076e+00 -2.96223581e-01 5.24350941e-01 -7.26490691e-02
8.06973279e-01 2.57037967e-01 3.11869234e-01 6.21120095e-01
8.35684016e-02 1.78020000e-01 1.22370124e+00 -3.53398323e-01
-3.80968034e-01 3.60935837e-01 -4.31725800e-01 -8.52814317e-02
-2.36186296e-01 -5.65013528e-01 -9.01298761e-01 5.49804747e-01
-1.46088564e+00 -7.79614389e-01 -6.36346936e-01 2.02287531e+00
7.39758670e-01 1.76745579e-02 2.29709834e-01 2.22343639e-01
5.32470345e-01 -1.97310094e-02 -6.89470172e-01 -9.00319591e-02
-3.66833031e-01 5.78271210e-01 1.19552064e+00 1.45606980e-01
-1.69802272e+00 1.26281524e+00 6.79360580e+00 4.87738729e-01
-1.04907405e+00 7.69649670e-02 9.14068639e-01 3.42358381e-01
7.26482630e-01 1.47315368e-01 -1.10362566e+00 1.12792067e-01
7.09201694e-01 3.74898732e-01 4.20723706e-01 1.01340890e+00
-2.10007891e-01 -3.83882761e-01 -8.41810107e-01 5.80844343e-01
-2.69572198e-01 -1.11037862e+00 -2.40064159e-01 -1.38696522e-01
4.95858490e-01 3.34467739e-01 -2.14435115e-01 3.65807801e-01
8.52619469e-01 -1.10828984e+00 2.85929531e-01 -5.74413463e-02
9.32060003e-01 -1.35324568e-01 6.17244661e-01 4.15870249e-01
-1.45351136e+00 -1.83137417e-01 -9.86825645e-01 2.11406559e-01
-1.23367466e-01 1.01619196e+00 -7.77481735e-01 2.39279270e-01
1.02356315e+00 4.94029999e-01 -1.14468741e+00 1.05132973e+00
-2.99088299e-01 8.99135947e-01 -7.18494892e-01 4.17131692e-01
2.56036639e-01 -1.36562124e-01 -1.54023871e-01 1.22872603e+00
3.02727222e-01 -1.49568141e-01 2.81485260e-01 1.43279624e+00
-3.70541513e-01 -3.42072934e-01 -3.46607089e-01 -4.51705694e-01
3.29363644e-01 2.24949980e+00 -1.30446315e+00 -3.56848955e-01
-2.41438180e-01 1.22643423e+00 3.50458175e-01 -1.58095658e-01
-9.72305655e-01 -1.72189295e-01 1.06174491e-01 -1.12096913e-01
7.43520319e-01 -1.80370912e-01 -4.82821703e-01 -7.56708980e-01
-3.15377116e-01 -3.69746238e-01 2.06322238e-01 -7.76218653e-01
-1.24515057e+00 2.33003393e-01 -2.77951777e-01 -5.77375591e-01
3.08085561e-01 -1.46013367e+00 -9.45883811e-01 8.83223057e-01
-1.48543704e+00 -1.73684037e+00 -1.22462308e+00 3.44535895e-02
4.67371672e-01 2.98205707e-02 1.33677602e+00 3.53946805e-01
-7.04775691e-01 2.62859732e-01 2.62697581e-02 2.16114700e-01
3.89864862e-01 -1.65610814e+00 7.76939631e-01 8.96014214e-01
2.75083661e-01 -1.14247859e-01 2.50802010e-01 -2.30913579e-01
-1.05325806e+00 -1.40078330e+00 5.69489717e-01 -7.17887938e-01
4.31591749e-01 -2.91817337e-01 -6.78277016e-01 7.34418988e-01
2.34754965e-01 8.63740206e-01 3.01053256e-01 -1.20889410e-01
-2.52098113e-01 -6.02840586e-03 -1.32298076e+00 4.57331657e-01
9.70770180e-01 1.44556249e-02 5.10470688e-01 7.97781646e-01
4.06184733e-01 -3.28443795e-01 -7.36191154e-01 8.14995170e-01
5.69066644e-01 -9.13220167e-01 1.19638419e+00 -4.11941528e-01
4.19944376e-01 -3.61085355e-01 -2.96024114e-01 -1.08114254e+00
-7.02008069e-01 -2.15730578e-01 -1.14681661e-01 1.41789532e+00
2.04142951e-03 2.11459860e-01 1.07449186e+00 -9.18798223e-02
7.28635192e-02 -2.52164662e-01 -3.37096542e-01 -7.46223927e-01
3.33671778e-01 9.00441259e-02 3.43717694e-01 4.84245628e-01
-1.11670959e+00 2.62206495e-01 4.60805893e-02 3.65297586e-01
8.64174306e-01 2.30625242e-01 7.81902790e-01 -1.66700494e+00
-1.16577176e-02 -4.76615906e-01 -1.61227927e-01 -7.19884157e-01
6.55865073e-02 -9.18822587e-01 2.05169097e-01 -1.52976298e+00
4.74528342e-01 -8.25897232e-02 1.29760072e-01 5.75439394e-01
-2.99580932e-01 5.22171795e-01 3.61220300e-01 -1.16810739e-01
-3.31060708e-01 8.75806585e-02 1.11238837e+00 -3.77281785e-01
2.53789097e-01 2.63918877e-01 -5.86129010e-01 9.59705412e-01
1.27754736e+00 -4.29344356e-01 -5.69763072e-02 -7.11198092e-01
-4.37367231e-01 -5.47059417e-01 9.41397607e-01 -1.30800128e+00
-3.19359064e-01 4.81995102e-03 8.52325022e-01 -1.09641349e+00
3.59043002e-01 -7.88799942e-01 -4.44144398e-01 7.57374108e-01
1.47823885e-01 -3.81159931e-01 6.34682775e-01 2.22670987e-01
4.95657086e-01 -6.61477923e-01 1.27943015e+00 -5.26928663e-01
-6.52398407e-01 4.36755091e-01 -3.05580676e-01 5.35003059e-02
9.03686583e-01 -9.96222198e-02 -5.87266445e-01 1.50352925e-01
-6.52274072e-01 1.47125095e-01 8.46430485e-04 4.84622717e-02
-1.68563217e-01 -1.06038511e+00 -8.53635192e-01 -1.12037525e-01
3.66484970e-01 3.18094373e-01 -2.32503012e-01 7.98223734e-01
-1.12696683e+00 5.51858127e-01 -5.66790044e-01 -1.02906561e+00
-1.21400154e+00 4.17301208e-01 4.01280493e-01 -6.05563700e-01
-2.59846509e-01 9.88505244e-01 2.45355591e-01 -8.32505882e-01
4.31122839e-01 -8.37285638e-01 -1.34826690e-01 -1.28325731e-01
4.08033252e-01 3.20854127e-01 2.11452186e-01 -4.97445697e-03
-3.17487746e-01 2.50668943e-01 2.28635177e-01 5.63648701e-01
1.33579969e+00 2.46688291e-01 -1.14867873e-01 4.73793983e-01
7.05791175e-01 -5.34369528e-01 -1.48494482e+00 8.41170475e-02
3.38995159e-01 -3.39122802e-01 2.61264354e-01 -1.26226115e+00
-1.37319565e+00 1.13918424e+00 1.16585612e+00 5.45225739e-01
1.18744338e+00 -1.78215399e-01 4.54315156e-01 5.28061509e-01
-1.40228063e-01 -8.00864816e-01 -1.61557496e-01 5.67276061e-01
7.06256986e-01 -1.91191638e+00 1.17494194e-02 -7.88457036e-01
-1.33780211e-01 1.18085492e+00 9.10157025e-01 -3.37149560e-01
6.05110407e-01 9.09804821e-01 -4.49304469e-02 -1.45999327e-01
-5.20985365e-01 -9.85537827e-01 -2.75303841e-01 9.94037509e-01
9.65698600e-01 1.66007027e-01 -1.10580266e-01 3.91490102e-01
4.17612612e-01 3.31423342e-01 4.87589687e-01 1.03407168e+00
-4.93050724e-01 -9.50414777e-01 -2.50491679e-01 3.61884981e-01
-5.43283820e-01 -1.30983055e-01 -8.26580524e-01 8.51619661e-01
3.73878717e-01 6.21716976e-01 5.04623260e-03 4.68868047e-01
2.24888682e-01 -9.47478190e-02 1.04640019e+00 -7.76580632e-01
-4.80635762e-01 -1.60230055e-01 -2.35161796e-01 -4.88455504e-01
-3.33567351e-01 -3.91394615e-01 -8.42681825e-01 -4.65473056e-01
-6.21524870e-01 -8.21714878e-01 8.40033889e-01 6.47523046e-01
1.26609683e-01 9.84470248e-01 1.28036603e-01 -1.11931634e+00
-7.03009665e-01 -1.22207034e+00 -8.33553612e-01 2.10591540e-01
-1.46045983e-01 -5.05518675e-01 -1.14709688e-02 4.24921632e-01] | [9.109793663024902, -1.4988937377929688] |
d89b1c1b-303e-4566-9a2f-9e57b965f0ed | self-supervised-3d-human-pose-estimation-in | 2210.04514 | null | https://arxiv.org/abs/2210.04514v1 | https://arxiv.org/pdf/2210.04514v1.pdf | Self-Supervised 3D Human Pose Estimation in Static Video Via Neural Rendering | Inferring 3D human pose from 2D images is a challenging and long-standing problem in the field of computer vision with many applications including motion capture, virtual reality, surveillance or gait analysis for sports and medicine. We present preliminary results for a method to estimate 3D pose from 2D video containing a single person and a static background without the need for any manual landmark annotations. We achieve this by formulating a simple yet effective self-supervision task: our model is required to reconstruct a random frame of a video given a frame from another timepoint and a rendered image of a transformed human shape template. Crucially for optimisation, our ray casting based rendering pipeline is fully differentiable, enabling end to end training solely based on the reconstruction task. | ['Bernhard Kainz', 'Athanasios Vlontzos', 'Benjamin Hou', 'Luca Schmidtke'] | 2022-10-10 | null | null | null | null | ['3d-human-pose-estimation'] | ['computer-vision'] | [ 4.72817183e-01 1.36096641e-01 2.97825873e-01 -2.76609600e-01
-6.19099140e-01 -3.06946248e-01 5.46966434e-01 -2.52408236e-01
-8.07759285e-01 4.48819041e-01 -8.75525996e-02 -1.35205641e-01
2.40353778e-01 -4.25001770e-01 -7.51169622e-01 -2.61224568e-01
-1.76054135e-01 8.72187972e-01 4.65043604e-01 -6.96796924e-02
-1.84980277e-02 5.48965633e-01 -1.46425569e+00 -2.94168025e-01
1.39384195e-01 8.95053506e-01 1.25514135e-01 1.25866520e+00
4.92360473e-01 2.71364212e-01 -4.57636952e-01 -4.72013742e-01
5.00018418e-01 -3.17613155e-01 -4.56909478e-01 7.87065744e-01
6.94431067e-01 -4.74181265e-01 -4.58973408e-01 4.86325473e-01
8.42823327e-01 4.54869330e-01 4.00953114e-01 -9.64905620e-01
3.66645247e-01 -5.04811764e-01 -4.77815866e-01 1.03580087e-01
8.88528526e-01 2.75243014e-01 4.40095842e-01 -6.65384471e-01
1.04909074e+00 1.09603488e+00 7.53882468e-01 6.78896010e-01
-1.21269715e+00 -3.84657905e-02 -9.05696116e-03 8.45772959e-03
-1.16962814e+00 -5.32667339e-01 7.48203754e-01 -6.69919431e-01
8.08855951e-01 3.01130593e-01 1.14657271e+00 9.44581509e-01
1.26363933e-01 5.43956161e-01 8.30391824e-01 -4.19298351e-01
2.11128935e-01 -3.43455046e-01 -3.33711147e-01 9.38748360e-01
-1.50745779e-01 1.02320239e-01 -4.32548791e-01 -4.39069979e-02
1.34076405e+00 2.53355433e-03 -3.05271536e-01 -1.00573254e+00
-1.20286012e+00 3.93505037e-01 6.25568554e-02 -3.32344770e-01
-5.74670613e-01 3.03427845e-01 1.40875936e-01 1.61420196e-01
3.97290319e-01 -7.81039074e-02 -3.48263949e-01 -4.40001577e-01
-8.83012414e-01 5.30382633e-01 6.14042580e-01 6.93383873e-01
3.58812064e-01 9.21327472e-02 3.68310586e-02 2.44551286e-01
1.94191068e-01 5.87254703e-01 1.09709084e-01 -1.29788387e+00
2.92315662e-01 2.89606929e-01 3.33212733e-01 -8.27457964e-01
-4.53606367e-01 -2.94584811e-01 -4.09162104e-01 5.87876976e-01
5.52421272e-01 -3.90682034e-02 -9.00384486e-01 1.40965176e+00
1.03106213e+00 5.12282193e-01 -4.34875399e-01 1.24944150e+00
5.77384114e-01 1.37783006e-01 -2.72159219e-01 -1.89191759e-01
1.24109709e+00 -6.98384821e-01 -2.93135971e-01 -5.31135023e-01
1.27527714e-01 -5.67269325e-01 7.01684952e-01 5.44608831e-01
-1.42561162e+00 -4.68773246e-01 -9.52396691e-01 -2.09175810e-01
2.58477479e-01 -2.96270121e-02 3.83509606e-01 6.49208128e-01
-7.15141952e-01 6.07868731e-01 -1.08952916e+00 -3.26109201e-01
1.30003661e-01 4.60354120e-01 -8.54405999e-01 2.72082491e-03
-6.50180399e-01 1.01390493e+00 -7.64517172e-04 2.08404943e-01
-6.81728363e-01 -3.84732068e-01 -1.05183554e+00 -3.59886736e-01
6.47484004e-01 -1.25750411e+00 1.20644879e+00 -5.88833451e-01
-1.74002326e+00 1.35978329e+00 -2.97098160e-01 -4.06444997e-01
1.20188272e+00 -4.99449134e-01 2.99399376e-01 2.70008326e-01
2.49010772e-02 2.11855352e-01 9.80649412e-01 -1.02919281e+00
-2.62781620e-01 -8.73684049e-01 7.29944855e-02 6.45090342e-01
5.52893877e-01 -7.40602911e-02 -1.03016448e+00 -5.93973696e-01
3.43509674e-01 -1.16080475e+00 -4.94834989e-01 3.86018425e-01
-2.39148319e-01 2.55560786e-01 4.96766567e-01 -1.02412963e+00
5.79029799e-01 -1.89173055e+00 6.57475293e-01 2.84172863e-01
1.14971213e-01 2.03123152e-01 2.20771119e-01 -1.41025513e-01
9.16344225e-02 -5.26536286e-01 -2.27488443e-01 -6.77834988e-01
-6.99443296e-02 1.25995383e-01 1.14623256e-01 9.44384694e-01
-2.97165364e-02 8.01086783e-01 -9.11045372e-01 -4.58639055e-01
7.73479939e-01 8.06902409e-01 -6.29723489e-01 5.12232959e-01
-1.69625029e-01 1.08601284e+00 -2.24039301e-01 3.81758302e-01
2.61801302e-01 -4.00748588e-02 2.00970657e-02 -4.72744107e-02
9.19598863e-02 5.03760315e-02 -1.60335445e+00 2.26049757e+00
-2.52498865e-01 4.41790134e-01 3.46748263e-01 -8.96861434e-01
4.29660261e-01 4.88215715e-01 7.97140002e-01 -3.61500770e-01
5.12885511e-01 -5.85575365e-02 -3.57342124e-01 -6.08339369e-01
2.79459745e-01 -2.76107043e-01 -5.68869673e-02 4.33003992e-01
-4.89635766e-03 -2.89277732e-01 -1.57136142e-01 -1.33254543e-01
1.08330238e+00 8.27630162e-01 2.85400808e-01 4.34492379e-01
3.84816766e-01 -6.60532638e-02 4.58942890e-01 3.29965800e-01
-2.18378365e-01 1.15745497e+00 8.76875147e-02 -5.84466100e-01
-1.16906703e+00 -1.24900603e+00 1.31521463e-01 6.05969727e-01
6.84888884e-02 -4.13826525e-01 -6.93087041e-01 -3.42241764e-01
-1.67321816e-01 1.52176440e-01 -4.59399134e-01 8.73847902e-02
-1.06308305e+00 -2.30861887e-01 1.39426365e-01 3.66635054e-01
2.31171548e-01 -6.98002577e-01 -1.24188340e+00 2.42970333e-01
-3.06347698e-01 -1.40301919e+00 -5.75497091e-01 -2.15961948e-01
-8.83399427e-01 -1.16702330e+00 -9.38040078e-01 -4.91676480e-01
6.60681486e-01 1.39141530e-01 1.19186437e+00 1.50445625e-01
-6.25678718e-01 7.69229889e-01 1.13255210e-01 -1.19288117e-01
-1.80258945e-01 -4.47429270e-01 1.38500541e-01 1.63537115e-02
-1.77936867e-01 -6.35190129e-01 -6.78553045e-01 2.80576468e-01
-4.60306913e-01 1.65286183e-01 1.43006086e-01 6.23842657e-01
7.45110929e-01 -3.38495076e-01 -3.23081881e-01 -4.62412983e-01
2.00983167e-01 1.11476414e-01 -6.40408695e-01 2.37593874e-02
2.30650827e-01 -2.79985607e-01 7.55216852e-02 -3.94470304e-01
-6.75466239e-01 8.46308351e-01 -2.99876720e-01 -7.36918807e-01
-2.31741637e-01 1.16852649e-01 -1.36293814e-01 -2.99067274e-02
6.26222551e-01 1.08535610e-01 4.48811591e-01 -3.82136613e-01
2.46774584e-01 1.84194878e-01 1.19150352e+00 -5.48055470e-01
8.04460347e-01 7.55205095e-01 3.36143643e-01 -9.88486707e-01
-5.23942888e-01 -6.44954681e-01 -1.28234684e+00 -5.87724805e-01
1.16264021e+00 -9.05966818e-01 -1.06168342e+00 2.67041057e-01
-1.22491205e+00 -3.91598284e-01 -2.24333256e-01 7.29650795e-01
-1.08139360e+00 5.22891283e-01 -1.99334413e-01 -8.17073762e-01
-1.46214738e-01 -1.14591813e+00 1.55424428e+00 -1.01485319e-01
-3.47153485e-01 -9.71563041e-01 1.51222497e-01 6.49974644e-01
-1.91115886e-01 9.30796504e-01 2.04733416e-01 1.86258629e-01
-5.98307014e-01 -5.73560715e-01 3.69390696e-01 8.89451951e-02
-6.99475184e-02 -2.73399204e-01 -8.85673583e-01 -2.79404044e-01
1.48493409e-01 -5.05012162e-02 3.63866895e-01 5.73100865e-01
6.93341434e-01 6.36660829e-02 -2.00136676e-01 8.02435338e-01
1.10300100e+00 -1.94403529e-01 4.15479183e-01 3.32526237e-01
8.68561804e-01 7.17399240e-01 4.61151928e-01 4.99676436e-01
4.88265902e-01 1.17194033e+00 3.39493424e-01 -5.38416915e-02
-1.12378843e-01 -1.45471632e-01 2.47343019e-01 2.29243025e-01
-6.59657598e-01 2.02156454e-01 -8.44449162e-01 1.91306859e-01
-1.87338889e+00 -9.19198990e-01 -1.35155499e-01 2.90452218e+00
4.37170297e-01 2.07305908e-01 4.29322302e-01 3.87557477e-01
4.45670813e-01 -2.05212668e-01 -6.24672115e-01 1.18885428e-01
2.92525470e-01 1.94978520e-01 3.81733119e-01 8.16291928e-01
-1.12807751e+00 7.13429093e-01 6.27889061e+00 2.41076318e-03
-1.04933202e+00 5.35605848e-02 1.73609689e-01 -3.60825717e-01
1.38728872e-01 -3.34764794e-02 -3.68961096e-01 1.29300699e-01
6.03838801e-01 4.97054383e-02 2.27295130e-01 5.77429712e-01
3.22284043e-01 -3.69836390e-01 -1.20250213e+00 1.25856686e+00
2.50791043e-01 -1.08752036e+00 -4.38087523e-01 1.77440435e-01
3.57104152e-01 -2.58932173e-01 -1.58309191e-01 -1.25551164e-01
-2.24341676e-02 -8.84970129e-01 8.72957349e-01 5.96733272e-01
7.05866873e-01 -5.19925535e-01 2.42114052e-01 6.38107300e-01
-1.01622558e+00 3.57233793e-01 2.05398332e-02 -2.93537021e-01
8.07238400e-01 3.12027991e-01 -7.85909951e-01 3.58076632e-01
5.25793314e-01 5.13411701e-01 -2.82895952e-01 1.13687205e+00
-1.24049038e-01 9.76109132e-02 -4.74846691e-01 6.54738843e-01
-2.05812097e-01 -3.69044691e-01 9.00073171e-01 8.07001472e-01
1.92685351e-01 2.85200655e-01 5.28844297e-01 2.48472437e-01
3.85680735e-01 -2.32012749e-01 -6.39141977e-01 5.62245727e-01
-1.98210329e-01 1.03324127e+00 -9.18433666e-01 -3.37118924e-01
-1.58295199e-01 1.43063378e+00 5.71368821e-02 1.74362496e-01
-8.23631644e-01 1.45310014e-01 6.46377861e-01 6.05402827e-01
2.60801673e-01 -8.42868507e-01 -2.24742576e-01 -1.42802346e+00
4.02834117e-01 -6.48696780e-01 2.72139996e-01 -8.47829223e-01
-6.27268434e-01 4.37367886e-01 8.68887678e-02 -1.46080494e+00
-7.97147393e-01 -5.01183510e-01 -3.24521542e-01 7.84155846e-01
-9.97700453e-01 -9.43397880e-01 -4.59870815e-01 7.99016476e-01
5.21689117e-01 1.05204061e-01 6.60429955e-01 2.16369495e-01
-2.49712929e-01 1.03901334e-01 -5.56734025e-01 1.42002597e-01
5.38225234e-01 -1.24794245e+00 7.01599300e-01 8.28706145e-01
2.03685984e-01 3.02021652e-01 9.96470928e-01 -6.38771057e-01
-1.77445912e+00 -5.34522951e-01 8.20225775e-01 -1.11041331e+00
1.15318246e-01 -4.86593574e-01 -4.77370203e-01 7.94938564e-01
-2.80172527e-01 3.47111255e-01 2.41415396e-01 -1.82492718e-01
1.73606694e-01 1.89932466e-01 -1.07224905e+00 6.07508838e-01
1.20502412e+00 -2.85241783e-01 -5.81847250e-01 3.67743671e-01
2.34048501e-01 -1.24666190e+00 -7.81916797e-01 1.94202155e-01
7.73846447e-01 -1.02247775e+00 1.49017727e+00 -4.74817365e-01
1.06652975e-01 -3.52873832e-01 -2.75709797e-02 -9.51285660e-01
4.63847490e-03 -8.26558352e-01 -2.93792367e-01 4.66949314e-01
-4.16920751e-01 -1.65459394e-01 1.20383453e+00 1.04372585e+00
1.21145800e-01 -4.42551851e-01 -1.21055615e+00 -5.41160345e-01
-4.60900366e-01 -6.06836021e-01 1.92954708e-02 4.75512296e-01
-4.55978125e-01 4.28185523e-01 -6.02418303e-01 3.11014533e-01
9.98877525e-01 -5.95975779e-02 1.42268419e+00 -1.35248649e+00
-5.89081347e-01 2.03143265e-02 -9.74169612e-01 -1.15401173e+00
9.84164327e-03 -4.52499747e-01 1.89293131e-01 -1.49316251e+00
-1.90847278e-01 -1.43162385e-02 4.49604660e-01 -1.97802391e-02
-2.18009308e-01 5.33309221e-01 2.55869567e-01 8.27461407e-02
-4.23633665e-01 3.32427591e-01 1.28714299e+00 2.37051532e-01
-1.19975328e-01 4.93978500e-01 5.13394177e-02 8.88700962e-01
1.88187256e-01 -3.30887228e-01 -5.18283606e-01 -4.85648453e-01
3.50738913e-02 5.98488808e-01 9.36834157e-01 -9.07288134e-01
4.86776419e-02 -1.80522949e-02 6.65970266e-01 -5.35239995e-01
9.18263912e-01 -9.48988259e-01 4.91850674e-01 5.26892006e-01
8.45595002e-02 3.52820545e-01 -9.86662041e-03 5.53621650e-01
2.34795392e-01 -4.23491709e-02 5.94075322e-01 -5.09994030e-01
-7.26376832e-01 4.82695192e-01 -1.13170035e-01 1.79101169e-01
1.08232820e+00 -7.23348856e-01 5.57402313e-01 -4.99361902e-01
-1.33299387e+00 5.79747278e-03 7.58471489e-01 3.33539993e-01
8.18470001e-01 -1.18360627e+00 -7.15665340e-01 4.75285769e-01
-3.13938469e-01 4.86631215e-01 1.82547837e-01 7.99755812e-01
-7.32010305e-01 1.09541439e-01 -2.36200199e-01 -1.05136943e+00
-1.54181743e+00 2.68601269e-01 5.27524471e-01 -1.15722403e-01
-1.09142828e+00 5.32136440e-01 -6.03176595e-04 -4.60937291e-01
2.53151238e-01 -8.38484317e-02 5.96972406e-02 -3.38972628e-01
5.50430834e-01 5.30280173e-01 2.48730093e-01 -1.05874193e+00
-3.57907236e-01 8.90587926e-01 4.36962932e-01 -6.64798021e-01
1.23997414e+00 -1.47366434e-01 3.20550740e-01 3.84386629e-01
1.15091395e+00 -2.20866337e-01 -1.77990806e+00 -1.33536845e-01
-3.35040778e-01 -8.86684775e-01 -6.45051673e-02 -3.10289413e-01
-8.81627440e-01 7.70087183e-01 5.98777115e-01 -3.77726465e-01
7.78766155e-01 -8.97911787e-02 6.97844326e-01 2.29175285e-01
6.69891357e-01 -9.53203261e-01 1.19826078e-01 4.11126852e-01
8.59142721e-01 -1.16290402e+00 2.80995965e-01 -4.55635220e-01
-5.51162720e-01 9.68056917e-01 3.37737620e-01 -4.50227201e-01
6.23573422e-01 1.67085886e-01 1.42375320e-01 -9.07250941e-02
-4.33545202e-01 -2.92800218e-01 5.85117161e-01 8.55060577e-01
2.96093106e-01 -2.78945923e-01 -9.67394039e-02 3.66659276e-02
-3.14381808e-01 1.59949526e-01 3.55141371e-01 1.30972862e+00
-2.14191005e-01 -9.30657923e-01 -5.57743788e-01 -5.99387139e-02
-3.86897445e-01 4.07546669e-01 -1.64201185e-01 6.14008546e-01
-1.00901902e-01 4.22358274e-01 1.28697038e-01 -9.62929428e-02
6.64431810e-01 2.24273838e-02 1.00985539e+00 -6.81908846e-01
-4.14550483e-01 4.32250470e-01 9.67252180e-02 -8.49090636e-01
-5.35369217e-01 -9.80241895e-01 -1.17645144e+00 -1.27418071e-01
3.84579562e-02 -1.98456764e-01 8.93482804e-01 1.00923276e+00
1.47180915e-01 3.39021206e-01 3.20366323e-01 -1.64850211e+00
-2.60258585e-01 -3.32283407e-01 -2.75936395e-01 5.71260929e-01
5.27438700e-01 -7.99862623e-01 2.23382246e-02 5.02068579e-01] | [7.122154712677002, -1.0201588869094849] |
de7833b7-4877-46fb-9d47-51f92e494382 | proknow-process-knowledge-for-safety | 2305.08010 | null | https://arxiv.org/abs/2305.08010v2 | https://arxiv.org/pdf/2305.08010v2.pdf | ProKnow: Process Knowledge for Safety Constrained and Explainable Question Generation for Mental Health Diagnostic Assistance | Current Virtual Mental Health Assistants (VMHAs) provide counseling and suggestive care. They refrain from patient diagnostic assistance because they lack training in safety-constrained and specialized clinical process knowledge. In this work, we define Proknow as an ordered set of information that maps to evidence-based guidelines or categories of conceptual understanding to experts in a domain. We also introduce a new dataset of diagnostic conversations guided by safety constraints and Proknow that healthcare professionals use. We develop a method for natural language question generation (NLG) that collects diagnostic information from the patient interactively. We demonstrate the limitations of using state-of-the-art large-scale language models (LMs) on this dataset. Our algorithm models the process knowledge through explicitly modeling safety, knowledge capture, and explainability. LMs augmented with ProKnow guided method generated 89% safer questions in the depression and anxiety domain. The Explainability of the generated question is assessed by computing similarity with concepts in depression and anxiety knowledge bases. Overall, irrespective of the type of LMs augmented with our ProKnow, we achieved an average 82% improvement over simple pre-trained LMs on safety, explainability, and process-guided question generation. We qualitatively and quantitatively evaluate the efficacy of the proposed ProKnow-guided methods by introducing three new evaluation metrics for safety, explainability, and process knowledge adherence. | ['Amit Sheth', 'Ashwin Kalyan', 'Vipula Rawte', 'Misagh Soltani', 'Manas Gaur', 'Kaushik Roy'] | 2023-05-13 | null | null | null | null | ['question-generation'] | ['natural-language-processing'] | [ 2.00303406e-01 1.25700128e+00 -3.07560444e-01 -5.59845090e-01
-8.97978425e-01 -5.22277176e-01 3.43347102e-01 9.01079655e-01
3.79753509e-03 7.59852290e-01 7.76782751e-01 -6.84824228e-01
-7.79414237e-01 -7.47904360e-01 -2.42089137e-01 1.82618737e-01
2.20602825e-01 1.05657375e+00 -2.64780879e-01 -2.81752974e-01
6.76455274e-02 9.89043191e-02 -9.89965260e-01 9.62251663e-01
1.74388051e+00 5.13304830e-01 -1.28682360e-01 6.86166167e-01
-3.97032380e-01 1.20109975e+00 -3.20630014e-01 -8.07582796e-01
-4.11322773e-01 -6.01530492e-01 -1.39397669e+00 -9.28681493e-02
1.59960717e-01 -5.93001544e-01 3.91333364e-02 5.40782988e-01
3.93606812e-01 -2.16304392e-01 6.67594373e-01 -1.47182536e+00
-8.14939618e-01 8.14600945e-01 4.61228430e-01 -3.37955981e-01
1.05325818e+00 2.30226144e-01 5.88663638e-01 -4.20159310e-01
9.15513933e-01 1.39722478e+00 8.38862360e-01 1.28358126e+00
-1.39564717e+00 -6.54539645e-01 3.18578571e-01 3.18774551e-01
-1.01596713e+00 -2.24282816e-01 4.60982099e-02 -9.13597465e-01
1.11325717e+00 4.25641447e-01 7.46441424e-01 1.20379508e+00
2.69052535e-01 3.29255581e-01 7.83990026e-01 -1.40301570e-01
6.72119617e-01 7.68570781e-01 8.00501645e-01 8.59373987e-01
3.46015036e-01 -9.53668579e-02 -5.20957947e-01 -8.44948471e-01
4.49501693e-01 7.62885287e-02 -3.84531200e-01 8.59073997e-02
-5.51650524e-01 9.47291613e-01 1.11802652e-01 -3.13893780e-02
-5.15722275e-01 -4.00929332e-01 3.25579673e-01 9.01490971e-02
1.33568972e-01 6.41657472e-01 -4.83892888e-01 -2.59902298e-01
-4.80787873e-01 2.71187425e-01 1.40485275e+00 1.12663007e+00
3.68365169e-01 -5.36578774e-01 -8.80923629e-01 5.10846555e-01
4.15984780e-01 4.96215373e-01 2.21554965e-01 -1.17180705e+00
1.95266709e-01 1.10483861e+00 4.06526297e-01 -7.76498020e-01
-6.26766205e-01 -1.02444757e-02 -6.20997429e-01 -5.45587718e-01
-8.81217793e-02 -5.66790521e-01 -7.48328030e-01 1.62771416e+00
2.47425601e-01 5.16708121e-02 4.10986245e-01 2.48115450e-01
1.24665582e+00 4.28639621e-01 7.56692886e-01 -5.64923584e-01
1.39594686e+00 -6.67496443e-01 -1.43603313e+00 -9.80641544e-02
9.94357884e-01 -2.23391622e-01 8.57163072e-01 5.04077017e-01
-1.01635993e+00 -1.84904575e-01 -5.21736801e-01 1.47666276e-01
-3.08011342e-02 -1.76204443e-01 6.79606974e-01 9.94851470e-01
-1.34431338e+00 4.56790924e-01 -5.66369474e-01 -7.09249437e-01
4.82787758e-01 2.08551720e-01 -2.14534506e-01 -7.87499696e-02
-1.23957622e+00 9.25311446e-01 7.80498013e-02 -3.69291723e-01
-9.89891052e-01 -1.38721251e+00 -9.37506080e-01 2.74637729e-01
4.05228883e-01 -1.20484436e+00 1.60179150e+00 -2.67433256e-01
-1.25839198e+00 6.47385120e-01 -2.60107905e-01 -4.49368566e-01
3.89768153e-01 -4.30603057e-01 -5.07911026e-01 4.14438009e-01
2.99642775e-02 6.79048359e-01 3.09492171e-01 -1.29262233e+00
-5.19377530e-01 -2.62149036e-01 2.87449926e-01 -6.21831305e-02
-4.40648705e-01 -1.46385282e-01 -3.14325020e-02 9.77461487e-02
-3.45307648e-01 -5.43617904e-01 -5.43985546e-01 1.77034855e-01
-4.66712654e-01 -1.95300132e-01 2.99892485e-01 -8.38731587e-01
1.62778759e+00 -1.66861129e+00 -1.86542302e-01 2.95432061e-01
5.35225451e-01 4.55085069e-01 3.82326096e-02 8.87136817e-01
-1.33514434e-01 5.03329575e-01 -2.15009689e-01 -6.32841289e-02
4.54828255e-02 1.89172149e-01 -2.43558183e-01 -3.59138459e-01
2.37489298e-01 8.35273445e-01 -1.18402398e+00 -5.37622094e-01
1.92169666e-01 5.95955193e-01 -1.28449261e+00 6.40569746e-01
-4.26845759e-01 4.41270679e-01 -6.16468310e-01 2.44439960e-01
4.73374039e-01 -4.92621481e-01 4.96352285e-01 3.86889800e-02
4.51757878e-01 4.39683437e-01 -8.45156968e-01 1.43857610e+00
-4.63492393e-01 -7.63901472e-02 -1.42241076e-01 -3.48325521e-01
7.12631404e-01 8.00265014e-01 4.70482260e-01 -3.16099018e-01
-3.31942476e-02 -8.40847194e-02 -1.44563690e-01 -1.11405945e+00
4.27463576e-02 -3.17997783e-01 1.87099010e-01 4.80326027e-01
-6.36550710e-02 -2.00335398e-01 -2.51284182e-01 7.91320920e-01
1.41966593e+00 -4.51493561e-01 5.68677843e-01 -2.96573728e-01
5.50349832e-01 1.84824280e-02 3.68604690e-01 9.80508685e-01
-1.99288979e-01 2.19091848e-01 7.04361022e-01 -3.10715050e-01
-3.64675015e-01 -9.39931989e-01 -2.51978844e-01 6.45270228e-01
-3.47404957e-01 -1.01189137e+00 -1.07793283e+00 -7.22917318e-01
-7.51989409e-02 1.32095826e+00 -6.53823674e-01 -5.25975108e-01
1.83979765e-01 -2.78569669e-01 5.53993881e-01 4.83456761e-01
1.67389885e-01 -1.18162000e+00 -5.65426350e-01 4.94100451e-01
-4.40683514e-01 -1.24326658e+00 -2.56934345e-01 -4.34850186e-01
-9.75069880e-01 -1.36502659e+00 -7.61510991e-03 -4.14830685e-01
6.78639770e-01 -3.59916210e-01 9.32635009e-01 2.34994307e-01
-1.92602381e-01 1.08846414e+00 -4.30638969e-01 -5.14307678e-01
-7.12491333e-01 -3.74274313e-01 -1.26262307e-01 -1.86526299e-01
4.20076847e-01 -3.29132259e-01 -7.71900654e-01 -7.76277483e-02
-9.59605634e-01 2.03540877e-01 3.96288559e-02 5.01876116e-01
1.92638248e-01 -7.30345011e-01 8.59115303e-01 -1.51295447e+00
1.45456827e+00 -6.69033229e-01 1.51455924e-01 4.53049600e-01
-9.39320207e-01 3.73699851e-02 2.58628428e-01 -1.51955530e-01
-1.21684515e+00 -1.67273849e-01 -5.17639756e-01 2.49616206e-01
-3.73192072e-01 5.25054216e-01 -4.72197589e-03 2.91041166e-01
1.10465705e+00 -2.15345263e-01 -6.67209998e-02 -1.84257343e-01
5.00216782e-01 9.23088551e-01 2.00050578e-01 -5.52813411e-01
2.60703206e-01 3.39875579e-01 -5.29393017e-01 -3.34985048e-01
-7.98341572e-01 -2.27451667e-01 8.12579319e-02 -4.88835946e-02
1.18252933e+00 -5.16125441e-01 -1.44037342e+00 -2.11985618e-01
-1.41296208e+00 -2.90172458e-01 -4.36627626e-01 1.52291104e-01
-4.76407200e-01 2.63450891e-01 -6.76560998e-01 -1.27727079e+00
-8.81000221e-01 -1.01289701e+00 1.01687372e+00 1.65789068e-01
-9.98371422e-01 -1.05765831e+00 2.54498348e-02 5.86836517e-01
3.93576264e-01 1.66235536e-01 1.61007023e+00 -1.10071564e+00
-1.97244048e-01 -5.56243993e-02 -1.41783372e-01 4.10186164e-02
4.80123639e-01 -4.11618143e-01 -9.72292304e-01 1.49743080e-01
2.19122827e-01 -1.85040891e-01 5.65109812e-02 3.94191235e-01
1.28949201e+00 -7.79502690e-01 -6.28397465e-01 1.38409376e-01
1.09440231e+00 7.09858894e-01 5.02816141e-01 -1.32890150e-01
4.35726583e-01 9.35757935e-01 5.02922654e-01 8.04743230e-01
9.41516995e-01 2.33387321e-01 -1.66941546e-02 2.95028035e-02
3.20349663e-01 -4.67688322e-01 3.29090245e-02 4.56283182e-01
8.50530565e-02 -1.73593145e-02 -1.42476737e+00 5.05353987e-01
-2.10708499e+00 -5.82921326e-01 -1.21187061e-01 1.74064744e+00
1.19784462e+00 -1.62225917e-01 -1.93775937e-01 -2.45762225e-02
3.69584560e-01 -8.82187605e-01 -5.33661246e-01 -7.53879905e-01
5.95643699e-01 3.15775305e-01 -2.45682895e-01 1.12800205e+00
-3.80475283e-01 6.80698752e-01 6.54623508e+00 3.02943021e-01
-2.77839005e-01 2.87992835e-01 7.88727939e-01 1.06947772e-01
-9.28169608e-01 -2.16923609e-01 -7.04728067e-01 1.06931217e-01
1.21137154e+00 -5.16532779e-01 1.61271214e-01 8.56675327e-01
6.46689475e-01 9.37568471e-02 -1.76471901e+00 9.86942887e-01
-9.80327185e-03 -1.65712667e+00 4.89070177e-01 -2.81534731e-01
7.77817070e-01 -9.72055554e-01 -2.39812285e-02 4.50996190e-01
4.64812666e-01 -1.29679501e+00 1.69705734e-01 9.97829258e-01
7.22752929e-01 -5.58930218e-01 7.55222321e-01 4.41006392e-01
-6.13416016e-01 -3.71723264e-01 5.04681394e-02 4.66355234e-02
1.23704068e-01 2.59196758e-01 -1.50495183e+00 6.76450968e-01
4.35578287e-01 3.71240735e-01 -2.53650665e-01 8.12143624e-01
-2.18652070e-01 5.66182196e-01 2.33753175e-01 7.28218928e-02
4.59248275e-02 5.70518337e-02 5.82082830e-02 1.34799373e+00
1.54842526e-01 9.03171122e-01 1.00550950e-01 9.61983740e-01
1.52709201e-01 2.98257202e-01 -7.36430109e-01 -1.01273276e-01
6.62942767e-01 8.73666346e-01 -3.88504006e-02 -6.67281210e-01
-1.43185943e-01 6.66425645e-01 9.18955207e-02 4.05607790e-01
-4.67784911e-01 1.41534686e-01 6.53515935e-01 5.24095833e-01
-6.01942420e-01 4.94520903e-01 -5.78511775e-01 -6.78075850e-01
-3.48639876e-01 -1.32466984e+00 6.33095324e-01 -8.39569032e-01
-1.26027274e+00 9.00331974e-01 1.54952973e-01 -6.18702471e-01
-5.95000565e-01 -2.83670694e-01 -2.84671545e-01 9.00094807e-01
-1.03477442e+00 -8.69731009e-01 -6.04776680e-01 5.82558215e-01
5.36680520e-01 6.67200089e-02 1.41544878e+00 1.16850428e-01
-2.64289826e-01 5.66587567e-01 -5.29719591e-01 -3.77459556e-01
8.14139664e-01 -9.53287899e-01 -2.01231409e-02 -6.03500977e-02
-7.96140432e-01 1.06838799e+00 7.04974174e-01 -1.13536286e+00
-1.22299552e+00 -1.34860539e+00 1.20620823e+00 -9.16947603e-01
4.02129769e-01 -1.89966306e-01 -1.11741030e+00 8.42294395e-01
2.32841343e-01 -8.68984640e-01 1.25259519e+00 8.34910423e-02
-2.23612390e-03 1.23770729e-01 -1.73965502e+00 7.30719805e-01
1.33028817e+00 -8.74104321e-01 -7.06935108e-01 8.05737317e-01
1.13345230e+00 -1.48468092e-01 -9.37856793e-01 4.67463285e-01
1.79900691e-01 -7.07538664e-01 7.31518626e-01 -1.24760008e+00
6.32730067e-01 2.48587906e-01 2.29917824e-01 -1.16064787e+00
-1.46148622e-01 -9.39505339e-01 -9.61543247e-02 8.94748390e-01
6.54406011e-01 -5.87783039e-01 5.90829074e-01 1.68277681e+00
-1.64487749e-01 -8.79666090e-01 -3.15707594e-01 -9.62476656e-02
-2.00147018e-01 -4.95701969e-01 6.88824475e-01 1.13930523e+00
8.94709945e-01 3.07943583e-01 -1.29406184e-01 3.67867619e-01
3.05511653e-01 -3.76391739e-01 8.38026255e-02 -1.34532714e+00
-2.96889752e-01 2.83286758e-02 -9.19999671e-04 -4.45709854e-01
-1.15613058e-01 -8.95678222e-01 -1.65646136e-01 -2.42205024e+00
5.58702886e-01 -1.27791926e-01 1.80889323e-01 7.31079817e-01
-4.85459894e-01 -7.20154941e-01 -9.97282267e-02 -3.33422929e-01
-4.38358694e-01 4.63910580e-01 1.15671992e+00 -4.71104011e-02
-5.73002934e-01 -2.31759503e-01 -1.31738436e+00 7.76818156e-01
7.28257895e-01 -4.55679238e-01 -1.08416152e+00 -2.35380247e-01
3.30004662e-01 7.29183912e-01 1.46856010e-01 -5.97904921e-01
3.49300385e-01 -2.91504890e-01 -1.85652241e-01 -1.73328042e-01
2.94197559e-01 -8.28125358e-01 3.03619713e-01 9.15959895e-01
-7.77877867e-01 -9.84729603e-02 6.00378692e-01 3.68616492e-01
2.10340157e-01 -3.01609606e-01 2.20203564e-01 4.47303355e-02
-1.61969841e-01 1.93425432e-01 -7.61586487e-01 5.57938963e-02
9.43194270e-01 8.06865022e-02 -5.57009220e-01 -9.27389562e-01
-1.32146835e+00 5.61394095e-01 -1.85096398e-01 3.55779320e-01
1.04106057e+00 -8.83913517e-01 -6.23184144e-01 -3.72067504e-02
3.15068245e-01 -3.22104365e-01 5.55534601e-01 7.04572976e-01
-3.78283739e-01 7.28936493e-01 -3.37263718e-02 -2.53226578e-01
-1.21796775e+00 4.71605331e-01 4.80627567e-01 -3.14758360e-01
-4.06291723e-01 6.50921881e-01 4.05545175e-01 -6.65458262e-01
6.21762514e-01 -5.77173054e-01 -5.68113446e-01 -1.57902330e-01
8.38700414e-01 1.69702381e-01 -1.60982367e-02 2.76878327e-01
-3.53221118e-01 2.88461484e-02 -3.52438271e-01 -2.10181981e-01
1.18127823e+00 9.34171453e-02 -2.92327777e-02 -7.52280056e-02
4.02140379e-01 -5.33482432e-01 -5.20263076e-01 6.98911250e-02
2.00868890e-01 -3.26197743e-02 -4.36838090e-01 -1.31932855e+00
-2.29142487e-01 7.72249222e-01 8.16907227e-01 6.38240874e-02
9.26999032e-01 7.61104822e-02 3.94388199e-01 6.74444675e-01
1.44809604e-01 -8.39530289e-01 2.34395683e-01 2.84987092e-01
1.12585151e+00 -9.95072961e-01 -6.36880696e-01 -1.00787926e+00
-1.21820605e+00 5.97604156e-01 9.16852951e-01 5.66987216e-01
7.89437294e-01 1.88269302e-01 2.54298031e-01 -4.38259661e-01
-1.16110122e+00 1.73175395e-01 1.54615976e-02 1.04655075e+00
4.53003287e-01 1.65946543e-01 -5.43089271e-01 1.36505985e+00
-1.38697118e-01 5.19152761e-01 6.99883401e-01 8.78957570e-01
-4.30313051e-01 -1.09754074e+00 -9.28520262e-02 6.75409377e-01
-9.70367789e-02 -3.90557379e-01 -6.94780648e-01 2.22293183e-01
6.06936477e-02 1.57223201e+00 -4.98047769e-01 -5.40909052e-01
7.52305627e-01 5.04114687e-01 1.34625003e-01 -1.08202875e+00
-1.08061314e+00 -4.09201890e-01 6.39411449e-01 -7.30816245e-01
1.26639664e-01 -2.18255028e-01 -1.69069600e+00 -2.48264864e-01
2.25052729e-01 5.16371548e-01 4.35329378e-01 1.04663229e+00
7.46461213e-01 8.39393377e-01 4.97649703e-03 2.73593485e-01
-5.94366193e-01 -8.31374943e-01 1.48684643e-02 5.20917416e-01
2.07452029e-01 -2.35720783e-01 2.33981878e-01 1.05444446e-01] | [9.341038703918457, 7.802137851715088] |
5a486e9c-c0ad-4325-87fb-ea3807fd1176 | improving-weakly-supervised-temporal-action | 2304.07978 | null | https://arxiv.org/abs/2304.07978v1 | https://arxiv.org/pdf/2304.07978v1.pdf | Improving Weakly Supervised Temporal Action Localization by Bridging Train-Test Gap in Pseudo Labels | The task of weakly supervised temporal action localization targets at generating temporal boundaries for actions of interest, meanwhile the action category should also be classified. Pseudo-label-based methods, which serve as an effective solution, have been widely studied recently. However, existing methods generate pseudo labels during training and make predictions during testing under different pipelines or settings, resulting in a gap between training and testing. In this paper, we propose to generate high-quality pseudo labels from the predicted action boundaries. Nevertheless, we note that existing post-processing, like NMS, would lead to information loss, which is insufficient to generate high-quality action boundaries. More importantly, transforming action boundaries into pseudo labels is quite challenging, since the predicted action instances are generally overlapped and have different confidence scores. Besides, the generated pseudo-labels can be fluctuating and inaccurate at the early stage of training. It might repeatedly strengthen the false predictions if there is no mechanism to conduct self-correction. To tackle these issues, we come up with an effective pipeline for learning better pseudo labels. Firstly, we propose a Gaussian weighted fusion module to preserve information of action instances and obtain high-quality action boundaries. Second, we formulate the pseudo-label generation as an optimization problem under the constraints in terms of the confidence scores of action instances. Finally, we introduce the idea of $\Delta$ pseudo labels, which enables the model with the ability of self-correction. Our method achieves superior performance to existing methods on two benchmarks, THUMOS14 and ActivityNet1.3, achieving gains of 1.9\% on THUMOS14 and 3.7\% on ActivityNet1.3 in terms of average mAP. | ['Hongsheng Li', 'Si Liu', 'Liang Wang', 'Linjiang Huang', 'Jingqiu Zhou'] | 2023-04-17 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Zhou_Improving_Weakly_Supervised_Temporal_Action_Localization_by_Bridging_Train-Test_Gap_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Zhou_Improving_Weakly_Supervised_Temporal_Action_Localization_by_Bridging_Train-Test_Gap_CVPR_2023_paper.pdf | cvpr-2023-1 | ['weakly-supervised-temporal-action', 'action-localization', 'action-recognition', 'pseudo-label'] | ['computer-vision', 'computer-vision', 'computer-vision', 'miscellaneous'] | [ 6.01905525e-01 1.14775412e-01 -3.19638878e-01 -5.16530752e-01
-9.16211307e-01 -3.02962065e-01 4.68410432e-01 9.65052471e-03
-4.16552871e-01 9.52416301e-01 1.37103319e-01 1.96660325e-01
-3.78100462e-02 -5.62914371e-01 -6.51224613e-01 -8.15084755e-01
1.59750953e-01 1.99246570e-01 6.45306766e-01 9.78492871e-02
1.91934153e-01 -8.89371186e-02 -1.67914414e+00 6.05017543e-01
1.11909425e+00 1.17496145e+00 5.29729947e-02 9.67442691e-02
-7.04080611e-02 8.91347885e-01 -7.40980506e-01 -3.11552078e-01
2.19234943e-01 -6.23348892e-01 -7.20971584e-01 1.77387819e-01
2.12089494e-01 -2.03759715e-01 -6.96979761e-02 1.14051211e+00
3.04784089e-01 2.00848013e-01 6.86255515e-01 -1.51596522e+00
-1.96121871e-01 7.56632745e-01 -6.27784252e-01 -1.66853204e-01
1.72356099e-01 3.49314421e-01 1.03729773e+00 -7.76029170e-01
4.93040681e-01 1.10197353e+00 4.64989215e-01 8.02995205e-01
-1.07310355e+00 -8.25845718e-01 4.22834963e-01 4.68367368e-01
-1.35732400e+00 -3.26815605e-01 4.82405633e-01 -3.87620270e-01
5.20344615e-01 1.54171869e-01 4.15060073e-01 1.32580221e+00
1.21873207e-01 1.10222208e+00 1.08154786e+00 -3.82573493e-02
4.26719815e-01 -7.16398731e-02 -1.42320201e-01 4.34594810e-01
-1.08269356e-01 -1.52534386e-02 -5.86921811e-01 1.16052166e-01
4.65584517e-01 1.41697139e-01 -2.30668440e-01 3.59994099e-02
-1.38935566e+00 3.37584108e-01 3.44862640e-01 3.89041305e-01
-3.43402177e-01 1.11785240e-01 3.76793355e-01 -1.63926244e-01
4.75403726e-01 2.96401381e-01 -5.48633873e-01 -3.84628028e-01
-9.04240668e-01 8.49310011e-02 1.67671159e-01 8.55124891e-01
6.05547667e-01 -1.75720051e-01 -7.07088053e-01 9.20198679e-01
2.40537763e-01 1.29464984e-01 6.75702989e-01 -9.61907089e-01
5.87311506e-01 8.52923691e-01 3.53701323e-01 -8.41647685e-01
-3.94880712e-01 -4.76821393e-01 -8.05413842e-01 2.41893142e-01
6.00805819e-01 1.41139026e-03 -9.99807715e-01 1.95287764e+00
3.78645629e-01 7.61286318e-01 2.69075390e-02 8.60174537e-01
4.08877462e-01 4.27724153e-01 2.32511729e-01 -5.08374095e-01
1.05958200e+00 -1.31743431e+00 -8.95264149e-01 -3.56421202e-01
9.51703250e-01 -6.33853853e-01 1.15436256e+00 3.98015738e-01
-8.03465188e-01 -8.05094481e-01 -9.06794667e-01 3.52214396e-01
-3.07065155e-02 3.87707293e-01 4.61171716e-01 1.01819023e-01
-6.90611243e-01 8.65829468e-01 -1.01135743e+00 1.95032954e-02
6.17877305e-01 1.31291375e-01 -3.12164724e-01 -2.30851546e-02
-1.28166258e+00 6.31831944e-01 6.98827505e-01 1.90981835e-01
-1.01723289e+00 -4.91125584e-01 -5.86443901e-01 -1.42129868e-01
8.20458889e-01 -1.12755202e-01 1.41589642e+00 -1.11352301e+00
-1.29042900e+00 4.34262216e-01 -8.96226466e-02 -4.54296261e-01
8.35110486e-01 -1.76040515e-01 -6.08185112e-01 -8.35063681e-02
4.37289685e-01 9.16283846e-01 7.19688594e-01 -9.36602056e-01
-1.26326311e+00 -1.92438617e-01 6.04669638e-02 3.54158103e-01
-2.43678391e-01 -3.45827192e-01 -5.36865890e-01 -6.95162594e-01
3.76676410e-01 -1.00928378e+00 -3.78867328e-01 -1.53004825e-01
-4.10642475e-01 -5.08554816e-01 5.08949757e-01 -4.52454776e-01
1.41074634e+00 -2.25307226e+00 -8.57802257e-02 -1.46785766e-01
1.28163293e-01 3.67148668e-01 -5.05928248e-02 -1.58562735e-01
-1.24851070e-01 1.69972345e-01 -3.90615851e-01 -2.98709571e-01
3.91754694e-02 1.21129677e-01 -6.45180941e-02 2.27912858e-01
2.85696805e-01 7.59302199e-01 -1.12997472e+00 -5.71130991e-01
1.39586419e-01 2.05449983e-01 -2.87751377e-01 1.52418777e-01
-5.13653636e-01 6.97602808e-01 -6.03183270e-01 7.21839249e-01
5.64401567e-01 -2.25920692e-01 1.44466115e-02 -2.48236418e-01
4.42769639e-02 1.66459113e-01 -1.31014371e+00 1.76470268e+00
-2.44334921e-01 9.18631777e-02 -4.11777586e-01 -9.00866866e-01
7.73175597e-01 3.31922919e-01 8.39713335e-01 -6.39653862e-01
1.36716679e-01 3.77620101e-01 -1.21020913e-01 -3.53931755e-01
2.32445821e-01 -9.62533578e-02 -8.43494684e-02 3.26960713e-01
-1.38387710e-01 1.92929521e-01 4.37970102e-01 7.86614884e-03
1.38374197e+00 6.15699112e-01 -6.42070919e-02 2.64715761e-01
5.69545567e-01 5.29843941e-02 1.10249555e+00 4.89995986e-01
-3.75143111e-01 5.65455616e-01 5.24614394e-01 -2.00474948e-01
-6.32178068e-01 -9.27143335e-01 -1.00298077e-01 8.21152091e-01
3.00588042e-01 -4.78094578e-01 -7.25637376e-01 -1.26440799e+00
-2.58897334e-01 8.43975127e-01 -4.71495152e-01 -5.48969984e-01
-4.23271686e-01 -7.45693505e-01 5.68319857e-01 6.88271523e-01
7.55974293e-01 -1.28664529e+00 -3.88045490e-01 5.06503761e-01
-5.02255797e-01 -1.13300383e+00 -5.37761211e-01 1.44150257e-01
-9.20312166e-01 -1.20690787e+00 -6.62751675e-01 -4.63259518e-01
8.36932003e-01 1.15755230e-01 8.54627013e-01 -1.06559977e-01
6.37920797e-02 -2.68088639e-01 -6.55528128e-01 -1.32902995e-01
-4.03395355e-01 -1.68479159e-01 1.93319350e-01 2.57178098e-01
4.44373727e-01 -5.11841178e-01 -5.68861723e-01 7.82609463e-01
-8.67903650e-01 3.21107894e-01 7.33709514e-01 9.25808012e-01
9.33123410e-01 4.51729506e-01 6.69428229e-01 -8.65646899e-01
1.77582383e-01 -3.16298306e-01 -4.86594528e-01 2.94544280e-01
-7.69945741e-01 -3.07291821e-02 7.34604895e-01 -6.20498180e-01
-1.20041525e+00 4.03525621e-01 -1.49560377e-01 -4.08052415e-01
-1.81288406e-01 2.88142920e-01 -3.87480974e-01 3.40939820e-01
7.70953536e-01 2.50683963e-01 -2.48111635e-01 -5.20914793e-01
2.62567759e-01 5.65471411e-01 3.36585492e-01 -3.22337747e-01
5.46708465e-01 4.52115238e-01 -1.37062460e-01 -9.56455022e-02
-1.30266762e+00 -3.50013971e-01 -5.74237347e-01 -2.38149807e-01
8.65944982e-01 -8.43610346e-01 -5.21381557e-01 6.96352005e-01
-8.94786179e-01 -5.30760884e-01 -2.94737458e-01 6.14861965e-01
-5.46500087e-01 3.88714164e-01 -4.54338253e-01 -6.44161880e-01
-1.03406884e-01 -1.24881303e+00 1.14129007e+00 3.70491266e-01
-9.47446153e-02 -5.44944882e-01 -2.53909647e-01 5.30438662e-01
2.29913425e-02 1.86913848e-01 4.48818028e-01 -6.04136169e-01
-5.19637465e-01 -1.44307643e-01 -2.18609080e-01 6.73712134e-01
3.63285482e-01 -1.93737745e-01 -8.61283064e-01 -1.42806679e-01
-2.11012319e-01 -4.82080340e-01 8.01219642e-01 2.75498629e-01
1.32524240e+00 -1.64037779e-01 -4.12187248e-01 2.68410951e-01
9.35308754e-01 3.22806448e-01 5.80105901e-01 1.69455335e-01
6.51572287e-01 5.23841202e-01 1.52086759e+00 5.40483296e-01
1.72958940e-01 8.52491498e-01 4.99466658e-01 1.19954325e-01
-1.74124226e-01 -3.96970600e-01 5.79648197e-01 4.93939579e-01
4.92154919e-02 -1.30725861e-01 -7.58941293e-01 3.43217164e-01
-2.16947269e+00 -8.77063870e-01 -1.60688579e-01 2.29501867e+00
1.15161490e+00 5.61294556e-01 -3.53361592e-02 3.03263366e-01
8.54727507e-01 1.19656458e-01 -7.75469780e-01 3.62534463e-01
9.87301767e-02 1.59310177e-02 3.34852993e-01 1.23207197e-01
-1.26258957e+00 1.09903252e+00 5.07363844e+00 1.24260998e+00
-9.55594778e-01 3.04995447e-01 8.02127361e-01 -1.31208912e-01
9.26240608e-02 6.59742281e-02 -9.97363627e-01 9.18414652e-01
7.08419502e-01 3.92669663e-02 7.45564774e-02 8.95324051e-01
4.72739190e-01 -3.61074209e-01 -1.14059317e+00 9.39723134e-01
-4.34129983e-02 -8.41070294e-01 -2.55315393e-01 -2.42673948e-01
9.73959863e-01 -2.39242911e-01 -1.81694046e-01 6.15514338e-01
2.53591567e-01 -7.49512196e-01 7.51914501e-01 4.92819756e-01
7.66546726e-01 -5.77177942e-01 7.92748690e-01 6.22068644e-01
-1.20612895e+00 -1.69863060e-01 -3.83096129e-01 3.90485860e-02
3.20879877e-01 8.20474148e-01 -7.89959371e-01 6.31056726e-01
5.60981631e-01 9.13123369e-01 -6.07214153e-01 1.12028539e+00
-6.83543742e-01 6.28088832e-01 -2.00418591e-01 9.33304131e-02
2.03837231e-01 -2.93195210e-02 1.40121982e-01 7.17969120e-01
4.57850009e-01 -1.09705087e-02 5.31020224e-01 6.15453243e-01
5.52940071e-02 1.59878671e-01 -1.62602201e-01 -6.26703724e-02
3.88838738e-01 1.18664432e+00 -7.35817432e-01 -4.65075761e-01
-2.01498196e-01 1.05431163e+00 4.95134369e-02 1.63095534e-01
-1.37359869e+00 -1.18323699e-01 4.53759909e-01 9.67198908e-02
-2.26984411e-01 4.46582437e-02 -1.87061518e-01 -1.17258132e+00
2.76672810e-01 -8.95344198e-01 4.48035479e-01 -6.60674691e-01
-9.63476539e-01 5.24100006e-01 -8.18096027e-02 -1.79580724e+00
-3.46855484e-02 -1.82841957e-01 -4.14232224e-01 4.96781200e-01
-1.19086361e+00 -9.16950822e-01 -4.79156494e-01 2.36079812e-01
9.29389954e-01 7.28793517e-02 4.53628391e-01 6.12350881e-01
-7.45622158e-01 6.24936461e-01 -1.80122480e-01 -7.52570182e-02
9.52769756e-01 -1.07743371e+00 9.99936610e-02 8.66573811e-01
1.50677621e-01 8.54875222e-02 5.13446033e-01 -8.90962660e-01
-5.21798491e-01 -1.52347875e+00 7.92164624e-01 -3.42274159e-01
4.97919232e-01 -3.59516516e-02 -8.51601303e-01 4.86504942e-01
-4.25390899e-01 1.17959552e-01 4.92133141e-01 -1.16673581e-01
-3.88240293e-02 -2.06335098e-01 -9.91191208e-01 7.42893994e-01
1.49964702e+00 -1.51774168e-01 -4.51493740e-01 6.81436718e-01
7.56771028e-01 -4.10269350e-01 -7.11891353e-01 7.75666714e-01
2.73315549e-01 -9.23016548e-01 6.11763299e-01 -4.04141963e-01
4.81893659e-01 -6.99875593e-01 1.46765873e-01 -1.13498199e+00
-2.77941197e-01 -2.39292443e-01 -5.93465334e-03 1.47080255e+00
6.68746471e-01 -3.65197062e-01 9.98162866e-01 6.20181799e-01
-2.93916643e-01 -9.18103397e-01 -9.47348714e-01 -8.53120327e-01
-6.02666080e-01 -6.93374693e-01 4.76429135e-01 7.29878008e-01
-8.91032740e-02 1.77947253e-01 -5.39398074e-01 -1.92032382e-02
3.85173440e-01 -1.44078940e-01 5.85052788e-01 -9.30201352e-01
-3.13864827e-01 -3.83804083e-01 -3.06858331e-01 -1.10072017e+00
2.21565828e-01 -7.26759791e-01 4.88130808e-01 -1.47548473e+00
4.55961153e-02 -8.02731693e-01 -6.88225806e-01 1.07924092e+00
-4.38345730e-01 3.98907036e-01 -6.84513524e-02 3.09069693e-01
-1.02793288e+00 7.87528455e-01 1.39656007e+00 -1.80028483e-01
-2.83596128e-01 3.90235901e-01 -4.31719869e-01 8.98385704e-01
8.10346544e-01 -6.52112544e-01 -6.48896456e-01 -1.30733028e-01
-7.25416001e-03 -2.85933521e-02 1.70647547e-01 -1.31516230e+00
8.25106129e-02 -4.18352276e-01 1.77974954e-01 -6.31054342e-01
3.78612459e-01 -7.83339918e-01 1.71777397e-01 3.76061350e-01
-4.05506462e-01 -3.26509088e-01 -2.22454458e-01 7.68777132e-01
-3.80880296e-01 -3.09062362e-01 8.33123505e-01 -1.67701766e-01
-9.11207318e-01 4.10069197e-01 -9.27373022e-02 6.12481721e-02
1.33387256e+00 -2.36470655e-01 -2.03661352e-01 -1.01006471e-01
-8.27670395e-01 4.99418288e-01 3.84637803e-01 5.81027031e-01
4.57350910e-01 -1.51555693e+00 -5.30307889e-01 -6.33989722e-02
2.07825765e-01 1.93243727e-01 4.76587802e-01 1.08136761e+00
-1.72575284e-02 1.19314656e-01 -1.13579988e-01 -5.69271028e-01
-1.14808261e+00 4.59519148e-01 1.07716344e-01 -4.16211337e-01
-4.22194034e-01 9.55094635e-01 2.24313065e-01 -2.23268285e-01
3.79682839e-01 -2.75673330e-01 -3.89128327e-01 2.28628457e-01
5.90174258e-01 3.01356226e-01 -9.56137627e-02 -6.35166764e-01
-4.15194690e-01 3.33352894e-01 2.27310695e-02 2.24621519e-02
1.04862154e+00 3.52599956e-02 1.27366006e-01 4.47437435e-01
8.33889067e-01 -2.41082788e-01 -1.64074850e+00 -2.61493325e-01
2.42741257e-01 -5.54148436e-01 -2.15804204e-01 -8.99588525e-01
-1.10228860e+00 7.04199731e-01 6.44613802e-01 -1.86632335e-01
1.30630839e+00 -1.36681631e-01 8.55397642e-01 6.11566529e-02
5.31565428e-01 -1.49389195e+00 4.35879648e-01 3.49480212e-01
5.56713760e-01 -1.32958317e+00 -1.69690475e-01 -5.48711240e-01
-8.37385297e-01 6.94187999e-01 1.14757705e+00 3.25215816e-01
2.87168294e-01 -5.65603301e-02 2.48010308e-02 1.95833519e-01
-7.77212083e-01 -3.32275540e-01 2.71501362e-01 2.41648912e-01
2.41129979e-01 1.46495804e-01 -8.35003674e-01 8.31552207e-01
3.29162955e-01 1.93010673e-01 2.17404976e-01 8.78520668e-01
-4.55768138e-01 -1.39371645e+00 -1.58042163e-01 6.33524716e-01
-4.66405898e-01 9.63158980e-02 -7.52422512e-02 3.36346477e-01
6.45504057e-01 1.02253544e+00 -2.92615116e-01 -8.29313219e-01
3.93057883e-01 1.87673092e-01 5.00852130e-02 -7.84984350e-01
-1.51161104e-01 2.53103465e-01 8.83505568e-02 -7.90986657e-01
-5.26564896e-01 -6.05849087e-01 -1.59061956e+00 2.33111754e-01
-5.95050871e-01 2.23061234e-01 2.63724387e-01 1.03980052e+00
3.80834639e-01 7.48135388e-01 6.78676009e-01 -4.77195829e-01
-6.84099019e-01 -1.14869022e+00 -6.18040621e-01 6.42672479e-01
-3.53372961e-01 -9.93741810e-01 -4.26749051e-01 6.75695017e-02] | [8.503715515136719, 0.6392882466316223] |
f7fb64ee-886c-4607-8ad7-3fa9c8ca6700 | challenges-and-trends-in-user-trust-discourse | 2305.11876 | null | https://arxiv.org/abs/2305.11876v2 | https://arxiv.org/pdf/2305.11876v2.pdf | Challenges and Trends in User Trust Discourse in AI | The Internet revolution in 1990, followed by the data-driven and information revolution, has transformed the world as we know it. Nowadays, what seam to be 10 to 20 years ago, a science fiction idea (i.e., machines dominating the world) is seen as possible. This revolution also brought a need for new regulatory practices where user trust and artificial Intelligence (AI) discourse has a central role. This work aims to clarify some misconceptions about user trust in AI discourse and fight the tendency to design vulnerable interactions that lead to further breaches of trust, both real and perceived. Findings illustrate the lack of clarity in understanding user trust and its effects on computer science, especially in measuring user trust characteristics. It argues for clarifying those notions to avoid possible trust gaps and misinterpretations in AI adoption and appropriation. | ['Paulo Martins', 'Jose Cravino', 'Sonia Sousa'] | 2023-05-05 | null | null | null | null | ['misconceptions'] | ['miscellaneous'] | [-2.15009347e-01 5.94711959e-01 -1.99498013e-01 -4.60334927e-01
2.57797956e-01 -5.32447994e-01 6.59202814e-01 5.86507618e-01
-4.94378924e-01 4.68228191e-01 4.21614796e-01 -7.51337469e-01
1.94260538e-01 -4.29940253e-01 -4.97903436e-01 -3.24690878e-01
5.29673517e-01 -9.20335501e-02 -2.29083344e-01 -3.96300673e-01
4.88196701e-01 1.30175650e-01 -1.24600923e+00 -1.47898421e-01
1.09011292e+00 4.33151156e-01 -2.02286437e-01 3.34142476e-01
4.68475491e-01 1.17657232e+00 -5.77495933e-01 -1.32419038e+00
-1.67908370e-01 -3.02425057e-01 -9.14868414e-01 -9.85909849e-02
-1.68849692e-01 -6.68060005e-01 6.90152077e-03 1.11916149e+00
2.15020820e-01 -6.10846460e-01 2.13882804e-01 -1.67632627e+00
-1.20519197e+00 1.02473080e+00 -3.49868536e-01 -8.58767480e-02
4.83817816e-01 2.40475953e-01 8.14592600e-01 -2.00984567e-01
5.52829027e-01 1.19923806e+00 8.07925761e-01 3.67623776e-01
-1.13800061e+00 -7.40923285e-01 -1.48195580e-01 1.38910830e-01
-1.14807880e+00 -4.49484259e-01 4.62981910e-01 -7.03111231e-01
7.02410281e-01 4.77621019e-01 1.42631257e+00 1.25478041e+00
4.14235234e-01 5.27527452e-01 1.37650180e+00 -7.28766799e-01
1.95176557e-01 1.15851259e+00 7.02195466e-01 3.29667091e-01
1.39251614e+00 1.06987376e-02 -1.81981102e-01 -7.00770259e-01
4.44265932e-01 -1.73864618e-01 -2.80791551e-01 -8.72954950e-02
-1.05404484e+00 9.37451184e-01 2.41692424e-01 1.07488060e+00
-6.74491584e-01 -1.88848421e-01 6.10741854e-01 5.82409561e-01
2.52919465e-01 8.72830570e-01 -2.07634896e-01 -8.48125696e-01
-1.47535816e-01 -1.72141999e-01 1.13676190e+00 5.05896688e-01
2.69969851e-01 -7.43955821e-02 5.26748836e-01 4.05050308e-01
6.57278478e-01 2.66177446e-01 3.08539033e-01 -9.28537011e-01
-5.57517886e-01 7.33496189e-01 4.83079404e-01 -1.30250812e+00
-1.25348419e-01 -3.95833582e-01 -3.94724399e-01 2.26026237e-01
4.14711714e-01 -2.86945581e-01 -8.16933960e-02 1.37712586e+00
1.74719885e-01 -2.63515532e-01 2.32919812e-01 8.38543534e-01
2.55962998e-01 3.24162155e-01 3.00828844e-01 -2.35096544e-01
1.16185272e+00 -3.59282851e-01 -1.00412989e+00 -2.37556279e-01
8.25312972e-01 -7.20359325e-01 1.10418212e+00 5.39681971e-01
-9.30620909e-01 -1.55222610e-01 -1.45058966e+00 1.00484341e-01
-2.59600163e-01 -5.07430017e-01 1.03839684e+00 1.62003553e+00
-9.26327527e-01 4.25300688e-01 -8.20348024e-01 -8.55309427e-01
3.99623722e-01 -6.26372173e-02 -1.32077113e-01 7.14195892e-02
-1.07300019e+00 1.38356328e+00 1.05449051e-01 2.93295532e-01
-2.88055152e-01 -3.93387735e-01 -5.07891715e-01 -3.29151571e-01
1.85845092e-01 -8.03525865e-01 1.18539786e+00 -1.86375391e+00
-1.30202603e+00 7.87676275e-01 6.62521482e-01 -6.26743257e-01
2.45789558e-01 -4.10486698e-01 -6.80602908e-01 -1.84233278e-01
-1.36833355e-01 1.65602311e-01 4.70697194e-01 -1.63658321e+00
-3.93995494e-01 -4.59034771e-01 3.50828052e-01 -1.09080181e-01
-5.68591058e-01 2.92383879e-01 4.33642179e-01 -8.10235888e-02
-2.15531543e-01 -1.06636143e+00 -1.71701968e-01 -1.25714347e-01
1.30132273e-01 -1.45347670e-01 5.96930802e-01 -4.11064237e-01
1.41151452e+00 -1.99830246e+00 -6.03135705e-01 1.71707109e-01
6.20668650e-01 3.09163362e-01 3.94768447e-01 9.84889328e-01
6.77691773e-02 7.61807084e-01 3.88431489e-01 2.88986832e-01
1.46177500e-01 1.78678080e-01 1.34704769e-01 8.54903519e-01
-2.88064122e-01 5.99750161e-01 -1.21022034e+00 -1.13687143e-01
-1.16751276e-01 7.56149173e-01 -3.63608897e-01 1.99433088e-01
3.52518022e-01 2.40172356e-01 -5.21100044e-01 5.31286001e-01
5.62080503e-01 -5.14695823e-01 2.54919171e-01 1.29164666e-01
-6.22786641e-01 3.08734179e-01 -3.34867716e-01 1.09983099e+00
-2.86736757e-01 5.64369082e-01 8.29803757e-03 -3.04173678e-01
6.67552650e-01 8.33709419e-01 3.16509187e-01 -4.89496797e-01
6.69757903e-01 2.24957049e-01 7.83540785e-01 -6.37792349e-01
4.10077572e-01 -3.84281948e-02 1.62927017e-01 7.18565881e-01
-5.93667865e-01 -4.76835012e-01 -6.60503864e-01 4.20476675e-01
8.42967033e-01 -5.27548902e-02 5.29981971e-01 -6.81759059e-01
1.28216445e-01 2.17551395e-01 4.49894041e-01 3.64668429e-01
-4.86928821e-01 -2.96242595e-01 5.13840973e-01 -5.35001338e-01
-1.02132714e+00 -4.18644279e-01 -1.30200863e-01 3.92322689e-01
1.92651346e-01 -4.34590161e-01 -8.60774517e-01 -4.35598493e-01
9.67667531e-03 1.63521945e+00 -6.08549476e-01 -7.57781863e-01
4.37950402e-01 -1.10033952e-01 4.26750451e-01 -1.21049240e-01
6.02816582e-01 -5.93420208e-01 -1.52476001e+00 -2.51348689e-02
1.45500647e-02 -7.85722077e-01 8.93468969e-03 -2.64864922e-01
-7.81940401e-01 -9.47923660e-01 -1.81748137e-01 -1.11663528e-01
8.00700665e-01 4.58263397e-01 1.26595318e+00 7.66696095e-01
9.84808952e-02 7.67427325e-01 -5.45843601e-01 -9.11777973e-01
-9.31694746e-01 -5.17571867e-01 3.87760773e-02 -3.27384949e-01
7.72444129e-01 -2.87613481e-01 -5.58122516e-01 2.03658164e-01
-9.92538273e-01 2.35400230e-01 6.59532547e-01 5.73542178e-01
-7.33924985e-01 -1.43615156e-01 8.57260227e-01 -1.32069194e+00
9.42800701e-01 -6.13342404e-01 -5.41248731e-02 2.71248370e-01
-1.36436987e+00 -4.55141276e-01 6.83841333e-02 -2.20467910e-01
-1.28158104e+00 -6.22547686e-01 5.21397926e-02 3.44568282e-01
-1.11111708e-01 7.92230248e-01 7.30852336e-02 -1.73636734e-01
1.04084504e+00 -4.43757355e-01 6.07767463e-01 2.78821826e-01
2.98654407e-01 1.39481652e+00 -3.90143186e-01 -5.53126335e-01
5.02005279e-01 4.12441581e-01 -9.61206317e-01 -1.09290445e+00
-6.27483904e-01 -1.23363987e-01 -2.93628246e-01 -6.05993450e-01
8.61039221e-01 -8.90261650e-01 -1.09040380e+00 3.36410582e-01
-1.10842407e+00 -1.03684843e-01 1.15143415e-03 7.07511783e-01
-1.73192233e-01 2.44088605e-01 -5.01377940e-01 -1.38133705e+00
-3.72962922e-01 -9.86487687e-01 -1.67082213e-02 4.53932047e-01
-9.76165414e-01 -9.03222859e-01 -1.67183951e-02 6.86764002e-01
6.91528976e-01 1.40117586e-01 7.03736007e-01 -6.67592466e-01
-5.92780635e-02 -3.80355924e-01 2.83565484e-02 5.81443787e-01
3.84844154e-01 2.94422179e-01 -9.58386242e-01 -3.36302578e-01
6.36975467e-01 -6.18081689e-01 -3.89937937e-01 -1.89081639e-01
1.21292762e-01 -7.83054650e-01 -2.11629122e-01 -6.40238285e-01
1.39477706e+00 8.06328297e-01 7.96197176e-01 4.31466579e-01
1.80799574e-01 6.49409294e-01 6.42235994e-01 6.16687000e-01
5.47717035e-01 5.02061509e-02 2.31502444e-01 -5.43610640e-02
7.04468966e-01 -1.26938611e-01 2.61949480e-01 8.98725629e-01
-2.46996939e-01 1.04989320e-01 -1.14742112e+00 3.92525345e-01
-1.65450013e+00 -5.60052812e-01 -3.98380607e-01 2.49836922e+00
7.50828803e-01 5.04065871e-01 7.77437240e-02 -8.85289814e-03
5.00402153e-01 -3.13557506e-01 -3.29501212e-01 -7.88518190e-01
3.66052568e-01 -4.16695684e-01 4.28097636e-01 3.48640025e-01
-1.12483621e-01 6.36196613e-01 6.33687258e+00 -4.67392951e-01
-1.02161539e+00 3.13168913e-01 1.00339413e+00 6.09039783e-01
-9.61787701e-01 3.46037477e-01 1.11113235e-01 1.86023280e-01
1.09381509e+00 -7.67148018e-01 -1.22631630e-02 7.86582410e-01
3.13122064e-01 -5.87474346e-01 -9.15934682e-01 6.43667519e-01
8.61740485e-02 -6.59898937e-01 -5.47948062e-01 -2.42991503e-02
4.01277423e-01 -2.47189373e-01 1.10243261e-01 -1.25484690e-01
5.83969235e-01 -7.50473559e-01 7.65185714e-01 3.49079490e-01
-7.17656612e-02 -8.56092632e-01 9.24114466e-01 2.93714762e-01
-6.59135133e-02 1.00774691e-03 -2.99144566e-01 -7.13156283e-01
-1.12402722e-01 5.61365664e-01 -8.24222684e-01 1.09044090e-01
5.61711192e-01 5.00029922e-01 -4.26953167e-01 5.56397915e-01
8.10522214e-02 8.26950073e-01 5.47720958e-03 -3.85545254e-01
2.04773352e-01 -5.87831199e-01 3.85917306e-01 5.51702499e-01
-1.53004229e-01 2.49521032e-01 -7.25555897e-01 7.42108107e-01
5.50139964e-01 -3.30687426e-02 -1.13104224e+00 -8.10558736e-01
6.91882253e-01 1.16895819e+00 -7.12324321e-01 -2.85566032e-01
-9.48213935e-01 1.06780112e+00 -1.38578817e-01 2.96980947e-01
-8.57653022e-01 1.84259206e-01 8.21866751e-01 5.26189804e-01
-6.59317315e-01 -3.36732566e-01 -4.49627906e-01 -9.12224293e-01
-5.52011251e-01 -1.19544554e+00 6.88070431e-04 -5.64112961e-01
-1.12961233e+00 4.71671283e-01 -1.87357038e-01 -7.29687750e-01
9.58968252e-02 -4.15340252e-02 -1.29171580e-01 4.73086774e-01
-1.01127613e+00 -1.04523599e+00 7.81136528e-02 1.00339167e-02
-1.74425274e-01 4.58092242e-01 9.75702643e-01 4.40552980e-02
-3.13076109e-01 5.32365441e-01 -1.21858753e-01 -3.39033127e-01
8.31949294e-01 -7.64598072e-01 3.26402068e-01 4.13541049e-01
-1.77522078e-01 1.22957349e+00 9.42346394e-01 -7.52390146e-01
-1.80167484e+00 -9.75683555e-02 1.06628513e+00 -5.59445620e-01
1.01835179e+00 -2.57377028e-01 -7.16955006e-01 8.89507353e-01
8.71595323e-01 -7.78226614e-01 1.18345296e+00 3.30658466e-01
-4.27021265e-01 7.58338720e-02 -1.41868162e+00 9.62967634e-01
7.62512386e-01 -5.03940880e-01 -3.45932156e-01 2.24553943e-01
7.19952941e-01 1.66356966e-01 -1.10139632e+00 -1.16096713e-01
9.71147358e-01 -1.38332713e+00 3.55275959e-01 -3.87004077e-01
9.51548740e-02 2.06252024e-01 2.33204201e-01 -1.19083333e+00
-4.94937479e-01 -1.03774822e+00 5.90845525e-01 1.10450613e+00
5.20100772e-01 -1.38769627e+00 4.98350173e-01 1.80365765e+00
-7.91149586e-02 -3.43527585e-01 -5.84540248e-01 -1.62441596e-01
-9.59224030e-02 -1.87729463e-01 5.37395120e-01 1.66997826e+00
9.31444347e-01 3.01217169e-01 -1.53180957e-01 1.52860463e-01
2.46233106e-01 -4.60632563e-01 6.81243181e-01 -1.50506663e+00
-2.68962979e-02 -2.16765597e-01 -5.51858425e-01 -4.44052696e-01
-3.84140372e-01 -2.41573155e-01 -3.93386990e-01 -1.41684413e+00
6.87124729e-02 -4.99918461e-01 9.51121375e-03 2.52483964e-01
2.80788932e-02 -4.67988044e-01 3.31282854e-01 2.03165799e-01
-1.81895748e-01 1.28175020e-01 1.30656970e+00 2.94882476e-01
-1.36918843e-01 -8.03508610e-02 -1.59434688e+00 7.08437860e-01
9.27041113e-01 -3.54307294e-01 -6.39131427e-01 -8.39574561e-02
9.52237487e-01 1.26801506e-01 2.13041425e-01 -9.92915154e-01
2.06534892e-01 -3.11481237e-01 -2.96461545e-02 2.42437810e-01
7.51468912e-02 -1.84283972e+00 9.03222382e-01 8.76366138e-01
-1.16758503e-01 1.50943503e-01 1.27626240e-01 4.41069081e-02
1.57218233e-01 -3.49718571e-01 7.08233893e-01 1.46673638e-02
1.14982806e-01 -4.04484570e-01 -7.45830834e-01 -5.81992984e-01
1.34614003e+00 -4.09765452e-01 -3.86646003e-01 -5.44228137e-01
-2.52766252e-01 1.42071709e-01 1.00508010e+00 5.17778277e-01
4.28503275e-01 -1.01524520e+00 -5.36804676e-01 -5.70584424e-02
2.62285054e-01 -5.98751545e-01 1.51939958e-01 1.00116456e+00
-5.46605170e-01 2.51824260e-01 -3.53858203e-01 -2.34844200e-02
-1.22880590e+00 4.23060507e-01 2.21395567e-01 6.19631588e-01
-6.15392804e-01 6.96037412e-01 -1.51953667e-01 1.92965433e-01
8.18301514e-02 -2.25066885e-01 -2.00209133e-02 -5.23653366e-02
3.91357780e-01 4.27133769e-01 -3.26770723e-01 -5.32563150e-01
-2.22835571e-01 -3.68799657e-01 -3.72107923e-01 -1.00848466e-01
1.11481309e+00 -3.13291311e-01 -5.72645307e-01 1.09932268e+00
7.00058103e-01 1.08279943e-01 -7.19321847e-01 3.12588453e-01
1.58759266e-01 -9.92925823e-01 5.70697561e-02 -1.17721760e+00
-5.46666503e-01 3.15562397e-01 4.59203511e-01 7.69385815e-01
5.85606396e-01 -1.71153441e-01 5.74958265e-01 3.53353918e-01
5.91398239e-01 -1.23730505e+00 -2.27268293e-01 -2.32213259e-01
9.39676881e-01 -1.27062917e+00 1.51224300e-01 -4.78105068e-01
-1.20092106e+00 8.77286494e-01 6.18145823e-01 4.23541695e-01
9.71389592e-01 9.17668119e-02 3.52084965e-01 -3.85396272e-01
-7.63169348e-01 3.31460416e-01 -5.21072388e-01 6.10950053e-01
1.33115149e+00 3.14177334e-01 -1.15143847e+00 6.95902407e-01
-1.12105943e-01 3.22942615e-01 1.06458652e+00 1.02360761e+00
-4.07156020e-01 -1.22243726e+00 -2.51869768e-01 3.57999027e-01
-5.53828180e-01 1.79619670e-01 -8.34515572e-01 1.04463685e+00
-1.32983709e-02 1.41059971e+00 -2.00179070e-01 -6.47812724e-01
1.49787441e-01 5.03436141e-02 6.96819052e-02 -3.33179146e-01
-1.14416325e+00 -1.80921867e-01 5.87740898e-01 -2.16034874e-01
-4.58354712e-01 -5.73288798e-01 -1.10157931e+00 -8.26393843e-01
-7.65123248e-01 7.48454750e-01 7.46178508e-01 4.79283333e-01
4.95906681e-01 -6.06414787e-02 5.80965459e-01 1.68228343e-01
-5.11658967e-01 -9.73872125e-01 -4.96252626e-01 1.27060622e-01
3.95651162e-02 -1.78970963e-01 -1.62523404e-01 -1.33319616e-01] | [9.063992500305176, 6.312570095062256] |
5c99ff66-c9e3-4603-99f3-773c2bb41ac5 | sampling-individually-fair-rankings-that-are | 2306.11964 | null | https://arxiv.org/abs/2306.11964v1 | https://arxiv.org/pdf/2306.11964v1.pdf | Sampling Individually-Fair Rankings that are Always Group Fair | Rankings on online platforms help their end-users find the relevant information -- people, news, media, and products -- quickly. Fair ranking tasks, which ask to rank a set of items to maximize utility subject to satisfying group-fairness constraints, have gained significant interest in the Algorithmic Fairness, Information Retrieval, and Machine Learning literature. Recent works, however, identify uncertainty in the utilities of items as a primary cause of unfairness and propose introducing randomness in the output. This randomness is carefully chosen to guarantee an adequate representation of each item (while accounting for the uncertainty). However, due to this randomness, the output rankings may violate group fairness constraints. We give an efficient algorithm that samples rankings from an individually-fair distribution while ensuring that every output ranking is group fair. The expected utility of the output ranking is at least $\alpha$ times the utility of the optimal fair solution. Here, $\alpha$ depends on the utilities, position-discounts, and constraints -- it approaches 1 as the range of utilities or the position-discounts shrinks, or when utilities satisfy distributional assumptions. Empirically, we observe that our algorithm achieves individual and group fairness and that Pareto dominates the state-of-the-art baselines. | ['Anand Louis', 'Amit Deshpande', 'Anay Mehrotra', 'Sruthi Gorantla'] | 2023-06-21 | null | null | null | null | ['fairness', 'retrieval', 'fairness', 'information-retrieval'] | ['computer-vision', 'methodology', 'miscellaneous', 'natural-language-processing'] | [-2.02865183e-01 1.00515792e-02 -6.47514939e-01 -5.83626628e-01
-9.31344628e-01 -7.91207016e-01 1.17346905e-01 3.85245144e-01
-6.84216201e-01 7.85077631e-01 4.23783094e-01 -1.64020896e-01
-5.48515201e-01 -7.12945461e-01 -1.87158465e-01 -3.78369898e-01
-2.03446046e-01 7.47950137e-01 -1.88031569e-01 -1.09988697e-01
6.48118734e-01 -1.63383424e-01 -1.58080101e+00 1.06413081e-01
1.34186709e+00 1.15845275e+00 -2.03618675e-01 3.15932512e-01
-1.16823718e-01 7.10466087e-01 -5.47600389e-01 -9.74803984e-01
7.75683224e-01 -3.96541774e-01 -8.37263346e-01 -4.55090046e-01
2.02590615e-01 -5.86029470e-01 1.22374564e-01 1.46507215e+00
2.77260691e-01 5.61311483e-01 8.67563009e-01 -1.49958754e+00
-8.31525683e-01 9.40571547e-01 -6.83419049e-01 -4.39673401e-02
1.23836808e-01 -2.06922695e-01 1.77352083e+00 -2.81287104e-01
4.67771083e-01 1.19548571e+00 3.08647063e-02 6.60159409e-01
-1.17371964e+00 -6.86396539e-01 2.81954199e-01 -1.75110385e-01
-1.08947980e+00 -3.48005563e-01 2.48673663e-01 -3.18652302e-01
1.98453903e-01 7.42821157e-01 1.92507729e-01 3.50566685e-01
2.36756317e-02 4.34229493e-01 9.43649769e-01 -3.34290653e-01
7.73886263e-01 1.17444597e-01 4.00912076e-01 1.06474526e-01
6.08161211e-01 -3.78030278e-02 -7.18356609e-01 -7.32836306e-01
2.23396957e-01 1.17920332e-01 -1.73496604e-01 -2.09491968e-01
-7.18265474e-01 9.06747341e-01 3.36482555e-01 -3.37286055e-01
-6.14132822e-01 3.78342897e-01 4.10996467e-01 5.63250840e-01
7.61116266e-01 8.52081418e-01 -2.70944744e-01 -8.08761865e-02
-9.90139186e-01 6.53822303e-01 7.21876085e-01 6.68419778e-01
5.33971608e-01 -4.87091750e-01 -7.52168715e-01 7.70713210e-01
7.32405931e-02 3.47052962e-01 2.47742578e-01 -1.53144026e+00
6.56641841e-01 4.38103408e-01 9.07495677e-01 -9.05577004e-01
2.02284697e-02 -2.76032269e-01 -5.03220618e-01 3.05772334e-01
6.38015628e-01 -2.13193297e-01 -3.57701391e-01 2.10439348e+00
1.70299178e-03 -5.30951738e-01 -4.61888492e-01 1.36538863e+00
9.06203240e-02 4.20552015e-01 3.30370456e-01 -5.42302907e-01
1.16897583e+00 -5.09167612e-01 -6.76286817e-01 -2.55022272e-02
1.27232090e-01 -7.48987913e-01 1.26196158e+00 3.23884875e-01
-1.36589921e+00 1.43655479e-01 -7.52351403e-01 4.60159816e-02
1.78464040e-01 -4.70833212e-01 4.47585285e-01 8.86136472e-01
-7.85040140e-01 8.59023213e-01 -4.25447943e-03 2.77055144e-01
3.90494734e-01 3.91564101e-01 1.36470376e-02 -2.77374103e-03
-1.24484348e+00 7.49477804e-01 -5.89967929e-02 -3.91318232e-01
-5.04968822e-01 -7.78254688e-01 -2.28264675e-01 6.41709208e-01
6.05736613e-01 -8.64390433e-01 1.43909466e+00 -1.07661295e+00
-1.24826956e+00 6.50748909e-01 8.05343986e-02 -3.55215400e-01
1.06069541e+00 -1.58598140e-01 5.44126220e-02 -2.28286922e-01
3.19800675e-01 2.37909064e-01 4.97274190e-01 -1.18829381e+00
-7.68060863e-01 -4.43437338e-01 3.46209973e-01 6.73571885e-01
-5.30422986e-01 3.04083019e-01 -3.28225531e-02 -3.71114343e-01
-1.31051496e-01 -7.97265887e-01 -5.29248357e-01 -3.92713174e-02
-3.45769495e-01 -2.78353125e-01 -1.47507802e-01 -3.86865050e-01
1.44547069e+00 -1.98690939e+00 -2.86206067e-01 6.75574720e-01
4.31585401e-01 -3.45829308e-01 -2.26127535e-01 1.12940118e-01
3.35703760e-01 6.02356315e-01 1.99642524e-01 -2.60941297e-01
6.17462039e-01 -1.24533653e-01 -2.52198875e-01 5.12509406e-01
-5.54957330e-01 5.26285052e-01 -9.26650643e-01 -7.24617764e-02
-2.04840824e-01 -2.98086494e-01 -7.77689338e-01 2.26355091e-01
-2.32896566e-01 -3.43733907e-01 -3.96255016e-01 1.91173628e-01
9.08845544e-01 -1.28525451e-01 2.99027324e-01 1.45727485e-01
-1.59756392e-01 4.84515607e-01 -1.31349587e+00 8.40126395e-01
-1.76294073e-01 5.58173805e-02 3.13736834e-02 -4.27670747e-01
5.66544175e-01 4.27061655e-02 5.81159472e-01 -7.75731981e-01
1.43057987e-01 3.40009421e-01 -8.54433179e-02 5.34688979e-02
9.08288598e-01 -3.68050605e-01 -3.41014802e-01 1.05798411e+00
-6.16878688e-01 1.95402935e-01 2.87040800e-01 5.12198031e-01
6.90198958e-01 -4.58411574e-01 3.02324325e-01 -6.83270335e-01
-1.38445318e-01 -3.38193744e-01 7.78408885e-01 9.93501544e-01
-3.29586476e-01 6.07689023e-01 8.45262289e-01 -2.55262464e-01
-9.58121359e-01 -1.12435937e+00 1.86435670e-01 1.61290121e+00
5.82927704e-01 -1.92860559e-01 -6.16863549e-01 -6.88146830e-01
4.10923868e-01 9.71870661e-01 -6.08855665e-01 -4.50595878e-02
6.52052984e-02 -6.96343243e-01 -2.39115972e-02 2.57105112e-01
1.28918856e-01 -6.49787903e-01 -6.23022556e-01 -1.61905065e-02
-5.55900693e-01 -5.18406868e-01 -1.12303054e+00 -7.65160695e-02
-5.05191088e-01 -1.03091729e+00 -6.79087579e-01 6.12080581e-02
7.32756853e-01 4.47870225e-01 1.28178906e+00 -1.50091629e-02
1.81697577e-01 1.71323237e-03 -1.01188719e-01 -4.16002631e-01
1.04017667e-01 -2.87593663e-01 2.12723568e-01 -6.60523474e-02
1.72287077e-01 -1.60701707e-01 -8.91043425e-01 4.69947457e-01
-7.48806179e-01 -3.34769398e-01 -1.26425952e-01 5.07952154e-01
4.69663113e-01 1.64039657e-01 8.66545141e-01 -1.10919213e+00
1.38768613e+00 -7.22280681e-01 -5.85595608e-01 3.86384159e-01
-9.35814738e-01 2.52211004e-01 6.76490545e-01 -1.15160152e-01
-9.81348455e-01 -7.39529848e-01 5.19246519e-01 -3.10836174e-02
6.68313086e-01 4.15689617e-01 -2.54932344e-01 3.65094125e-01
7.10036397e-01 -5.61036170e-01 -3.69207598e-02 -2.26950198e-01
5.70446432e-01 6.40478015e-01 2.48576105e-01 -8.88323724e-01
4.02312636e-01 1.57897621e-01 -1.98105365e-01 2.01880008e-01
-9.87105787e-01 -4.67194438e-01 3.24739307e-01 -2.02785850e-01
2.80283302e-01 -7.13106990e-01 -1.11996949e+00 -1.36133984e-01
-7.36700714e-01 2.47969665e-02 -6.37764215e-01 3.15815091e-01
-4.63303477e-01 2.77762264e-01 -3.81040365e-01 -1.41821849e+00
-6.28454804e-01 -9.32853699e-01 3.53986710e-01 3.79186600e-01
-5.62191784e-01 -4.46975559e-01 -8.07971954e-02 2.39172667e-01
7.12102711e-01 3.10395565e-02 1.00384724e+00 -6.61117494e-01
-4.54749167e-01 -3.88364315e-01 -3.49840611e-01 2.18455747e-01
3.03031709e-02 -2.91715302e-02 -5.53833246e-01 -6.81143925e-02
-3.58580202e-01 -3.30730438e-01 8.27610552e-01 5.50481737e-01
1.40697300e+00 -1.04422390e+00 6.25592694e-02 -2.38707718e-02
1.19057226e+00 -6.85496256e-02 3.17779183e-01 3.01530153e-01
1.98956486e-02 8.28091085e-01 6.96428716e-01 1.12407959e+00
4.90622431e-01 6.69218898e-01 6.39388382e-01 3.86683077e-01
6.00054681e-01 -1.91008374e-01 2.23811924e-01 1.67545930e-01
-2.34635159e-01 -4.14338440e-01 -4.54993665e-01 3.12441140e-01
-2.11963677e+00 -1.31516159e+00 2.13670835e-01 3.19152832e+00
8.32480311e-01 1.90674216e-01 4.76913691e-01 -3.74186859e-02
8.71506572e-01 5.91170602e-02 -6.25963032e-01 -7.80946553e-01
6.86204731e-02 -1.54465392e-01 8.10102105e-01 7.08935559e-01
-7.36357689e-01 3.70391876e-01 6.49445152e+00 7.91311383e-01
-5.38557231e-01 -1.49329647e-01 1.16558361e+00 -9.26641822e-01
-1.13474500e+00 -1.02901369e-01 -3.17292303e-01 8.82361829e-01
7.81610429e-01 -1.07722247e+00 8.40012670e-01 8.70757937e-01
3.86954278e-01 -1.08967118e-01 -1.09914386e+00 6.77904308e-01
-4.46169406e-01 -1.06022751e+00 -1.17439434e-01 3.25631291e-01
9.95793402e-01 -2.76002288e-01 4.51380461e-01 2.54494816e-01
1.03379047e+00 -9.69124079e-01 1.04568422e+00 4.65662748e-01
7.92932034e-01 -1.22282934e+00 7.73714006e-01 4.06675667e-01
-7.61733413e-01 -2.81200200e-01 -4.71948951e-01 -2.20864862e-01
1.79219022e-01 1.22294414e+00 -2.05580607e-01 3.84102046e-01
4.94253635e-01 -7.68503100e-02 -7.23951310e-02 1.03368294e+00
6.19941726e-02 3.07771713e-01 -3.09987813e-01 -1.64466679e-01
2.81701647e-02 -2.59237945e-01 2.21896484e-01 7.31924176e-01
2.82731801e-01 1.98149115e-01 3.05972934e-01 7.46696234e-01
-8.47734630e-01 4.38426137e-01 -1.41668290e-01 2.10996136e-01
9.14322495e-01 1.09311914e+00 -2.80466735e-01 -1.67594522e-01
1.31047279e-01 5.99041045e-01 3.72381538e-01 2.95279860e-01
-7.09344327e-01 -2.20463127e-01 1.05564106e+00 2.26198554e-01
-4.87032592e-01 3.34641904e-01 -8.08920979e-01 -8.97035480e-01
1.11164004e-01 -7.49705851e-01 8.01501811e-01 -3.66888613e-01
-1.77533519e+00 1.95028633e-01 -1.49145508e-02 -1.00607824e+00
-2.71474689e-01 -1.55089632e-01 -7.12276042e-01 1.01669383e+00
-1.44000661e+00 -1.92267925e-01 7.65867857e-03 1.20578207e-01
9.30122957e-02 -7.36902934e-03 5.47313511e-01 1.57904893e-01
-2.70237595e-01 9.00008082e-01 4.53193754e-01 -4.68272835e-01
9.30259883e-01 -1.26238561e+00 -2.13020630e-02 5.82577765e-01
-3.63095462e-01 8.23228598e-01 8.69149208e-01 -4.08799440e-01
-8.42850387e-01 -7.78166533e-01 1.21223330e+00 -2.60061443e-01
2.43242472e-01 -4.01609987e-02 -5.21907747e-01 1.66281641e-01
1.02449551e-01 1.33733647e-02 1.11063457e+00 5.17131925e-01
-6.32703483e-01 -2.79217333e-01 -1.65713525e+00 9.36149597e-01
9.53993201e-01 -3.73554885e-01 -1.39899701e-01 1.88650683e-01
5.61219692e-01 -9.96215492e-02 -4.89268512e-01 6.24687262e-02
1.07165337e+00 -1.04692829e+00 7.36735344e-01 -8.32278728e-01
7.06057131e-01 -1.13878183e-01 -3.81712675e-01 -1.58325112e+00
-5.72106063e-01 -7.27926493e-01 1.21310819e-02 9.73711610e-01
4.29220557e-01 -5.15936017e-01 7.68187463e-01 1.70986700e+00
3.63090903e-01 -5.40871322e-01 -8.91253233e-01 -7.42336154e-01
1.72568113e-01 -1.53350383e-01 9.66318369e-01 6.68230832e-01
4.28174704e-01 2.35446286e-03 -5.22319078e-01 -2.86024630e-01
1.00635862e+00 2.85071671e-01 3.28694165e-01 -1.32753968e+00
-2.19275787e-01 -1.00521660e+00 2.37133473e-01 -6.49328589e-01
9.86879095e-02 -6.30051255e-01 -4.20997571e-03 -1.44224739e+00
7.42231488e-01 -8.26003671e-01 -7.82845140e-01 3.61776590e-01
-6.40408099e-01 -4.28436417e-03 5.87472796e-01 3.25958014e-01
-1.04438949e+00 3.51906449e-01 9.02276516e-01 1.28680486e-02
-1.33897960e-01 3.26694667e-01 -1.66461885e+00 3.84906471e-01
8.57122004e-01 -5.53051054e-01 -4.29174781e-01 -2.24933386e-01
7.07728148e-01 5.53903207e-02 -8.38792920e-02 -1.60128400e-01
1.04714133e-01 -8.40929687e-01 8.16378891e-02 -2.05860585e-01
-1.41606793e-01 -6.66821361e-01 8.77821818e-02 2.47268558e-01
-1.16739345e+00 8.24955404e-02 -4.69231993e-01 5.29734313e-01
6.34694993e-02 -4.55962151e-01 7.92308867e-01 -2.17243984e-01
2.03801901e-03 3.56082290e-01 -1.42885596e-01 4.42123950e-01
8.13988328e-01 1.51461408e-01 -5.97122848e-01 -8.09581518e-01
-1.14623092e-01 5.79478621e-01 6.57277048e-01 2.27738455e-01
2.08800256e-01 -1.37594366e+00 -8.61693859e-01 -3.86832118e-01
2.62042910e-01 -3.87782872e-01 2.09213048e-01 2.20019877e-01
-4.34220508e-02 9.89592895e-02 -2.97763884e-01 2.78418660e-01
-9.82054174e-01 1.87691376e-01 2.11438894e-01 -6.53656304e-01
2.27933064e-01 7.70056248e-01 2.55528092e-01 -1.22155920e-01
3.34235609e-01 1.10608086e-01 3.15557569e-02 2.23256782e-01
7.51935542e-01 9.10719752e-01 -3.32148862e-03 -7.96953365e-02
-3.04383367e-01 -4.61859144e-02 -7.88425580e-02 -4.58455056e-01
1.00227880e+00 -3.31782252e-01 -1.39938056e-01 -2.92643849e-02
8.14872980e-01 4.05446857e-01 -1.16993129e+00 -1.58940822e-01
2.40361802e-02 -1.12882519e+00 -1.37810344e-02 -9.66818333e-01
-9.38040614e-01 3.85218978e-01 1.01640068e-01 4.99005705e-01
9.85877872e-01 -4.78461564e-01 4.05068904e-01 -7.20534027e-02
7.57858753e-01 -1.37780881e+00 1.56356543e-02 2.26417229e-01
7.07492948e-01 -1.07280278e+00 1.95505440e-01 -6.03148974e-02
-1.05857968e+00 7.42669225e-01 6.27224863e-01 -1.37245253e-01
4.63381618e-01 -6.83847740e-02 -1.60073027e-01 2.77413636e-01
-8.47199619e-01 5.16922362e-02 3.74410480e-01 2.39265531e-01
5.95193744e-01 7.57630348e-01 -8.92290652e-01 1.19693911e+00
-2.83304423e-01 -1.18022710e-01 5.56773007e-01 5.04036844e-01
-6.72775149e-01 -1.11033750e+00 -2.78192729e-01 1.03925514e+00
-8.75834107e-01 -1.74431503e-01 -3.00118953e-01 -8.18593800e-03
-7.72647336e-02 1.09090340e+00 9.32875127e-02 -1.24281257e-01
3.58335942e-01 -2.46959329e-01 2.35741347e-01 -4.76817012e-01
-8.15831721e-01 -1.74236387e-01 1.82585493e-01 -5.58575809e-01
5.30456752e-02 -5.14992476e-01 -1.10871089e+00 -9.49604571e-01
-3.44780266e-01 5.96446693e-01 3.22939396e-01 6.88274741e-01
3.20719689e-01 7.74883553e-02 1.05451429e+00 -2.70192236e-01
-1.36306763e+00 -4.72633362e-01 -9.91155446e-01 7.64872015e-01
9.79658738e-02 -4.95371431e-01 -5.58657706e-01 -5.41073859e-01] | [9.364383697509766, 5.538438320159912] |
afc3af89-e7c1-46fe-88cb-fc0a9b4371f3 | inter-patient-ecg-heartbeat-classification | null | null | https://doi.org/10.1038/s41598-017-09837-3 | https://www.nature.com/articles/s41598-017-09837-3.pdf | Inter-Patient ECG Heartbeat Classification with Temporal VCG Optimized by PSO | Classifying arrhythmias can be a tough task for a human being and automating this task is highly desirable. Nevertheless fully automatic arrhythmia classification through Electrocardiogram (ECG) signals is a challenging task when the inter-patient paradigm is considered. For the inter-patient paradigm, classifiers are evaluated on signals of unknown subjects, resembling the real world scenario. In this work, we explore a novel ECG representation based on vectorcardiogram (VCG), called temporal vectorcardiogram (TVCG), along with a complex network for feature extraction. We also fine-tune the SVM classifier and perform feature selection with a particle swarm optimization (PSO) algorithm. Results for the inter-patient paradigm show that the proposed method achieves the results comparable to state-of-the-art in MIT-BIH database (53% of Positive predictive (+P) for the Supraventricular ectopic beat (S) class and 87.3% of Sensitivity (Se) for the Ventricular ectopic beat (V) class) that TVCG is a richer representation of the heartbeat and that it could be useful for problems involving the cardiac signal and pattern recognition.*
Source code available from http://www.decom.ufop.br/csilab/site_media/uploads/code/tvcg_pso.zip | ['Gladston Moreira', 'Eduardo Luz', 'David Menotti', 'Gabriel Garcia'] | 2017-09-05 | null | null | null | scientific-reports-2017-9 | ['arrhythmia-detection', 'ecg-classification', 'heartbeat-classification', 'electrocardiography-ecg'] | ['medical', 'medical', 'medical', 'methodology'] | [ 4.63287473e-01 -3.29728872e-01 8.68025050e-02 -2.02619005e-02
-3.04618299e-01 -5.03921330e-01 7.60152265e-02 4.51076776e-01
-3.46909106e-01 9.09077585e-01 -3.58571917e-01 -4.16381478e-01
-5.38470089e-01 -4.60435539e-01 -1.77304193e-01 -8.92197549e-01
-3.50352794e-01 5.87112784e-01 -2.93222219e-02 -7.04910383e-02
2.93887198e-01 7.07143843e-01 -1.40216160e+00 2.01170743e-01
7.53407776e-01 1.13108861e+00 7.33139440e-02 9.09754932e-01
5.63630104e-01 1.10697441e-01 -9.41281140e-01 2.27897584e-01
2.01971486e-01 -8.41147900e-01 -4.39416915e-01 -2.27113992e-01
-3.98672402e-01 5.22862375e-01 3.76689255e-01 7.23727286e-01
1.01899958e+00 -8.25926512e-02 6.95206404e-01 -1.15865219e+00
3.45404029e-01 3.12947452e-01 -1.60755798e-01 7.17367232e-01
4.26183134e-01 5.79146843e-04 6.07462883e-01 -7.62526035e-01
4.58628565e-01 3.90372217e-01 1.03155804e+00 8.41413662e-02
-1.19168496e+00 -5.23217320e-01 -6.04556859e-01 3.65948439e-01
-1.54302096e+00 -1.05558240e-04 9.66944098e-01 -6.28233731e-01
6.76650107e-01 9.38069999e-01 1.27481413e+00 8.47255945e-01
4.82656717e-01 2.47358784e-01 1.15251827e+00 -5.15599489e-01
2.86046714e-01 2.78328180e-01 5.15732706e-01 4.13647354e-01
4.97798443e-01 1.56346798e-01 -2.92773157e-01 -7.19039321e-01
7.30389595e-01 -5.04244640e-02 -7.18380988e-01 -1.59219489e-01
-1.44489706e+00 7.23540187e-01 -1.64635092e-01 6.86428845e-01
-7.46393740e-01 -3.08533937e-01 5.35928130e-01 4.63658273e-01
1.43396333e-01 8.97596955e-01 -5.23461878e-01 -5.46117187e-01
-9.55558479e-01 4.16832000e-01 9.79819179e-01 2.92422563e-01
1.14241183e-01 2.07344547e-01 -1.78383067e-01 7.66923428e-01
5.95096536e-02 4.93352145e-01 7.27917433e-01 -5.76679289e-01
1.61662802e-01 5.36949575e-01 7.88555965e-02 -1.08013105e+00
-8.58161271e-01 -1.06221962e+00 -1.29525971e+00 -1.34368904e-03
4.77624595e-01 -3.35418046e-01 -3.68642658e-01 1.12953532e+00
2.11516649e-01 1.77299216e-01 1.47244036e-01 7.17656910e-01
6.70979083e-01 5.16914606e-01 -1.77344412e-01 -9.01918232e-01
1.36732399e+00 -3.66638064e-01 -6.33589387e-01 1.42166734e-01
5.14087498e-01 -6.43130362e-01 6.12882078e-01 8.32959473e-01
-7.95851290e-01 -5.83295286e-01 -1.02876043e+00 1.01299417e+00
-4.94134352e-02 4.85086739e-01 3.55229586e-01 9.05523658e-01
-7.03974485e-01 8.11804891e-01 -9.10923362e-01 -4.42346036e-01
2.85594463e-01 5.53905129e-01 -1.87837407e-01 4.69293386e-01
-1.20133972e+00 7.44254410e-01 3.98305535e-01 2.60729462e-01
-7.10682049e-02 -5.48471749e-01 -5.69434106e-01 -9.12847519e-02
-6.53400272e-02 -5.78913569e-01 5.52961469e-01 -6.77746117e-01
-1.28874111e+00 6.80370867e-01 1.00486306e-02 -5.35610855e-01
6.45831227e-01 1.77473739e-01 -4.54503328e-01 1.42414525e-01
-2.32903391e-01 -2.84092903e-01 9.70271707e-01 -8.75528455e-01
-3.14330250e-01 -5.70927799e-01 -5.95070839e-01 1.93282992e-01
-3.69656906e-02 -6.56653196e-03 1.61365926e-01 -8.99296880e-01
3.98602724e-01 -1.16255558e+00 -1.42293319e-01 -5.71348429e-01
-3.78278404e-01 2.90078223e-02 5.64084113e-01 -8.21233511e-01
1.71613610e+00 -2.05148005e+00 3.68929625e-01 8.39085579e-01
1.82531819e-01 5.89318514e-01 4.47936893e-01 6.08311296e-01
-3.21692288e-01 -2.18454413e-02 -5.73412538e-01 2.43569613e-01
-5.59960306e-01 9.91530940e-02 1.04622439e-01 6.86132252e-01
-1.62860796e-01 5.39811194e-01 -6.39845431e-01 -3.39985907e-01
2.15243086e-01 4.10259515e-01 -1.24480739e-01 1.06510771e-02
5.32138586e-01 7.56670952e-01 -4.10820603e-01 5.93353927e-01
3.25692147e-01 -2.89846480e-01 4.05011714e-01 -2.86960572e-01
1.22905888e-01 -5.18772721e-01 -1.53398347e+00 1.26212764e+00
2.98179872e-02 4.69954282e-01 -1.89745709e-01 -1.28866029e+00
1.23582995e+00 6.89352393e-01 8.63944292e-01 -1.07631549e-01
3.37974578e-01 3.16295743e-01 6.07944369e-01 -6.70923054e-01
-1.83276296e-01 -1.06657542e-01 1.69011459e-01 3.71749759e-01
-8.16470683e-02 -9.76568535e-02 1.83408931e-01 -3.14424664e-01
1.10903585e+00 -6.34871125e-02 8.73744369e-01 -5.91048658e-01
8.59379530e-01 -8.45439732e-02 7.01953888e-01 6.96332216e-01
-2.77396590e-01 7.33098269e-01 5.20861864e-01 -7.13596463e-01
-5.71768522e-01 -8.00926805e-01 -6.57810211e-01 9.65520293e-02
-1.09648846e-01 -3.46830696e-01 -5.60955286e-01 -3.77922267e-01
1.92016121e-02 4.17079359e-01 -3.83297831e-01 -1.16281345e-01
-6.23022377e-01 -1.46732116e+00 6.24182701e-01 2.44940355e-01
7.70825520e-02 -1.18619430e+00 -1.20648170e+00 5.35600424e-01
-2.55706340e-01 -6.89909816e-01 1.28822297e-01 3.74151617e-01
-1.18697107e+00 -1.24077260e+00 -8.27086091e-01 -3.26403648e-01
3.34415764e-01 -6.20091975e-01 1.00303030e+00 2.07083002e-01
-7.08890617e-01 2.14239553e-01 -5.14532566e-01 -7.30741382e-01
-5.70043623e-01 -2.98027936e-02 2.22266570e-01 3.27431113e-01
1.96956228e-02 -7.66184509e-01 -6.50863230e-01 4.22449082e-01
-3.27059805e-01 -1.55362532e-01 2.85969138e-01 9.99316216e-01
6.89425707e-01 1.11490317e-01 9.12594855e-01 -8.62063468e-01
8.56294632e-01 -3.07481349e-01 -5.03289223e-01 1.18294232e-01
-8.50089490e-01 -4.91074175e-01 5.93589783e-01 -3.44564855e-01
-2.04405770e-01 2.63328999e-01 -2.46348396e-01 -2.97884792e-01
-2.65588552e-01 4.10968751e-01 2.39580795e-01 -2.25564390e-01
9.73137796e-01 5.03073812e-01 1.10334538e-01 -2.94025600e-01
-6.87293470e-01 7.41135776e-01 2.62992740e-01 -2.06440225e-01
5.94357073e-01 1.36394814e-01 3.93834054e-01 -1.20021498e+00
-2.46272266e-01 -6.76028848e-01 -5.90172589e-01 -3.67812037e-01
7.21768796e-01 -4.52585399e-01 -7.23538935e-01 4.06459361e-01
-9.35606122e-01 -4.08941275e-03 -4.44699705e-01 7.88851440e-01
-5.86211085e-01 4.10982937e-01 -2.23123282e-01 -1.06626987e+00
-7.36713588e-01 -1.02046597e+00 5.71784973e-01 3.14762779e-02
-6.84029341e-01 -1.01683903e+00 1.92937747e-01 1.66381761e-01
4.56702888e-01 9.54001427e-01 7.67163575e-01 -9.68003154e-01
1.66566461e-01 -5.53598583e-01 4.12416637e-01 5.85415781e-01
1.40081450e-01 -1.28753990e-01 -8.94407690e-01 -3.33019465e-01
4.69536096e-01 3.69577914e-01 2.94566840e-01 5.35231471e-01
1.14303672e+00 1.01573221e-01 -3.89563292e-01 5.99810243e-01
1.37778056e+00 6.88997507e-01 4.33051586e-01 1.99028865e-01
3.86775225e-01 1.53234676e-01 6.12757742e-01 8.11186910e-01
-9.00312960e-02 7.67124772e-01 7.54079521e-02 -2.35088989e-01
2.99856722e-01 4.93983060e-01 -2.86676195e-02 7.56234527e-01
-6.46729946e-01 -1.12040199e-01 -1.24098599e+00 2.24440515e-01
-1.91091025e+00 -9.41493511e-01 -5.69663167e-01 2.31270027e+00
5.84548235e-01 1.75962657e-01 3.00791115e-01 1.22112262e+00
7.26822734e-01 -2.62399971e-01 -2.53771096e-01 -5.16447842e-01
-2.07795933e-01 4.74215537e-01 8.53838250e-02 3.93525630e-01
-1.01009583e+00 -7.73757398e-02 5.07495689e+00 6.16200030e-01
-1.35853279e+00 1.03457808e-01 6.20756805e-01 2.75280684e-01
3.82879794e-01 -3.16542625e-01 -4.80250776e-01 6.18582547e-01
9.88980114e-01 -2.46197164e-01 2.29518965e-01 3.77121210e-01
3.61492962e-01 -6.38046265e-02 -8.56851876e-01 1.64965963e+00
9.09406543e-02 -1.26738870e+00 -4.76111263e-01 -9.99112725e-02
2.26355419e-01 -1.83339804e-01 -3.42817068e-01 -6.78105280e-02
-1.07463741e+00 -8.64379585e-01 3.01479608e-01 6.71427071e-01
8.71614158e-01 -5.64614594e-01 1.11190999e+00 4.90709662e-01
-1.15985322e+00 -2.89907098e-01 8.63765702e-02 -4.49956544e-02
3.97959165e-02 8.34418476e-01 -9.45164084e-01 8.23891342e-01
6.85252964e-01 9.05145407e-01 -6.94801807e-01 1.38634086e+00
1.90209925e-01 9.33959007e-01 -4.63208109e-01 -8.84291753e-02
-3.58006746e-01 -3.10917228e-01 1.09290302e+00 1.13242054e+00
5.51890314e-01 1.36971414e-01 6.06492348e-02 5.06705523e-01
5.74945807e-01 4.37126040e-01 -5.69792569e-01 1.13185555e-01
2.65980899e-01 1.18138385e+00 -1.04408872e+00 -3.31083119e-01
1.73994571e-01 7.01527655e-01 -3.51451457e-01 1.76188067e-01
-7.35646546e-01 -8.29149187e-01 6.84891269e-02 1.48744673e-01
-6.67889565e-02 3.85411650e-01 -5.83638847e-01 -1.06524491e+00
2.05849171e-01 -9.77397621e-01 5.44080853e-01 -4.95084643e-01
-8.65976632e-01 9.90880907e-01 -3.25662047e-02 -1.66643143e+00
-4.30390656e-01 -4.91064966e-01 -6.32379770e-01 9.53613520e-01
-8.21326554e-01 -4.39916641e-01 -2.96610475e-01 5.66844046e-01
2.23710492e-01 -3.35177392e-01 1.29022443e+00 2.76570439e-01
-3.20064992e-01 3.92559886e-01 3.17340195e-02 -1.65265515e-01
2.85555303e-01 -1.32871282e+00 -2.04354197e-01 5.88003993e-01
1.19226545e-01 2.30288640e-01 8.93207073e-01 -4.87810135e-01
-1.21857762e+00 -8.00623298e-01 1.08867216e+00 -5.95113873e-01
1.19126201e-01 -6.98956549e-02 -6.73382998e-01 6.06320091e-02
-1.63671583e-01 8.08424130e-02 9.04695749e-01 -1.59589678e-01
5.81808805e-01 -2.53726542e-01 -1.26463914e+00 1.30483478e-01
5.33517897e-01 1.61128130e-03 -5.80619574e-01 4.32895988e-01
-2.86906093e-01 -4.00301039e-01 -1.36405611e+00 6.78425133e-01
7.98947930e-01 -1.15936434e+00 1.00536776e+00 -6.88540787e-02
-2.66626865e-01 -3.96719277e-01 1.30914077e-01 -1.20691431e+00
-1.30363241e-01 -9.12521482e-01 -2.03888252e-01 6.59237385e-01
4.01541531e-01 -9.05276597e-01 7.01870084e-01 -5.38442172e-02
1.04762027e-02 -1.17929983e+00 -9.81378913e-01 -8.46924901e-01
-5.81648350e-01 -4.19060230e-01 2.36051351e-01 1.08396626e+00
1.17048673e-01 1.18513577e-01 -2.08103031e-01 8.98868144e-02
7.16743648e-01 1.64499730e-01 2.90646166e-01 -1.80114293e+00
-7.11453319e-01 -2.59804219e-01 -7.84193158e-01 2.10256442e-01
-3.49266142e-01 -1.02531493e+00 -4.40875560e-01 -1.24106419e+00
-2.78959274e-01 -6.39664114e-01 -6.29075944e-01 2.16369256e-01
-1.76251709e-01 4.61915851e-01 2.32727602e-01 2.36716956e-01
3.44926976e-02 -3.10385022e-02 9.84182715e-01 2.03211680e-01
-6.95675790e-01 7.33512878e-01 -2.69285113e-01 6.66968822e-01
1.07031572e+00 -7.58818746e-01 -3.72897595e-01 4.02062982e-01
1.17944330e-01 7.53426433e-01 1.91811562e-01 -1.45634127e+00
-1.91652775e-02 2.62860328e-01 5.22254527e-01 -4.51048642e-01
4.80060309e-01 -7.51991212e-01 6.33880138e-01 1.16549993e+00
7.87682384e-02 4.51117754e-01 1.27170295e-01 4.07088667e-01
-3.55011463e-01 -3.08755338e-01 6.76373720e-01 5.50764836e-02
4.73345779e-02 4.18947488e-02 -7.16830194e-01 8.68187100e-02
1.14285314e+00 -5.97833395e-01 2.17378139e-01 -6.49076402e-02
-1.18551111e+00 -2.00268447e-01 -3.05410363e-02 6.73942342e-02
6.78059876e-01 -9.87151027e-01 -8.26595545e-01 4.25761104e-01
2.31435344e-01 -4.71246332e-01 2.47576922e-01 1.47044623e+00
-8.14779162e-01 3.45595449e-01 -5.08254528e-01 -9.24827993e-01
-1.67833388e+00 3.02923582e-02 5.23898721e-01 -2.93840975e-01
-6.34090960e-01 5.37412941e-01 -4.91689235e-01 3.59222922e-03
1.09116919e-01 -3.43641400e-01 -7.64927149e-01 2.60805905e-01
3.15353692e-01 6.56365812e-01 3.29720318e-01 -4.87678617e-01
-5.85652411e-01 7.21056402e-01 5.83846450e-01 2.18835086e-01
1.33193302e+00 2.32095987e-01 -2.71082997e-01 6.67970896e-01
7.77829826e-01 -1.52108029e-01 -2.46788979e-01 2.98378646e-01
-1.16060460e-02 -1.85670301e-01 -5.24648800e-02 -9.52525675e-01
-8.35149944e-01 7.97547400e-01 1.16933918e+00 4.47991192e-01
1.34868801e+00 -5.51387429e-01 2.43049443e-01 5.16672075e-01
2.92128980e-01 -7.10641921e-01 -4.35362756e-01 3.09583824e-02
9.45286572e-01 -8.75244200e-01 1.33027688e-01 -4.98126477e-01
-7.61637330e-01 1.34452903e+00 -1.17713235e-01 -2.57621288e-01
1.21777415e+00 8.26622173e-02 2.35087767e-01 -1.37590900e-01
-4.64304358e-01 2.96302974e-01 2.98647255e-01 5.71182430e-01
4.44474071e-01 3.02441806e-01 -9.56864476e-01 9.03030217e-01
-2.39937410e-01 2.40497559e-01 6.94802105e-01 1.04748285e+00
-5.06416112e-02 -1.16865563e+00 -5.20616949e-01 9.47934031e-01
-8.06066334e-01 4.36375476e-02 -6.32424280e-02 6.23739481e-01
1.20925263e-01 8.53307486e-01 -3.38478804e-01 -3.60690087e-01
3.64081442e-01 4.65748310e-01 4.64034170e-01 -5.74688911e-01
-1.21231282e+00 1.69585541e-01 1.77026257e-01 -2.97242910e-01
-2.48749822e-01 -8.41800928e-01 -1.06734848e+00 3.17178190e-01
-2.37482131e-01 5.28793633e-01 8.01155806e-01 5.91335177e-01
3.60232741e-01 6.79441810e-01 6.04716837e-01 -7.74586976e-01
-4.99478519e-01 -9.41298246e-01 -1.02117729e+00 2.67005533e-01
3.64364237e-01 -5.13692081e-01 -4.96781558e-01 1.54900759e-01] | [14.20549488067627, 3.1997601985931396] |
b1808b1f-c6f4-40ad-a3d6-974c87232d2f | fast-training-method-for-stochastic | null | null | http://proceedings.neurips.cc/paper/2021/hash/d5397f1497b5cdaad7253fdc92db610b-Abstract.html | http://proceedings.neurips.cc/paper/2021/file/d5397f1497b5cdaad7253fdc92db610b-Paper.pdf | Fast Training Method for Stochastic Compositional Optimization Problems | The stochastic compositional optimization problem covers a wide range of machine learning models, such as sparse additive models and model-agnostic meta-learning. Thus, it is necessary to develop efficient methods for its optimization. Existing methods for the stochastic compositional optimization problem only focus on the single machine scenario, which is far from satisfactory when data are distributed on different devices. To address this problem, we propose novel decentralized stochastic compositional gradient descent methods to efficiently train the large-scale stochastic compositional optimization problem. To the best of our knowledge, our work is the first one facilitating decentralized training for this kind of problem. Furthermore, we provide the convergence analysis for our methods, which shows that the convergence rate of our methods can achieve linear speedup with respect to the number of devices. At last, we apply our decentralized training methods to the model-agnostic meta-learning problem, and the experimental results confirm the superior performance of our methods. | ['Heng Huang', 'Hongchang Gao'] | 2021-12-01 | null | null | null | neurips-2021-12 | ['additive-models'] | ['methodology'] | [-6.33398667e-02 -3.23861897e-01 -6.24868810e-01 -2.58765340e-01
-9.95876312e-01 -1.48671582e-01 2.45433912e-01 -1.57323942e-01
-8.87854844e-02 7.52894342e-01 6.38781637e-02 -3.74831975e-01
-3.41493301e-02 -5.56585312e-01 -1.10242891e+00 -8.07695150e-01
8.17226395e-02 6.29785061e-01 4.22443449e-02 -1.72151159e-02
1.01779088e-01 1.33612037e-01 -1.31629956e+00 2.96404392e-01
8.45898032e-01 1.10810137e+00 3.98568034e-01 5.95789433e-01
-1.94083631e-01 9.46616113e-01 -4.63718414e-01 -3.86087656e-01
5.08775041e-02 -4.05466467e-01 -6.33303404e-01 -1.12137266e-01
3.17809016e-01 -3.76791716e-01 -1.55312762e-01 1.03858054e+00
6.77912235e-01 -1.17643006e-01 4.66380775e-01 -1.32507789e+00
-2.95587093e-01 1.25764334e+00 -5.44327021e-01 -6.40411377e-02
-3.10856730e-01 -9.14728716e-02 1.04362905e+00 -7.03296900e-01
2.21201196e-01 1.16228855e+00 6.84657812e-01 7.19015181e-01
-1.11056936e+00 -7.14873374e-01 7.84316540e-01 -1.31001174e-02
-1.27955079e+00 -4.83215004e-01 9.75010991e-01 -1.85222566e-01
8.09832871e-01 8.81333128e-02 4.90898430e-01 1.08307850e+00
-8.26150775e-02 1.44977403e+00 1.15192890e+00 -4.22047228e-01
6.61967456e-01 3.59313130e-01 -1.14772715e-01 8.64863217e-01
1.75596073e-01 -2.32178062e-01 -6.33435190e-01 -5.62814295e-01
6.33090615e-01 2.46665478e-01 1.63183570e-01 -4.25443828e-01
-1.11392999e+00 7.59733617e-01 2.41004080e-01 2.61234432e-01
-2.47243881e-01 7.02531993e-01 4.71539289e-01 3.04444760e-01
7.32030809e-01 8.92781168e-02 -8.21526885e-01 -1.71715528e-01
-1.10192716e+00 1.54834554e-01 1.13799441e+00 1.18075633e+00
7.72819519e-01 2.95165181e-01 1.71644151e-01 7.99132526e-01
4.55250770e-01 6.32354558e-01 5.72048724e-01 -9.62030351e-01
5.69010258e-01 4.05153185e-01 1.02399588e-02 -8.54034662e-01
-2.28708759e-01 -7.69385517e-01 -1.18669665e+00 -3.71853948e-01
1.08927436e-01 -4.58057940e-01 -2.62727499e-01 1.84068036e+00
5.39738119e-01 4.68864679e-01 -6.25457615e-02 7.67951906e-01
4.92260367e-01 7.47444630e-01 -2.53776520e-01 -5.06375492e-01
8.66458833e-01 -1.44416642e+00 -5.73917031e-01 -1.34729072e-01
9.49256420e-01 -5.26502728e-01 1.17603779e+00 2.09706768e-01
-1.16746783e+00 -1.64471492e-01 -1.03221250e+00 2.49872118e-01
-1.55048911e-02 1.90871939e-01 1.09265530e+00 8.21467996e-01
-8.48626673e-01 5.34611166e-01 -1.15963697e+00 -1.07308567e-01
2.96549290e-01 5.31652510e-01 2.83332556e-01 4.61331196e-02
-7.33710468e-01 3.65208328e-01 3.26670706e-02 8.24710801e-02
-1.09382331e+00 -8.61643016e-01 -5.46155453e-01 -5.21877073e-02
4.95381981e-01 -1.16929102e+00 1.46806490e+00 -1.09361744e+00
-2.03185701e+00 4.50733274e-01 -5.52032053e-01 -6.03748262e-01
5.89383125e-01 -1.22397467e-01 -4.92495708e-02 -3.82174820e-01
-1.18340619e-01 -1.10451147e-01 8.80670488e-01 -1.19645000e+00
-6.56398296e-01 -3.72626275e-01 1.13439955e-01 3.95542756e-02
-9.07384098e-01 5.38448468e-02 -4.50133979e-01 -6.89739048e-01
-8.67847428e-02 -9.72968161e-01 -6.27395809e-01 -2.18412966e-01
-4.73669469e-01 7.20137209e-02 7.74160147e-01 -1.86195910e-01
1.40719855e+00 -2.03253508e+00 4.19640154e-01 1.28341213e-01
2.93980479e-01 -2.28249226e-02 8.58479291e-02 5.42256534e-01
4.64719772e-01 1.57529682e-01 -3.92490804e-01 -9.72397685e-01
2.82317221e-01 1.51712552e-01 -4.26274002e-01 4.50346619e-01
-4.83561724e-01 9.76247370e-01 -8.81292403e-01 -5.83104253e-01
-9.89733711e-02 1.13161191e-01 -7.54515588e-01 1.56855211e-01
-6.29802108e-01 2.31634259e-01 -6.97521687e-01 6.52241349e-01
5.45403302e-01 -7.61887908e-01 4.78316188e-01 -1.31981567e-01
2.35218719e-01 2.93227375e-01 -1.29704893e+00 1.94479501e+00
-9.27002966e-01 1.40989169e-01 4.48382080e-01 -1.25706804e+00
4.42888081e-01 2.31774703e-01 7.85772145e-01 -3.67810696e-01
5.03778458e-02 5.72222412e-01 -3.81150126e-01 -1.50012583e-01
3.09444904e-01 -1.79031700e-01 -1.28985286e-01 9.64601159e-01
-8.41823965e-02 -1.23132415e-01 -1.18746087e-01 2.67497987e-01
8.03886473e-01 -1.42639905e-01 1.46564636e-02 -3.11086088e-01
2.26263911e-01 -2.40518987e-01 6.16689861e-01 1.03231132e+00
2.05320150e-01 2.83741087e-01 2.43897483e-01 -4.81227696e-01
-7.87484646e-01 -7.28943110e-01 3.38583022e-01 1.21215558e+00
5.78355566e-02 -6.68280363e-01 -8.76591206e-01 -7.52127588e-01
7.41222352e-02 4.07692373e-01 -3.10120046e-01 -7.73423631e-03
-5.49201071e-01 -1.20799923e+00 1.43082500e-01 5.23055613e-01
4.13842976e-01 -3.90310705e-01 -2.13530049e-01 2.84755945e-01
-5.90287484e-02 -1.01744282e+00 -5.86383998e-01 1.36184782e-01
-1.44019544e+00 -7.26539969e-01 -5.10420501e-01 -8.35412204e-01
6.44321382e-01 3.16796333e-01 1.01182365e+00 2.76670139e-02
1.13840960e-01 3.43212575e-01 2.34258715e-02 -4.41715628e-01
-4.63931859e-01 5.49341440e-01 1.47301286e-01 1.69131979e-01
-2.01318488e-01 -9.52669919e-01 -5.89623272e-01 3.68043572e-01
-6.86447620e-01 3.25270236e-01 5.82370400e-01 9.11076009e-01
8.33819866e-01 1.86041504e-01 5.25532544e-01 -1.20842695e+00
5.30488670e-01 -5.71465611e-01 -7.32307434e-01 2.63520807e-01
-6.83030903e-01 2.59591520e-01 1.04617846e+00 -7.16048539e-01
-9.15565670e-01 2.33041495e-01 3.72836776e-02 -4.17478025e-01
3.27566177e-01 6.21991217e-01 -2.67245620e-01 -1.65041804e-01
3.82685542e-01 5.08557796e-01 -1.03509873e-01 -6.31019235e-01
4.97635394e-01 6.78906620e-01 -3.43386605e-02 -1.04475355e+00
5.92318535e-01 8.17326427e-01 8.69860947e-02 -6.82063162e-01
-7.81645060e-01 -2.27191299e-01 1.59361027e-02 1.30977303e-01
1.05455510e-01 -1.16852701e+00 -8.38710129e-01 4.50866371e-01
-9.24307764e-01 -7.99492896e-01 -2.86310077e-01 4.31524277e-01
-8.67001235e-01 2.27941662e-01 -7.90523589e-01 -9.48958337e-01
-6.49816990e-01 -1.30308688e+00 1.19832778e+00 -2.27114424e-01
1.02974728e-01 -1.22578847e+00 -1.34838130e-02 2.31358424e-01
6.51394546e-01 -1.39945641e-01 9.96378541e-01 -3.65369231e-01
-8.62388611e-01 -1.47971034e-01 1.37483239e-01 1.73262820e-01
8.78875554e-02 -1.47241995e-01 -7.22436011e-01 -4.26565081e-01
4.06470746e-01 -4.23014015e-01 7.43842423e-01 5.08984089e-01
1.44217813e+00 -5.56963027e-01 -4.54280108e-01 8.07750940e-01
1.43246388e+00 -3.56114596e-01 5.80868311e-02 1.10538818e-01
8.13355565e-01 7.78442174e-02 2.28345081e-01 6.40356660e-01
6.81547761e-01 7.13713586e-01 4.02355105e-01 -1.42117351e-01
6.88303113e-02 -4.81611252e-01 3.97518873e-01 1.35433424e+00
1.78367928e-01 -4.97923605e-02 -5.38189709e-01 5.08873045e-01
-2.46750665e+00 -7.70548105e-01 8.82324204e-02 2.14722013e+00
7.99848795e-01 -1.68608069e-01 3.48429888e-01 -4.68486585e-02
5.30686557e-01 2.86283612e-01 -7.36379981e-01 -2.27181524e-01
2.37920741e-03 -1.04340322e-01 5.53722799e-01 2.94494748e-01
-9.07693803e-01 1.03955698e+00 6.99180126e+00 1.25509858e+00
-1.25336576e+00 6.81408405e-01 8.28675330e-01 -7.25666225e-01
-5.21867394e-01 -1.89525168e-02 -9.11121964e-01 8.43126476e-01
1.05633891e+00 -1.49869695e-01 7.68185496e-01 1.25947034e+00
2.69698530e-01 2.25522682e-01 -1.18511105e+00 1.31341898e+00
-4.18193303e-02 -1.53651249e+00 -9.78386104e-02 -1.63323979e-03
1.36056256e+00 3.27899516e-01 2.96462566e-01 2.04744592e-01
6.33698940e-01 -6.09088004e-01 7.56323516e-01 2.17109725e-01
1.46020263e-01 -7.42355525e-01 3.93852443e-01 7.66202748e-01
-1.00235331e+00 -2.66941041e-01 -3.71783167e-01 -1.31429881e-01
2.55896658e-01 1.05375373e+00 -1.47921577e-01 2.79965371e-01
5.04587531e-01 5.94805539e-01 -1.86068654e-01 8.84651542e-01
-2.81531531e-02 1.00605762e+00 -6.74810350e-01 -4.53362584e-01
3.49253565e-02 -4.86422405e-02 3.17564398e-01 1.00160825e+00
3.30491096e-01 -3.79634976e-01 2.17203096e-01 7.09478140e-01
-5.25383234e-01 3.36424440e-01 -4.49813604e-01 -7.69272586e-03
3.94790888e-01 1.02426422e+00 -4.39191490e-01 -4.38172221e-01
-5.71503997e-01 8.94111991e-01 5.49617350e-01 3.26141655e-01
-8.68125975e-01 2.24642053e-01 5.54771602e-01 -2.68211681e-02
1.34750172e-01 -3.39686900e-01 -5.98731637e-01 -1.55703700e+00
3.44055086e-01 -1.05804050e+00 2.42831439e-01 -1.91898122e-01
-1.35615277e+00 1.37816474e-01 -1.79469958e-01 -8.42372060e-01
-3.81531507e-01 -2.51662016e-01 -4.60999101e-01 4.88269418e-01
-1.26101506e+00 -1.19896054e+00 1.72866836e-01 5.11651278e-01
4.28374499e-01 -3.05265576e-01 7.24237382e-01 4.53318387e-01
-6.77915812e-01 8.65110815e-01 7.92330921e-01 -3.60820234e-01
4.14695084e-01 -1.04224098e+00 3.55519742e-01 6.73712611e-01
1.04098588e-01 7.98607469e-01 6.29063189e-01 -4.35241520e-01
-2.18870711e+00 -1.10082209e+00 6.34032845e-01 -2.23634928e-01
9.58346665e-01 -6.81810319e-01 -3.93334687e-01 6.49424016e-01
-5.32069840e-02 1.51678817e-02 8.43888998e-01 3.74671102e-01
-2.38990113e-01 -4.75308299e-01 -8.88968647e-01 7.29575634e-01
1.23698807e+00 -5.80555320e-01 2.88927734e-01 8.62630606e-01
9.86775756e-01 -4.32172030e-01 -8.83499503e-01 2.46454939e-01
5.41216195e-01 -5.43370426e-01 8.29729140e-01 -5.52408218e-01
5.09500027e-01 -1.83409691e-01 -4.78933752e-01 -1.25924945e+00
1.04642421e-01 -8.47113550e-01 -8.53677273e-01 1.19316936e+00
5.80659926e-01 -7.81400919e-01 1.25804341e+00 5.70896745e-01
-2.04023588e-02 -1.09488857e+00 -8.91590118e-01 -9.13992524e-01
1.78021371e-01 -6.32898152e-01 9.02172089e-01 7.69865692e-01
-1.64438128e-01 2.37576365e-01 -8.92375171e-01 4.03338298e-02
7.52986729e-01 6.71192169e-01 1.10957456e+00 -7.89048791e-01
-1.08612573e+00 -5.76050818e-01 5.00780204e-03 -1.47834194e+00
3.98045421e-01 -8.46905231e-01 -1.74152046e-01 -1.26341295e+00
5.27056217e-01 -6.67212307e-01 -3.35533917e-01 1.43616199e-01
-2.02164635e-01 -1.89266086e-01 3.75099145e-02 4.50880796e-01
-9.47957397e-01 7.63783157e-01 9.45294082e-01 -4.93774086e-01
-2.34698087e-01 3.53853852e-01 -1.01592290e+00 5.44705629e-01
9.16509688e-01 -5.98343790e-01 -6.57022476e-01 -8.99950624e-01
3.52654904e-01 5.46626397e-04 7.91791733e-03 -7.46272683e-01
4.65958565e-01 -3.01876158e-01 -2.16967300e-01 -2.83441067e-01
2.44176984e-01 -1.01314616e+00 2.21855909e-01 5.00126243e-01
-2.23480895e-01 -3.25571410e-02 6.05912367e-03 6.14551306e-01
-1.82046100e-01 -3.93655756e-03 6.02931976e-01 -1.87164530e-01
-6.46479726e-02 6.81707442e-01 -2.70777285e-01 1.69435963e-01
7.31505871e-01 2.51235306e-01 -1.00407496e-01 -4.55541790e-01
-4.16400164e-01 3.84801775e-01 5.67182124e-01 -8.36782996e-03
3.54316205e-01 -1.33626270e+00 -5.35980046e-01 -7.76651353e-02
-1.43731818e-01 1.37900203e-01 3.83355767e-01 1.00854218e+00
-1.71066999e-01 2.04058364e-01 6.11404181e-01 -6.43187344e-01
-8.96973789e-01 8.32075596e-01 3.89468640e-01 -4.79202360e-01
-2.40610495e-01 7.40723133e-01 1.49959818e-01 -5.65256953e-01
2.79669940e-01 -2.04485983e-01 7.45097339e-01 -3.59161586e-01
5.84265530e-01 5.09482086e-01 1.08164266e-01 -2.64880061e-01
-3.00968051e-01 3.95606548e-01 -6.39182106e-02 -1.59536466e-01
1.34165871e+00 -5.47461361e-02 -4.01039124e-01 6.48507595e-01
1.52056980e+00 6.25625327e-02 -1.03542483e+00 -6.85139716e-01
-3.19966167e-01 -2.01514527e-01 6.95116119e-03 -3.19526881e-01
-1.24295545e+00 9.34748292e-01 3.67512256e-01 -1.06628574e-01
1.14155376e+00 2.45849025e-02 8.50726247e-01 6.34933412e-01
7.81708241e-01 -1.21772504e+00 -7.61104375e-02 3.54838550e-01
4.18181956e-01 -1.25068593e+00 9.59905982e-02 -3.02964509e-01
-3.30674499e-01 6.10328734e-01 4.41913784e-01 3.05695776e-02
8.48227143e-01 4.95524436e-01 -2.74049580e-01 9.79658961e-02
-9.75760520e-01 1.51862979e-01 -1.86345011e-01 2.90637463e-01
2.18292609e-01 3.44671547e-01 -2.51679659e-01 6.82300210e-01
-7.29799494e-02 6.05554990e-02 -2.62823831e-02 1.00245523e+00
-1.97495177e-01 -1.28749251e+00 -9.76842046e-02 5.23098171e-01
-4.62337315e-01 -1.71043143e-01 -2.65608698e-01 3.41036469e-01
-4.10336047e-01 9.45937216e-01 -3.35344791e-01 -3.11370283e-01
1.23352848e-03 -1.06883846e-01 6.09685481e-01 -5.33490002e-01
-5.96235752e-01 2.12557316e-01 -1.22419305e-01 -5.32428920e-01
-4.10644323e-01 -4.54969496e-01 -9.40741181e-01 -5.29052198e-01
-3.67129356e-01 2.64723808e-01 1.04198658e+00 9.73826349e-01
5.75881362e-01 3.72091532e-01 8.63766074e-01 -8.30897689e-01
-9.05034423e-01 -5.51471055e-01 -6.23703539e-01 8.28419551e-02
1.31545737e-01 -3.39565039e-01 -2.12839708e-01 -2.48463303e-01] | [6.251470565795898, 5.006959438323975] |
e95cbb6e-d5a9-4e86-b557-91e283a6b066 | ultra-light-deep-mir-by-trimming-lottery | 2007.16187 | null | https://arxiv.org/abs/2007.16187v1 | https://arxiv.org/pdf/2007.16187v1.pdf | Ultra-light deep MIR by trimming lottery tickets | Current state-of-the-art results in Music Information Retrieval are largely dominated by deep learning approaches. These provide unprecedented accuracy across all tasks. However, the consistently overlooked downside of these models is their stunningly massive complexity, which seems concomitantly crucial to their success. In this paper, we address this issue by proposing a model pruning method based on the lottery ticket hypothesis. We modify the original approach to allow for explicitly removing parameters, through structured trimming of entire units, instead of simply masking individual weights. This leads to models which are effectively lighter in terms of size, memory and number of operations. We show that our proposal can remove up to 90% of the model parameters without loss of accuracy, leading to ultra-light deep MIR models. We confirm the surprising result that, at smaller compression ratios (removing up to 85% of a network), lighter models consistently outperform their heavier counterparts. We exhibit these results on a large array of MIR tasks including audio classification, pitch recognition, chord extraction, drum transcription and onset estimation. The resulting ultra-light deep learning models for MIR can run on CPU, and can even fit on embedded devices with minimal degradation of accuracy. | ['Theis Bazin', 'Philippe Esling', 'Adrien Bitton', 'Tristan Carsault', 'Ninon Devis'] | 2020-07-31 | null | null | null | null | ['drum-transcription'] | ['music'] | [ 2.94460118e-01 -7.48061910e-02 2.39319131e-02 1.45791486e-01
-7.57876337e-01 -6.12653434e-01 3.50416183e-01 1.62250727e-01
-6.41098738e-01 3.90310585e-01 2.33096983e-02 -1.84450328e-01
-2.55379468e-01 -4.93520260e-01 -5.21975040e-01 -6.52418196e-01
-1.07979812e-01 3.05020928e-01 2.10363746e-01 -2.85410464e-01
2.87465513e-01 2.17472881e-01 -1.91666806e+00 4.10592943e-01
3.62588406e-01 1.08845007e+00 -1.53884813e-02 9.32551265e-01
1.61154777e-01 7.60836780e-01 -5.56065023e-01 -5.20129323e-01
2.08300963e-01 -2.40925819e-01 -7.86034405e-01 -4.20319170e-01
4.90559220e-01 -1.95149496e-01 -2.61392117e-01 7.16487944e-01
8.30264747e-01 -6.58041704e-03 5.40731847e-01 -6.79420531e-01
-9.74883437e-02 1.01097500e+00 -5.89051664e-01 -3.24732885e-02
1.88690469e-01 -1.29886672e-01 1.54239869e+00 -8.49040806e-01
3.02236646e-01 7.79306114e-01 1.03674304e+00 3.89649749e-01
-1.47567606e+00 -6.42535806e-01 -1.23468488e-01 4.16599900e-01
-1.61204708e+00 -8.31725180e-01 7.33647704e-01 -2.64085561e-01
9.89068151e-01 5.67864358e-01 7.33900011e-01 8.95935535e-01
2.16592271e-02 7.09361851e-01 7.20993400e-01 -7.44540691e-01
2.06886217e-01 -1.62076965e-01 6.03422150e-02 4.82029200e-01
1.38718709e-01 -2.00365245e-01 -9.58797276e-01 -1.06939390e-01
5.06543219e-01 -3.10315579e-01 -2.25683242e-01 -1.57602653e-01
-1.15854609e+00 5.81043065e-01 8.50308016e-02 2.90807694e-01
-9.89969373e-02 3.72988611e-01 7.55118370e-01 5.09532869e-01
2.70766407e-01 7.77259171e-01 -6.31529033e-01 -4.44396466e-01
-1.33366728e+00 3.89279664e-01 8.77112091e-01 5.80722511e-01
2.80571193e-01 2.67948717e-01 6.15992062e-02 1.03753245e+00
-1.26224995e-01 3.09885629e-02 5.78686953e-01 -1.05299926e+00
1.43837228e-01 8.53457972e-02 -1.20199345e-01 -9.14793730e-01
-6.50856972e-01 -1.03921032e+00 -1.00597620e+00 1.57403961e-01
5.50304711e-01 1.33781493e-01 -5.30451894e-01 1.90421879e+00
9.20673683e-02 1.49524687e-02 -3.43562335e-01 8.27757478e-01
5.15576839e-01 4.66811061e-01 -2.88610518e-01 -2.93058574e-01
1.44644606e+00 -8.49165380e-01 -6.32185042e-01 -6.56903088e-02
5.29682159e-01 -1.07839167e+00 1.25640512e+00 1.22487581e+00
-1.40092528e+00 -6.31524801e-01 -1.42241251e+00 -2.29948759e-01
2.43812110e-02 1.72168508e-01 8.49448919e-01 5.47922134e-01
-8.96180034e-01 1.30367351e+00 -6.50342882e-01 7.63147604e-03
1.59293115e-01 6.15111530e-01 -3.83202657e-02 2.87697643e-01
-1.02863717e+00 6.54670000e-01 2.74269283e-01 5.70715033e-02
-5.12577593e-01 -6.95883334e-01 -4.80704725e-01 4.29750890e-01
4.47762758e-01 -6.61197126e-01 1.63012516e+00 -8.04614961e-01
-1.51078987e+00 6.52530253e-01 -3.89597192e-02 -7.98708558e-01
3.30285966e-01 -5.45322061e-01 -2.30076343e-01 9.77724567e-02
-5.69137871e-01 5.84442079e-01 1.17250526e+00 -8.94660711e-01
-2.97832191e-01 -1.75338343e-01 -1.79778207e-02 9.99091100e-03
-7.10332692e-01 -1.03324383e-01 -2.87489772e-01 -8.76193583e-01
2.02473402e-01 -1.04973805e+00 2.61123851e-02 -2.81701505e-01
-2.72368670e-01 -6.79955631e-02 3.29142720e-01 -6.37494743e-01
1.77650154e+00 -2.07896471e+00 2.94932991e-01 -1.66736003e-02
3.77190679e-01 2.25545600e-01 -1.24896824e-01 4.62364614e-01
-1.15910895e-01 3.29291672e-02 -2.07547173e-01 -5.08393228e-01
1.92783609e-01 -1.14723518e-02 -5.03524542e-01 2.85041213e-01
-3.65918018e-02 6.83248818e-01 -4.66792941e-01 -3.36125851e-01
6.41204938e-02 4.60192472e-01 -9.02490377e-01 -2.19719455e-01
-3.82524468e-02 1.45238385e-01 3.41200262e-01 4.88622099e-01
2.26156101e-01 -7.55620375e-02 2.58131087e-01 -3.03424239e-01
-1.66157201e-01 8.65107477e-01 -1.41456282e+00 1.86682165e+00
-5.01440108e-01 8.21941972e-01 1.31897897e-01 -9.65010762e-01
6.59126639e-01 4.11168277e-01 5.02581537e-01 -5.40494204e-01
1.16746515e-01 4.02485579e-01 4.55244124e-01 -8.68205503e-02
7.11918950e-01 -1.77475303e-01 -1.88771814e-01 5.79354942e-01
1.43881932e-01 -1.38952270e-01 2.26555809e-01 -4.60634418e-02
1.10416043e+00 1.16284434e-02 1.63527623e-01 -3.27196568e-01
2.65693188e-01 -3.65888298e-01 5.37635028e-01 8.49580348e-01
2.24715136e-02 7.79016554e-01 4.98629719e-01 -5.22159874e-01
-1.19345355e+00 -9.03513134e-01 -3.24168324e-01 1.39086962e+00
-3.41640145e-01 -9.41927910e-01 -8.11702132e-01 -7.70431943e-03
-1.08118445e-01 2.53047824e-01 -3.72371465e-01 -2.30140597e-01
-9.31757867e-01 -8.97930324e-01 9.65710044e-01 3.31157267e-01
5.11583798e-02 -1.08625007e+00 -7.90948629e-01 3.56095463e-01
-6.10767975e-02 -8.42381120e-01 -2.35646337e-01 4.73794192e-01
-9.87615585e-01 -7.51952887e-01 -5.79938054e-01 -6.78227782e-01
-9.26534384e-02 -2.29826849e-02 1.19977546e+00 1.49145514e-01
-6.64233446e-01 -1.65802971e-01 -2.39320680e-01 -4.11668628e-01
-2.72407234e-01 5.21117330e-01 3.94775778e-01 -1.60831809e-01
1.34640753e-01 -1.02334058e+00 -7.35802352e-01 -9.84921604e-02
-7.75161564e-01 6.50896132e-02 6.33461595e-01 8.28180373e-01
5.09068370e-01 -1.16299978e-02 8.62016141e-01 -6.33793116e-01
5.18279791e-01 7.56629780e-02 -2.35763043e-01 -1.92021176e-01
-7.61128187e-01 6.39077649e-02 8.59089732e-01 -6.19756162e-01
-3.26711744e-01 -5.38343228e-02 -3.65043938e-01 -3.34925413e-01
9.51914489e-02 4.07850772e-01 2.82273680e-01 4.53069396e-02
6.05674446e-01 8.72362033e-02 -2.93011755e-01 -8.66261721e-01
3.03979933e-01 5.75791121e-01 6.36967063e-01 -4.32068020e-01
5.75535834e-01 3.16963822e-01 2.00575113e-01 -9.81274486e-01
-8.67193699e-01 -3.34270656e-01 -7.00734556e-01 -7.36490488e-02
4.95784402e-01 -8.10525835e-01 -9.52733219e-01 3.25544298e-01
-1.04523098e+00 -1.52239382e-01 -4.16519463e-01 5.22473514e-01
-6.19605184e-01 4.23162192e-01 -1.09951794e+00 -8.44668150e-01
-7.51988709e-01 -8.31834316e-01 8.63898754e-01 -2.23571584e-01
-7.15449631e-01 -4.34822142e-01 1.45659864e-01 2.15947956e-01
4.40823078e-01 -1.38358951e-01 1.22035336e+00 -5.46744823e-01
-2.35157251e-01 -2.66753286e-01 1.81850456e-02 4.94243085e-01
-2.32346848e-01 -1.18174478e-01 -1.48745990e+00 -2.87803352e-01
1.48372605e-01 -3.71815026e-01 1.23788452e+00 3.77539545e-01
1.38882685e+00 -2.00274140e-01 1.67032942e-01 6.63374424e-01
1.21431518e+00 -3.16158533e-02 5.99007905e-01 1.65020391e-01
5.95336914e-01 4.93376285e-01 1.74827218e-01 5.75058579e-01
-1.73162401e-01 9.29790914e-01 3.03647071e-01 3.13358777e-03
-3.24074149e-01 -8.40700418e-02 2.78868884e-01 1.57513833e+00
-4.22714174e-01 -2.01846547e-02 -7.67558634e-01 4.43415105e-01
-1.77520990e+00 -1.00070727e+00 -1.10748656e-01 2.53333831e+00
1.05639040e+00 5.00039577e-01 2.62510836e-01 9.92928088e-01
2.80874312e-01 2.70360678e-01 -3.82909447e-01 -6.72066569e-01
-9.07437317e-03 8.63483489e-01 2.43152946e-01 3.49802822e-01
-1.14684474e+00 7.48576403e-01 6.79561520e+00 1.13494647e+00
-1.10000408e+00 1.35830075e-01 3.90096217e-01 -8.13650846e-01
3.89697291e-02 -1.64836138e-01 -5.36178887e-01 1.91884503e-01
1.09571612e+00 1.05220512e-01 7.29130983e-01 6.10555112e-01
8.71700719e-02 8.67066458e-02 -1.28035069e+00 1.24772799e+00
9.28170886e-03 -1.26365364e+00 7.02307597e-02 -1.33510008e-02
3.80240083e-01 2.37106271e-02 4.51853499e-02 3.94304961e-01
-4.30178761e-01 -1.14367199e+00 9.87248361e-01 3.89881104e-01
7.41433561e-01 -1.05802417e+00 5.85258305e-01 2.90174901e-01
-1.16353440e+00 -2.42647529e-01 -4.05034959e-01 -5.62400222e-01
-1.50922406e-02 7.01742470e-01 -4.25674558e-01 3.30020249e-01
6.16864860e-01 3.47943187e-01 -4.83734697e-01 1.09379101e+00
1.28301501e-01 7.73028851e-01 -4.13930774e-01 9.44127887e-02
-7.44219795e-02 8.14870149e-02 5.24251640e-01 1.39429474e+00
3.52339476e-01 -3.20375562e-01 -2.10446745e-01 5.37229538e-01
-1.98056370e-01 1.84681430e-01 -3.56002420e-01 1.54379949e-01
4.67657775e-01 1.20368779e+00 -5.95499873e-01 -3.02060068e-01
-4.72410619e-02 7.91657507e-01 3.53745401e-01 4.62545687e-03
-7.28544652e-01 -6.35618031e-01 6.47026360e-01 2.18359739e-01
3.68316770e-01 -3.06374401e-01 -6.21224701e-01 -1.05632269e+00
1.10510796e-01 -9.74321663e-01 1.12507865e-01 -3.86921823e-01
-9.68094587e-01 4.62053567e-01 -3.90701979e-01 -1.26249051e+00
-3.67323369e-01 -6.14798605e-01 -5.21635413e-01 5.24493277e-01
-1.24102139e+00 -6.30717695e-01 2.06917763e-01 2.94026226e-01
5.67844629e-01 5.18831089e-02 1.14465165e+00 8.20793808e-01
-4.79805827e-01 8.45041335e-01 5.81336301e-03 -4.99418750e-02
8.64328623e-01 -1.18161356e+00 3.47039074e-01 3.48712146e-01
7.38925815e-01 7.07886100e-01 7.80930996e-01 4.21109758e-02
-1.42244148e+00 -4.93039012e-01 9.95171249e-01 -2.97614813e-01
6.44467533e-01 -7.34636068e-01 -8.31566513e-01 2.00780913e-01
8.90707076e-02 -4.83897299e-01 7.72225559e-01 6.88473165e-01
-5.60924590e-01 -3.31920594e-01 -4.87596780e-01 7.82351434e-01
1.05905223e+00 -6.83752835e-01 -5.67674816e-01 1.77786037e-01
7.46525168e-01 -8.29768777e-02 -8.26629519e-01 6.04333758e-01
1.13430119e+00 -1.14183831e+00 9.94121611e-01 -4.96606797e-01
4.70856607e-01 -5.66681325e-02 -1.60070211e-01 -8.07246149e-01
-2.83107430e-01 -9.19238687e-01 -5.93129277e-01 9.93250072e-01
4.49360251e-01 -9.53027532e-02 6.97347045e-01 8.94536674e-02
-3.84505749e-01 -9.67409015e-01 -1.17770445e+00 -8.29460025e-01
5.52294701e-02 -8.01985025e-01 3.25295419e-01 8.94861639e-01
3.55803937e-01 6.64059222e-01 -6.38342261e-01 -4.33621287e-01
5.18161595e-01 1.79395512e-01 7.13400066e-01 -1.44114530e+00
-8.73504519e-01 -8.95372450e-01 -2.48157293e-01 -8.93694341e-01
-2.44492009e-01 -8.45486462e-01 -1.15668073e-01 -1.06498766e+00
1.99756950e-01 -3.78204525e-01 -6.24709070e-01 4.66082692e-01
1.94737405e-01 8.74520540e-01 3.85306776e-01 4.11181003e-01
-5.17593861e-01 2.92934150e-01 7.86545396e-01 1.97521709e-02
-2.07838222e-01 5.70935532e-02 -6.14686430e-01 1.15578532e+00
8.96393657e-01 -4.60115701e-01 -1.60542443e-01 -5.92433751e-01
6.64623082e-01 -1.60026148e-01 3.08038473e-01 -1.39437139e+00
1.92709520e-01 4.89844680e-01 3.72364163e-01 -4.32325780e-01
7.68520713e-01 -4.28310126e-01 -1.72680046e-03 5.37636995e-01
-5.68576217e-01 4.63312976e-02 3.84123921e-01 2.91893214e-01
-1.68975383e-01 -3.52912337e-01 6.92559183e-01 4.51865904e-02
-1.78415418e-01 -7.87467957e-02 -2.82968342e-01 -2.26187557e-01
4.12604034e-01 3.00085098e-02 -7.49612674e-02 -3.43553662e-01
-1.04448593e+00 -4.67151582e-01 1.32555738e-01 3.19131553e-01
2.78079361e-01 -1.16325891e+00 -6.57509565e-01 2.87585050e-01
-2.30212197e-01 -2.33870670e-01 3.16954345e-01 9.85387087e-01
-5.05687237e-01 6.75566852e-01 -3.47426336e-04 -4.98131305e-01
-1.47385621e+00 5.29956639e-01 5.90403713e-02 -4.31525081e-01
-8.12762618e-01 8.05134892e-01 4.61167004e-03 -1.32737383e-01
6.02826357e-01 -3.33432347e-01 7.91143328e-02 3.70461434e-01
5.72247982e-01 4.12458867e-01 4.93945897e-01 -1.16766572e-01
-1.75232470e-01 6.31437838e-01 -1.43471733e-01 -1.97797745e-01
1.39190388e+00 2.35042080e-01 -8.11865926e-02 7.05916405e-01
8.68302464e-01 3.61778796e-01 -9.48220730e-01 -1.52620086e-02
9.44100693e-02 -1.01947598e-01 1.90022230e-01 -7.28730261e-01
-7.53802240e-01 1.25097632e+00 4.94542658e-01 3.41926277e-01
1.21129811e+00 -1.37505189e-01 1.04021871e+00 5.52393854e-01
4.29049671e-01 -1.27164805e+00 -6.22758344e-02 5.29621005e-01
7.11204767e-01 -7.44925618e-01 1.76389843e-01 -3.65373380e-02
-1.33404031e-01 1.10229015e+00 1.77996740e-01 -3.24253201e-01
4.48816895e-01 4.16525275e-01 -2.24695519e-01 6.81612119e-02
-1.01111400e+00 -4.13104966e-02 3.25997055e-01 9.29903388e-02
7.30014086e-01 2.49371380e-02 -4.63031501e-01 9.39596951e-01
-6.41738296e-01 -2.45171815e-01 3.99429888e-01 6.60034478e-01
-5.74776411e-01 -1.15373254e+00 -2.51423150e-01 4.45345730e-01
-8.69215965e-01 -5.31337738e-01 -3.02541941e-01 6.39945269e-01
3.28635015e-02 8.11177254e-01 3.02569102e-02 -6.52913749e-01
1.78687513e-01 4.55125093e-01 7.64766335e-01 -3.35679621e-01
-8.43860805e-01 4.06516820e-01 1.09725475e-01 -3.64807338e-01
-1.28031760e-01 -5.38592339e-01 -9.25763249e-01 -4.79402214e-01
-3.24235737e-01 1.30521869e-02 7.40224600e-01 7.56520152e-01
3.67739499e-01 7.18980610e-01 3.36263448e-01 -1.14677870e+00
-8.34097028e-01 -1.15168715e+00 -6.28019273e-01 1.48950621e-01
4.26435411e-01 -3.82525951e-01 -3.85596693e-01 -6.08849265e-02] | [15.766778945922852, 5.30109977722168] |
01592014-b79e-4a5a-92a4-78ba2d6f32aa | advancing-direct-convolution-using | 2303.04739 | null | https://arxiv.org/abs/2303.04739v1 | https://arxiv.org/pdf/2303.04739v1.pdf | Advancing Direct Convolution using Convolution Slicing Optimization and ISA Extensions | Convolution is one of the most computationally intensive operations that must be performed for machine-learning model inference. A traditional approach to compute convolutions is known as the Im2Col + BLAS method. This paper proposes SConv: a direct-convolution algorithm based on a MLIR/LLVM code-generation toolchain that can be integrated into machine-learning compilers . This algorithm introduces: (a) Convolution Slicing Analysis (CSA) - a convolution-specific 3D cache-blocking analysis pass that focuses on tile reuse over the cache hierarchy; (b) Convolution Slicing Optimization (CSO) - a code-generation pass that uses CSA to generate a tiled direct-convolution macro-kernel; and (c) Vector-Based Packing (VBP) - an architecture-specific optimized input-tensor packing solution based on vector-register shift instructions for convolutions with unitary stride. Experiments conducted on 393 convolutions from full ONNX-MLIR machine-learning models indicate that the elimination of the Im2Col transformation and the use of fast packing routines result in a total packing time reduction, on full model inference, of 2.0x - 3.9x on Intel x86 and 3.6x - 7.2x on IBM POWER10. The speed-up over an Im2Col + BLAS method based on current BLAS implementations for end-to-end machine-learning model inference is in the range of 9% - 25% for Intel x86 and 10% - 42% for IBM POWER10 architectures. The total convolution speedup for model inference is 12% - 27% on Intel x86 and 26% - 46% on IBM POWER10. SConv also outperforms BLAS GEMM, when computing pointwise convolutions, in more than 83% of the 219 tested instances. | ['Guido Araujo', 'José Moreira', 'José Nelson Amaral', 'João P. L. de Carvalho', 'Marcio Pereira', 'Rafael Sousa', 'Victor Ferrari'] | 2023-03-08 | null | null | null | null | ['blocking'] | ['natural-language-processing'] | [ 6.80616423e-02 -2.21861809e-01 -2.16714218e-01 -4.35669601e-01
-3.88001233e-01 -3.25289428e-01 3.42592150e-01 1.14748538e-01
-6.44981623e-01 3.43666762e-01 -1.17419370e-01 -1.29866874e+00
-1.04883406e-02 -9.20343637e-01 -8.04008543e-01 -6.15166128e-01
-1.55277058e-01 4.91568863e-01 7.55251572e-02 1.54249236e-01
1.49202213e-01 6.30773008e-01 -1.77237010e+00 8.31441581e-01
3.34468395e-01 1.22401595e+00 9.89466533e-02 1.17157972e+00
-4.81089920e-01 7.57502913e-01 -2.13763803e-01 -4.49199557e-01
4.65887010e-01 2.68141270e-01 -9.33360517e-01 -2.19890326e-01
4.10002649e-01 -1.86118588e-01 -4.46870364e-03 8.23208869e-01
1.47227541e-01 -2.86056012e-01 4.57327634e-01 -9.23719406e-01
2.98878998e-01 8.84146273e-01 -6.35259628e-01 2.06166565e-01
-3.98047119e-01 2.49866307e-01 7.33826041e-01 -5.73261380e-01
2.39907876e-01 1.12021148e+00 7.35593915e-01 2.12771103e-01
-1.53330374e+00 -6.26875758e-01 -5.86445928e-01 -2.47668587e-02
-1.48783219e+00 7.13775381e-02 -1.04820858e-04 -6.95326746e-01
1.72929215e+00 9.87235427e-01 7.79173493e-01 4.09528345e-01
6.50139153e-01 3.22614789e-01 1.30986500e+00 -6.38578951e-01
5.84590614e-01 -2.93014292e-03 7.16740370e-01 8.19390357e-01
3.59045923e-01 4.36756432e-01 -8.61069560e-02 -3.71536762e-01
8.12140524e-01 -2.38346353e-01 3.59139562e-01 2.74983972e-01
-1.15336955e+00 9.66678739e-01 3.51749122e-01 2.56599765e-02
-2.92513907e-01 6.09112620e-01 1.00939202e+00 2.29583949e-01
4.40647155e-01 5.09495378e-01 -8.81340444e-01 -4.46482629e-01
-1.22537446e+00 5.85186839e-01 1.00442290e+00 1.05758667e+00
1.00688314e+00 2.77370512e-01 -2.20781416e-01 3.54957074e-01
1.48880318e-01 3.52568388e-01 4.03886378e-01 -5.45718372e-01
2.89604366e-01 6.03343606e-01 -3.08489293e-01 -2.59776562e-01
-5.94316781e-01 -7.62297750e-01 -8.25465679e-01 4.58091557e-01
2.19787970e-01 -3.19330066e-01 -9.33719099e-01 9.42109168e-01
3.78604203e-01 6.96494356e-02 1.36445835e-01 6.67606413e-01
6.49188817e-01 6.49308383e-01 1.23854689e-01 1.62543833e-01
1.98073339e+00 -1.23168039e+00 -1.56776428e-01 9.99819413e-02
1.29453468e+00 -1.35213852e+00 7.79209495e-01 4.29862648e-01
-1.03493977e+00 -7.49244630e-01 -1.16348994e+00 -9.05144587e-02
-2.29526743e-01 3.42958212e-01 1.24738812e+00 1.03119040e+00
-9.21556890e-01 6.16488636e-01 -1.14577901e+00 2.09220767e-01
3.34087610e-01 6.86379135e-01 -5.48571870e-02 7.35826567e-02
-3.85844082e-01 6.29987001e-01 7.69145489e-01 -1.98725387e-01
-4.42324966e-01 -1.52132165e+00 -5.36566794e-01 1.40870467e-01
7.32811689e-02 -6.99966490e-01 1.27951717e+00 -8.20150375e-01
-1.16728616e+00 7.10025072e-01 -3.14695418e-01 -1.03265071e+00
1.89050451e-01 -1.39563158e-01 -2.26087585e-01 -3.35263461e-01
-3.26887727e-01 4.55659837e-01 5.65755010e-01 -6.35009289e-01
-6.43249929e-01 -3.72755826e-02 4.66153547e-02 -2.30277434e-01
1.60268843e-02 2.34499753e-01 -2.46208072e-01 -6.47226572e-01
-2.17900142e-01 -1.19243586e+00 -4.36905980e-01 -4.26782101e-01
-5.55968702e-01 6.25666529e-02 6.21345043e-01 -5.91143548e-01
1.17591131e+00 -1.90845108e+00 -4.78469878e-02 4.07283962e-01
2.17273504e-01 6.00173473e-01 1.49403915e-01 2.10225716e-01
-5.14332831e-01 -1.77699178e-01 2.01576516e-01 -2.62639254e-01
-3.17938626e-02 3.51873368e-01 -4.62192625e-01 1.39344886e-01
2.98814359e-03 7.60295928e-01 -3.07406038e-01 -1.80515990e-01
4.44516510e-01 4.92895156e-01 -9.48839843e-01 2.88735598e-01
-3.49425614e-01 -1.24581493e-01 -1.40256792e-01 4.00102973e-01
1.01887250e+00 -1.49730012e-01 2.89829616e-02 -7.07437158e-01
-7.16181934e-01 3.85719270e-01 -1.12777281e+00 1.37583292e+00
-6.82392180e-01 5.31661332e-01 -4.71611731e-02 -8.25130641e-01
8.33768070e-01 3.48445475e-02 1.42065808e-01 -4.10438418e-01
3.01203460e-01 2.76028186e-01 3.58416885e-01 -1.89589262e-01
6.16810739e-01 1.60018817e-01 8.59197751e-02 3.24620694e-01
-6.03883192e-02 6.72054514e-02 4.24573421e-01 -9.33210179e-02
1.31180871e+00 2.36920994e-02 8.94720852e-02 -1.00022376e+00
4.74771529e-01 6.62536204e-01 2.30048463e-01 4.34090227e-01
6.44630373e-01 8.77671838e-02 7.08965659e-01 -1.01177287e+00
-1.34204388e+00 -8.01303029e-01 -5.90049803e-01 1.22771037e+00
-6.81425393e-01 -9.01069403e-01 -1.26959741e+00 1.89044781e-03
-6.17115125e-02 1.08722329e+00 -3.07092041e-01 2.75164664e-01
-6.56017900e-01 -1.20198750e+00 6.91282511e-01 6.72856510e-01
5.36395073e-01 -4.89150763e-01 -1.00456071e+00 2.78498262e-01
6.03323281e-01 -1.05678916e+00 -1.74006715e-01 8.59699070e-01
-1.16675866e+00 -7.89715827e-01 1.65975422e-01 -4.49634761e-01
6.52039886e-01 -5.28740175e-02 1.11589849e+00 1.75300509e-01
-8.54686439e-01 -3.37084711e-01 -2.84314156e-01 -4.65714157e-01
-6.34457767e-01 2.20143452e-01 -2.31513560e-01 -4.26708043e-01
6.29180074e-01 -5.07078171e-01 -1.92555070e-01 8.19805041e-02
-6.92639589e-01 8.86896908e-01 7.36018002e-01 9.85799074e-01
7.62068391e-01 1.04155831e-01 -5.76262772e-01 -9.49336946e-01
-7.54068047e-02 -1.78486407e-02 -1.22677886e+00 -6.48632944e-02
-6.42450511e-01 5.40071249e-01 7.35316634e-01 -4.32150900e-01
-7.42947340e-01 3.63112956e-01 -3.98789018e-01 -6.62460566e-01
-6.92489445e-02 3.74292672e-01 1.12018637e-01 -2.21991837e-01
7.39584744e-01 6.44841939e-02 1.29020184e-01 -5.48997700e-01
1.40391365e-01 4.67869043e-01 3.35149378e-01 -8.36433828e-01
6.66002333e-01 1.97154507e-01 3.84597003e-01 -1.00198090e+00
-5.23712993e-01 -2.96866536e-01 -3.93472642e-01 3.73342454e-01
1.11018991e+00 -1.01977313e+00 -9.91803229e-01 4.21023697e-01
-1.15898263e+00 -7.97110617e-01 -3.53701562e-02 5.30839741e-01
-1.68031514e-01 -7.54832998e-02 -7.55853117e-01 -5.17618835e-01
-8.43995273e-01 -1.77105939e+00 9.20463800e-01 -4.89750020e-02
-3.24748486e-01 -7.70796001e-01 -3.65974128e-01 5.16681194e-01
5.55860639e-01 5.83629571e-02 1.21505725e+00 -5.34505486e-01
-8.41430008e-01 1.04230635e-01 -5.91434538e-01 5.34904540e-01
-5.22867918e-01 1.26839846e-01 -9.52325463e-01 -2.10753366e-01
-2.83859283e-01 9.71575230e-02 6.80913627e-01 3.85686904e-01
1.38321924e+00 -2.88497835e-01 -2.94152319e-01 1.18264759e+00
1.87690687e+00 -3.07327751e-02 8.10547292e-01 2.05695897e-01
8.29237223e-01 -7.15487730e-03 4.97307897e-01 4.53394502e-01
-9.51206125e-03 8.63321066e-01 5.77737570e-01 -2.48032555e-01
-5.80470674e-02 2.37454787e-01 1.10420004e-01 8.76571178e-01
-4.02693689e-01 6.03759766e-01 -9.60342884e-01 -1.37491468e-02
-1.47411335e+00 -5.77308297e-01 -9.10197735e-01 2.17495775e+00
9.27175879e-01 5.06464303e-01 -2.70287152e-02 2.50956386e-01
2.02081412e-01 -3.59544009e-01 1.77742839e-02 -1.29935181e+00
4.62093979e-01 1.12366509e+00 1.25242531e+00 5.64045727e-01
-1.09984970e+00 9.95087564e-01 5.11194849e+00 1.21991241e+00
-1.35800123e+00 4.01385039e-01 6.02777481e-01 -1.97532937e-01
9.03871208e-02 1.25422433e-01 -1.43539858e+00 5.00101984e-01
1.22656393e+00 1.65375665e-01 6.33585334e-01 1.44031322e+00
-3.60761061e-02 -2.98836589e-01 -1.08381224e+00 1.02712023e+00
-3.71774584e-01 -1.88332069e+00 -2.38302410e-01 4.09138471e-01
5.38846135e-01 3.30042064e-01 -2.34222159e-01 2.96576619e-01
9.02466699e-02 -1.19715011e+00 8.25495899e-01 9.49291214e-02
9.74065900e-01 -1.16997981e+00 8.98522437e-01 1.36195600e-01
-1.40008080e+00 1.61015987e-01 -6.95402086e-01 -3.60840410e-01
-2.68246919e-01 8.64069700e-01 -1.03939962e+00 2.82753855e-01
6.55896902e-01 -1.75714210e-01 -5.39224684e-01 6.69553757e-01
2.99319148e-01 7.77603388e-01 -1.46210387e-01 -2.57197976e-01
4.33288544e-01 -1.46208972e-01 8.99290964e-02 1.78146327e+00
3.75257470e-02 -1.46359608e-01 -8.66430849e-02 7.99779594e-01
3.83106679e-01 9.96103231e-03 1.35637343e-01 8.79784971e-02
2.73871243e-01 1.53315461e+00 -7.05937564e-01 -5.35659432e-01
-4.47210580e-01 6.62203610e-01 1.51277542e-01 -9.28327069e-02
-1.16344070e+00 -5.34207702e-01 1.24375451e+00 3.03804696e-01
6.84899867e-01 -3.34323823e-01 -7.87696779e-01 -7.86536336e-01
-2.94947743e-01 -8.53028417e-01 -3.39552499e-02 -4.72978890e-01
-6.48075521e-01 8.17691028e-01 7.26029649e-02 -6.47977293e-01
-1.45699412e-01 -1.17331278e+00 -5.52211463e-01 1.23847401e+00
-1.13348651e+00 -1.12035120e+00 -2.21639365e-01 2.12157950e-01
2.93072492e-01 -4.35935467e-01 1.18971717e+00 3.50978315e-01
-4.43844169e-01 8.76090348e-01 2.23886997e-01 4.88465726e-02
-1.92620516e-01 -1.16180742e+00 7.09247291e-01 5.71947515e-01
-1.69505358e-01 8.00718963e-01 6.63781226e-01 -5.16467750e-01
-2.06511950e+00 -1.59286761e+00 8.65711987e-01 8.75324979e-02
6.98168039e-01 -3.86530191e-01 -8.33738446e-01 7.47035384e-01
3.75937000e-02 1.17690086e-01 7.40119040e-01 2.57778108e-01
-6.11644566e-01 -2.68029571e-01 -9.77640450e-01 5.27038991e-01
6.65960431e-01 -1.69846117e-01 -1.40719861e-01 6.10435605e-01
7.23008037e-01 -1.03314507e+00 -1.35267544e+00 1.25871181e-01
4.79080319e-01 -9.32855666e-01 1.07139254e+00 -4.70341951e-01
5.80264568e-01 -2.81023741e-01 -4.95644599e-01 -6.31649315e-01
-2.63969719e-01 -5.18455207e-01 -8.96772072e-02 7.39358068e-01
2.11955771e-01 -5.66147387e-01 7.46636569e-01 4.03991491e-01
-5.47632098e-01 -1.00736475e+00 -8.89595449e-01 -6.30213320e-01
5.98874316e-02 -9.92054045e-01 7.36805737e-01 5.67487895e-01
-3.30337018e-01 2.25015059e-01 1.07597604e-01 9.29733738e-03
5.37252724e-01 7.81520549e-03 1.04817128e+00 -7.15623081e-01
-9.49153423e-01 -5.49342334e-01 -6.31764174e-01 -7.49129236e-01
-2.05105799e-03 -1.21912634e+00 -5.10372043e-01 -8.93672824e-01
2.76631892e-01 -6.91634774e-01 3.58586699e-01 6.71378314e-01
3.38646889e-01 1.75537705e-01 -6.66994555e-03 -1.79772332e-01
1.33525029e-01 -4.61866319e-01 8.24257255e-01 2.52784967e-01
1.23868674e-01 -8.62408206e-02 -2.55470484e-01 6.85124576e-01
6.21895254e-01 -2.89180428e-01 -2.39113286e-01 -4.96777594e-01
2.65572429e-01 -2.13369921e-01 5.44459462e-01 -1.20248079e+00
-8.83250087e-02 7.63356164e-02 3.48255694e-01 -9.54729736e-01
3.31056237e-01 -5.49971640e-01 6.95616364e-01 9.67646182e-01
1.60775315e-02 3.71756494e-01 7.85770357e-01 -1.98333129e-01
2.44012162e-01 -4.81495023e-01 8.15446436e-01 -3.65013517e-02
-6.62417173e-01 -5.08163050e-02 -4.06083792e-01 -2.60804325e-01
8.90448928e-01 4.40071933e-02 -2.64144450e-01 6.64342403e-01
-4.16389465e-01 -4.28214133e-01 2.56070286e-01 -1.19203612e-01
1.89163908e-01 -1.13881195e+00 -6.81894541e-01 6.58399105e-01
-2.83242613e-01 3.64467390e-02 3.63825738e-01 8.23077798e-01
-1.37020826e+00 8.52268994e-01 -1.19059518e-01 -8.31311107e-01
-1.81380916e+00 5.13801515e-01 1.50641218e-01 -5.03042996e-01
-6.39470279e-01 9.73629832e-01 6.97621852e-02 -1.78879231e-01
1.64752752e-02 -8.27358425e-01 5.90392649e-01 -3.97568315e-01
9.65707362e-01 5.72590292e-01 6.60679460e-01 -3.65384340e-01
-2.51166642e-01 2.30846614e-01 -7.01493099e-02 1.96688727e-01
1.12659895e+00 8.87113035e-01 -7.07551956e-01 -7.81420842e-02
1.59920228e+00 -3.09804648e-01 -7.14330435e-01 -1.84911229e-02
-1.69233441e-01 -2.28755206e-01 4.71982986e-01 -5.97135007e-01
-9.89914119e-01 8.74882936e-01 6.62167668e-01 -1.05574876e-01
9.91705596e-01 -2.78302789e-01 7.06545830e-01 3.38421613e-01
3.52468222e-01 -8.09014916e-01 -5.51275373e-01 6.95727825e-01
4.85423267e-01 -4.83495682e-01 3.70125920e-01 -6.14351928e-01
-1.79090112e-01 1.48151171e+00 5.52870154e-01 -1.16118342e-01
8.26813996e-01 1.08192849e+00 -3.19730699e-01 3.17351483e-02
-9.41541612e-01 7.58560821e-02 2.72798359e-01 1.00038938e-01
6.25922441e-01 6.87314630e-01 -3.51503670e-01 4.55108404e-01
-7.82275319e-01 -8.44065193e-03 2.94470280e-01 9.65476990e-01
-1.99903503e-01 -1.36732912e+00 -5.64711928e-01 8.50477457e-01
-3.24011356e-01 -5.97730517e-01 6.08164370e-01 9.11859751e-01
4.78001744e-01 3.65410060e-01 7.33487010e-01 -7.69084215e-01
-8.95765796e-02 7.81803727e-02 7.71995485e-01 -6.33389711e-01
-1.23496580e+00 2.99376637e-01 3.58701468e-01 -5.29332399e-01
1.16198450e-01 -5.81347704e-01 -1.37615824e+00 -1.02682984e+00
-2.09155992e-01 -2.34812815e-02 1.30668521e+00 8.95020247e-01
3.93272638e-01 9.35812533e-01 1.04571171e-01 -8.50445330e-01
-7.40106583e-01 -7.19727635e-01 -4.55257386e-01 -2.05064386e-01
-2.35349998e-01 -3.78557771e-01 -1.79626182e-01 7.04022571e-02] | [8.422927856445312, 3.0197155475616455] |
f30da945-479d-4fda-8a1e-77023b2ecd96 | online-video-super-resolution-with | 2208.02470 | null | https://arxiv.org/abs/2208.02470v1 | https://arxiv.org/pdf/2208.02470v1.pdf | Online Video Super-Resolution with Convolutional Kernel Bypass Graft | Deep learning-based models have achieved remarkable performance in video super-resolution (VSR) in recent years, but most of these models are less applicable to online video applications. These methods solely consider the distortion quality and ignore crucial requirements for online applications, e.g., low latency and low model complexity. In this paper, we focus on online video transmission, in which VSR algorithms are required to generate high-resolution video sequences frame by frame in real time. To address such challenges, we propose an extremely low-latency VSR algorithm based on a novel kernel knowledge transfer method, named convolutional kernel bypass graft (CKBG). First, we design a lightweight network structure that does not require future frames as inputs and saves extra time costs for caching these frames. Then, our proposed CKBG method enhances this lightweight base model by bypassing the original network with ``kernel grafts'', which are extra convolutional kernels containing the prior knowledge of external pretrained image SR models. In the testing phase, we further accelerate the grafted multi-branch network by converting it into a simple single-path structure. Experiment results show that our proposed method can process online video sequences up to 110 FPS, with very low model complexity and competitive SR performance. | ['Kin-Man Lam', 'Dongsheng Li', 'Yuqing Yang', 'Yifan Yang', 'Huan Yang', 'Ningxin Zheng', 'Xinyang Jiang', 'Jun Xiao'] | 2022-08-04 | null | null | null | null | ['video-super-resolution'] | ['computer-vision'] | [ 2.95140028e-01 -3.19116265e-01 -2.07963377e-01 -3.13466012e-01
-4.99522418e-01 -4.09741104e-02 -1.08981943e-02 -4.02686298e-01
-4.99169618e-01 5.83785355e-01 -1.17537603e-01 -4.15666848e-01
-6.10666573e-02 -9.32109296e-01 -1.01973128e+00 -4.29208666e-01
-7.33932406e-02 -2.93408066e-01 9.67077672e-01 -2.36296132e-01
1.29102180e-02 4.27279711e-01 -1.33437467e+00 4.45165426e-01
8.38873744e-01 1.22340631e+00 6.45585895e-01 6.79548621e-01
6.05392121e-02 1.35275173e+00 -3.54273140e-01 -4.24797475e-01
3.24355453e-01 -3.33745658e-01 -5.74730992e-01 5.33475215e-03
3.07961047e-01 -1.05545580e+00 -8.94810379e-01 8.90603960e-01
5.55698812e-01 -2.21081078e-02 -8.99438709e-02 -8.29147696e-01
-6.39191508e-01 8.11659217e-01 -7.57256269e-01 4.97644991e-01
1.78604797e-01 1.53395265e-01 5.72824538e-01 -8.96846890e-01
4.27523047e-01 1.14681900e+00 5.55193484e-01 6.47068322e-01
-8.98545861e-01 -1.01662409e+00 2.32840031e-01 8.04607689e-01
-1.49080694e+00 -7.12524176e-01 6.56080842e-01 -3.07600386e-02
7.16851115e-01 2.03421600e-02 5.49376845e-01 6.55301630e-01
-1.08498074e-01 7.82781482e-01 8.31086218e-01 -1.14928879e-01
1.64885089e-01 -3.02528501e-01 -2.34330252e-01 6.89903975e-01
6.58259466e-02 -4.79299240e-02 -7.45904982e-01 1.79793403e-01
1.53554535e+00 2.49749850e-02 -6.31014168e-01 -9.68270972e-02
-1.29543126e+00 5.35243332e-01 4.57846135e-01 3.62588093e-02
-3.32864255e-01 4.08160508e-01 4.49417174e-01 3.62204731e-01
3.77707779e-01 -4.81478572e-01 -3.34754050e-01 -1.55411020e-01
-1.02773309e+00 -5.60674034e-02 3.28961492e-01 1.12436152e+00
8.12270045e-01 3.64211977e-01 -2.11001653e-02 9.05763805e-01
1.83370598e-02 2.96038747e-01 3.01333308e-01 -1.09116817e+00
6.17226362e-01 1.29133001e-01 8.93319398e-02 -1.04482913e+00
-6.28015772e-02 -3.70634913e-01 -1.12735903e+00 -1.58681476e-03
-1.17886983e-01 -9.45977122e-02 -6.71081901e-01 1.43848455e+00
3.04969937e-01 7.64725983e-01 1.42059356e-01 1.19957459e+00
7.92842507e-01 1.01004314e+00 -2.30036929e-01 -4.77174550e-01
1.00515401e+00 -1.19545531e+00 -6.17407799e-01 1.43382609e-01
4.92568910e-01 -7.69855857e-01 9.75278258e-01 5.27511358e-01
-1.56130564e+00 -8.49386334e-01 -1.18870032e+00 -1.32491946e-01
4.24004316e-01 2.92877495e-01 6.06710255e-01 3.43320101e-01
-1.26051295e+00 7.86475360e-01 -8.53122652e-01 1.66462287e-01
5.69721878e-01 4.78793114e-01 -9.70066339e-02 -2.35846817e-01
-1.17806423e+00 2.81502545e-01 5.23329198e-01 3.09642702e-01
-9.89351928e-01 -7.41575539e-01 -7.28889465e-01 2.91100442e-01
6.95238531e-01 -4.58805442e-01 1.33785403e+00 -1.20503271e+00
-1.96148968e+00 1.84541002e-01 -1.85914561e-01 -5.75707376e-01
5.47316194e-01 -3.42263192e-01 -6.36436760e-01 5.16069353e-01
-4.56812352e-01 5.06160796e-01 1.05488026e+00 -9.37471271e-01
-8.88026774e-01 1.05065204e-01 4.89882678e-01 7.76817352e-02
-5.62006474e-01 3.51388514e-01 -7.69594491e-01 -8.13879609e-01
7.04197362e-02 -6.83081150e-01 -2.20791608e-01 2.63464570e-01
1.77413911e-01 3.09207924e-02 8.96609247e-01 -8.73355567e-01
1.37695432e+00 -2.13872504e+00 -1.85491738e-03 -1.46465832e-02
3.40054959e-01 6.07371449e-01 -1.07503578e-01 -9.44511816e-02
3.36951166e-02 -1.33817986e-01 -7.11031035e-02 8.42935443e-02
-3.57234508e-01 8.13556239e-02 -4.21338826e-01 1.88353598e-01
1.67038813e-01 6.24413192e-01 -7.60807574e-01 -6.71904683e-01
9.96454135e-02 7.38301098e-01 -7.76150465e-01 2.29084134e-01
-7.74700344e-02 1.89436093e-01 -3.21506500e-01 3.85280281e-01
9.38259900e-01 -4.32112932e-01 1.28564507e-01 -6.45157993e-01
-2.31886357e-01 1.88776851e-01 -1.26349413e+00 1.67496026e+00
-6.65398836e-01 4.83791232e-01 1.59063369e-01 -8.84291172e-01
9.36260879e-01 2.66481608e-01 2.35268667e-01 -8.43985319e-01
2.24406943e-02 2.22625643e-01 -8.53596404e-02 -3.17017436e-01
6.09053373e-01 2.25513846e-01 4.82428223e-01 1.78097948e-01
-1.79653287e-01 5.72193801e-01 1.87129065e-01 3.50567162e-01
1.06951010e+00 3.18228215e-01 -2.21587461e-03 1.03598736e-01
8.40780437e-01 -2.78836638e-01 1.01639712e+00 3.97323281e-01
3.61369923e-02 5.24852395e-01 2.27829799e-01 -5.07956028e-01
-1.16927743e+00 -1.00115263e+00 2.14545444e-01 8.56141090e-01
4.61201996e-01 -4.82234478e-01 -7.87602961e-01 -2.42549106e-01
-4.94273961e-01 1.75315797e-01 -1.39530306e-03 -9.28535387e-02
-1.13937259e+00 -5.06581306e-01 3.53790760e-01 7.07205355e-01
1.08345497e+00 -7.91678786e-01 -6.85965061e-01 3.99919003e-01
-2.36852735e-01 -1.60269678e+00 -6.37766421e-01 -6.06501758e-01
-1.06275725e+00 -7.89237738e-01 -9.04615164e-01 -9.28191900e-01
4.46602523e-01 7.71826744e-01 8.56293201e-01 3.17815632e-01
-1.48414195e-01 7.24430708e-03 -5.33889055e-01 2.42389828e-01
-2.65416473e-01 -6.23650849e-02 -2.22185580e-03 3.95066947e-01
8.78494233e-03 -7.32228220e-01 -8.95509362e-01 5.45560181e-01
-1.04124629e+00 6.59366429e-01 8.07437360e-01 6.59122825e-01
8.24913502e-01 2.67168522e-01 6.10121727e-01 -5.38930714e-01
1.15706459e-01 -3.05572331e-01 -8.60286236e-01 1.70404851e-01
-2.86295146e-01 -2.58686543e-01 1.10840917e+00 -6.01670980e-01
-1.31415749e+00 -8.29192400e-02 -1.50919378e-01 -9.63169396e-01
3.55847806e-01 2.96424627e-01 -1.05500467e-01 -3.89434874e-01
2.65841991e-01 6.30297005e-01 -1.08268425e-01 -4.89917487e-01
1.18358478e-01 7.91934311e-01 6.55458152e-01 -4.60162520e-01
9.15046930e-01 5.66223085e-01 -1.29017718e-02 -6.48717701e-01
-6.00335479e-01 -5.50142564e-02 -2.63767987e-01 -2.97701359e-01
7.13922560e-01 -1.37950253e+00 -9.24437523e-01 5.45374334e-01
-1.11857736e+00 -4.86793995e-01 1.08525857e-01 6.51449084e-01
-7.27864265e-01 6.37970567e-01 -1.07982624e+00 -4.67713177e-01
-5.34339190e-01 -1.17459810e+00 6.18354917e-01 3.20258200e-01
6.49427891e-01 -4.21211571e-01 -5.30100763e-01 3.48703265e-01
7.88687885e-01 -1.89967200e-01 5.99512339e-01 7.83747733e-02
-1.22845244e+00 2.86624193e-01 -8.53283465e-01 5.53198814e-01
1.30930930e-01 -2.08454043e-01 -5.56802809e-01 -5.77828467e-01
-1.53117314e-01 -2.30949938e-01 7.84413874e-01 1.11076944e-01
1.72234869e+00 -2.89998382e-01 2.35050432e-02 1.00573003e+00
1.66384554e+00 3.01648527e-01 8.75492811e-01 2.46153593e-01
8.94068897e-01 8.90944526e-02 7.09286392e-01 5.20682096e-01
3.80225062e-01 7.63661683e-01 2.91644752e-01 -1.22426137e-01
-1.66549981e-01 -1.18315987e-01 7.32446969e-01 1.35407424e+00
-5.35633743e-01 7.13842213e-02 -4.13467705e-01 2.89248437e-01
-1.96477473e+00 -9.65964317e-01 1.57384768e-01 2.32130766e+00
9.05147731e-01 2.74888843e-01 1.22959819e-02 3.45962271e-02
9.24991071e-01 2.29732111e-01 -6.37785554e-01 -2.36955851e-01
-3.72727290e-02 3.76585931e-01 5.61093628e-01 2.34562322e-01
-8.27420652e-01 1.02459085e+00 5.23258543e+00 1.28937006e+00
-1.17684722e+00 2.50036776e-01 4.86232221e-01 -2.56942153e-01
-4.05211821e-02 7.29014352e-02 -8.12196612e-01 5.67434132e-01
1.07744622e+00 -2.73604989e-01 6.87940717e-01 8.85393023e-01
5.55827439e-01 1.66736081e-01 -8.48979354e-01 1.26222122e+00
3.26535925e-02 -1.70900619e+00 3.31574678e-01 -2.65883118e-01
4.44040924e-01 -2.57309020e-01 -1.83720872e-01 1.72907397e-01
-2.51945313e-02 -6.84700012e-01 6.34087622e-01 4.34572011e-01
1.13195443e+00 -1.10895169e+00 5.21773994e-01 9.43547189e-02
-1.55738902e+00 -1.84395537e-01 -7.82967210e-01 8.51912275e-02
3.56465220e-01 5.12160361e-01 -1.61126599e-01 7.15711832e-01
9.94836986e-01 7.78365612e-01 -2.99631536e-01 1.01166987e+00
1.18819825e-01 6.61393046e-01 -9.77374092e-02 4.27562833e-01
1.78206917e-02 -1.20114617e-01 2.49021247e-01 1.06069756e+00
4.89776552e-01 4.36597198e-01 1.23914696e-01 5.44165254e-01
-2.71821290e-01 9.52320769e-02 -1.08396575e-01 2.97505707e-01
3.97430480e-01 1.10539496e+00 -5.38689554e-01 -5.48227251e-01
-8.60249519e-01 1.33640862e+00 2.85270482e-01 2.06032708e-01
-1.23250806e+00 -5.01053393e-01 5.79057217e-01 2.39044800e-01
6.63283587e-01 -3.64432693e-01 3.64158928e-01 -1.42585993e+00
3.38584304e-01 -9.40247118e-01 8.66195485e-02 -8.83009791e-01
-8.46718550e-01 5.61643660e-01 -1.67434305e-01 -1.49035418e+00
5.39825223e-02 -3.19686353e-01 -3.10517758e-01 5.57586074e-01
-2.09340596e+00 -9.86486912e-01 -6.51424468e-01 1.03974771e+00
8.02782238e-01 -4.37175520e-02 2.22882494e-01 7.95302153e-01
-6.11770689e-01 7.48936832e-01 -2.15872228e-02 2.05060646e-01
6.51525378e-01 -5.99266171e-01 3.65476429e-01 9.38968599e-01
-2.73676485e-01 4.82489765e-01 1.81783929e-01 -4.99810725e-01
-1.66548038e+00 -1.33440149e+00 3.69014084e-01 4.90675300e-01
5.92635989e-01 -1.85671762e-01 -1.06755376e+00 5.15438974e-01
-1.74278066e-01 4.04332191e-01 3.55314642e-01 -5.53135931e-01
-3.38803858e-01 -5.91390073e-01 -8.61822188e-01 6.30174220e-01
1.44640696e+00 -3.71667504e-01 -7.75936022e-02 2.06791803e-01
1.23619783e+00 -6.20762885e-01 -1.02647614e+00 5.04862964e-01
4.28557366e-01 -1.12482822e+00 1.05455649e+00 -2.35462591e-01
7.92628527e-01 -6.13133550e-01 -1.16787344e-01 -6.79967284e-01
-4.58310574e-01 -7.01783180e-01 -5.16287267e-01 9.92898345e-01
7.13259950e-02 -4.87413913e-01 8.73222351e-01 2.15768531e-01
-2.63189822e-01 -8.74427676e-01 -7.71442592e-01 -8.89920652e-01
-4.09731716e-01 -3.05404127e-01 7.06856012e-01 7.99517870e-01
-5.47988296e-01 3.96769159e-02 -8.32665324e-01 3.97319019e-01
7.56737173e-01 1.42385751e-01 8.21393073e-01 -7.35665798e-01
-6.89316988e-01 1.48629900e-02 -5.03827631e-01 -1.54954946e+00
-1.73409790e-01 -5.55920660e-01 -2.26156861e-01 -1.28268671e+00
3.00813794e-01 -5.06214917e-01 -3.79062235e-01 2.27813348e-01
-4.55873348e-02 3.61492008e-01 3.89589459e-01 3.75916332e-01
-8.36121440e-01 6.12906575e-01 1.32755804e+00 2.73885608e-01
-7.35202953e-02 -2.90642709e-01 -3.05747718e-01 8.41102123e-01
7.22970486e-01 -1.01835400e-01 -5.94192028e-01 -8.81022751e-01
4.34742793e-02 5.32374918e-01 4.34660941e-01 -1.30514061e+00
3.44851643e-01 -2.09948286e-01 3.73833537e-01 -4.94658142e-01
3.05750489e-01 -8.47039759e-01 2.84233361e-01 3.55184823e-01
-1.47257000e-01 1.38156250e-01 1.94017272e-02 6.17349267e-01
-1.83317557e-01 8.65715444e-02 9.51146781e-01 -8.60231835e-03
-1.05520713e+00 7.23256290e-01 -7.23415390e-02 -2.79282779e-01
9.36830819e-01 -3.80203009e-01 -2.39050224e-01 -2.44503438e-01
-5.72872221e-01 1.55373320e-01 5.21910429e-01 4.11583006e-01
1.04903519e+00 -1.27407801e+00 -7.88577855e-01 -2.89910473e-03
-3.00937831e-01 2.08286688e-01 8.30794275e-01 6.79500282e-01
-9.73882556e-01 1.82141080e-01 -1.84318125e-01 -4.63630021e-01
-1.29043460e+00 6.74557984e-01 1.65705547e-01 -8.63576159e-02
-9.64546442e-01 5.94876707e-01 3.41657221e-01 2.45507419e-01
1.37070149e-01 -1.83043387e-02 -1.36430949e-01 -4.73839581e-01
9.79429007e-01 4.91984040e-01 -1.78338885e-01 -5.16498506e-01
-5.68116307e-02 6.84264779e-01 -3.76842797e-01 2.22662643e-01
1.31758237e+00 -2.76785642e-01 -1.31327748e-01 8.85352790e-02
1.02383244e+00 -1.88403368e-01 -1.58378994e+00 -6.00507319e-01
-3.93998355e-01 -9.29281831e-01 3.23817492e-01 -2.98742503e-01
-1.62980437e+00 7.22330987e-01 6.94051862e-01 -3.38232487e-01
1.58743370e+00 -3.69575948e-01 1.51855409e+00 3.25259805e-01
7.76926816e-01 -1.07665217e+00 9.50393677e-02 2.46760875e-01
4.94511306e-01 -9.64814365e-01 2.58046333e-02 -7.70704091e-01
-2.99595714e-01 1.30010045e+00 8.63290370e-01 -1.40681639e-01
4.94158775e-01 2.21493527e-01 -2.48019114e-01 4.24130231e-01
-9.26595509e-01 1.32930782e-02 -3.17890465e-01 3.85538042e-01
9.22313556e-02 -1.56188324e-01 -4.67715621e-01 5.75051785e-01
1.42939553e-01 7.64997780e-01 8.32324266e-01 8.84528160e-01
-5.05901039e-01 -9.16627586e-01 -6.38897792e-02 3.23919863e-01
-5.94700694e-01 -3.02923292e-01 4.76397008e-01 4.32443142e-01
3.71930143e-03 6.68660104e-01 1.77338481e-01 -5.74216664e-01
1.79404393e-01 -6.00523472e-01 4.92474943e-01 -1.82539687e-01
-2.33781889e-01 6.11131899e-02 1.48385968e-02 -8.16974938e-01
-6.09822333e-01 -1.48188427e-01 -1.27128649e+00 -6.25570118e-01
-4.09335434e-01 -2.71557327e-02 4.34690297e-01 6.95223272e-01
5.92718482e-01 5.96410692e-01 8.92316699e-01 -8.63294065e-01
-4.19156134e-01 -4.69501257e-01 -2.61048704e-01 -7.58771896e-02
2.54396945e-01 -3.22615057e-01 -4.93387133e-02 3.02890331e-01] | [11.099864959716797, -1.773682713508606] |
495bc1c2-003c-4f6a-8e4d-2e46003f5518 | direction-aware-feature-level-frequency | 2106.07941 | null | https://arxiv.org/abs/2106.07941v1 | https://arxiv.org/pdf/2106.07941v1.pdf | Direction-aware Feature-level Frequency Decomposition for Single Image Deraining | We present a novel direction-aware feature-level frequency decomposition network for single image deraining. Compared with existing solutions, the proposed network has three compelling characteristics. First, unlike previous algorithms, we propose to perform frequency decomposition at feature-level instead of image-level, allowing both low-frequency maps containing structures and high-frequency maps containing details to be continuously refined during the training procedure. Second, we further establish communication channels between low-frequency maps and high-frequency maps to interactively capture structures from high-frequency maps and add them back to low-frequency maps and, simultaneously, extract details from low-frequency maps and send them back to high-frequency maps, thereby removing rain streaks while preserving more delicate features in the input image. Third, different from existing algorithms using convolutional filters consistent in all directions, we propose a direction-aware filter to capture the direction of rain streaks in order to more effectively and thoroughly purge the input images of rain streaks. We extensively evaluate the proposed approach in three representative datasets and experimental results corroborate our approach consistently outperforms state-of-the-art deraining algorithms. | ['Jing Qin', 'Xiao-Ping Zhang', 'Jonathan Li', 'Yiping Chen', 'Haoran Xie', 'Mingqiang Wei', 'Yidan Feng', 'Sen Deng'] | 2021-06-15 | null | null | null | null | ['single-image-deraining'] | ['computer-vision'] | [ 1.83120459e-01 -1.97683781e-01 4.00652021e-01 -2.92423695e-01
-3.56447786e-01 -5.82871735e-01 2.04476580e-01 -9.18303654e-02
-1.78224668e-01 7.74827838e-01 4.38934714e-01 1.46308377e-01
-3.33196640e-01 -1.19034851e+00 -9.09554422e-01 -8.88771713e-01
-5.00468731e-01 -3.95104229e-01 2.21460134e-01 -4.19075280e-01
1.47818968e-01 6.78253114e-01 -1.95044982e+00 3.47393394e-01
1.22479010e+00 7.50320077e-01 3.22173625e-01 7.56455064e-01
7.42497221e-02 5.89781463e-01 -4.87806350e-01 1.02450632e-01
3.41972679e-01 -4.21712548e-01 -6.01256788e-01 3.21289957e-01
1.00973439e+00 -5.21923184e-01 -4.51585919e-01 1.22992015e+00
3.77158612e-01 2.11127162e-01 3.29145640e-01 -4.49713737e-01
-7.25030065e-01 2.54863769e-01 -7.33821392e-01 5.56392252e-01
3.04591924e-01 2.15939179e-01 8.63590598e-01 -1.17329264e+00
7.79692829e-01 1.23079872e+00 6.16991878e-01 6.72170818e-02
-1.44354677e+00 -6.00804687e-01 5.29425800e-01 2.28021175e-01
-1.01867545e+00 -2.10922971e-01 1.01705468e+00 -2.28666037e-01
8.85117054e-01 3.76886189e-01 1.10489655e+00 6.78569734e-01
1.60548702e-01 5.32765508e-01 1.09213829e+00 -3.63989711e-01
-8.50040391e-02 -3.73964667e-01 3.21834981e-01 6.57424450e-01
3.50470662e-01 3.49017143e-01 -6.44999862e-01 9.34674069e-02
9.01597738e-01 -9.87242442e-04 -1.15332782e+00 -1.23731956e-01
-1.21988416e+00 6.49435520e-01 7.96813488e-01 4.94145334e-01
-7.21652031e-01 -3.84011073e-03 -1.37848839e-01 3.84313524e-01
5.19510686e-01 3.79414201e-01 -2.75088519e-01 3.41089159e-01
-1.03046596e+00 4.40442145e-01 4.49941099e-01 4.48095024e-01
1.31061578e+00 3.62190813e-01 -3.14873576e-01 7.92989492e-01
-9.91566689e-04 5.91287553e-01 1.66998699e-01 -6.90202177e-01
3.68815660e-01 3.55087548e-01 2.53383905e-01 -1.18694425e+00
-4.99634266e-01 -7.78603911e-01 -1.35271835e+00 2.96825022e-01
1.10950395e-01 -1.03095934e-01 -1.10240519e+00 1.38847566e+00
3.03770334e-01 6.67274833e-01 8.80092531e-02 1.37339747e+00
8.70409846e-01 8.74227643e-01 -1.36744171e-01 -2.71620899e-01
1.51136923e+00 -6.96734846e-01 -6.72034264e-01 -2.62199581e-01
-2.65616197e-02 -8.68404984e-01 1.00185156e+00 4.14117396e-01
-9.22812819e-01 -1.06672442e+00 -1.13651657e+00 9.76222903e-02
-6.02288768e-02 2.87270635e-01 7.52222121e-01 1.89902663e-01
-1.10252082e+00 7.28152692e-01 -5.79490840e-01 -2.09232032e-01
3.62622261e-01 -1.16759324e-02 -4.31569666e-01 -2.88846254e-01
-1.32260466e+00 6.29429162e-01 4.18248147e-01 6.45456493e-01
-7.33825445e-01 -1.06427062e+00 -1.05229497e+00 2.51453042e-01
-1.17073707e-01 -7.37474322e-01 4.54109758e-01 -8.31589878e-01
-1.14433587e+00 3.50895643e-01 -2.46712714e-01 -3.22777689e-01
2.59220693e-02 -5.87618291e-01 -6.04337275e-01 4.69784200e-01
-1.89676613e-01 5.31800032e-01 1.39392352e+00 -1.46002734e+00
-8.66726398e-01 -3.73275280e-02 1.92089394e-01 2.35353678e-01
-4.17439252e-01 -7.05971122e-01 -3.10192496e-01 -1.13626933e+00
4.79550362e-01 -3.67701888e-01 -3.85599174e-02 6.44362047e-02
-1.73352197e-01 4.41354215e-01 1.10049450e+00 -8.07064414e-01
1.25649846e+00 -2.36217356e+00 3.47799182e-01 1.16089247e-01
6.21626437e-01 4.43994582e-01 -3.22594792e-01 1.69745609e-01
-3.50903809e-01 -2.91852534e-01 -4.54042733e-01 6.12647161e-02
-4.76288348e-01 2.32184529e-01 -4.85715032e-01 5.87584853e-01
6.72560930e-01 5.57833195e-01 -8.63183141e-01 -1.73586700e-02
6.05189323e-01 8.60143423e-01 -8.08027029e-01 2.39147067e-01
8.06714073e-02 4.88691837e-01 -3.04809064e-01 5.90922654e-01
1.42883456e+00 8.13793112e-03 -2.02381790e-01 -8.38162839e-01
-2.94613361e-01 -1.28997928e-02 -1.37818432e+00 1.31338966e+00
-6.31598771e-01 4.99995261e-01 4.18738753e-01 -9.88830328e-01
8.23557317e-01 2.44755093e-02 2.68627495e-01 -6.45448685e-01
-2.48101294e-01 6.51928857e-02 -2.97149330e-01 -6.01004243e-01
4.67403173e-01 -1.42008841e-01 3.55475426e-01 1.00455172e-01
3.73208255e-01 -8.68495703e-02 1.53003782e-01 3.46943387e-03
8.25148761e-01 2.07304537e-01 1.51866600e-01 -4.52269882e-01
7.34457374e-01 -2.14558497e-01 5.62227190e-01 6.51664615e-01
-5.99648021e-02 8.90165091e-01 1.19943675e-02 -7.44634748e-01
-5.73983788e-01 -1.22061014e+00 -4.34254467e-01 8.43685389e-01
5.27514875e-01 -2.17500016e-01 -5.36787927e-01 -4.85692859e-01
1.84703171e-01 3.59654963e-01 -9.18351769e-01 -6.74608424e-02
-7.41843402e-01 -9.96142507e-01 1.89493641e-01 2.45243952e-01
1.03166997e+00 -1.03907645e+00 -7.81142592e-01 4.80714470e-01
-4.74722177e-01 -9.24438477e-01 -4.74593669e-01 3.80162328e-01
-8.59515846e-01 -9.98136580e-01 -7.94593334e-01 -9.14662719e-01
7.37584949e-01 7.55884171e-01 1.34132636e+00 3.15976262e-01
-6.33334756e-01 1.38591975e-02 -4.71370906e-01 4.03363407e-02
1.80147350e-01 -2.99176931e-01 -5.61158001e-01 3.04716617e-01
2.37727147e-02 -1.12014949e+00 -9.74820971e-01 7.26678148e-02
-9.77662086e-01 -8.31784308e-03 8.16004932e-01 1.13969147e+00
6.25135422e-01 4.28401977e-01 4.58834559e-01 -8.73671114e-01
2.84503579e-01 -2.18484804e-01 -5.77624202e-01 -5.83425798e-02
-6.67611510e-02 -4.10554893e-02 8.10546100e-01 -4.76508647e-01
-1.06178367e+00 7.35556409e-02 -1.62460715e-01 -4.80716079e-01
-2.52746344e-01 2.62156069e-01 -3.79236639e-02 -3.62179130e-01
7.94617891e-01 6.42609715e-01 -2.54436105e-01 -5.52615225e-01
4.15673822e-01 8.31455216e-02 8.94675612e-01 -3.11696112e-01
1.32670045e+00 7.60389030e-01 -1.98668972e-01 -1.07875955e+00
-8.91897023e-01 -5.11024773e-01 -6.22533381e-01 -4.58151072e-01
6.94149613e-01 -1.26893532e+00 -5.99049747e-01 6.46411240e-01
-9.53330636e-01 1.24016767e-02 -3.53794277e-01 5.68147063e-01
-1.15585655e-01 2.98862904e-01 -5.10450125e-01 -5.25961518e-01
-4.21680957e-01 -7.97758877e-01 1.16380215e+00 4.85269397e-01
2.48984128e-01 -9.03586149e-01 -1.01449732e-02 -3.18685263e-01
8.05214226e-01 4.39881027e-01 8.88879299e-01 3.53961647e-01
-7.44412959e-01 2.53937274e-01 -2.86070287e-01 2.74996608e-01
3.74731719e-01 -7.75997415e-02 -1.04979312e+00 -5.84954143e-01
1.46007061e-01 -4.68091965e-02 1.34786713e+00 5.66691577e-01
9.59269226e-01 -4.28968459e-01 -1.91815704e-01 1.12616861e+00
1.66703808e+00 -4.47507501e-01 9.11664069e-01 3.96603912e-01
6.25621796e-01 5.09049952e-01 4.64389861e-01 4.66289699e-01
2.60148883e-01 3.05729300e-01 5.55607796e-01 -7.07888722e-01
-5.34560204e-01 8.50356668e-02 1.36529446e-01 4.25757557e-01
-2.36262307e-01 -1.16078757e-01 -1.28386796e-01 7.70623028e-01
-1.63429809e+00 -1.11581671e+00 -1.17372841e-01 1.87946427e+00
7.05522895e-01 -1.41055239e-02 -1.48519725e-01 1.39641717e-01
5.99590838e-01 6.48444653e-01 -3.46292824e-01 1.09794647e-01
-4.62129325e-01 5.36484897e-01 3.40733290e-01 6.80954218e-01
-1.36618483e+00 7.06461549e-01 5.88960361e+00 5.07150531e-01
-1.25288832e+00 -3.39075267e-01 3.55438352e-01 -1.75777718e-01
-4.67730731e-01 -1.60312057e-01 -5.18942416e-01 3.61478120e-01
2.56512612e-01 9.70196724e-02 5.38917422e-01 5.37879527e-01
1.91167533e-01 -1.71964183e-01 -6.09190524e-01 7.26969779e-01
-1.96981519e-01 -1.42226493e+00 2.90786564e-01 -1.68185577e-01
6.72892034e-01 -3.10076863e-01 -1.77096203e-01 1.99638277e-01
2.66419470e-01 -9.65965688e-01 6.58945918e-01 6.55144036e-01
7.14155972e-01 -1.01099789e+00 5.96478462e-01 6.07248656e-02
-1.58624804e+00 4.52683195e-02 -5.41488171e-01 1.86776906e-01
-5.10431491e-02 1.31612527e+00 -1.98527113e-01 1.03298080e+00
1.18776596e+00 9.29639280e-01 -3.44237208e-01 9.88515735e-01
-2.91109234e-01 6.60016537e-01 -4.66226578e-01 6.37546122e-01
2.33075306e-01 -2.10129246e-01 6.92152917e-01 1.40585899e+00
3.64317209e-01 2.39250600e-01 1.43458933e-01 7.98655927e-01
9.37787071e-02 -2.41502047e-01 -4.58099246e-01 4.09118265e-01
2.85641372e-01 1.39199376e+00 -5.22965133e-01 -1.83465123e-01
-4.30094808e-01 1.10443079e+00 2.08076209e-01 6.44404411e-01
-7.20044076e-01 -7.51235247e-01 9.93514419e-01 2.06158802e-01
7.63005435e-01 -2.02705696e-01 6.10991679e-02 -1.34127152e+00
2.34274536e-01 -7.58671999e-01 2.52610534e-01 -5.96219301e-01
-1.26545286e+00 7.95523524e-01 -3.16256702e-01 -1.51900458e+00
1.60415560e-01 -2.86249161e-01 -6.39667630e-01 1.29275870e+00
-2.34974265e+00 -1.12708402e+00 -8.90183985e-01 8.39675725e-01
4.14564550e-01 2.78329760e-01 8.10690403e-01 3.83731812e-01
-4.68587205e-02 2.85560369e-01 4.00906093e-02 6.40248880e-02
7.13205874e-01 -1.23052573e+00 3.44193101e-01 1.43932223e+00
1.32457882e-01 6.37017906e-01 7.70187020e-01 -6.08974278e-01
-1.48761117e+00 -1.35462666e+00 6.15612626e-01 3.23847622e-01
2.59175509e-01 -4.08199728e-01 -1.40874767e+00 2.87968189e-01
-3.21963243e-02 4.44540769e-01 1.55419022e-01 -5.59946448e-02
-4.00867790e-01 -3.22529942e-01 -1.05168271e+00 3.15700054e-01
1.00089931e+00 -2.36272857e-01 -5.10371149e-01 2.89512277e-01
5.36317706e-01 -5.27413726e-01 -6.47032380e-01 6.92461789e-01
3.64586562e-01 -1.37307334e+00 1.19890475e+00 1.11264540e-02
4.30038065e-01 -8.30031931e-01 -9.01351497e-02 -1.67527032e+00
-9.64348555e-01 -5.91406286e-01 -3.29343110e-01 1.04015362e+00
3.40881534e-02 -7.41506755e-01 5.91651499e-01 -6.70583308e-01
-2.17081174e-01 -4.13664252e-01 -7.81696796e-01 -3.50506753e-01
-2.12173030e-01 5.61668575e-02 4.53910798e-01 1.00873935e+00
-5.20183265e-01 -2.35062353e-02 -5.26132703e-01 9.46331084e-01
8.79698753e-01 8.17607224e-01 4.81431782e-01 -1.22346365e+00
-4.27513570e-01 -1.88591599e-01 -3.26616704e-01 -1.19761419e+00
-7.44031370e-02 -4.57190156e-01 2.59281695e-01 -1.62741053e+00
-1.16798855e-01 -1.80317461e-02 -1.21054254e-01 4.34542716e-01
-5.78576148e-01 7.13244200e-01 4.83552702e-02 2.82932639e-01
4.77654114e-03 7.23249495e-01 1.72216666e+00 -8.80805328e-02
-2.32385486e-01 -2.06371784e-01 -7.77559757e-01 6.92310572e-01
3.45114976e-01 -1.50081828e-01 -2.30168998e-01 -4.88367975e-01
-2.66010791e-01 -5.54518625e-02 5.80407977e-01 -1.18495226e+00
-3.93039435e-02 -1.65114745e-01 9.95042682e-01 -6.73141360e-01
6.71476647e-02 -7.93276787e-01 8.90495405e-02 4.21989352e-01
1.46055266e-01 -4.72271442e-01 4.67264742e-01 7.40109086e-01
-6.83723152e-01 2.02294245e-01 1.22635365e+00 -1.09669179e-01
-9.30694342e-01 1.39595181e-01 -5.68556607e-01 -2.21254334e-01
7.66084254e-01 -2.86489695e-01 -4.70907152e-01 -2.58708656e-01
-9.96964395e-01 2.02593103e-01 2.82503754e-01 3.22273731e-01
6.58123434e-01 -1.00078762e+00 -1.02615917e+00 8.24254334e-01
-1.07539698e-01 2.24144384e-01 8.57487977e-01 6.83596790e-01
-6.19219184e-01 6.57175332e-02 -5.00832856e-01 -5.74936628e-01
-9.99177516e-01 2.58355528e-01 4.43864405e-01 -2.73217469e-01
-1.41690266e+00 9.15289521e-01 6.82916105e-01 -6.14094958e-02
6.30276427e-02 -4.73353475e-01 -4.23148930e-01 1.07367404e-01
8.60081971e-01 -6.71682507e-02 3.57816160e-01 -4.45927918e-01
-3.79485756e-01 9.46804523e-01 -1.65346965e-01 2.52368450e-01
1.86547470e+00 -2.25736797e-01 -1.02574132e-01 -9.10316557e-02
9.88491595e-01 1.68932930e-01 -1.57859147e+00 -2.80718714e-01
-5.40557683e-01 -7.82421529e-01 4.19676095e-01 -7.81800389e-01
-1.52039123e+00 6.39605939e-01 7.40753353e-01 1.88122064e-01
1.93471825e+00 -4.16206360e-01 6.35306001e-01 1.49913251e-01
1.42241195e-01 -6.06948137e-01 1.68498665e-01 5.32355666e-01
9.97978270e-01 -7.69871593e-01 1.84681386e-01 -7.42027938e-01
-2.03258768e-01 1.34822023e+00 5.49271703e-01 -6.36343122e-01
7.72740722e-01 5.85655868e-01 7.17454553e-02 -2.91782320e-01
-4.79680419e-01 -5.00887156e-01 4.30024981e-01 8.45479131e-01
2.82835573e-01 -3.92640084e-01 8.24760720e-02 3.80927354e-01
-1.71728969e-01 -1.41129404e-01 5.61563015e-01 9.29143906e-01
-8.21373224e-01 -7.24645913e-01 -6.91201806e-01 3.52771848e-01
-2.30022654e-01 -3.48542005e-01 6.78192526e-02 7.76116014e-01
3.14535111e-01 7.96212971e-01 1.93102613e-01 -2.53690958e-01
6.41964555e-01 -2.02958956e-01 4.93209332e-01 -3.93188536e-01
-6.92635477e-01 3.88176531e-01 -6.51789978e-02 -6.39949501e-01
-6.90262556e-01 -2.37529725e-01 -9.36867118e-01 -1.41607240e-01
-2.54659623e-01 2.67817136e-02 1.53983042e-01 5.44453442e-01
4.41556036e-01 9.23475921e-01 7.41974950e-01 -1.46283424e+00
2.99811512e-02 -1.01615918e+00 -8.11056197e-01 4.00753915e-01
1.10966086e+00 -6.39604211e-01 -7.48223722e-01 4.38870072e-01] | [10.920806884765625, -3.2304866313934326] |
1d0b175e-81c7-4065-9984-47909a9155b4 | dual-generator-face-reenactment | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Hsu_Dual-Generator_Face_Reenactment_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Hsu_Dual-Generator_Face_Reenactment_CVPR_2022_paper.pdf | Dual-Generator Face Reenactment | We propose the Dual-Generator (DG) network for large-pose face reenactment. Given a source face and a reference face as inputs, the DG network can generate an output face that has the same pose and expression as of the reference face, and has the same identity as of the source face. As most approaches do not particularly consider large-pose reenactment, the proposed approach addresses this issue by incorporating a 3D landmark detector into the framework and considering a loss function to capture visible local shape variation across large pose. The DG network consists of two modules, the ID-preserving Shape Generator (IDSG) and the Reenacted Face Generator (RFG). The IDSG encodes the 3D landmarks of the reference face into a reference landmark code, and encodes the source face into a source face code. The reference landmark code and the source face code are concatenated and decoded to a set of target landmarks that exhibits the pose and expression of the reference face and preserves the identity of the source face. | ['Hung-Yi Wu', 'Chun-Hung Tsai', 'Gee-Sern Hsu'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['face-reenactment'] | ['computer-vision'] | [ 2.87396789e-01 6.50316536e-01 1.79967657e-01 -2.76867568e-01
-6.60068810e-01 -7.29855597e-01 4.35743332e-01 -5.77651322e-01
2.65127212e-01 2.50174463e-01 2.06263751e-01 3.62226784e-01
2.93257058e-01 -7.88478732e-01 -8.56130481e-01 -8.51279914e-01
1.91048846e-01 3.93178284e-01 -7.63739496e-02 -1.18905909e-01
-1.20378509e-01 1.01159608e+00 -1.59687150e+00 5.00569306e-02
4.60974187e-01 1.21784520e+00 -2.00159296e-01 4.33493167e-01
2.57155627e-01 3.51207107e-02 -7.63237119e-01 -5.45327902e-01
7.43753135e-01 -7.21264541e-01 -3.94923836e-01 1.94808409e-01
9.72012281e-01 -1.38931975e-01 -4.51542914e-01 9.96830940e-01
6.93468511e-01 -2.98694186e-02 5.12145340e-01 -1.55934465e+00
-8.66221666e-01 8.63102600e-02 -7.65767455e-01 -5.73840559e-01
6.93584740e-01 -2.29052082e-01 6.63797915e-01 -1.18526936e+00
1.03175330e+00 1.55512357e+00 7.16116369e-01 9.05725360e-01
-1.30500221e+00 -9.59880769e-01 8.40701684e-02 -4.42323238e-01
-1.75960279e+00 -7.90965796e-01 1.10948324e+00 -2.23612934e-01
1.19956173e-01 2.08502993e-01 6.69365227e-01 1.04005110e+00
2.71433204e-01 2.46150658e-01 7.14121878e-01 -3.66352230e-01
1.18615240e-01 2.78636310e-02 -5.89978874e-01 9.07971919e-01
-1.55561334e-02 3.21705461e-01 -5.68443298e-01 -2.45102480e-01
9.45812464e-01 -1.15495259e-02 -3.82062584e-01 -8.56000006e-01
-8.80376816e-01 5.35722375e-01 6.88921988e-01 1.75287828e-01
-2.47218430e-01 2.62111396e-01 -7.01491460e-02 1.81782022e-01
3.48143727e-01 2.78575644e-02 -9.79517493e-03 4.23568875e-01
-8.86074364e-01 1.57319576e-01 6.86952710e-01 1.17349601e+00
9.87154067e-01 2.10704669e-01 -3.04795444e-01 7.69972920e-01
5.53793311e-01 8.85957360e-01 1.58169672e-01 -1.14126980e+00
3.09431016e-01 7.97847450e-01 -1.96626753e-01 -1.38188815e+00
3.61961685e-02 -4.23982769e-01 -8.29315066e-01 4.62229371e-01
8.70480537e-02 -1.22639164e-01 -9.01035845e-01 2.15855241e+00
5.43394506e-01 1.65918380e-01 -2.51475181e-02 6.81083560e-01
1.00234151e+00 4.40862834e-01 -4.75948364e-01 -3.34362648e-02
1.11247218e+00 -6.38492882e-01 -4.02283520e-01 -2.49377400e-01
9.31208059e-02 -8.69831204e-01 6.56560302e-01 -1.00517839e-01
-1.24098134e+00 -5.87210655e-01 -1.20936847e+00 -1.52224362e-01
-2.40363963e-02 4.62958604e-01 -1.86131775e-01 6.39417648e-01
-1.55723369e+00 2.95378536e-01 -2.19929308e-01 -3.85130882e-01
3.35106879e-01 5.16921282e-01 -8.85737062e-01 -3.31645459e-02
-7.52135873e-01 5.45292199e-01 -4.52555977e-02 1.87835634e-01
-1.07763958e+00 -6.60485506e-01 -1.14594030e+00 -4.85944822e-02
-4.33260873e-02 -7.09284425e-01 9.43785071e-01 -1.33678818e+00
-1.53252590e+00 1.20891368e+00 -3.25154006e-01 2.51756012e-01
6.41639769e-01 2.54908502e-01 -4.03898537e-01 1.17192857e-01
4.00369287e-01 8.31287503e-01 1.54024684e+00 -1.75906634e+00
-2.46077359e-01 -6.78406060e-01 -2.39231721e-01 1.90462440e-01
1.83923207e-02 -3.06506664e-01 -6.50756240e-01 -8.88345838e-01
5.30624509e-01 -9.68660653e-01 3.09853256e-01 5.92912078e-01
-5.76915145e-01 1.88588873e-01 1.16232383e+00 -6.13824606e-01
7.81474113e-01 -2.51166415e+00 3.42374057e-01 5.10350943e-01
1.59535632e-01 1.33703500e-01 -6.05487704e-01 3.66064429e-01
-5.45391917e-01 -1.83516756e-01 -2.54157960e-01 -6.75031006e-01
-1.80430815e-01 -8.47908203e-03 -4.16051388e-01 7.44480014e-01
6.90869316e-02 9.38305616e-01 -8.41601372e-01 -2.61265963e-01
-3.86237316e-02 8.52172852e-01 -5.89952767e-01 3.40757698e-01
4.93551232e-03 5.49600005e-01 -2.27966875e-01 5.25935888e-01
1.10929644e+00 2.67015576e-01 -4.93022241e-03 -4.50624943e-01
1.83335915e-02 -3.64617050e-01 -1.31181014e+00 1.57236743e+00
-3.39976758e-01 1.91013724e-01 5.77229559e-01 -1.18789621e-01
1.51026654e+00 3.03535819e-01 5.26978076e-01 -5.17150760e-01
1.37433201e-01 2.51139343e-01 -2.68236190e-01 -4.46655937e-02
2.29604080e-01 -2.10598290e-01 5.30412011e-02 6.13401115e-01
1.40046701e-01 -2.37566978e-01 -3.51861775e-01 2.13631064e-01
6.61902964e-01 2.38285705e-01 5.07930145e-02 -1.60114661e-01
8.67158294e-01 -6.37372375e-01 4.56292599e-01 2.07760155e-01
1.17294900e-01 1.03399777e+00 4.91501808e-01 -3.57532918e-01
-1.16700506e+00 -1.45318627e+00 4.86006178e-02 8.54102790e-01
1.75377041e-01 -2.12998316e-01 -1.08570480e+00 -6.70835912e-01
1.41680300e-01 2.87138045e-01 -8.74209464e-01 -4.86802548e-01
-5.93533099e-01 -1.12713061e-01 7.24976957e-01 2.41491348e-01
7.63328254e-01 -8.99162054e-01 -2.68108070e-01 -1.46586329e-01
6.15784302e-02 -7.40502417e-01 -1.17574191e+00 -3.81984204e-01
-5.01883149e-01 -1.23327947e+00 -7.61783600e-01 -1.11115289e+00
1.36551929e+00 1.24995494e-02 6.32786393e-01 2.62058973e-01
-1.08815283e-01 4.28052962e-01 -1.09986015e-01 -2.22116932e-02
-7.35701323e-01 -3.32929581e-01 1.73647106e-01 8.10669959e-01
-3.54104638e-01 -6.94444954e-01 -5.66387415e-01 4.27427083e-01
-9.12154794e-01 -5.94225191e-02 2.37107202e-01 6.10178471e-01
6.78167284e-01 -3.15534323e-01 5.18103719e-01 -6.25798166e-01
3.96771401e-01 -1.24485925e-01 -4.23518986e-01 1.99219957e-01
-4.30834413e-01 -9.52432230e-02 7.04587281e-01 -2.55268008e-01
-7.76948512e-01 5.24510443e-01 -2.91650921e-01 -6.97824180e-01
1.15494005e-01 -2.52314240e-01 -8.34782660e-01 -4.06670243e-01
6.95669472e-01 3.94217938e-01 3.65578532e-01 -5.25291860e-01
6.28353238e-01 4.18461591e-01 8.86181414e-01 -4.94435966e-01
1.28659463e+00 6.69126391e-01 1.56368285e-01 -5.49165070e-01
-4.79707301e-01 1.15433782e-01 -1.10276020e+00 -2.98126578e-01
5.09831548e-01 -8.20099950e-01 -6.32847369e-01 6.82334185e-01
-1.32264960e+00 1.82790786e-01 -5.86974621e-01 -3.36310029e-01
-6.43667519e-01 2.15195283e-01 -4.47134636e-02 -4.30702955e-01
-3.98915082e-01 -1.03072667e+00 1.59339702e+00 3.54681373e-01
-6.81830943e-02 -6.92549407e-01 1.71518713e-01 -1.76333398e-01
2.48285279e-01 7.06407011e-01 9.18371081e-01 -2.14122519e-01
-5.77412605e-01 -5.52069068e-01 -4.60402779e-02 2.87415922e-01
4.51319575e-01 9.26037133e-02 -1.15009129e+00 -7.41737962e-01
1.42440289e-01 -1.70168113e-02 3.96745831e-01 2.02619448e-01
6.18924081e-01 -5.05986750e-01 -3.61081272e-01 1.10510302e+00
1.15103173e+00 3.73377591e-01 6.77153707e-01 -1.57049507e-01
7.31105447e-01 6.64316177e-01 4.65497784e-02 1.85825095e-01
2.91154176e-01 7.76382804e-01 3.98650467e-01 -2.60223240e-01
-5.61368465e-01 -9.02972579e-01 5.37069321e-01 3.35068643e-01
1.33325875e-01 -2.66832635e-02 -5.78582108e-01 3.27316940e-01
-1.41405046e+00 -8.75078321e-01 3.31068486e-01 2.35687733e+00
5.29439747e-01 -5.66956580e-01 -8.07934552e-02 5.52535243e-02
9.91201401e-01 2.99677879e-01 -7.52472878e-01 -3.96800637e-01
-1.40541375e-01 1.27986372e-01 -2.27473751e-02 5.25714636e-01
-6.93022788e-01 7.86464930e-01 6.24369097e+00 5.91529906e-01
-1.23600411e+00 6.25266135e-02 6.10395968e-01 2.66229175e-02
-3.82078528e-01 -1.33759290e-01 -7.21018672e-01 2.49135718e-01
3.97180349e-01 -4.84742165e-01 5.23355484e-01 8.42295110e-01
-3.61201987e-02 2.02417165e-01 -1.23509741e+00 1.12886441e+00
6.43042803e-01 -1.12387133e+00 2.90857732e-01 1.26219362e-01
7.89252996e-01 -5.44861972e-01 4.29766983e-01 -5.61391301e-02
-1.14145532e-01 -1.11017740e+00 1.03375387e+00 4.86959666e-01
1.52676451e+00 -9.25297141e-01 3.13040227e-01 4.40537855e-02
-1.50580454e+00 -1.69301797e-02 -2.32304022e-01 3.98815811e-01
2.51366720e-02 1.75621286e-01 -7.91130602e-01 6.32679701e-01
4.40421462e-01 6.19244576e-01 -6.26934111e-01 7.64014006e-01
-3.00575942e-01 -1.85304940e-01 -7.31213465e-02 6.70912981e-01
-2.16636479e-01 -3.50144327e-01 9.43679929e-01 5.65441072e-01
5.71863413e-01 -2.55710542e-01 1.28755584e-01 1.07725394e+00
-5.69041848e-01 2.64155447e-01 -9.25636053e-01 3.32175255e-01
6.08169794e-01 1.27654338e+00 -6.23757660e-01 1.23876904e-03
-1.90706570e-02 1.27831888e+00 1.11919492e-01 3.21282089e-01
-5.01136303e-01 -2.34346181e-01 9.15687680e-01 3.61216545e-01
1.15300141e-01 1.87740147e-01 -9.09358710e-02 -6.83085501e-01
2.58404255e-01 -8.59790206e-01 2.09949985e-01 -9.12886024e-01
-1.05315483e+00 1.11366582e+00 -1.28267676e-01 -1.27831292e+00
-3.01733524e-01 -2.49052897e-01 -8.27386737e-01 1.08014119e+00
-9.72501695e-01 -1.65101326e+00 -5.25802732e-01 8.13918829e-01
1.22334976e-02 -3.06984156e-01 9.09458041e-01 1.86123580e-01
-3.04789811e-01 1.00614715e+00 -4.65837121e-02 3.94480228e-01
5.76034784e-01 -6.51170433e-01 5.45569539e-01 6.69830620e-01
3.48975584e-02 7.34341979e-01 3.15210760e-01 -7.74885297e-01
-1.44006467e+00 -1.52768683e+00 7.82378256e-01 -5.61469734e-01
5.05991615e-02 -6.68590307e-01 -5.42776763e-01 8.73297036e-01
-1.64983898e-01 1.95814639e-01 2.40749925e-01 -4.91632819e-01
-6.11676633e-01 -3.48050952e-01 -1.59596169e+00 6.15110934e-01
1.19850516e+00 -8.22103620e-01 -2.64181226e-01 -4.27567326e-02
6.87055469e-01 -7.64649510e-01 -5.70782065e-01 2.13663548e-01
7.86828399e-01 -8.37074578e-01 1.02053344e+00 -6.38563186e-02
2.39617214e-01 -6.92911267e-01 -1.03749186e-01 -1.25000000e+00
-2.38301694e-01 -6.80075407e-01 1.33160189e-01 1.49114037e+00
1.16746753e-01 -6.58665240e-01 7.76594877e-01 4.11470771e-01
-7.47502893e-02 -6.88304842e-01 -1.39836037e+00 -7.72622943e-01
3.47206928e-02 8.23567286e-02 1.03943896e+00 5.90074122e-01
-4.93752033e-01 1.42591968e-01 -3.93490613e-01 3.07488561e-01
6.22020006e-01 3.67890358e-01 7.54971564e-01 -1.12605393e+00
3.29879910e-01 -1.01162985e-01 -5.25011182e-01 -8.01615655e-01
3.80614638e-01 -1.33409429e+00 5.68430871e-02 -1.19253075e+00
-1.13108754e-02 -3.14471394e-01 3.09202224e-01 5.19418299e-01
3.55660737e-01 7.36431062e-01 3.19821477e-01 -3.47005129e-02
8.91999006e-02 8.59927416e-01 1.44249809e+00 -8.82575661e-02
-2.06190079e-01 4.98512313e-02 -9.50500131e-01 6.44727290e-01
3.12125623e-01 -4.94301587e-01 -4.15101200e-01 -2.85683483e-01
-1.26054138e-01 1.39004603e-01 5.71165502e-01 -8.98041189e-01
2.34854445e-01 2.30988473e-01 7.51691878e-01 -3.66795391e-01
5.35305142e-01 -1.02686799e+00 8.37292135e-01 3.19670975e-01
-9.03664157e-02 2.84186661e-01 8.48547518e-02 4.29359883e-01
-9.80432034e-02 -9.65204388e-02 1.18087983e+00 1.07202441e-01
-1.88098356e-01 7.52131820e-01 2.83822030e-01 -1.47168487e-01
1.20841777e+00 -6.44978404e-01 -8.87529105e-02 -5.22787452e-01
-7.44858980e-01 -5.30316085e-02 9.70749021e-01 7.79705048e-01
1.00905871e+00 -1.83353031e+00 -8.36282849e-01 1.14430475e+00
1.31914541e-01 1.23728387e-01 4.40381430e-02 2.95721263e-01
-4.41023171e-01 6.28429949e-02 -2.97193855e-01 -4.67904240e-01
-1.26954710e+00 3.90743554e-01 8.04808259e-01 4.33024853e-01
-6.43510818e-01 9.26560462e-01 4.98597294e-01 -6.68064833e-01
7.71883577e-02 1.42255276e-01 -4.21177670e-02 1.30546868e-01
6.20863616e-01 2.46633619e-01 -3.44794169e-02 -1.56731367e+00
-3.64631653e-01 1.15893638e+00 3.24023843e-01 -1.91193610e-01
1.27422714e+00 -8.39461461e-02 -5.04263163e-01 -1.12866439e-01
1.50473428e+00 3.87226999e-01 -1.30754030e+00 -5.78848496e-02
-3.39599818e-01 -7.84866452e-01 -3.38121682e-01 -6.02959037e-01
-1.43420041e+00 5.13472855e-01 6.89360857e-01 -4.52400059e-01
1.27352321e+00 1.11035213e-01 5.25468588e-01 -2.00067505e-01
8.05337012e-01 -6.31266832e-01 2.50369340e-01 4.59605396e-01
1.48297417e+00 -4.58606660e-01 -3.61576855e-01 -5.36913753e-01
-4.12025154e-01 1.03025067e+00 6.59993172e-01 -2.09525242e-01
7.55923033e-01 1.41547024e-01 1.32015884e-01 -3.46157342e-01
-2.62410134e-01 2.33053654e-01 5.75585246e-01 9.65180516e-01
6.81370944e-02 -1.69711530e-01 2.70276427e-01 3.83466780e-01
-6.04377270e-01 -1.48582131e-01 3.00038159e-01 4.73124802e-01
-9.04922932e-02 -1.18579125e+00 -7.15911269e-01 1.61020644e-02
5.21711074e-02 1.20702945e-01 -7.06649244e-01 4.98319805e-01
5.07445097e-01 6.58497572e-01 2.08172053e-01 -6.19240463e-01
6.84303820e-01 1.70754522e-01 5.18347442e-01 -8.01652908e-01
-4.24098462e-01 -6.51129112e-02 -5.25603831e-01 -7.23926127e-01
-1.64981872e-01 -4.31679666e-01 -1.22640967e+00 -4.34219152e-01
6.71014488e-02 1.64460689e-02 3.75512421e-01 4.69280690e-01
5.99487007e-01 8.49857703e-02 1.11086202e+00 -1.08892131e+00
-2.17408821e-01 -5.46677589e-01 -7.90140092e-01 5.29260695e-01
6.21763885e-01 -5.71116269e-01 -3.94518495e-01 1.43144861e-01] | [12.690739631652832, -0.1153566986322403] |
4f93a29a-84e0-4f48-8ab2-5fa328c6ca7a | ils-summ-iterated-local-search-for | 1912.03650 | null | https://arxiv.org/abs/1912.03650v1 | https://arxiv.org/pdf/1912.03650v1.pdf | ILS-SUMM: Iterated Local Search for Unsupervised Video Summarization | In recent years, there has been an increasing interest in building video summarization tools, where the goal is to automatically create a short summary of an input video that properly represents the original content. We consider shot-based video summarization where the summary consists of a subset of the video shots which can be of various lengths. A straightforward approach to maximize the representativeness of a subset of shots is by minimizing the total distance between shots and their nearest selected shots. We formulate the task of video summarization as an optimization problem with a knapsack-like constraint on the total summary duration. Previous studies have proposed greedy algorithms to solve this problem approximately, but no experiments were presented to measure the ability of these methods to obtain solutions with low total distance. Indeed, our experiments on video summarization datasets show that the success of current methods in obtaining results with low total distance still has much room for improvement. In this paper, we develop ILS-SUMM, a novel video summarization algorithm to solve the subset selection problem under the knapsack constraint. Our algorithm is based on the well-known metaheuristic optimization framework -- Iterated Local Search (ILS), known for its ability to avoid weak local minima and obtain a good near-global minimum. Extensive experiments show that our method finds solutions with significantly better total distance than previous methods. Moreover, to indicate the high scalability of ILS-SUMM, we introduce a new dataset consisting of videos of various lengths. | ['Daniel Rotman', 'Yair Shemer', 'Nahum Shimkin'] | 2019-12-08 | null | null | null | null | ['unsupervised-video-summarization', 'metaheuristic-optimization'] | ['computer-vision', 'methodology'] | [ 3.81818473e-01 -5.01232669e-02 -3.63410234e-01 -1.39623970e-01
-9.53591764e-01 -5.31704307e-01 9.33524072e-02 3.86296302e-01
-2.44615257e-01 9.32732940e-01 3.89007777e-01 1.57226309e-01
-3.72824669e-01 -6.38896286e-01 -8.23551178e-01 -6.32596195e-01
-1.88198283e-01 1.48357719e-01 4.45437610e-01 -3.56385186e-02
7.72024870e-01 2.55872726e-01 -1.64061022e+00 3.12420309e-01
1.20281255e+00 7.47389793e-01 4.05592978e-01 7.75644600e-01
-1.10881120e-01 6.60300314e-01 -9.39307749e-01 -1.16376482e-01
2.90662706e-01 -1.01356041e+00 -9.45898771e-01 5.78327835e-01
4.21179503e-01 -2.15595111e-01 -9.07183625e-03 9.69355822e-01
4.60855663e-01 5.79231799e-01 4.41452950e-01 -1.28627944e+00
-1.95328400e-01 4.98417288e-01 -5.82905829e-01 3.70614052e-01
8.60647082e-01 -1.86056152e-01 1.13471901e+00 -3.79429728e-01
9.26172376e-01 8.39944184e-01 4.17983919e-01 2.78358489e-01
-1.15282142e+00 -3.68904173e-02 1.94084138e-01 4.52647686e-01
-1.45092499e+00 -3.59386951e-01 5.72443426e-01 -7.99518544e-03
1.14167953e+00 8.37434113e-01 8.67751718e-01 3.72006387e-01
2.04077572e-01 9.44898903e-01 4.78674412e-01 -5.90524852e-01
5.86582661e-01 -6.03620056e-03 2.48752296e-01 5.59837401e-01
5.54905891e-01 -6.96745455e-01 -6.25889063e-01 -3.42837721e-01
1.95735604e-01 -1.55773312e-01 -5.51037729e-01 -5.31753302e-01
-1.10816741e+00 8.37034047e-01 -2.59584002e-03 4.06881452e-01
-4.89974499e-01 -8.96608829e-03 4.44346279e-01 2.42384836e-01
5.30599117e-01 7.75220990e-01 -8.44524354e-02 -2.92685986e-01
-1.39612603e+00 7.40053236e-01 1.06617081e+00 9.47630703e-01
6.58640027e-01 -8.45584795e-02 -2.87816852e-01 7.23853648e-01
-3.01914841e-01 1.41209081e-01 3.49353790e-01 -1.38704967e+00
4.85770315e-01 5.95972598e-01 1.96671560e-01 -1.32254863e+00
-6.37085363e-02 -4.81883213e-02 -4.31343973e-01 -2.38741711e-01
1.13502696e-01 -1.49861649e-01 -5.02631247e-01 1.55286860e+00
1.38495207e-01 2.17463136e-01 3.30726081e-03 8.22260737e-01
5.87830126e-01 1.12948740e+00 -4.45222408e-01 -1.12996042e+00
8.64352047e-01 -1.09260654e+00 -8.03110003e-01 2.53930986e-02
5.38025975e-01 -4.97648329e-01 6.40616298e-01 4.27837342e-01
-1.51340199e+00 -1.75339475e-01 -1.22861779e+00 4.31823015e-01
4.67032976e-02 -2.11182863e-01 2.40293130e-01 5.97345412e-01
-1.21989763e+00 7.05892205e-01 -5.43002069e-01 -7.57817805e-01
3.71477604e-02 3.76286596e-01 -2.23031566e-01 -2.65839070e-01
-6.22203767e-01 5.40125966e-01 8.28083575e-01 -3.01793963e-01
-3.64486992e-01 -5.06671250e-01 -6.73294425e-01 3.14559668e-01
1.03796661e+00 -7.49633551e-01 1.11257291e+00 -1.13930106e+00
-1.25425041e+00 4.44602102e-01 -5.93224585e-01 -5.26003480e-01
3.35535198e-01 1.56527536e-03 -1.03809556e-03 5.50181031e-01
1.37662888e-01 4.86523688e-01 7.51435757e-01 -1.17041421e+00
-8.15247715e-01 -8.63191485e-02 1.35675028e-01 4.37414855e-01
-5.54670572e-01 1.76024679e-02 -7.24828064e-01 -5.85004985e-01
-1.94563735e-02 -9.60678697e-01 -4.77501094e-01 -7.74989307e-01
-5.31301916e-01 -1.95012435e-01 4.90151316e-01 -4.19800311e-01
1.92911851e+00 -1.82207298e+00 6.29766047e-01 2.08718330e-01
-6.37831688e-02 2.93636203e-01 -1.71486676e-01 8.36880982e-01
2.00627014e-01 2.40998447e-01 -3.68798256e-01 -1.28291920e-01
-3.60932946e-01 1.97573677e-01 -1.70293391e-01 2.72091776e-01
-2.05417097e-01 4.83759999e-01 -1.02237153e+00 -6.94186866e-01
-2.79084388e-02 -1.38978064e-01 -8.08971524e-01 2.27856472e-01
-3.22063029e-01 -7.16561228e-02 -3.30070049e-01 5.33205926e-01
5.92416108e-01 1.34639442e-01 4.09252048e-01 -1.40301183e-01
-3.28884691e-01 -2.90444255e-01 -1.37853479e+00 1.58468616e+00
1.36889204e-01 5.80410600e-01 -1.24146067e-01 -1.22120798e+00
6.12887859e-01 1.26926035e-01 9.98853564e-01 -4.04967427e-01
6.46585226e-02 1.59778342e-01 -3.91673476e-01 -6.92322791e-01
1.08744502e+00 9.87024903e-02 -1.09640479e-01 4.83169824e-01
-2.09205776e-01 -7.23103285e-02 1.06742656e+00 4.70405906e-01
1.26548886e+00 -2.16175765e-01 5.88061035e-01 -2.32624739e-01
5.10590672e-01 3.70711952e-01 6.16648734e-01 9.13495779e-01
-2.29208507e-02 8.53982210e-01 8.08738768e-01 -1.53861091e-01
-9.68226969e-01 -7.44852960e-01 3.75827998e-01 8.32408905e-01
3.90062034e-01 -9.64084506e-01 -1.15005136e+00 -5.08774281e-01
-3.30700070e-01 7.85450339e-01 -3.37419659e-01 -1.54517695e-01
-7.07215071e-01 -7.65871704e-01 8.19329247e-02 1.71121642e-01
3.38218629e-01 -9.56557035e-01 -1.06921577e+00 3.80767226e-01
-5.73568940e-01 -9.56610858e-01 -7.15458214e-01 -2.02199325e-01
-1.00924230e+00 -1.11464798e+00 -9.29115415e-01 -7.05820262e-01
7.86091506e-01 7.49781966e-01 8.79284859e-01 8.93130079e-02
-2.83579141e-01 5.80004573e-01 -6.78202569e-01 -1.74490273e-01
-3.54037225e-01 2.58770585e-01 -3.43605429e-02 -4.46276143e-02
-1.33151526e-03 -2.46683314e-01 -3.01523149e-01 2.84965813e-01
-1.26885557e+00 -1.04466930e-01 1.85086071e-01 4.28500831e-01
7.24140704e-01 4.81964678e-01 6.82816863e-01 -6.08673096e-01
9.94008958e-01 -5.09447157e-01 -2.33556867e-01 5.29831529e-01
-2.48989552e-01 3.83482985e-02 5.54604888e-01 -3.35155308e-01
-8.07931900e-01 5.32551073e-02 2.35761642e-01 -3.20946842e-01
2.18928248e-01 7.38053799e-01 -1.47152677e-01 8.24543014e-02
3.60758841e-01 4.32223052e-01 6.40715286e-02 -1.20841488e-01
1.52605414e-01 5.12895882e-01 4.02532637e-01 -3.05873871e-01
2.55308509e-01 2.83497602e-01 3.80754285e-02 -1.30945373e+00
-7.22579300e-01 -7.72613406e-01 -3.04161251e-01 -4.54049677e-01
6.54647887e-01 -4.81030732e-01 -5.36652088e-01 2.71385610e-01
-9.46107328e-01 3.79737355e-02 -3.55183154e-01 2.52808243e-01
-8.70166004e-01 9.21704113e-01 -4.62179072e-02 -7.60357738e-01
-3.88923168e-01 -1.13112366e+00 7.11948216e-01 3.89388651e-01
-5.07924736e-01 -6.57203257e-01 2.56980598e-01 3.28142941e-01
2.38167405e-01 4.95343328e-01 7.10667312e-01 -6.57169223e-01
-6.52683020e-01 -1.74895123e-01 2.08734095e-01 3.12945247e-01
1.25989780e-01 3.00380915e-01 -3.86926644e-02 -4.63733315e-01
-8.46657157e-02 4.33261357e-02 9.24067616e-01 8.20084274e-01
9.46722209e-01 -4.84812796e-01 -2.60774970e-01 3.83453518e-01
1.68765986e+00 4.19145346e-01 6.24583304e-01 5.53705752e-01
2.95863569e-01 4.81333852e-01 1.08162892e+00 7.64721930e-01
9.85675082e-02 8.44741464e-01 3.83742094e-01 4.12519097e-01
2.22030222e-01 -1.26048140e-02 3.82959396e-01 8.03009212e-01
-2.57640749e-01 -8.72856140e-01 -4.88168031e-01 6.96496546e-01
-2.24997020e+00 -1.45061553e+00 2.35285368e-02 2.46007586e+00
4.27740037e-01 1.69442035e-02 5.87584496e-01 2.83247262e-01
8.52243841e-01 3.43918741e-01 -3.51684242e-01 -8.15210462e-01
-4.40964922e-02 -2.91916579e-01 3.70239764e-01 3.10238570e-01
-9.34653878e-01 7.65780091e-01 6.56422997e+00 9.50791001e-01
-8.43955874e-01 -3.11807007e-01 3.78252745e-01 -6.47336304e-01
-2.04003960e-01 -1.16152436e-01 -5.38228869e-01 6.29935563e-01
8.84504020e-01 -8.64733338e-01 3.97228599e-01 6.56666696e-01
4.49097306e-01 -7.81458020e-01 -7.99899578e-01 8.88767064e-01
5.97158134e-01 -1.66687810e+00 2.39493370e-01 3.06827389e-02
1.14160669e+00 -5.52892983e-01 -3.34891260e-01 -5.13552688e-02
-2.98725396e-01 -5.55134058e-01 6.82044327e-01 4.01396930e-01
4.76481915e-01 -1.24456620e+00 6.09229445e-01 4.35135782e-01
-1.16719711e+00 -2.52701968e-01 -5.07306337e-01 7.77401924e-02
5.64277232e-01 4.09870118e-01 -6.54489696e-01 9.01472092e-01
5.16670942e-01 5.89262664e-01 -3.50157708e-01 1.69319737e+00
2.32323617e-01 3.97609174e-01 -3.06849629e-01 -2.75531471e-01
3.92544359e-01 -4.86996353e-01 1.08875799e+00 1.18172896e+00
5.49908876e-01 2.95209944e-01 4.07610953e-01 2.48720691e-01
4.82153939e-03 5.09675860e-01 -6.22632861e-01 -8.81105885e-02
3.03642362e-01 8.55991304e-01 -1.16505945e+00 -3.71867806e-01
-8.62374157e-02 1.02093279e+00 6.22901656e-02 2.34276697e-01
-1.05187643e+00 -5.84917307e-01 3.83155584e-01 1.33194968e-01
4.45060343e-01 -1.54131323e-01 -1.11956902e-01 -1.06544411e+00
1.38049796e-01 -8.40985358e-01 5.09015620e-01 -5.75718403e-01
-5.04064143e-01 6.69309855e-01 4.57997113e-01 -1.19008112e+00
-4.18408453e-01 1.77704394e-01 -8.22924912e-01 1.67099804e-01
-8.30366433e-01 -4.37558979e-01 -2.98566252e-01 3.02787095e-01
1.12952280e+00 -5.88529035e-02 3.01122516e-01 -3.27333137e-02
-6.41677916e-01 3.62080395e-01 3.10428709e-01 -6.52930915e-01
5.09336293e-01 -1.03918612e+00 -2.45474294e-01 1.12147129e+00
-7.88234621e-02 4.09104675e-01 1.23562586e+00 -7.78552532e-01
-1.56150174e+00 -8.11457336e-01 8.75464499e-01 4.38422821e-02
2.32647806e-01 2.97233433e-01 -7.58682609e-01 3.85243446e-01
3.93423975e-01 -7.28760779e-01 6.24047816e-01 -2.97863930e-01
3.87752503e-01 -9.45717618e-02 -1.06783426e+00 6.04250610e-01
1.07104719e+00 2.69479990e-01 -6.92097723e-01 3.73628139e-01
6.61354423e-01 -2.31863916e-01 -4.59225297e-01 3.40495318e-01
3.72296482e-01 -1.23300457e+00 8.65335584e-01 -3.88132989e-01
5.21369755e-01 -1.60069019e-01 -1.33082911e-01 -1.49696350e+00
-1.77198336e-01 -7.66031265e-01 -2.06451178e-01 1.20795548e+00
1.01529069e-01 -1.92946345e-01 8.93618643e-01 5.05584598e-01
-1.19699501e-01 -9.44059193e-01 -8.20798635e-01 -1.03217006e+00
-5.42828560e-01 -4.31847572e-02 2.80383676e-01 4.78545725e-01
4.36071783e-01 1.97222114e-01 -5.44915318e-01 -1.87847778e-01
6.68113470e-01 4.24746752e-01 7.23562539e-01 -8.61096919e-01
-9.24370438e-02 -4.43106860e-01 -3.52749884e-01 -8.22106004e-01
1.61323205e-01 -5.67257702e-01 8.81347433e-02 -1.98065305e+00
5.79505742e-01 1.57095313e-01 6.08094558e-02 3.63617800e-02
-2.33929709e-01 8.86154324e-02 4.50554550e-01 1.62581205e-01
-1.17277598e+00 3.15129489e-01 9.48109567e-01 -2.17874069e-02
-7.30359793e-01 1.85229883e-01 -6.78918421e-01 5.78765988e-01
7.62256622e-01 -5.59758127e-01 -6.26908302e-01 -1.07128732e-01
3.18055868e-01 4.25826937e-01 -3.11923206e-01 -1.08992445e+00
3.89937669e-01 -3.66008341e-01 -3.51493716e-01 -9.14359808e-01
2.06578627e-01 -6.69874728e-01 3.81002814e-01 4.84710842e-01
-3.68274689e-01 1.33850753e-01 -1.03447251e-02 7.10857093e-01
-4.00962830e-01 -7.39005148e-01 6.20902777e-01 -1.91281781e-01
-8.49090755e-01 8.77757818e-02 -7.77433753e-01 5.92161939e-02
1.54294276e+00 -7.32821524e-01 -2.66232956e-02 -5.81932068e-01
-4.84038174e-01 3.87611866e-01 7.68261194e-01 2.20494136e-01
6.84119940e-01 -1.00156736e+00 -7.02583671e-01 -1.85968846e-01
-8.90000015e-02 -2.45480224e-01 3.67161542e-01 7.99062550e-01
-8.69148493e-01 5.10348558e-01 -2.47711822e-01 -4.67741072e-01
-1.84508467e+00 6.39059007e-01 -1.55198157e-01 -2.39186749e-01
-4.71576601e-01 5.83363473e-01 -2.28193298e-01 3.68514299e-01
2.54223675e-01 -1.76678315e-01 -1.92657843e-01 4.92702216e-01
7.53808260e-01 1.04007256e+00 -4.67592217e-02 -5.82301259e-01
-3.51260364e-01 4.76372868e-01 9.74284671e-03 -1.07714079e-01
1.56020820e+00 -2.84819365e-01 -3.58531207e-01 1.35239422e-01
1.28781343e+00 1.45798579e-01 -8.82447302e-01 1.66550279e-01
6.20008968e-02 -7.54938960e-01 -1.96513444e-01 -2.57069647e-01
-9.33061838e-01 7.87558258e-02 -1.07993715e-01 6.79599226e-01
1.44744313e+00 -1.24375634e-01 9.78771269e-01 4.59000826e-01
4.90088195e-01 -1.21375406e+00 2.43459716e-01 3.67747277e-01
8.08266461e-01 -7.42990196e-01 3.85905474e-01 -4.79916781e-01
-8.05792153e-01 1.14564538e+00 3.89723897e-01 -1.50830224e-01
4.59098779e-02 1.17953062e-01 -5.46640575e-01 2.93480046e-02
-8.32942545e-01 -1.83273032e-01 1.03274450e-01 1.99097991e-01
-1.54603794e-02 -1.06707640e-01 -9.02986169e-01 3.55822742e-01
5.95675856e-02 1.23226233e-01 1.09242630e+00 1.20410275e+00
-1.09411657e+00 -9.77076113e-01 -4.88870174e-01 5.32313466e-01
-5.23863196e-01 2.48468637e-01 -5.00983179e-01 5.26929617e-01
-2.39986613e-01 1.18538916e+00 -4.63287272e-02 -2.77405828e-01
4.28142995e-01 -8.82740095e-02 6.01547837e-01 -5.38864851e-01
-5.14962256e-01 2.07976371e-01 2.16504395e-01 -6.67857945e-01
-7.08060682e-01 -8.48967910e-01 -1.12396634e+00 -5.12353122e-01
-3.26123595e-01 5.76859891e-01 3.89010668e-01 8.70466113e-01
2.36721024e-01 4.47605044e-01 6.75902188e-01 -1.01860416e+00
-4.32337493e-01 -4.10288244e-01 -6.91792905e-01 2.38623068e-01
4.25353721e-02 -2.69256502e-01 -3.08293909e-01 9.07178745e-02] | [10.437253952026367, 0.4373369812965393] |
75ab3287-3387-4b31-8513-567792acf3b1 | malicious-network-traffic-detection-via-deep | 2009.07753 | null | https://arxiv.org/abs/2009.07753v1 | https://arxiv.org/pdf/2009.07753v1.pdf | Malicious Network Traffic Detection via Deep Learning: An Information Theoretic View | The attention that deep learning has garnered from the academic community and industry continues to grow year over year, and it has been said that we are in a new golden age of artificial intelligence research. However, neural networks are still often seen as a "black box" where learning occurs but cannot be understood in a human-interpretable way. Since these machine learning systems are increasingly being adopted in security contexts, it is important to explore these interpretations. We consider an Android malware traffic dataset for approaching this problem. Then, using the information plane, we explore how homeomorphism affects learned representation of the data and the invariance of the mutual information captured by the parameters on that data. We empirically validate these results, using accuracy as a second measure of similarity of learned representations. Our results suggest that although the details of learned representations and the specific coordinate system defined over the manifold of all parameters differ slightly, the functional approximations are the same. Furthermore, our results show that since mutual information remains invariant under homeomorphism, only feature engineering methods that alter the entropy of the dataset will change the outcome of the neural network. This means that for some datasets and tasks, neural networks require meaningful, human-driven feature engineering or changes in architecture to provide enough information for the neural network to generate a sufficient statistic. Applying our results can serve to guide analysis methods for machine learning engineers and suggests that neural networks that can exploit the convolution theorem are equally accurate as standard convolutional neural networks, and can be more computationally efficient. | ['Erick Galinkin'] | 2020-09-16 | null | null | null | null | ['information-plane'] | ['methodology'] | [ 2.93027908e-01 1.81894675e-01 -1.55732960e-01 -4.13577735e-01
8.13926905e-02 -6.63685501e-01 6.17084444e-01 -1.37320071e-01
-4.22159255e-01 5.07549822e-01 9.89824347e-03 -4.55627948e-01
-4.15939689e-01 -8.45311582e-01 -8.14265847e-01 -7.39681900e-01
-2.19787136e-01 1.30735338e-01 -8.67348462e-02 -3.37567180e-01
2.30401903e-01 6.59069300e-01 -1.57359862e+00 1.18746273e-01
5.33537865e-01 9.68944728e-01 -2.45829523e-01 5.50062180e-01
2.02240154e-01 7.10839748e-01 -5.01133740e-01 -2.88348615e-01
6.20876849e-01 -4.32144105e-01 -9.11094844e-01 -3.28837901e-01
3.83581728e-01 -9.28304046e-02 -4.19382036e-01 1.18697798e+00
1.02871731e-01 2.38833115e-01 7.57088840e-01 -1.14734697e+00
-6.25630200e-01 4.20713574e-01 -2.19870478e-01 2.03937262e-01
-5.77979572e-02 1.61409378e-01 1.01082492e+00 -4.65787053e-01
5.19763529e-01 1.14228785e+00 9.51524913e-01 4.83857840e-01
-1.38288951e+00 -7.62437522e-01 -1.49195358e-01 2.19032571e-01
-1.24160433e+00 -3.16596001e-01 8.07812929e-01 -4.94914383e-01
7.47182846e-01 3.45815867e-01 8.63142550e-01 1.06777132e+00
6.16264343e-01 1.83556974e-01 9.52562749e-01 -2.66133070e-01
2.33019084e-01 3.71260971e-01 1.08509906e-01 7.66128659e-01
6.57334983e-01 1.57864019e-01 -1.88233286e-01 -4.05858979e-02
7.40630805e-01 1.86802387e-01 -2.58306146e-01 -7.64009714e-01
-9.60756660e-01 1.11594391e+00 5.88843167e-01 5.25548577e-01
-1.47727653e-01 3.59088778e-01 4.47486192e-01 5.43712616e-01
3.31716239e-01 1.00516748e+00 -5.24527431e-01 -3.84317726e-01
-6.60076201e-01 1.38954476e-01 1.03583860e+00 2.94469178e-01
8.67249131e-01 -4.53153141e-02 5.99208951e-01 5.18554986e-01
1.53625414e-01 2.93069035e-01 6.14211857e-01 -1.13289154e+00
-3.48964520e-02 8.81170869e-01 -3.65358025e-01 -1.55857706e+00
-5.08933604e-01 -5.56624532e-01 -7.50920653e-01 5.15208781e-01
8.04160118e-01 -1.91303045e-01 -5.54563105e-01 1.84533572e+00
-6.08417541e-02 -2.71434244e-02 6.78230152e-02 6.34596586e-01
5.02600595e-02 4.51361060e-01 -3.90395850e-01 1.99734196e-01
1.11351728e+00 -2.48627827e-01 -4.17915493e-01 -1.12633035e-01
9.63902414e-01 -4.57761526e-01 1.15158403e+00 3.54146421e-01
-8.06006193e-01 -3.16446424e-01 -1.37149835e+00 8.28501359e-02
-7.32908726e-01 -3.70956838e-01 7.29529619e-01 8.68115485e-01
-9.72098589e-01 1.04164493e+00 -1.03565061e+00 -5.66438913e-01
6.35898232e-01 7.69635141e-01 -4.31764424e-01 2.60349065e-01
-1.01224625e+00 1.17812347e+00 4.08845216e-01 -5.10599352e-02
-4.64881420e-01 -7.21099436e-01 -6.66726828e-01 2.59960890e-01
9.27930176e-02 -4.71274167e-01 1.02249587e+00 -1.30582738e+00
-1.20037985e+00 4.82376963e-01 1.77129745e-01 -5.96027553e-01
2.33166769e-01 1.88181028e-02 -2.42945686e-01 -4.84876931e-02
-3.35678369e-01 6.55467391e-01 7.67924428e-01 -1.22143185e+00
-4.21293736e-01 -4.97654498e-01 2.30344608e-01 -1.66587576e-01
-6.86448932e-01 -2.11277053e-01 2.85251111e-01 -3.65209013e-01
1.40324339e-01 -1.20154834e+00 -6.44201934e-02 1.13145016e-01
-1.91435054e-01 5.16799688e-02 1.26581013e+00 -5.16934931e-01
9.31646943e-01 -2.24937487e+00 -1.82149205e-02 5.69231391e-01
4.14324135e-01 2.84023464e-01 5.58567867e-02 1.65468946e-01
-3.02746356e-01 5.26185036e-01 -3.27856123e-01 7.73959532e-02
-7.55457133e-02 2.30815217e-01 -2.75934488e-01 6.42586648e-01
3.37297052e-01 7.41316974e-01 -7.03925669e-01 4.17464003e-02
2.46998463e-02 9.34161842e-01 -8.20676088e-01 -1.70307383e-01
7.11111501e-02 3.68350744e-01 -2.60681629e-01 7.32162083e-03
3.61719877e-01 -3.12791795e-01 1.82730943e-01 -2.08501741e-01
1.41875759e-01 3.20418119e-01 -8.90467227e-01 1.24051094e+00
-5.09988427e-01 1.29614425e+00 -1.84191838e-01 -1.26470351e+00
9.17948663e-01 -1.80233698e-02 6.30233109e-01 -5.63327432e-01
3.88598144e-01 3.07069421e-02 8.02858055e-01 -3.33010107e-01
2.94673294e-01 -4.31868657e-02 1.58331960e-01 6.95467472e-01
1.32843452e-02 -4.71085915e-03 -2.70636350e-01 -9.37499553e-02
1.29288769e+00 -3.12033981e-01 1.23810656e-01 -5.17146468e-01
2.34254152e-01 -1.04372832e-03 3.78294528e-01 4.73444045e-01
2.41826777e-03 3.97084355e-01 7.23795712e-01 -7.22831070e-01
-1.26559150e+00 -8.64201903e-01 -3.40608358e-01 7.11760044e-01
-6.93854615e-02 -2.02802166e-01 -1.10292172e+00 -6.62847579e-01
-3.96363586e-02 6.33303642e-01 -1.04354441e+00 -8.21545184e-01
-5.78704536e-01 -5.61457157e-01 6.91941977e-01 2.52295762e-01
5.63986003e-01 -9.09841657e-01 -9.06585813e-01 -9.57632884e-02
3.58745933e-01 -8.56541872e-01 -1.13814488e-01 4.27913636e-01
-9.86907423e-01 -1.28947818e+00 -1.57631278e-01 -4.61654007e-01
8.00047040e-01 1.87437564e-01 7.49761879e-01 1.95732608e-01
-2.09452555e-01 4.49224979e-01 -1.25604331e-01 -5.25521994e-01
-5.63640893e-01 2.28261635e-01 2.56446302e-01 -1.06494660e-02
5.41704059e-01 -8.98278177e-01 -4.74218965e-01 3.97730410e-01
-1.02206755e+00 -2.36427173e-01 5.68590403e-01 7.61815429e-01
5.05301580e-02 4.22512144e-01 5.09144366e-01 -7.34536171e-01
7.46321440e-01 -5.72353542e-01 -3.62350017e-01 1.93035528e-02
-8.13821852e-01 5.02567470e-01 8.49896967e-01 -5.48071802e-01
-7.00446486e-01 -9.82180685e-02 1.74797028e-01 -3.28601182e-01
-1.01602137e-01 4.30280238e-01 -1.65713266e-01 -3.03987265e-01
1.01887012e+00 -1.12779647e-01 5.46515346e-01 -3.40766639e-01
1.72210380e-01 5.03859699e-01 2.74835497e-01 -4.41964239e-01
8.95518959e-01 6.29582942e-01 1.70267776e-01 -9.19045568e-01
-4.25649077e-01 1.84361711e-01 -7.14550316e-01 5.01496438e-03
9.13508773e-01 -2.86664277e-01 -8.68737042e-01 1.85583770e-01
-9.91014361e-01 -2.70552456e-01 -6.18960559e-01 4.43048954e-01
-5.13929784e-01 -6.61171377e-02 -9.15648788e-02 -6.74258113e-01
4.62260507e-02 -1.27899683e+00 4.70339864e-01 2.11688265e-01
-4.92697209e-01 -1.39999259e+00 -7.50715062e-02 6.57135341e-03
6.53059542e-01 2.62726605e-01 1.19541514e+00 -1.06358814e+00
-3.78389150e-01 -2.24997178e-01 -2.64612496e-01 6.73585415e-01
4.71987069e-01 1.00304775e-01 -1.10879123e+00 -2.50442952e-01
2.40070686e-01 8.61938223e-02 7.69537687e-01 2.21953094e-01
1.25650418e+00 -6.52624726e-01 -1.66970372e-01 6.74356222e-01
1.09433079e+00 3.46455991e-01 5.55090845e-01 4.09512460e-01
7.94419944e-01 8.80363107e-01 -1.07228488e-01 -2.30032019e-02
8.28828067e-02 4.82439339e-01 4.67220098e-01 3.52829695e-02
2.32128382e-01 -1.64038137e-01 5.09385645e-01 7.11812198e-01
-1.29506737e-01 3.19289505e-01 -1.07569528e+00 1.95120439e-01
-1.61866176e+00 -1.06348038e+00 3.67542475e-01 2.29211092e+00
6.99437797e-01 3.39014173e-01 1.24864250e-01 2.56342500e-01
5.11034966e-01 -2.23438516e-02 -8.45667005e-01 -7.35733271e-01
6.68371618e-02 3.34966242e-01 5.89519322e-01 3.09472442e-01
-8.24317932e-01 5.15260041e-01 6.40369415e+00 5.09131074e-01
-1.51504922e+00 -2.33447477e-01 7.55515695e-01 5.61361462e-02
-2.45146781e-01 1.71215400e-01 -3.44686210e-01 3.45101178e-01
1.19155490e+00 -3.58864993e-01 6.69111013e-01 1.00935161e+00
1.70453459e-01 1.18857518e-01 -1.40804243e+00 7.82309234e-01
-9.39751789e-03 -1.40904498e+00 -8.32464695e-02 6.12789333e-01
5.43909490e-01 -3.55279595e-02 5.33032894e-01 7.83938169e-02
3.19644898e-01 -1.48619342e+00 3.12215328e-01 6.73214376e-01
4.06167060e-01 -8.04069161e-01 6.05887234e-01 2.73700535e-01
-8.01439881e-01 -3.41020107e-01 -3.85809839e-01 -3.59869391e-01
-6.03126526e-01 4.26481873e-01 -1.00671256e+00 -1.39986604e-01
6.18098319e-01 7.69729137e-01 -8.25188637e-01 7.54637003e-01
2.26903394e-01 6.84762418e-01 -2.76195288e-01 -1.95925206e-01
1.11107282e-01 -2.50648677e-01 4.62873995e-01 8.62558961e-01
2.45484620e-01 -3.06534141e-01 -4.06060517e-01 1.09809315e+00
-9.78843570e-02 -1.58669621e-01 -1.14586401e+00 -3.21456760e-01
3.82806212e-01 1.15948141e+00 -8.63355219e-01 -1.42357135e-02
-3.75174493e-01 4.57222819e-01 1.03563845e-01 3.19358796e-01
-5.52758694e-01 -4.21234012e-01 1.27003753e+00 2.11698145e-01
1.26224831e-01 -2.99778163e-01 -5.15692115e-01 -9.62446451e-01
1.85406723e-04 -1.04543304e+00 -1.96260184e-01 -3.41578633e-01
-1.03352630e+00 3.33240986e-01 5.68972640e-02 -1.04468405e+00
-5.36550820e-01 -9.92443383e-01 -6.99537158e-01 3.53133947e-01
-8.89478624e-01 -5.15203297e-01 -1.26379982e-01 3.54024261e-01
7.64687732e-02 -3.68782729e-01 9.23520684e-01 2.40113623e-02
-4.78346288e-01 6.53681397e-01 2.62732476e-01 4.24789429e-01
1.99398354e-01 -1.06766546e+00 4.34961289e-01 5.23048639e-01
4.20177281e-01 1.09950459e+00 8.33633184e-01 -3.42312574e-01
-1.49836648e+00 -8.89990747e-01 2.53206760e-01 -6.23235464e-01
8.62720311e-01 -4.80647683e-01 -1.03561366e+00 7.03108490e-01
4.05515842e-02 -1.14900926e-02 6.64362192e-01 1.18355483e-01
-5.01771271e-01 -1.96749538e-01 -1.21582985e+00 6.53081059e-01
9.54326034e-01 -6.60750687e-01 -3.20493758e-01 1.74931716e-02
7.03390300e-01 1.18136302e-01 -9.33958352e-01 3.93637091e-01
8.51987720e-01 -1.01171267e+00 7.93521047e-01 -9.28187132e-01
3.10170531e-01 6.92536384e-02 -2.82048821e-01 -1.39317501e+00
-1.11971609e-01 -5.32461166e-01 -4.03837636e-02 8.17498982e-01
5.90834320e-01 -9.55744386e-01 9.73967671e-01 9.38087463e-01
1.19694255e-01 -9.64151978e-01 -8.19668591e-01 -8.89233828e-01
5.06775022e-01 -5.36815226e-01 6.45422816e-01 1.08608818e+00
5.78580871e-02 1.70371622e-01 1.08939409e-01 -1.71887830e-01
3.22598577e-01 -4.20780897e-01 7.53761292e-01 -1.54902840e+00
-1.88781664e-01 -8.11691105e-01 -1.05432129e+00 -5.03429174e-01
3.20676297e-01 -9.33730721e-01 -4.16987062e-01 -9.31899130e-01
1.00622058e-01 -4.94828492e-01 -2.33663067e-01 5.28586090e-01
3.39622527e-01 1.31565869e-01 1.98223025e-01 1.34855330e-01
2.16023345e-02 2.68485159e-01 9.49696839e-01 -7.28745461e-02
-2.66120821e-01 -6.09274209e-02 -9.94902551e-01 1.01623881e+00
1.15843892e+00 -3.97951871e-01 -5.69122970e-01 -1.72590807e-01
4.81038481e-01 -7.52529383e-01 5.68747759e-01 -1.27756786e+00
1.36892140e-01 -2.67727375e-02 5.89234352e-01 3.24909210e-01
3.56459469e-01 -1.20972216e+00 8.33425969e-02 6.02717638e-01
-5.51291347e-01 1.05074920e-01 1.81965575e-01 2.49860957e-01
7.31784105e-02 -2.33669937e-01 7.77470529e-01 6.13866076e-02
-2.99223334e-01 1.25940844e-01 -2.87710190e-01 1.18981987e-01
9.26938653e-01 -3.87904227e-01 -3.24676424e-01 -5.83939552e-01
-4.13258523e-01 -3.89336884e-01 6.81430995e-01 4.34807420e-01
4.45109785e-01 -1.28013849e+00 -2.76083738e-01 5.25337756e-01
-1.98425114e-01 -1.88429147e-01 -1.84177563e-01 7.21123517e-01
-4.52986509e-01 3.28898609e-01 -4.30399716e-01 -6.03986144e-01
-9.64666486e-01 3.70455891e-01 6.07638896e-01 1.98951289e-02
-5.81417859e-01 4.13066059e-01 6.70146868e-02 -5.30811608e-01
1.42214313e-01 -4.73218799e-01 -2.54428126e-02 3.85324173e-02
3.91838878e-01 3.45677853e-01 3.36912908e-02 -6.19812667e-01
-1.83429122e-01 4.97163355e-01 -1.25183374e-01 1.27167371e-03
1.46988523e+00 3.50958884e-01 -5.61471209e-02 5.41705310e-01
1.63152027e+00 -1.56943128e-01 -1.29876041e+00 2.12891027e-01
7.51496479e-02 -3.06539327e-01 3.36766168e-02 -3.62296909e-01
-1.25549042e+00 1.10516882e+00 7.89510190e-01 7.32170343e-01
9.04523671e-01 -1.75746456e-01 4.24560279e-01 8.84982407e-01
6.69521019e-02 -1.02218449e+00 9.26352888e-02 5.31385958e-01
7.10254014e-01 -1.07360423e+00 -1.09892050e-02 1.63851294e-03
-2.85214722e-01 1.28754127e+00 4.83560413e-01 -4.46652770e-01
9.62729335e-01 2.20571712e-01 -7.78879598e-02 -2.08817646e-01
-5.50962329e-01 2.21628651e-01 2.70036489e-01 6.55419469e-01
4.78840500e-01 -1.72543868e-01 9.25464183e-02 2.94006288e-01
-8.50048423e-01 -3.05885226e-01 5.70161104e-01 6.77674353e-01
-5.62667668e-01 -1.00903571e+00 -2.16328904e-01 7.08522201e-01
-4.13508058e-01 1.88288346e-01 -6.15221739e-01 1.21120536e+00
9.87777337e-02 7.12433040e-01 2.94197559e-01 -9.08953905e-01
7.29200840e-02 1.46797389e-01 3.59993756e-01 -5.24275362e-01
-3.58592987e-01 -7.40246832e-01 -2.16325536e-01 -6.18308485e-01
-1.29993841e-01 -6.93388224e-01 -1.41255629e+00 -6.02706254e-01
-3.02241921e-01 1.78245232e-01 1.09388530e+00 1.11404324e+00
4.89233971e-01 4.11316931e-01 6.67457163e-01 -9.10036862e-01
-6.49150252e-01 -8.18409741e-01 -3.16272020e-01 3.20295721e-01
4.84458387e-01 -7.79806137e-01 -6.65116251e-01 -1.13851339e-01] | [8.129749298095703, 3.7217400074005127] |
b4333e7a-51f1-4978-bcf7-596540099746 | decorate-the-examples-a-simple-method-of | 2204.10360 | null | https://arxiv.org/abs/2204.10360v1 | https://arxiv.org/pdf/2204.10360v1.pdf | Decorate the Examples: A Simple Method of Prompt Design for Biomedical Relation Extraction | Relation extraction is a core problem for natural language processing in the biomedical domain. Recent research on relation extraction showed that prompt-based learning improves the performance on both fine-tuning on full training set and few-shot training. However, less effort has been made on domain-specific tasks where good prompt design can be even harder. In this paper, we investigate prompting for biomedical relation extraction, with experiments on the ChemProt dataset. We present a simple yet effective method to systematically generate comprehensive prompts that reformulate the relation extraction task as a cloze-test task under a simple prompt formulation. In particular, we experiment with different ranking scores for prompt selection. With BioMed-RoBERTa-base, our results show that prompting-based fine-tuning obtains gains by 14.21 F1 over its regular fine-tuning baseline, and 1.14 F1 over SciFive-Large, the current state-of-the-art on ChemProt. Besides, we find prompt-based learning requires fewer training examples to make reasonable predictions. The results demonstrate the potential of our methods in such a domain-specific relation extraction task. | ['Pierre Zweigenbaum', 'Thomas Lavergne', 'Hui-Syuan Yeh'] | 2022-04-21 | null | https://aclanthology.org/2022.lrec-1.403 | https://aclanthology.org/2022.lrec-1.403.pdf | lrec-2022-6 | ['cloze-test'] | ['natural-language-processing'] | [ 4.40703183e-01 6.75643742e-01 -5.26757658e-01 -5.09148836e-01
-1.33249795e+00 -3.20106000e-01 5.71339548e-01 7.34636307e-01
-5.44654906e-01 1.03219485e+00 4.18992847e-01 -4.02972102e-01
-2.96619207e-01 -5.73678672e-01 -4.57081914e-01 -2.89282084e-01
7.25858063e-02 6.54491603e-01 2.47864008e-01 -3.28339010e-01
-1.01006664e-01 1.58489466e-01 -8.44638586e-01 5.32891929e-01
7.15683341e-01 5.88332236e-01 -1.43863901e-01 5.52937984e-01
1.18889257e-01 8.65407288e-01 -6.46907866e-01 -5.40555239e-01
-1.33359149e-01 -2.71669000e-01 -1.41554976e+00 -2.67907262e-01
1.45946458e-01 -9.13994536e-02 -1.99400529e-01 6.46036923e-01
6.96092486e-01 7.08623743e-03 6.84660316e-01 -8.64796996e-01
-3.70451152e-01 9.28256631e-01 -4.05889124e-01 3.03592920e-01
5.81794620e-01 1.13279782e-01 1.35865498e+00 -6.60159111e-01
9.69448864e-01 8.24654698e-01 5.80986321e-01 7.73561060e-01
-1.48830903e+00 -6.50952220e-01 1.71213690e-02 2.97211289e-01
-1.21605301e+00 -6.60066843e-01 3.73897076e-01 -3.77624393e-01
1.44219601e+00 2.89067537e-01 2.23488867e-01 1.14992929e+00
1.81860641e-01 7.83003628e-01 9.43150401e-01 -5.96074045e-01
1.48275316e-01 -1.95993185e-02 7.98679590e-01 5.31155050e-01
5.23622751e-01 -2.99815889e-02 -6.84730470e-01 -4.62947965e-01
2.01778695e-01 -3.61963838e-01 -3.88072729e-01 1.95779428e-01
-1.16365099e+00 6.95207000e-01 1.21646740e-01 2.54780084e-01
-4.91219699e-01 -3.05609465e-01 5.15778244e-01 3.77972752e-01
5.79068661e-01 1.22491419e+00 -1.16989040e+00 -1.80198729e-01
-7.80301630e-01 3.61705929e-01 1.19195712e+00 1.10969591e+00
3.40189636e-01 -6.77003264e-01 -7.83195138e-01 8.01635325e-01
-2.22159341e-01 -7.59867802e-02 3.26071173e-01 -5.84069669e-01
4.59494859e-01 6.03453577e-01 4.30320539e-02 -4.67364490e-01
-8.99393380e-01 -3.34114313e-01 -6.65622413e-01 -3.61724079e-01
5.71274161e-01 -5.89340448e-01 -8.60667109e-01 1.70512664e+00
3.27575028e-01 1.71183437e-01 2.14687198e-01 5.22396445e-01
1.36557913e+00 3.25365812e-01 6.22840047e-01 -6.11194849e-01
1.84372652e+00 -8.89591396e-01 -1.02359056e+00 -2.42569432e-01
1.10264206e+00 -7.79508889e-01 9.72616017e-01 5.75170398e-01
-8.52441490e-01 -2.05434605e-01 -1.03829134e+00 -2.69329190e-01
-2.56938905e-01 3.16626951e-03 1.06796241e+00 4.03848380e-01
-5.10465920e-01 7.95114458e-01 -7.82128811e-01 -5.77784300e-01
4.68200356e-01 2.97533393e-01 -4.60424453e-01 -1.38953894e-01
-1.42945802e+00 1.10580230e+00 4.95963603e-01 -5.00551462e-01
-4.31421906e-01 -1.29010439e+00 -7.01077998e-01 2.06474930e-01
8.12403023e-01 -8.69266152e-01 1.58644819e+00 2.55548358e-01
-1.22233105e+00 9.32183027e-01 -2.33233884e-01 -6.51077807e-01
1.21702269e-01 -4.29991543e-01 -4.51376975e-01 1.13801248e-01
1.98938340e-01 5.61394691e-01 1.03838436e-01 -5.85701585e-01
-5.26111543e-01 3.73267122e-02 1.12745585e-02 1.80955864e-02
-2.94068843e-01 3.57414901e-01 -3.67394745e-01 -4.36546087e-01
-3.99249077e-01 -6.62087917e-01 -5.29844880e-01 -4.45721030e-01
-6.16600096e-01 -7.37286568e-01 1.88007459e-01 -4.51590627e-01
1.40989208e+00 -1.81423950e+00 -4.34165359e-01 -8.32654610e-02
4.62064713e-01 4.21600044e-01 -2.90076047e-01 4.70642120e-01
-4.54258025e-01 2.29100406e-01 2.17757039e-02 1.86098307e-01
-2.09593311e-01 3.61877345e-02 -7.78990313e-02 1.01271749e-01
6.75394595e-01 1.15982652e+00 -1.22662008e+00 -6.22547388e-01
-3.28836024e-01 1.67732269e-01 -4.97902602e-01 2.63679266e-01
-3.83609265e-01 2.52768427e-01 -6.29057825e-01 6.28822386e-01
1.80008337e-01 -7.14624643e-01 4.89871114e-01 -3.12710732e-01
2.60979295e-01 7.61575997e-01 -6.39723480e-01 1.82208335e+00
-2.96316631e-02 1.87083602e-01 -3.21013063e-01 -8.60696614e-01
7.73294628e-01 7.20177948e-01 6.23178005e-01 -5.41656435e-01
9.06499326e-02 1.07819267e-01 3.08503389e-01 -7.56489158e-01
3.24995697e-01 -2.07841426e-01 -1.68081522e-01 4.22150105e-01
2.98941493e-01 -4.06077728e-02 6.07877970e-01 3.96359384e-01
1.85758436e+00 -3.10851354e-02 1.07291090e+00 -2.62566924e-01
2.32760146e-01 2.09035352e-01 7.95028746e-01 6.92238867e-01
-1.64220050e-01 3.61607790e-01 6.43687308e-01 -4.10070181e-01
-5.39675295e-01 -4.43868220e-01 -3.34688425e-01 1.15644014e+00
-4.18345183e-01 -1.01001418e+00 -6.51047647e-01 -1.11822319e+00
-2.56206188e-02 7.58057415e-01 -5.55069327e-01 -1.83518574e-01
-5.21938682e-01 -1.17882049e+00 6.29828453e-01 4.07471359e-01
1.48078194e-02 -1.16119397e+00 -4.64234799e-01 4.73097235e-01
-3.92038405e-01 -1.47605002e+00 -4.12273288e-01 7.09518433e-01
-7.46440768e-01 -1.36532176e+00 -4.77793425e-01 -6.25200748e-01
4.61623132e-01 -1.73645355e-02 1.47043252e+00 -1.42742753e-01
-3.71761620e-01 -3.66476066e-02 -5.11994720e-01 -6.84882760e-01
-2.41631955e-01 5.68413675e-01 -1.53024569e-01 -6.31429732e-01
1.01758862e+00 -3.62467349e-01 -4.97150004e-01 9.52183679e-02
-6.18460953e-01 -2.77066380e-02 7.75255382e-01 1.17164826e+00
5.37623405e-01 -2.93959230e-01 8.66329670e-01 -1.68820202e+00
1.00780022e+00 -5.07802486e-01 -8.51771012e-02 4.14128125e-01
-1.01270080e+00 2.82549202e-01 2.86011517e-01 -3.72697055e-01
-9.62216735e-01 2.23728880e-01 -2.55071551e-01 3.12767684e-01
-3.70016396e-01 7.73584485e-01 -1.01388693e-01 2.93019295e-01
1.33700359e+00 -3.19924146e-01 -3.08539689e-01 -5.51808596e-01
4.68080640e-01 5.94906926e-01 5.51257849e-01 -8.66115332e-01
4.77999359e-01 -1.41155794e-01 2.29546577e-02 -6.36314154e-01
-1.36082923e+00 -6.97515905e-01 -5.12608409e-01 5.23462296e-01
8.24016869e-01 -9.69231904e-01 -7.17872858e-01 -2.58063406e-01
-1.24212396e+00 -5.30298471e-01 -4.23350960e-01 4.10249025e-01
-3.84099334e-01 2.04776049e-01 -9.26657319e-01 -2.81066149e-01
-8.16918492e-01 -8.07058036e-01 1.05333376e+00 -1.45773702e-05
-1.02672625e+00 -7.23677218e-01 2.81079203e-01 3.43575746e-01
-5.54015040e-02 1.70595691e-01 1.11974406e+00 -1.24839652e+00
-2.30737850e-01 -8.68435353e-02 -2.99940646e-01 -2.37248063e-01
3.67865980e-01 -4.08787876e-01 -1.04507887e+00 4.59189862e-02
-2.58152068e-01 -5.88941872e-01 7.17471838e-01 2.04792053e-01
1.03039730e+00 -3.38895798e-01 -7.62166977e-01 4.18896228e-01
9.66687679e-01 1.41585901e-01 3.41223150e-01 2.46408939e-01
4.33025837e-01 6.95433617e-01 1.12010860e+00 2.69528300e-01
3.27422798e-01 6.36138678e-01 -3.03969860e-01 -3.01251918e-01
-1.82932585e-01 -1.05593391e-01 -1.78230599e-01 4.66941804e-01
-3.68766673e-02 5.78341223e-02 -1.02017307e+00 5.88783622e-01
-1.87253225e+00 -6.30922258e-01 -9.94936824e-02 1.78144550e+00
1.82450175e+00 2.48709083e-01 1.85319021e-01 2.08809331e-01
2.25334615e-01 -2.40312830e-01 -3.44363749e-01 -2.95027345e-01
2.00387686e-01 7.59020209e-01 2.41855398e-01 3.44362617e-01
-1.24782431e+00 1.21844959e+00 6.42659664e+00 1.01504147e+00
-7.90630996e-01 1.12238219e-02 6.40433252e-01 -1.33501083e-01
-1.03360586e-01 -7.17113093e-02 -1.11975920e+00 1.90210380e-02
1.26006043e+00 -4.25720870e-01 -1.15934014e-01 7.76637554e-01
1.87284410e-01 7.87722412e-03 -1.72754812e+00 8.53322983e-01
-2.87142545e-01 -1.54649591e+00 -1.97968185e-01 -1.56968534e-02
5.76129973e-01 -1.15176685e-01 -5.51721990e-01 6.16029203e-01
5.27059913e-01 -1.08295965e+00 -5.42710572e-02 2.20697790e-01
9.31800067e-01 -4.54008490e-01 8.59074235e-01 3.22519869e-01
-7.86027908e-01 3.31968725e-01 -3.36300582e-01 -1.15773231e-01
2.16475159e-01 1.11996889e+00 -1.51996064e+00 7.39666760e-01
3.39005679e-01 6.50737524e-01 -5.71580648e-01 1.02025962e+00
-5.88637829e-01 7.42264688e-01 -8.79794434e-02 -2.07470760e-01
-1.73602536e-01 4.75081593e-01 2.98136622e-01 1.72338271e+00
-1.87372744e-01 6.40329421e-01 4.19898391e-01 3.70654941e-01
-3.10924411e-01 4.22363013e-01 -5.10141790e-01 -2.60952532e-01
4.90923017e-01 1.45990539e+00 -5.85227728e-01 -5.79829335e-01
-2.50332117e-01 4.16532278e-01 6.23767436e-01 1.88557863e-01
-5.08479178e-01 -5.42712271e-01 5.20141721e-01 1.09570473e-02
1.16856471e-01 2.06116632e-01 -5.10448277e-01 -9.41174090e-01
-2.94666201e-01 -1.09716237e+00 6.92678332e-01 -3.50352466e-01
-1.47867823e+00 7.76703417e-01 1.13312222e-01 -8.83877099e-01
-4.01766390e-01 -6.33174539e-01 -3.07049751e-01 7.55130768e-01
-1.34705591e+00 -9.67587531e-01 4.63568158e-02 3.05809677e-01
4.87409502e-01 5.35167046e-02 1.41035092e+00 4.07372147e-01
-5.81439912e-01 9.21429217e-01 -5.86745322e-01 1.76895306e-01
1.25892127e+00 -1.38801980e+00 5.34805059e-01 6.12378895e-01
1.11547641e-01 1.05180299e+00 7.75562763e-01 -8.28523755e-01
-1.11686134e+00 -1.05686796e+00 1.50391293e+00 -6.54968023e-01
7.17920125e-01 -2.35238791e-01 -9.16129053e-01 6.48059309e-01
2.76978761e-01 1.66556034e-02 1.22588956e+00 9.11531270e-01
-4.32067662e-01 4.02891375e-02 -1.05908084e+00 4.78643149e-01
1.10445154e+00 -3.79358739e-01 -8.75077009e-01 6.85788751e-01
9.90295291e-01 -4.43259716e-01 -1.29175615e+00 6.07410133e-01
3.61531019e-01 -1.80054158e-01 8.10417235e-01 -1.15728307e+00
5.35058916e-01 5.06892148e-03 2.49625355e-01 -1.24991000e+00
-6.85111642e-01 -1.02555633e+00 -3.89159083e-01 1.19845498e+00
1.00603807e+00 -2.65726566e-01 8.39776576e-01 8.35941076e-01
-3.93757038e-02 -1.14818132e+00 -4.87486929e-01 -6.79239571e-01
1.08505785e-01 -2.69817561e-01 4.90650922e-01 1.14743960e+00
7.17964828e-01 1.31269205e+00 -1.88090995e-01 -2.31281832e-01
3.39187443e-01 -1.60826240e-02 7.97103405e-01 -1.44668806e+00
-6.30534589e-01 -2.29714677e-01 8.46748650e-02 -8.79027545e-01
-3.39138359e-02 -9.29377019e-01 2.31324106e-01 -1.59286165e+00
5.68475425e-01 -2.64083266e-01 -3.32311571e-01 1.07386601e+00
-8.58533561e-01 -1.30886629e-01 -1.52034193e-01 -1.94801733e-01
-7.37298310e-01 1.36931031e-03 1.31367195e+00 -2.11449787e-01
-2.97117263e-01 -6.19377308e-02 -1.37295711e+00 4.57685709e-01
7.56219506e-01 -6.21996522e-01 -4.78145540e-01 -9.29207820e-03
3.10224533e-01 9.12398472e-02 -3.08043122e-01 -5.19180834e-01
2.72740602e-01 -2.35834181e-01 9.43028927e-02 -4.34319109e-01
8.67124498e-02 -2.60257900e-01 -1.81725457e-01 3.96089524e-01
-6.74044788e-01 -4.04041409e-01 3.07966799e-01 4.47555721e-01
1.08487591e-01 -1.21862188e-01 4.51180816e-01 -5.58198243e-02
-5.44655025e-01 3.47547919e-01 -1.11680724e-01 4.94837612e-01
7.36387491e-01 2.67732233e-01 -7.12165773e-01 1.11619523e-02
-9.40286458e-01 2.23296255e-01 -1.50664479e-01 1.97709799e-01
3.45748723e-01 -9.62935448e-01 -1.03619850e+00 -1.93167806e-01
5.51890373e-01 6.44076839e-02 -1.13147676e-01 9.89511669e-01
-1.15257334e-02 7.36736894e-01 1.12327456e-01 -2.91304171e-01
-1.57274640e+00 6.64749801e-01 -2.07954779e-01 -9.77443695e-01
-6.92690194e-01 1.15412819e+00 1.95673853e-01 -2.94719338e-01
2.47017235e-01 -5.26687205e-01 -4.20810819e-01 9.90021229e-02
7.53052294e-01 3.55699100e-02 4.94417220e-01 2.56711900e-01
-6.08627915e-01 6.90479726e-02 -7.05189109e-01 8.16150010e-02
1.45103705e+00 4.99799311e-01 6.39076205e-03 1.32168859e-01
9.33265388e-01 1.28809754e-02 -6.94853127e-01 -4.81441706e-01
6.21113896e-01 7.02774897e-02 -1.72043115e-01 -1.28919172e+00
-4.81665343e-01 4.87567365e-01 -1.95258092e-02 4.56335396e-02
8.75536382e-01 2.41151959e-01 7.89428294e-01 7.66675115e-01
3.80472898e-01 -8.77160907e-01 -1.34548470e-01 6.23701215e-01
6.86497569e-01 -1.21766198e+00 3.85689288e-01 -8.90329182e-01
-6.25238836e-01 7.83574283e-01 5.54205239e-01 2.11720690e-01
6.88048244e-01 4.98954594e-01 4.78989631e-02 -5.07622480e-01
-1.18840921e+00 -2.80050099e-01 5.19436836e-01 6.61146522e-01
1.12616169e+00 1.94064379e-01 -7.77387440e-01 1.15465856e+00
-3.23587656e-01 3.04209888e-01 3.50114882e-01 1.03494418e+00
-2.95815557e-01 -1.52856231e+00 8.03018212e-02 8.28769863e-01
-8.17938924e-01 -4.84980345e-01 -5.23492277e-01 6.42906606e-01
-8.27185288e-02 1.25266540e+00 -5.93134344e-01 -4.30812150e-01
5.41078866e-01 3.37866962e-01 6.96942210e-01 -1.31641126e+00
-7.20670402e-01 1.20559983e-01 8.95447433e-01 -5.77807724e-01
-2.99377114e-01 -5.50913334e-01 -1.31324279e+00 -4.89788763e-02
-4.22400326e-01 5.48884153e-01 -3.09425760e-02 1.14343929e+00
4.47290331e-01 9.36545432e-01 1.54605499e-02 -1.24359198e-01
-6.34655535e-01 -1.25866699e+00 -2.47585773e-01 3.35890085e-01
9.53597277e-02 -5.65463126e-01 1.08010158e-01 1.56604484e-01] | [8.670714378356934, 8.71225357055664] |
28120caf-1161-4ac6-81e4-2327192e2077 | scar-sentence-compression-using-autoencoders | null | null | https://aclanthology.org/2020.acl-srw.13 | https://aclanthology.org/2020.acl-srw.13.pdf | SCAR: Sentence Compression using Autoencoders for Reconstruction | Sentence compression is the task of shortening a sentence while retaining its meaning. Most methods proposed for this task rely on labeled or paired corpora (containing pairs of verbose and compressed sentences), which is often expensive to collect. To overcome this limitation, we present a novel unsupervised deep learning framework (SCAR) for deletion-based sentence compression. SCAR is primarily composed of two encoder-decoder pairs: a compressor and a reconstructor. The compressor masks the input, and the reconstructor tries to regenerate it. The model is entirely trained on unlabeled data and does not require additional inputs such as explicit syntactic information or optimal compression length. SCAR{'}s merit lies in the novel Linkage Loss function, which correlates the compressor and its effect on reconstruction, guiding it to drop inferable tokens. SCAR achieves higher ROUGE scores on benchmark datasets than the existing state-of-the-art methods and baselines. We also conduct a user study to demonstrate the application of our model as a text highlighting system. Using our model to underscore salient information facilitates speed-reading and reduces the time required to skim a document. | ['Manish Shrivastava', 'Tirth Maniar', 'Chanakya Malireddy'] | 2020-07-01 | null | null | null | acl-2020-6 | ['sentence-compression'] | ['natural-language-processing'] | [ 6.69877291e-01 4.26255941e-01 -3.05442214e-01 -6.01080120e-01
-9.96262312e-01 -3.55749041e-01 2.75215328e-01 6.25043392e-01
-6.51225030e-01 7.10100949e-01 7.12884426e-01 -4.26962525e-01
3.75225872e-01 -6.13382220e-01 -8.43530774e-01 -2.95213223e-01
1.37096643e-01 4.27632540e-01 -1.81391567e-01 -6.43010065e-02
5.20858228e-01 -1.58697709e-01 -1.30848134e+00 6.33513808e-01
1.14697075e+00 4.78710175e-01 7.62668192e-01 7.03227878e-01
-3.71589541e-01 8.84711981e-01 -8.21816027e-01 -6.58537269e-01
3.86797488e-02 -6.08332992e-01 -8.87014806e-01 -2.00418577e-01
3.07255387e-01 -5.53222299e-01 -5.23691952e-01 1.04122496e+00
6.76550567e-01 1.18053248e-02 5.20224869e-01 -6.12597108e-01
-8.42421532e-01 1.28606212e+00 -6.45161629e-01 1.16872586e-01
4.45756853e-01 -4.77980115e-02 1.34476054e+00 -9.71651495e-01
7.03024507e-01 1.14090252e+00 5.10167897e-01 5.26333988e-01
-1.17991626e+00 -4.20197010e-01 3.57696451e-02 7.94595256e-02
-1.15535748e+00 -6.95791125e-01 6.19965553e-01 -4.62256297e-02
1.29092991e+00 3.34343702e-01 3.95664692e-01 8.58291388e-01
1.24178886e-01 1.23689938e+00 3.39561552e-01 -7.90534437e-01
2.27317214e-01 5.27525768e-02 3.96759361e-02 6.01747632e-01
1.22486144e-01 -3.27648044e-01 -6.71040177e-01 1.48958504e-01
9.64415371e-02 4.45308425e-02 -3.46813917e-01 -7.13571087e-02
-8.43025327e-01 6.52722061e-01 3.02270174e-01 -1.41734749e-01
-1.26715839e-01 1.78496204e-02 7.43413210e-01 3.42259318e-01
6.68947637e-01 5.25327861e-01 -2.29671031e-01 -2.97205716e-01
-1.37003899e+00 3.38581353e-01 7.70133197e-01 1.26407409e+00
4.08406675e-01 -4.24421430e-01 -5.30674219e-01 1.01479244e+00
3.79984006e-02 2.32501879e-01 4.40257639e-01 -6.20206714e-01
1.09088254e+00 6.47159636e-01 -1.31411791e-01 -7.49487877e-01
-1.32817492e-01 -4.44799423e-01 -6.93573415e-01 -2.74187565e-01
-7.92158842e-02 -2.88436431e-02 -7.08110452e-01 1.73965907e+00
-1.61556199e-01 -1.60822883e-01 6.52052788e-03 8.17630827e-01
7.48897314e-01 6.71272516e-01 -4.77049984e-02 -1.87817022e-01
1.04507363e+00 -1.13612080e+00 -7.29615510e-01 -4.40823734e-01
1.10270286e+00 -8.23543906e-01 1.51724374e+00 2.28430495e-01
-1.45929468e+00 -2.27747053e-01 -1.13068569e+00 -7.88742185e-01
-1.15737930e-01 3.83024573e-01 3.21408689e-01 3.30676138e-01
-9.53872740e-01 8.81257892e-01 -7.35060990e-01 -1.10189296e-01
5.74034035e-01 1.31247878e-01 -2.85771549e-01 -2.85531074e-01
-1.13706934e+00 7.07412481e-01 3.75520349e-01 -9.30418968e-02
-6.16150022e-01 -6.61850154e-01 -1.07172096e+00 6.32406950e-01
2.08428249e-01 -7.28211224e-01 1.46475697e+00 -7.16200173e-01
-1.24212623e+00 8.12036633e-01 -4.89931166e-01 -7.90726364e-01
4.88383472e-01 -5.69270492e-01 -7.21914023e-02 3.15077871e-01
2.83614308e-01 6.74576223e-01 8.39505017e-01 -7.77606249e-01
-4.58655298e-01 -1.62140906e-01 -2.28797011e-02 3.84694964e-01
-5.79983950e-01 -7.31762946e-02 -7.93079853e-01 -9.69413817e-01
-1.39204606e-01 -5.16276896e-01 4.62539233e-02 -1.39445886e-01
-9.18653131e-01 -9.62943286e-02 7.56545305e-01 -1.05183291e+00
1.64146054e+00 -2.17092228e+00 3.33964020e-01 -2.27321684e-01
2.63695061e-01 3.08730960e-01 -3.23346376e-01 8.17092419e-01
-4.40658480e-02 2.93982148e-01 -6.64119780e-01 -9.73375261e-01
4.43766788e-02 -1.38023004e-01 -5.40130615e-01 2.32941955e-01
2.77470797e-01 9.27706540e-01 -1.03088677e+00 -6.26197040e-01
-8.03322718e-02 1.82868391e-01 -7.07688034e-01 2.75358528e-01
-4.25220877e-01 2.62900610e-02 1.85806043e-02 4.87235278e-01
7.44732618e-01 -1.51466429e-01 2.90853322e-01 7.53287273e-03
8.07465687e-02 1.07973731e+00 -5.85382164e-01 2.01479387e+00
-3.54998797e-01 9.09855902e-01 -1.39115483e-01 -9.60423648e-01
8.31075728e-01 -3.73999290e-02 7.29539096e-02 -8.76980722e-01
8.69070832e-03 -2.17707567e-02 -3.20829391e-01 -6.74014330e-01
1.01923275e+00 1.86695158e-01 -1.84888393e-01 9.23443019e-01
-1.12539485e-01 -1.44650742e-01 5.92384160e-01 9.03027236e-01
1.38678312e+00 1.32830590e-01 -3.58531177e-02 -5.34354970e-02
3.83529097e-01 -3.13154608e-01 4.63153541e-01 8.04730594e-01
1.84858128e-01 7.05573082e-01 8.31474543e-01 2.09168009e-02
-1.42695534e+00 -7.66668260e-01 1.11359142e-01 1.11366928e+00
1.34907007e-01 -9.51724052e-01 -1.06613803e+00 -8.33354056e-01
9.10811573e-02 1.18058157e+00 -3.39638531e-01 -5.20885348e-01
-6.28486097e-01 -2.88228244e-01 6.56563401e-01 5.71983099e-01
2.59041727e-01 -1.07227027e+00 -5.70234656e-01 2.69609075e-02
-5.23354292e-01 -7.79748321e-01 -9.44366455e-01 1.97437748e-01
-9.37357128e-01 -8.62029314e-01 -4.51571137e-01 -9.32912350e-01
1.04136562e+00 3.80438417e-01 1.32459819e+00 1.97646245e-01
-1.91175327e-01 -1.99704453e-01 -6.01538837e-01 -4.52563137e-01
-3.89578998e-01 3.58796179e-01 -2.96065092e-01 -4.16800708e-01
5.20909071e-01 -4.18109030e-01 -6.65312469e-01 -2.29123950e-01
-1.11258292e+00 4.35849249e-01 7.10497141e-01 1.03533947e+00
6.71784639e-01 -3.89895737e-02 5.73545337e-01 -1.37960577e+00
9.89333034e-01 -3.00563872e-01 -2.04001471e-01 3.41860831e-01
-5.83791375e-01 3.49162012e-01 8.03683102e-01 6.04428500e-02
-1.03069818e+00 6.61189482e-02 -1.11927763e-01 -1.88777655e-01
3.31514567e-01 8.36506248e-01 -1.47023052e-01 6.86812043e-01
4.98314202e-01 3.83396477e-01 -4.84107509e-02 -8.80886197e-01
4.44075346e-01 9.56799984e-01 8.18895519e-01 -1.81099951e-01
4.34229791e-01 1.52230248e-01 -4.05208051e-01 -5.93955576e-01
-8.76067996e-01 -2.26465389e-01 -6.98990226e-01 1.75015941e-01
3.41820896e-01 -8.67715418e-01 -3.49052995e-01 1.39109448e-01
-1.32706475e+00 -2.39788249e-01 -5.01201868e-01 1.57096669e-01
-3.67029697e-01 6.66510761e-01 -8.46326888e-01 -7.08753407e-01
-7.78690279e-01 -8.84159565e-01 1.16139758e+00 -2.42177490e-02
-2.73358226e-01 -6.46448553e-01 -1.77915484e-01 3.53889793e-01
1.84882253e-01 -1.50687709e-01 1.24086308e+00 -5.94499409e-01
-4.18993980e-01 -3.70342463e-01 -2.29232326e-01 4.47844863e-01
-6.87230602e-02 -1.65890709e-01 -7.67086148e-01 -5.18841207e-01
-3.69759426e-02 -5.07096350e-01 1.45948040e+00 1.71593517e-01
1.71886623e+00 -5.63260436e-01 -2.85722822e-01 6.46227181e-01
1.03372478e+00 -3.82957533e-02 8.70214522e-01 1.04115255e-01
5.50045788e-01 7.12683558e-01 6.03501141e-01 4.30278718e-01
4.73254859e-01 3.22455734e-01 4.87308115e-01 -8.70018676e-02
-2.58229375e-01 -7.69332409e-01 4.26642567e-01 1.10230732e+00
4.48096126e-01 -6.10353351e-01 -6.38277233e-01 4.98140246e-01
-1.76095188e+00 -1.00025761e+00 2.50467181e-01 2.32074451e+00
1.28676736e+00 4.57700491e-01 -1.50221646e-01 2.77363211e-01
7.27434516e-01 2.58580834e-01 -7.26188898e-01 -4.66929644e-01
-1.33295611e-01 2.25496262e-01 3.49251181e-01 7.43169665e-01
-1.04586816e+00 9.52246130e-01 6.19070387e+00 7.27480352e-01
-1.02057338e+00 -1.59260035e-01 8.38166893e-01 -6.11456513e-01
-7.12338567e-01 2.11112183e-02 -7.06685901e-01 6.83832288e-01
8.92279267e-01 -2.66157657e-01 3.62131834e-01 7.79151559e-01
3.97635937e-01 -1.73520476e-01 -1.41704011e+00 8.51835847e-01
2.51640916e-01 -1.45794415e+00 1.68072149e-01 -3.41729015e-01
2.79045731e-01 -2.07352415e-01 -4.41421382e-02 3.81093502e-01
7.26873651e-02 -1.13554263e+00 8.00550818e-01 5.08633971e-01
1.14668798e+00 -9.20517147e-01 6.71040297e-01 3.57276797e-01
-7.65173674e-01 7.68833682e-02 -5.76255798e-01 -1.70706883e-01
2.61206746e-01 8.31076980e-01 -1.03049648e+00 2.70799190e-01
3.47306907e-01 1.03075182e+00 -5.38244009e-01 8.68855953e-01
-6.30191326e-01 6.48261309e-01 -5.55665083e-02 -1.59580857e-01
-1.77895859e-01 -4.69878092e-02 5.90234697e-01 1.69843268e+00
6.85864985e-02 -1.00419983e-01 -2.96503417e-02 9.26079571e-01
-6.59011185e-01 1.39865354e-01 -4.88125235e-01 -4.10134554e-01
8.31400692e-01 9.73510921e-01 -4.80166405e-01 -4.27997559e-01
-3.74069333e-01 1.31569552e+00 7.19876349e-01 2.91177720e-01
-5.21012187e-01 -1.04775381e+00 2.77731657e-01 1.21944817e-02
4.21382189e-01 -3.38592753e-02 -7.31654525e-01 -1.29526889e+00
4.47287142e-01 -7.53993571e-01 2.06895187e-01 -6.61062419e-01
-9.80264723e-01 5.05984187e-01 -2.90085912e-01 -1.18876553e+00
-2.37246186e-01 1.06937876e-02 -5.89548051e-01 1.08787608e+00
-1.57157743e+00 -7.62192369e-01 -3.49985622e-02 1.15393087e-01
8.60486567e-01 -1.14842251e-01 7.28658080e-01 3.69886398e-01
-8.25059712e-01 1.13071895e+00 2.39848644e-01 1.24338470e-01
8.30209434e-01 -1.33192444e+00 8.21948349e-01 1.14473867e+00
-5.82281165e-02 7.93795824e-01 6.19577706e-01 -8.48908782e-01
-1.34008241e+00 -1.13370323e+00 1.50408351e+00 -1.78007349e-01
1.57119498e-01 -8.91178608e-01 -9.38769937e-01 5.62817335e-01
1.86161742e-01 -4.37443912e-01 7.66448975e-01 3.01901400e-02
-3.98582339e-01 -6.95968419e-02 -1.01378095e+00 6.74933255e-01
1.08666885e+00 -6.42830908e-01 -7.73621380e-01 5.32947123e-01
1.19413865e+00 -5.32831013e-01 -3.19270760e-01 8.77067894e-02
3.31163704e-01 -9.64996457e-01 6.58689380e-01 -6.32077813e-01
1.30906761e+00 1.23525649e-01 1.21161379e-01 -1.21661508e+00
-2.53276259e-01 -6.78784549e-01 -5.41372120e-01 1.31604159e+00
6.23412907e-01 -4.67674844e-02 9.07647848e-01 6.19625509e-01
-4.99974817e-01 -1.08196282e+00 -6.58053756e-01 -5.06095886e-01
-4.71624210e-02 -3.56415659e-01 7.21744061e-01 6.53236866e-01
4.92592067e-01 5.91763139e-01 -3.83241624e-01 -2.65851885e-01
3.85676771e-01 9.13203508e-02 5.96396208e-01 -7.61038005e-01
-3.74765635e-01 -3.66822153e-01 1.71047747e-01 -1.57000315e+00
1.64259776e-01 -1.25916135e+00 1.45353690e-01 -1.57828188e+00
5.20015955e-01 -3.51989925e-01 -3.85091803e-03 5.25145710e-01
-4.46116745e-01 -3.00559968e-01 2.15493575e-01 4.84512895e-01
-7.32143223e-01 8.87088001e-01 1.03312385e+00 -3.58035117e-01
-3.39666963e-01 -1.26657158e-01 -9.82316136e-01 5.34876525e-01
6.51333869e-01 -6.17110014e-01 -5.66949248e-01 -1.02870643e+00
2.98237979e-01 8.19914490e-02 -3.17354798e-02 -6.10084772e-01
4.92399663e-01 2.56330907e-01 2.90882826e-01 -9.90385473e-01
1.11204283e-02 -4.11568552e-01 -5.11762738e-01 4.01359439e-01
-1.04624939e+00 1.61833242e-01 -9.01965946e-02 5.69372773e-01
-2.03693852e-01 -6.37913942e-01 5.88181257e-01 -1.37191549e-01
-1.98559567e-01 7.45085180e-02 -7.13088885e-02 2.06391215e-01
5.97510755e-01 1.44606963e-01 -3.98894191e-01 -6.10775471e-01
-3.28676134e-01 4.54840779e-01 5.16555727e-01 3.34502101e-01
9.11028862e-01 -9.92539644e-01 -8.43072474e-01 3.05421054e-01
-6.33186474e-02 3.25874865e-01 4.98542674e-02 4.87989455e-01
-7.81777382e-01 5.57727635e-01 1.39525428e-01 -2.57111639e-01
-1.18138635e+00 6.93512738e-01 -5.08007556e-02 -3.72009069e-01
-9.59006488e-01 9.90266860e-01 -1.11527458e-01 -3.83319467e-01
6.88847244e-01 -2.62311667e-01 -6.83087856e-02 -9.14226174e-02
8.18302870e-01 3.20239216e-01 1.71222493e-01 -8.87729898e-02
-1.20427690e-01 -1.40698045e-01 -7.01696336e-01 -7.28879794e-02
1.46426177e+00 -4.16239530e-01 -3.39754134e-01 1.30491927e-01
1.35273349e+00 1.05985321e-01 -1.17005670e+00 -4.93567050e-01
1.82812631e-01 -5.86255670e-01 9.15369298e-03 -7.89500535e-01
-7.70664692e-01 9.95404184e-01 -2.34802943e-02 1.81418695e-02
1.26265824e+00 -1.88916102e-01 1.24705076e+00 5.19333303e-01
-2.46034395e-02 -1.39978921e+00 7.87378103e-02 6.83357239e-01
9.30806994e-01 -1.14205325e+00 1.52240291e-01 -4.45992321e-01
-5.69476545e-01 1.12427485e+00 5.09003222e-01 5.24334535e-02
1.98670775e-01 3.64341944e-01 -3.08065027e-01 -4.41478603e-02
-9.96759593e-01 1.73386589e-01 -3.60653363e-02 4.15873975e-01
7.71745503e-01 -1.62930936e-01 -7.08480179e-01 7.04638541e-01
-5.51946700e-01 -2.61789095e-02 5.72018564e-01 1.19170451e+00
-4.50096518e-01 -1.30212116e+00 1.83160946e-01 8.58769178e-01
-4.25410241e-01 -6.01884007e-01 -5.17227769e-01 1.98316082e-01
-2.14240581e-01 8.41040611e-01 2.39631489e-01 -5.40627778e-01
2.64831930e-01 -1.84825491e-02 1.54690564e-01 -9.91046906e-01
-5.25450766e-01 -4.89058830e-02 2.14436337e-01 -3.33372861e-01
2.66878121e-02 -6.40685141e-01 -1.25754964e+00 -4.56113249e-01
-2.44045213e-01 1.54298380e-01 5.65183878e-01 8.30691278e-01
7.55680025e-01 5.20452917e-01 6.88439965e-01 -7.09425986e-01
-7.62363374e-01 -1.04982996e+00 -4.56296951e-01 6.36486948e-01
4.41927463e-01 -2.70571485e-02 -8.14463198e-02 2.16812775e-01] | [12.121528625488281, 9.207356452941895] |
972e7436-2396-4494-93a1-8d8cf11eb9da | learning-accurate-template-matching-with | 2303.08438 | null | https://arxiv.org/abs/2303.08438v1 | https://arxiv.org/pdf/2303.08438v1.pdf | Learning Accurate Template Matching with Differentiable Coarse-to-Fine Correspondence Refinement | Template matching is a fundamental task in computer vision and has been studied for decades. It plays an essential role in manufacturing industry for estimating the poses of different parts, facilitating downstream tasks such as robotic grasping. Existing methods fail when the template and source images have different modalities, cluttered backgrounds or weak textures. They also rarely consider geometric transformations via homographies, which commonly exist even for planar industrial parts. To tackle the challenges, we propose an accurate template matching method based on differentiable coarse-to-fine correspondence refinement. We use an edge-aware module to overcome the domain gap between the mask template and the grayscale image, allowing robust matching. An initial warp is estimated using coarse correspondences based on novel structure-aware information provided by transformers. This initial alignment is passed to a refinement network using references and aligned images to obtain sub-pixel level correspondences which are used to give the final geometric transformation. Extensive evaluation shows that our method is significantly better than state-of-the-art methods and baselines, providing good generalization ability and visually plausible results even on unseen real data. | ['Kai Xu', 'Chenyang Zhu', 'Yunfan Ye', 'Zheng Qin', 'Renjiao Yi', 'Zhirui Gao'] | 2023-03-15 | null | null | null | null | ['template-matching', 'robotic-grasping'] | ['computer-vision', 'robots'] | [ 8.32413435e-01 -1.69147477e-01 1.09891169e-01 -3.01428318e-01
-5.57260573e-01 -6.85413718e-01 6.16355598e-01 -9.89762172e-02
-1.46713838e-01 3.81062120e-01 -1.53900743e-01 8.76128301e-02
-1.53338864e-01 -7.01446831e-01 -9.14963901e-01 -5.82488537e-01
2.92920858e-01 6.43659174e-01 5.24261832e-01 -3.38970661e-01
4.36190546e-01 7.93375552e-01 -1.52982950e+00 3.34846377e-01
8.43243241e-01 1.01203465e+00 5.67427576e-01 1.22244433e-01
-9.00443941e-02 9.62730274e-02 -3.34719509e-01 -4.10782605e-01
8.85943532e-01 -1.19408928e-01 -5.91034353e-01 5.56387782e-01
8.43544781e-01 -1.46024555e-01 -3.68936300e-01 1.39191842e+00
1.09758779e-01 1.29804432e-01 5.49333215e-01 -1.11385465e+00
-4.29814070e-01 2.07850903e-01 -8.22544038e-01 -3.98329914e-01
4.19604987e-01 1.16469868e-01 6.36989772e-01 -9.70196247e-01
9.55880523e-01 1.42702150e+00 6.02156639e-01 3.44644427e-01
-1.38297009e+00 -6.56703711e-01 1.53073490e-01 2.61534631e-01
-1.21792519e+00 -1.87569827e-01 1.02222502e+00 -3.75541449e-01
5.11424243e-01 1.39704198e-01 4.79683578e-01 8.48759353e-01
4.65449780e-01 3.47624213e-01 1.12023973e+00 -4.72306490e-01
-2.99834162e-02 -1.88761294e-01 -4.60736036e-01 6.89170361e-01
1.24553762e-01 2.42095217e-01 -4.11053717e-01 1.52206630e-01
1.43262053e+00 2.53407985e-01 -4.73290712e-01 -8.62725616e-01
-1.73336744e+00 4.45947140e-01 7.22993553e-01 2.16343850e-01
-6.33638561e-01 -2.62877010e-02 -1.95178017e-01 3.40318888e-01
1.42426863e-01 6.15423918e-01 -2.38075778e-01 3.77839029e-01
-7.05602646e-01 7.94517919e-02 6.77334487e-01 1.23985720e+00
1.00863147e+00 -1.11208789e-01 1.27025498e-02 8.55721533e-01
9.54123661e-02 4.31869447e-01 -1.15399929e-02 -1.23833716e+00
6.72065198e-01 7.43854940e-01 2.39404947e-01 -1.24659050e+00
-3.05941790e-01 -3.95720750e-01 -9.70159829e-01 6.18063450e-01
6.80148900e-01 3.70132446e-01 -1.09649563e+00 1.23834229e+00
5.05661905e-01 1.67608216e-01 -2.80441880e-01 8.93957198e-01
3.02356809e-01 3.61742049e-01 -4.00388539e-01 9.91922319e-02
1.35306942e+00 -9.42383766e-01 -4.75784183e-01 -5.61921716e-01
-1.41838878e-01 -1.32980514e+00 7.44549274e-01 4.71447259e-01
-1.08283985e+00 -6.91705942e-01 -1.17479789e+00 -9.02408287e-02
-2.97717720e-01 8.99509415e-02 3.50967854e-01 2.26725504e-01
-6.61385596e-01 8.16752970e-01 -5.77344656e-01 -2.96331167e-01
3.05493951e-01 4.94980633e-01 -7.95905828e-01 -5.13126969e-01
-6.26728714e-01 1.05610251e+00 4.59654599e-01 2.81315625e-01
-4.59569305e-01 -6.29912436e-01 -9.53848243e-01 -2.29329005e-01
7.56835997e-01 -8.20662796e-01 1.04374266e+00 -9.93218839e-01
-1.53581989e+00 7.99048722e-01 -2.48297267e-02 2.23069917e-02
8.48774970e-01 -1.16544239e-01 -6.68083951e-02 7.34487083e-04
3.12108845e-02 6.74869299e-01 1.11946952e+00 -1.57602167e+00
-7.80242443e-01 -4.52848524e-01 1.11542568e-01 1.98730394e-01
3.08642060e-01 -2.80143619e-01 -8.03681254e-01 -8.08856726e-01
7.68675804e-01 -9.16321635e-01 -2.93379903e-01 4.30205613e-01
-4.66431648e-01 2.32319579e-01 1.06540608e+00 -8.11017513e-01
4.84409571e-01 -1.99045432e+00 3.31925422e-01 4.37940449e-01
-9.61645916e-02 -1.37003418e-02 -3.42069298e-01 2.92142749e-01
-5.40136322e-02 -6.26012385e-01 -2.97131181e-01 -1.79894358e-01
8.60467926e-03 1.02711022e-01 -4.80578430e-02 6.04284704e-01
4.07013357e-01 9.02430713e-01 -9.04688954e-01 -2.77818531e-01
5.94700515e-01 4.93714839e-01 -2.59247363e-01 2.01032415e-01
-1.77185491e-01 6.62817955e-01 -2.33151302e-01 8.35535169e-01
8.09481561e-01 -1.06477516e-03 7.22781122e-02 -8.70102167e-01
-2.41371423e-01 -2.38391951e-01 -1.53974271e+00 1.94025135e+00
-4.54102874e-01 4.54553336e-01 4.26423430e-01 -9.08653378e-01
1.11460173e+00 2.69855210e-03 5.24370551e-01 -6.31416678e-01
2.43674472e-01 4.34747577e-01 5.44897467e-02 -1.26545370e-01
3.65877599e-01 1.30620196e-01 8.68714973e-02 8.69989246e-02
-1.40849710e-01 -8.26282501e-01 4.18162420e-02 -2.85831600e-01
7.43868172e-01 5.20246804e-01 2.28487357e-01 -2.21094221e-01
5.48861861e-01 1.33218125e-01 5.71966469e-01 4.28865761e-01
2.68465728e-01 1.06243873e+00 1.82103422e-02 -5.91165841e-01
-1.24761569e+00 -1.09227753e+00 5.50457686e-02 3.31895649e-01
7.57321715e-01 1.75971165e-02 -7.87231743e-01 -4.33749974e-01
1.48001269e-01 3.02487314e-01 -5.65672994e-01 -1.87337883e-02
-7.85294652e-01 -6.91961870e-02 -3.40058625e-01 5.91584325e-01
7.65076995e-01 -9.53750312e-01 -6.20033324e-01 4.83299583e-01
-1.67508826e-01 -1.39021015e+00 -7.26955891e-01 -2.27334183e-02
-9.12169158e-01 -1.18710005e+00 -8.01360250e-01 -1.03511214e+00
1.01793611e+00 5.40008307e-01 9.24513400e-01 5.64378425e-02
-5.31231403e-01 1.22242443e-01 -8.19147378e-02 -2.87915498e-01
-4.25984144e-01 -2.71648765e-01 -8.55524093e-02 1.81337953e-01
-4.42678295e-02 -5.01984179e-01 -6.72014475e-01 8.60751688e-01
-8.98972392e-01 3.99150878e-01 9.89508271e-01 9.81706858e-01
7.86544561e-01 -4.56672609e-02 7.20205978e-02 -6.99801385e-01
2.42330566e-01 1.76200166e-01 -9.78554666e-01 4.37472612e-01
-1.77173987e-01 1.75403997e-01 3.58467698e-01 -6.07727885e-01
-9.13793862e-01 6.26423359e-01 1.62665352e-01 -6.24201357e-01
-1.23110265e-01 3.28590674e-03 -4.21831876e-01 -4.04751360e-01
3.31369758e-01 -2.36980543e-02 2.02345505e-01 -4.71592516e-01
3.17518324e-01 4.43177670e-01 9.77533877e-01 -5.77067912e-01
1.29311621e+00 6.33508921e-01 1.67133510e-01 -7.20698595e-01
-5.60200572e-01 -5.10350406e-01 -1.18552351e+00 -2.67206639e-01
8.13620925e-01 -5.55065930e-01 -4.70057726e-01 5.87574899e-01
-1.44003284e+00 -2.24999204e-01 -1.29612416e-01 4.60311711e-01
-7.57736266e-01 3.37687522e-01 -4.01486903e-01 -2.85101175e-01
-2.61382699e-01 -1.43919337e+00 1.32692909e+00 1.82205096e-01
-1.20264301e-02 -6.92046225e-01 -4.78152573e-01 2.63852030e-01
4.20474112e-01 4.26472306e-01 8.00789475e-01 -6.71442226e-02
-1.08482003e+00 -2.27261648e-01 -4.00873065e-01 1.30190283e-01
6.34475052e-01 -4.49529067e-02 -6.98003352e-01 -1.69644192e-01
-7.87423030e-02 1.76128790e-01 5.38187325e-01 2.61953354e-01
9.31600988e-01 2.92693055e-03 -4.76040810e-01 5.78411996e-01
1.39433444e+00 2.21835449e-01 7.56306708e-01 5.23616731e-01
8.80405366e-01 9.61753666e-01 9.51795578e-01 7.10840002e-02
3.06385998e-02 1.02959037e+00 6.62134230e-01 -3.27225894e-01
-3.30948651e-01 -8.39526877e-02 -8.37893877e-03 5.03892362e-01
-3.14322382e-01 3.05438817e-01 -6.66379452e-01 3.46872121e-01
-1.86906087e+00 -7.78213739e-01 -8.62540305e-02 2.43960285e+00
6.95831060e-01 2.05437854e-01 -3.13142061e-01 -7.32880011e-02
9.07056272e-01 -1.53379411e-01 -6.68981969e-01 2.81163510e-02
-1.15043469e-01 2.58038282e-01 5.99954903e-01 5.06186068e-01
-1.01422167e+00 9.50529933e-01 5.66779041e+00 6.50459826e-01
-1.06252158e+00 -2.34446958e-01 2.97903061e-01 5.20598292e-01
-1.57322899e-01 -1.48526922e-01 -4.13111925e-01 1.36877552e-01
-8.23121667e-02 7.81777948e-02 4.33755845e-01 6.14397168e-01
-3.85126472e-02 -2.80772168e-02 -1.28734207e+00 1.09389210e+00
1.77756965e-01 -1.33786809e+00 -7.23017156e-02 3.74066718e-02
1.03543019e+00 -1.46808892e-01 -1.34381294e-01 -2.79559523e-01
3.07921052e-01 -8.72595310e-01 8.83814096e-01 5.27299166e-01
6.41043603e-01 -5.43523192e-01 7.16311812e-01 2.05941647e-01
-1.37039053e+00 1.17035367e-01 -5.34062505e-01 2.78624475e-01
3.44655424e-01 4.01868641e-01 -5.57479501e-01 8.64308238e-01
4.79146361e-01 5.86098611e-01 -2.44239837e-01 1.14426196e+00
-1.98230579e-01 -3.76065969e-01 -4.08313930e-01 4.70814973e-01
1.73610169e-02 -5.20248652e-01 3.84384602e-01 7.91498721e-01
5.49142718e-01 -1.16823651e-01 4.40542012e-01 1.04242289e+00
-6.97543323e-02 -1.45307451e-01 -5.47238171e-01 2.98486650e-01
4.50996131e-01 1.40021169e+00 -1.06709874e+00 -1.30377725e-01
-3.83520871e-01 1.41319549e+00 -7.95985572e-03 3.37051809e-01
-5.68318605e-01 -2.72001445e-01 6.51909053e-01 1.45449281e-01
4.46914107e-01 -4.79984581e-01 -2.22287670e-01 -9.72664773e-01
1.80361524e-01 -1.02339256e+00 -9.80803296e-02 -7.60491788e-01
-1.27300131e+00 6.56158090e-01 -8.07568897e-03 -1.67316985e+00
-3.38735014e-01 -8.81350577e-01 -4.50283229e-01 9.75827277e-01
-1.44585085e+00 -1.32424283e+00 -7.48313606e-01 5.16939580e-01
9.04258966e-01 1.08307362e-01 5.66111505e-01 1.48176640e-01
-1.18316740e-01 2.41906047e-01 -1.72913317e-02 1.79213146e-03
8.70565593e-01 -1.00975418e+00 6.36976182e-01 9.90421772e-01
2.66707718e-01 5.20737588e-01 7.11883664e-01 -7.22786427e-01
-1.64144349e+00 -9.69339430e-01 3.71095449e-01 -3.81578535e-01
3.96743327e-01 -3.55639309e-01 -9.43682134e-01 4.77576613e-01
1.86526239e-01 1.84478313e-01 -2.58404166e-01 -5.64033210e-01
-3.55717838e-01 -1.82691902e-01 -1.11773884e+00 5.98396897e-01
1.14433765e+00 -2.97641218e-01 -5.24136841e-01 1.63069949e-01
3.54828268e-01 -8.58218193e-01 -8.76320124e-01 5.89290679e-01
8.95387828e-01 -9.06711400e-01 1.17402220e+00 -9.99730602e-02
3.06687146e-01 -6.57605171e-01 -8.74513984e-02 -1.36678970e+00
-4.00779903e-01 -6.50375783e-01 2.96564013e-01 1.04213655e+00
2.01746523e-01 -6.27467394e-01 7.36252725e-01 6.37107253e-01
-1.92912102e-01 -4.80800480e-01 -7.76581109e-01 -9.13419664e-01
-4.06080276e-01 -1.90471068e-01 7.10986257e-01 9.44695413e-01
-4.59933758e-01 -2.54841372e-02 -1.20691724e-01 4.07476634e-01
8.90861452e-01 5.81691265e-01 9.27147210e-01 -1.37687171e+00
-7.51307085e-02 -5.11084735e-01 -7.32269645e-01 -1.12046897e+00
1.39335366e-02 -4.47419405e-01 4.87926364e-01 -1.60916483e+00
1.74835287e-02 -4.07649606e-01 9.90972482e-03 4.85141903e-01
-1.26927234e-02 6.55016363e-01 2.52077848e-01 1.50049597e-01
5.36314137e-02 2.85289913e-01 1.60034680e+00 -3.88529450e-01
-6.65630996e-02 -3.36501040e-02 -2.61373490e-01 7.52812624e-01
6.10200942e-01 -1.04362607e-01 -7.11715966e-02 -5.29667139e-01
-3.35506976e-01 -2.51499474e-01 4.70162332e-01 -1.07470977e+00
3.23344469e-01 -3.60073477e-01 5.34138501e-01 -5.44511557e-01
5.77099681e-01 -1.31708980e+00 6.68669939e-01 5.39898932e-01
1.01660721e-01 2.48863980e-01 7.94304088e-02 4.00109112e-01
-2.11958483e-01 -1.95412800e-01 7.05111027e-01 -1.41743049e-01
-9.01141047e-01 4.30363387e-01 3.59123170e-01 -4.29169327e-01
1.14610898e+00 -7.50731051e-01 -2.08942786e-01 -2.81007558e-01
-4.26047623e-01 -2.54376978e-02 1.08290279e+00 7.01149285e-01
8.23281765e-01 -1.35685551e+00 -7.42848992e-01 5.50368249e-01
1.63029149e-01 2.64832705e-01 6.90384731e-02 8.81006598e-01
-7.98189878e-01 1.25892773e-01 -5.35843790e-01 -8.72387588e-01
-1.31875992e+00 4.81598496e-01 2.78350025e-01 8.15214366e-02
-7.97396362e-01 5.79984486e-01 5.10225296e-01 -3.25491220e-01
1.66157559e-01 -6.04048133e-01 1.89676106e-01 -2.80672312e-01
4.30299878e-01 3.00549716e-01 2.63120681e-01 -8.85480583e-01
-2.16712981e-01 1.18436301e+00 -5.64564578e-02 -6.62368611e-02
1.32874918e+00 1.50746116e-02 -6.52348921e-02 6.90532476e-02
9.75581050e-01 1.36396140e-01 -1.72096348e+00 -3.34455043e-01
-3.80323045e-02 -9.05990303e-01 -1.77859351e-01 -5.02783239e-01
-1.28048360e+00 7.98107505e-01 5.87313175e-01 -3.50460522e-02
9.62189138e-01 4.22692932e-02 6.91328883e-01 2.09975496e-01
8.37320268e-01 -9.69795883e-01 4.43720259e-02 4.18662399e-01
1.28859878e+00 -1.26707244e+00 1.06893063e-01 -1.02007282e+00
-2.80938655e-01 1.41489565e+00 5.75748324e-01 -1.85690328e-01
1.79223433e-01 2.58682191e-01 2.64148474e-01 -6.81780577e-02
-1.11641943e-01 -1.93484023e-01 7.49144256e-01 8.06004465e-01
-3.44049297e-02 -2.23095179e-01 5.19278534e-02 -9.02583301e-02
-6.21169619e-02 -2.97191292e-01 1.52353615e-01 1.06021535e+00
-2.97306210e-01 -1.52369535e+00 -6.64004505e-01 2.55321264e-01
-5.37444912e-02 2.08650708e-01 -3.20038110e-01 8.26561332e-01
7.04336725e-03 6.58903837e-01 1.14406891e-01 -3.80622387e-01
7.33737171e-01 -3.57998788e-01 9.97940540e-01 -6.04441166e-01
-4.41822231e-01 1.97685003e-01 -1.31377444e-01 -9.64890480e-01
-5.67562282e-01 -6.24805033e-01 -1.00295460e+00 7.28357136e-02
-4.51394141e-01 -3.26146394e-01 8.15297961e-01 8.59881103e-01
2.78253347e-01 4.53785121e-01 4.47099924e-01 -1.65635478e+00
-4.76593286e-01 -6.90808594e-01 -3.92627180e-01 7.63290465e-01
8.60194787e-02 -9.52462971e-01 -1.71633989e-01 2.82011360e-01] | [8.625284194946289, -2.391019821166992] |
78954bd6-f035-426c-b7bc-79ca934792b9 | the-hidden-language-of-diffusion-models | 2306.00966 | null | https://arxiv.org/abs/2306.00966v2 | https://arxiv.org/pdf/2306.00966v2.pdf | The Hidden Language of Diffusion Models | Text-to-image diffusion models have demonstrated an unparalleled ability to generate high-quality, diverse images from a textual concept (e.g., "a doctor", "love"). However, the internal process of mapping text to a rich visual representation remains an enigma. In this work, we tackle the challenge of understanding concept representations in text-to-image models by decomposing an input text prompt into a small set of interpretable elements. This is achieved by learning a pseudo-token that is a sparse weighted combination of tokens from the model's vocabulary, with the objective of reconstructing the images generated for the given concept. Applied over the state-of-the-art Stable Diffusion model, this decomposition reveals non-trivial and surprising structures in the representations of concepts. For example, we find that some concepts such as "a president" or "a composer" are dominated by specific instances (e.g., "Obama", "Biden") and their interpolations. Other concepts, such as "happiness" combine associated terms that can be concrete ("family", "laughter") or abstract ("friendship", "emotion"). In addition to peering into the inner workings of Stable Diffusion, our method also enables applications such as single-image decomposition to tokens, bias detection and mitigation, and semantic image manipulation. Our code will be available at: https://hila-chefer.github.io/Conceptor/ | ['Lior Wolf', 'Inbar Mosseri', 'Michal Irani', 'Assaf Shocher', 'Volodymyr Polosukhin', 'Mor Geva', 'Oran Lang', 'Hila Chefer'] | 2023-06-01 | null | null | null | null | ['image-manipulation', 'bias-detection'] | ['computer-vision', 'natural-language-processing'] | [ 2.68667340e-01 1.30964473e-01 -5.85245080e-02 -2.94758558e-01
-5.08774579e-01 -5.95222235e-01 1.03768456e+00 2.37452567e-01
1.31294981e-01 4.83020067e-01 6.11401320e-01 -6.76447302e-02
3.95632610e-02 -8.01728487e-01 -7.69693255e-01 -8.51181149e-01
2.25919276e-01 4.13687646e-01 -4.75288332e-01 -3.99664849e-01
9.16568562e-02 1.24900162e-01 -1.72074664e+00 7.79278815e-01
5.55852830e-01 8.45852017e-01 4.46132988e-01 4.56191540e-01
-6.29677355e-01 1.01658762e+00 -7.98577964e-01 -5.25313675e-01
-1.55747518e-01 -5.90837300e-01 -8.19411159e-01 2.66493946e-01
4.11237597e-01 -1.17024951e-01 -1.96254969e-01 1.14839518e+00
5.45457117e-02 -7.63752460e-02 7.40044832e-01 -1.26180851e+00
-1.17772794e+00 7.59498835e-01 -7.03023136e-01 -1.56602189e-01
4.53754991e-01 5.11023849e-02 1.10270190e+00 -1.13804424e+00
1.11909330e+00 1.58184564e+00 3.53349477e-01 6.56726301e-01
-1.69863856e+00 -6.01888180e-01 2.45198339e-01 4.96274270e-02
-1.43303812e+00 -4.47993577e-01 8.12607408e-01 -8.03514004e-01
5.28468847e-01 4.87179369e-01 9.11060870e-01 1.58989155e+00
1.01705538e-02 9.41773236e-01 1.09273839e+00 -4.49116200e-01
1.53229341e-01 4.49701160e-01 -1.56017557e-01 6.32145464e-01
8.46989732e-03 -2.24544749e-01 -8.07714403e-01 -8.90326872e-02
6.15258157e-01 1.43007085e-01 -2.09479243e-01 -2.54621327e-01
-1.46035612e+00 8.54851663e-01 4.47515368e-01 3.89712453e-01
-4.29227829e-01 3.04228574e-01 1.81710154e-01 1.35583133e-01
8.13573062e-01 2.13789016e-01 -1.06086746e-01 -1.27579737e-02
-9.98458982e-01 3.34568471e-01 6.36392117e-01 8.82773995e-01
9.90665138e-01 5.45905903e-02 -1.83875456e-01 8.43023598e-01
2.09960982e-01 4.60815430e-01 4.80119497e-01 -8.98228645e-01
1.26378581e-01 4.96280730e-01 2.05737457e-01 -1.11130118e+00
-1.49060160e-01 -3.96724164e-01 -9.88774180e-01 3.62201124e-01
2.75224686e-01 1.63692050e-02 -9.67757702e-01 1.73164511e+00
2.07745224e-01 5.29155992e-02 9.24786851e-02 9.70961094e-01
1.00346565e+00 8.04614246e-01 3.03152919e-01 -9.17806290e-03
1.70293760e+00 -8.08632791e-01 -5.29501677e-01 -2.39351794e-01
1.88169777e-01 -8.45861137e-01 1.31672919e+00 2.66564637e-01
-1.22311151e+00 -5.27834535e-01 -7.59840727e-01 -3.98405343e-01
-5.10592818e-01 1.60418108e-01 5.75843275e-01 3.90923768e-01
-9.38823938e-01 5.13641059e-01 -5.51759422e-01 -5.62604785e-01
6.89162731e-01 -2.19319358e-01 -4.39937413e-01 -2.23495528e-01
-8.55476320e-01 6.20053351e-01 2.38741301e-02 -3.24626029e-01
-1.06690586e+00 -9.45160806e-01 -7.36932039e-01 1.21257514e-01
2.70371199e-01 -1.10395133e+00 8.49187374e-01 -1.69097030e+00
-9.89870489e-01 1.21341193e+00 -4.78855401e-01 -3.00462335e-01
4.17183459e-01 -6.71236366e-02 -3.52964550e-01 2.91605771e-01
3.75008404e-01 1.03087807e+00 1.34741294e+00 -1.72686708e+00
-2.91050762e-01 -4.07012731e-01 2.32175693e-01 2.27982104e-01
-4.96131063e-01 1.30708203e-01 -2.39456668e-01 -1.09296823e+00
1.95969015e-01 -6.47840440e-01 5.50553575e-02 4.53111649e-01
-5.24568796e-01 3.54574770e-02 8.46649587e-01 -6.23929024e-01
1.06002378e+00 -2.34915113e+00 3.80548030e-01 1.65199280e-01
6.43679202e-01 -3.35842431e-01 -3.77913974e-02 7.06449389e-01
-2.76152194e-01 3.72198641e-01 -3.94663662e-01 -6.23854995e-01
1.72173865e-02 2.70323336e-01 -3.67964059e-01 2.28646293e-01
1.77252918e-01 9.09472346e-01 -1.06025147e+00 -4.66281384e-01
5.92350811e-02 8.49050522e-01 -3.57358426e-01 -7.64203593e-02
-3.92556071e-01 5.85870683e-01 -2.41017625e-01 6.43328428e-01
4.91316885e-01 -5.42042077e-01 2.13470757e-01 -4.08556312e-01
-3.01675498e-02 1.24587372e-01 -1.01942992e+00 1.83145225e+00
-4.25677806e-01 8.28561485e-01 4.07898545e-01 -7.61064768e-01
9.25250471e-01 1.82519957e-01 4.29630935e-01 -4.90786642e-01
-7.44988248e-02 -1.59493580e-01 -2.98704982e-01 -5.11306047e-01
7.76874244e-01 -4.77864355e-01 1.09967507e-01 6.23319328e-01
-1.24569058e-01 -2.11464047e-01 2.38932133e-01 6.61015093e-01
7.67649829e-01 2.07766294e-01 8.71935189e-02 -3.57992321e-01
1.81361929e-01 1.72165722e-01 1.20216586e-01 5.19538760e-01
2.37076491e-01 9.60404217e-01 5.91629684e-01 -3.19703788e-01
-1.15293539e+00 -1.26771343e+00 -6.77514682e-03 8.82704496e-01
-1.54777998e-02 -8.55532110e-01 -7.39862621e-01 -2.47904390e-01
1.38383523e-01 8.38896275e-01 -8.94154906e-01 -8.17321648e-04
-1.42195192e-03 -5.89229703e-01 4.51000422e-01 1.41224489e-01
2.95367748e-01 -9.08899665e-01 -4.48885888e-01 1.09991647e-01
-5.21753192e-01 -7.77465701e-01 -3.25350970e-01 -1.75665185e-01
-4.48354334e-01 -7.00892568e-01 -8.07992458e-01 -5.55353522e-01
1.02989686e+00 4.27092791e-01 1.52350807e+00 2.36765727e-01
-5.67989588e-01 6.41379416e-01 -3.31528157e-01 -3.59837174e-01
-5.90430975e-01 -3.12634379e-01 -1.01585105e-01 4.73839253e-01
1.69367358e-01 -5.14792204e-01 -8.08512747e-01 3.43139172e-02
-1.17497337e+00 4.99721199e-01 2.94812530e-01 9.18157518e-01
6.41037464e-01 -1.38635993e-01 3.48083168e-01 -1.02441096e+00
8.91595900e-01 -8.93910587e-01 5.00676595e-02 1.90607101e-01
-4.61967230e-01 -4.67905141e-02 4.60661471e-01 -6.28657818e-01
-1.16717267e+00 -9.73844714e-03 1.33856922e-01 -6.77847624e-01
-1.18712187e-01 5.84547222e-01 1.27990216e-01 4.24225867e-01
9.53054428e-01 3.38023424e-01 1.79717824e-01 -5.00622809e-01
6.89038277e-01 3.86467665e-01 4.49865073e-01 -9.24940825e-01
6.87735558e-01 1.01059353e+00 -1.96741626e-01 -9.89794910e-01
-6.26706481e-01 -3.13562930e-01 -3.28786522e-01 -2.53656507e-01
9.08671975e-01 -1.16316330e+00 -4.89952803e-01 3.53128135e-01
-1.16588545e+00 -1.91944361e-01 -7.34146416e-01 3.11976392e-02
-4.19605762e-01 3.01310062e-01 -5.63100040e-01 -5.88253736e-01
-1.84411898e-01 -8.24799180e-01 1.07349920e+00 1.40192538e-01
-6.38211966e-01 -1.11459625e+00 -2.97674716e-01 4.58579957e-01
4.22168940e-01 3.04715663e-01 1.10825694e+00 -4.41244878e-02
-5.65047026e-01 1.56744823e-01 -2.22056925e-01 3.58703285e-01
3.88851278e-02 3.31515282e-01 -9.78011966e-01 -1.78182885e-01
-3.67243588e-03 -3.61217916e-01 9.21896279e-01 2.44081557e-01
1.16344500e+00 -7.10623085e-01 -3.83434176e-01 4.49730009e-01
1.43283296e+00 -2.46636003e-01 8.18082750e-01 -8.68173875e-03
7.37980485e-01 7.70029366e-01 1.67107850e-01 7.17618108e-01
5.65044165e-01 4.51179355e-01 2.88031459e-01 -2.82030523e-01
-4.73757744e-01 -5.87378800e-01 3.83737266e-01 4.80233848e-01
-1.09804958e-01 -2.36087441e-01 -8.35083425e-01 6.14276528e-01
-1.71283460e+00 -1.28971314e+00 -2.54694402e-01 1.90186322e+00
7.79688537e-01 -2.01172337e-01 -1.64707601e-01 -1.44842908e-01
5.83917439e-01 3.86417121e-01 -4.07030761e-01 -2.37738982e-01
-4.26457405e-01 8.01138878e-02 -1.92383211e-02 4.87833798e-01
-6.73522055e-01 9.84495103e-01 5.65866327e+00 1.03609061e+00
-1.10947144e+00 2.94513255e-01 9.10851955e-01 -2.03273967e-01
-9.77590382e-01 3.18034321e-01 -4.54253942e-01 4.07377750e-01
5.11081159e-01 -3.45208526e-01 6.78182900e-01 6.36172235e-01
2.06875548e-01 -2.24228919e-01 -9.38493550e-01 1.11036849e+00
3.48609984e-01 -1.70859838e+00 5.27099848e-01 -1.12798057e-01
9.17427599e-01 -3.52705896e-01 3.17033798e-01 -7.92692602e-02
3.59364122e-01 -1.21059966e+00 1.30081689e+00 6.75151646e-01
1.05280459e+00 -2.90203303e-01 -1.46037698e-01 1.57293245e-01
-9.84815240e-01 -2.14617383e-02 -3.97103608e-01 -8.62921104e-02
1.28670022e-01 7.95803607e-01 -5.78547060e-01 2.76814759e-01
7.50534713e-01 9.72345829e-01 -4.87266392e-01 2.63352215e-01
-3.17972332e-01 3.28763157e-01 -7.03021660e-02 7.55546466e-02
7.37845376e-02 -3.41532350e-01 6.40653193e-01 1.37853181e+00
4.81945783e-01 2.07300514e-01 -1.01557583e-01 1.58204651e+00
-3.13256949e-01 9.52951089e-02 -7.35282958e-01 -1.84581071e-01
2.56100833e-01 1.46931660e+00 -6.84967458e-01 -6.25471532e-01
-3.37850511e-01 1.15916610e+00 2.94081151e-01 7.91257203e-01
-6.43622279e-01 5.79265431e-02 9.24576223e-01 5.70629716e-01
8.56134966e-02 -1.63688615e-01 -4.82503027e-01 -1.26004100e+00
4.89923134e-02 -9.06023085e-01 1.00772358e-01 -1.37110102e+00
-1.22983432e+00 4.12954092e-01 8.90179500e-02 -1.08053255e+00
-2.21696403e-02 -4.32725847e-01 -6.17012560e-01 9.66547966e-01
-1.09032071e+00 -1.39009392e+00 -4.97331351e-01 8.32964301e-01
6.13532126e-01 -1.07911296e-01 8.92826557e-01 1.09767042e-01
-7.35626817e-02 2.19808668e-01 1.04638934e-01 -3.96317840e-02
5.79428554e-01 -1.10056019e+00 2.89328307e-01 6.10840142e-01
5.04508555e-01 7.43285060e-01 7.80690670e-01 -6.16671205e-01
-1.31989241e+00 -8.25315833e-01 9.08866644e-01 -5.08327484e-01
7.99356341e-01 -4.70502734e-01 -9.06162262e-01 6.90837979e-01
3.08338255e-01 -2.33233988e-01 5.79455793e-01 9.35331658e-02
-7.76967525e-01 -5.48654720e-02 -1.08507812e+00 9.50811088e-01
1.07678139e+00 -7.68287599e-01 -2.51731932e-01 5.19778311e-01
5.42860746e-01 -3.02881628e-01 -6.96393907e-01 -3.59566867e-01
5.30330181e-01 -9.81033981e-01 1.24175477e+00 -5.61603427e-01
1.03263783e+00 -2.42008522e-01 -8.79053306e-03 -1.45846081e+00
-2.92995840e-01 -5.96865833e-01 6.04401007e-02 1.33213758e+00
3.97549182e-01 -3.09123278e-01 7.02681959e-01 6.61079943e-01
5.36381863e-02 -5.00934005e-01 -8.36673796e-01 -3.57633710e-01
3.47647220e-02 -3.08505148e-01 4.80801880e-01 1.22850430e+00
1.61826655e-01 4.04610157e-01 -4.32708651e-01 -3.13896656e-01
5.77129960e-01 6.16285428e-02 5.80053389e-01 -1.10206783e+00
-2.42770880e-01 -5.82961619e-01 -1.44119889e-01 -9.08616006e-01
3.46753933e-02 -1.17543077e+00 -2.85286129e-01 -1.75471294e+00
3.65909666e-01 -5.33317029e-01 1.55877126e-02 4.65660870e-01
4.78760041e-02 2.05889001e-01 3.37329835e-01 3.85021299e-01
-2.36343414e-01 4.57054287e-01 1.47819221e+00 -6.31412685e-01
1.91596285e-01 -4.47121501e-01 -1.12357581e+00 6.19836628e-01
7.78431296e-01 -5.42833745e-01 -4.36102509e-01 -6.99923933e-01
6.34712875e-01 1.06126517e-01 7.97405541e-01 -5.68586290e-01
1.51037931e-01 -2.57985413e-01 4.33888048e-01 -7.95707256e-02
7.20122159e-01 -5.97624302e-01 5.92027009e-01 2.74440259e-01
-4.02791828e-01 1.35308385e-01 -1.31913992e-02 5.39803565e-01
-3.98055792e-01 -2.02697128e-01 5.87260962e-01 -6.20321214e-01
-6.15916252e-01 8.42073634e-02 -3.16483557e-01 1.25129670e-01
7.78113663e-01 -1.08574592e-01 -6.84864819e-01 -8.29754293e-01
-8.14886570e-01 -1.43895611e-01 6.16967797e-01 6.07560635e-01
9.04202819e-01 -1.31695116e+00 -9.58886623e-01 2.33216211e-01
2.63131529e-01 -4.33953293e-02 4.04174894e-01 5.10040164e-01
-2.76487261e-01 -2.37217456e-01 -2.72549480e-01 -4.98532444e-01
-1.24180889e+00 5.71000874e-01 2.99243256e-03 9.76306573e-02
-7.11458266e-01 8.46469879e-01 6.20685041e-01 1.35388643e-01
3.45707648e-02 -2.47382239e-01 4.95047346e-02 3.84630173e-01
5.10260999e-01 1.00905307e-01 -4.72364962e-01 -8.66976321e-01
-1.60165146e-01 4.74602044e-01 4.45308946e-02 -2.39480183e-01
1.15770769e+00 -2.49036163e-01 -3.92951250e-01 5.62069297e-01
1.20639062e+00 5.28754033e-02 -1.16186035e+00 -1.08893283e-01
-3.60882550e-01 -6.20845139e-01 -2.25020245e-01 -9.07293797e-01
-1.05285871e+00 9.95768368e-01 3.33592802e-01 2.16418564e-01
9.64555025e-01 3.75029922e-01 3.69428635e-01 1.18639423e-02
1.71082690e-01 -1.02831709e+00 3.51063192e-01 2.25273281e-01
1.37988746e+00 -1.17851913e+00 1.52041823e-01 -3.14894378e-01
-9.55092013e-01 9.71349061e-01 1.90022752e-01 1.56543300e-01
5.88472426e-01 1.30085930e-01 1.20764352e-01 -4.11987305e-01
-9.14117038e-01 -1.57520771e-01 2.12409049e-01 6.54614568e-01
3.93736422e-01 4.74126935e-01 -1.46176800e-01 4.56421912e-01
-2.37261340e-01 -1.77123964e-01 6.87848687e-01 5.52673221e-01
-2.23649755e-01 -9.31367218e-01 -5.01238883e-01 3.52401674e-01
-3.48636568e-01 -3.47733736e-01 -4.46456999e-01 5.13123453e-01
3.36383462e-01 8.57841372e-01 2.31407017e-01 -2.72161841e-01
-1.94671203e-03 1.77677915e-01 3.29220891e-01 -7.37070918e-01
-7.40076900e-01 4.59188037e-02 1.40958488e-01 -2.76322484e-01
-7.17693985e-01 -7.07315743e-01 -1.20932138e+00 -6.71870887e-01
3.16160619e-01 -1.52196914e-01 8.22593570e-01 6.57462955e-01
3.97381961e-01 4.37584251e-01 1.30043492e-01 -7.71547735e-01
-4.62541021e-02 -7.49335885e-01 -5.85362613e-01 1.10479391e+00
2.17825577e-01 -4.75651443e-01 -3.23277116e-01 5.56637347e-01] | [11.337724685668945, 0.14457295835018158] |
7d88c929-0111-404a-b75d-3b5a50b7a2d5 | analysing-lexical-semantic-change-with | 2004.14118 | null | https://arxiv.org/abs/2004.14118v1 | https://arxiv.org/pdf/2004.14118v1.pdf | Analysing Lexical Semantic Change with Contextualised Word Representations | This paper presents the first unsupervised approach to lexical semantic change that makes use of contextualised word representations. We propose a novel method that exploits the BERT neural language model to obtain representations of word usages, clusters these representations into usage types, and measures change along time with three proposed metrics. We create a new evaluation dataset and show that the model representations and the detected semantic shifts are positively correlated with human judgements. Our extensive qualitative analysis demonstrates that our method captures a variety of synchronic and diachronic linguistic phenomena. We expect our work to inspire further research in this direction. | ['Raquel Fernández', 'Marco del Tredici', 'Mario Giulianelli'] | 2020-04-29 | analysing-lexical-semantic-change-with-1 | https://aclanthology.org/2020.acl-main.365 | https://aclanthology.org/2020.acl-main.365.pdf | acl-2020-6 | ['contextualised-word-representations'] | ['natural-language-processing'] | [ 2.32597873e-01 -1.94554366e-02 -5.38879216e-01 -6.23705268e-01
-1.34637386e-01 -7.13708162e-01 1.20790446e+00 6.15206122e-01
-6.82789385e-01 5.14564633e-01 8.09493482e-01 -2.60730475e-01
-6.22295402e-02 -9.36908901e-01 -3.19742531e-01 -3.45555335e-01
-6.94817603e-02 1.99094549e-01 2.13431731e-01 -6.10909283e-01
7.67033994e-01 2.25505441e-01 -1.77657986e+00 3.49427313e-01
5.28999448e-01 5.77301681e-01 3.96640599e-02 3.04860383e-01
-5.78607082e-01 7.25345433e-01 -5.03644645e-01 -2.28257224e-01
-1.29793435e-01 -4.23067063e-01 -1.15974343e+00 -1.25557948e-02
2.04068515e-03 4.00784194e-01 3.60348038e-02 1.05773914e+00
2.56984770e-01 4.58273083e-01 8.66892993e-01 -6.75984442e-01
-1.01338279e+00 9.56519842e-01 -3.33134294e-01 8.10110569e-01
5.80196202e-01 -1.22696333e-01 1.20360363e+00 -8.65775764e-01
9.87724602e-01 1.50723577e+00 8.62735271e-01 2.17032537e-01
-1.20848417e+00 -3.50777119e-01 3.95105451e-01 3.44239324e-01
-1.15120244e+00 -2.26922631e-01 9.28226054e-01 -7.02774048e-01
1.38213098e+00 -2.63444963e-03 6.97917342e-01 1.29445946e+00
4.24887873e-02 4.65338528e-01 1.42520189e+00 -9.78597581e-01
3.25087994e-01 2.39853784e-01 7.79785156e-01 4.59864348e-01
2.92161584e-01 2.55602039e-02 -6.19767010e-01 -3.50875817e-02
2.04315111e-01 2.76168417e-02 8.69117305e-02 -2.16319516e-01
-1.12133288e+00 1.02777696e+00 3.84201169e-01 1.27202284e+00
-6.06915355e-01 3.07120740e-01 5.26290774e-01 5.16841650e-01
8.03654134e-01 6.44944429e-01 -5.11292696e-01 -5.28264344e-01
-7.99719572e-01 -2.17825994e-02 5.88447511e-01 4.78223443e-01
9.94660854e-01 -9.91417691e-02 1.23212300e-01 1.04904962e+00
3.03940535e-01 1.61609843e-01 1.25098228e+00 -7.05176294e-01
-1.31361380e-01 7.41814315e-01 -1.00843772e-01 -1.32089543e+00
-3.58828425e-01 1.88772622e-02 -1.49950460e-01 -3.88533056e-01
-2.46476933e-01 2.60447770e-01 -6.79000199e-01 1.90072739e+00
-3.90203781e-02 -1.98526308e-03 -1.81250691e-01 3.53915662e-01
5.80416381e-01 5.25650442e-01 5.72587073e-01 -5.01679182e-01
1.25443435e+00 -4.20747846e-01 -1.02040374e+00 -1.80753805e-02
9.07864332e-01 -5.24283469e-01 1.24181890e+00 9.15440917e-02
-1.00169849e+00 -5.06046295e-01 -9.12921011e-01 2.19961032e-01
-8.99900973e-01 -4.81147379e-01 6.94535315e-01 6.98901534e-01
-1.16170621e+00 8.18528235e-01 -8.45172703e-01 -1.15790498e+00
1.04424752e-01 8.27849954e-02 7.84817152e-03 5.18678367e-01
-1.58805811e+00 1.14161420e+00 6.35035753e-01 -2.66834438e-01
-3.56697798e-01 -3.74239981e-01 -9.01309133e-01 -4.31169361e-01
1.93444788e-01 -4.29501504e-01 1.31221604e+00 -1.25673735e+00
-1.33751547e+00 1.21296692e+00 -3.48886788e-01 -4.72579598e-01
-7.56540298e-02 7.73636922e-02 -7.07075417e-01 -2.64263600e-01
1.00447543e-01 3.70240301e-01 5.26210964e-01 -1.32382643e+00
-5.21598756e-01 -4.54121679e-01 -1.07854493e-01 1.43079907e-01
-9.13690627e-01 5.05353473e-02 1.79685161e-01 -1.00691068e+00
-1.48058906e-01 -6.75111115e-01 -1.79480314e-01 -6.65311873e-01
1.22790132e-02 -6.47876918e-01 2.91867584e-01 -3.55203182e-01
1.79645574e+00 -1.89476776e+00 2.65495449e-01 3.85043412e-01
1.08840197e-01 2.45734751e-02 -2.68990193e-02 8.95969093e-01
-1.88356176e-01 5.58985114e-01 -3.56504083e-01 -2.88736761e-01
9.57502350e-02 3.81802082e-01 -4.58359748e-01 3.60892206e-01
-2.59861946e-01 1.09381461e+00 -1.16863990e+00 -3.16183478e-01
3.05428535e-01 3.30543250e-01 -4.80957061e-01 -1.64051980e-01
-1.47074059e-01 -2.84847077e-02 -4.71101791e-01 4.07790869e-01
1.52286977e-01 3.87133136e-02 4.54967886e-01 -6.84647262e-02
-4.02177960e-01 6.68737233e-01 -7.28836715e-01 1.75109398e+00
-5.68626940e-01 8.54516268e-01 -8.47004414e-01 -1.21140134e+00
1.05598676e+00 -4.42917051e-04 2.32421279e-01 -9.92118716e-01
2.42664278e-01 4.76799756e-02 -1.30188942e-01 -6.01677835e-01
8.34665000e-01 -3.59571874e-01 -4.70200956e-01 5.66997290e-01
1.80665225e-01 -2.30836235e-02 1.24454595e-01 1.40183335e-02
1.01649642e+00 3.84174734e-02 1.05243802e+00 -6.71674132e-01
4.00868773e-01 -3.89647111e-02 1.83975935e-01 5.59729636e-01
-2.88540184e-01 -3.07055209e-02 1.81255355e-01 -6.12358332e-01
-7.33588934e-01 -9.80050683e-01 -1.88948885e-01 1.60097718e+00
5.08499667e-02 -6.10701859e-01 -5.82920790e-01 -5.73095143e-01
4.80708927e-02 1.29857802e+00 -9.85993087e-01 -4.06412363e-01
-5.24492502e-01 -5.52516937e-01 5.03818274e-01 5.40667415e-01
-8.76964815e-03 -1.65257871e+00 -6.79229975e-01 2.26789370e-01
5.25237359e-02 -5.30035734e-01 -4.66321446e-02 2.16399908e-01
-9.85366046e-01 -7.86679745e-01 -1.24841079e-01 -1.00142205e+00
4.20545369e-01 2.21155852e-01 1.17924142e+00 7.02260360e-02
-2.77571827e-01 5.76861441e-01 -7.69586861e-01 -2.71358788e-01
-5.39120972e-01 2.78030306e-01 1.74424902e-01 -1.84995323e-01
1.08431816e+00 -7.56859481e-01 -4.54399973e-01 -2.03955263e-01
-1.09092855e+00 -4.79283214e-01 1.28817573e-01 6.02751136e-01
2.60507703e-01 -1.16352849e-01 6.16541505e-01 -1.16795361e+00
1.20612156e+00 -9.84552801e-01 -5.73076084e-02 1.01153582e-01
-1.13455999e+00 2.43996158e-01 2.02574119e-01 -3.78165036e-01
-1.14998758e+00 -3.21087003e-01 -2.92057022e-02 1.12396784e-01
-3.52579653e-01 8.96093965e-01 4.20969874e-01 3.99437219e-01
9.97054935e-01 1.78671002e-01 -3.30048621e-01 -5.39244771e-01
9.03687060e-01 9.55895483e-01 3.53050947e-01 -3.52051258e-01
4.58871990e-01 5.70098579e-01 -7.92020261e-01 -1.12989068e+00
-6.86389387e-01 -7.25794494e-01 -8.94465506e-01 -2.43693307e-01
7.04244554e-01 -6.06292784e-01 -1.86594903e-01 2.74094075e-01
-1.16906095e+00 -2.22476006e-01 -5.43608427e-01 2.74049968e-01
-5.24469435e-01 6.20295167e-01 -4.09619212e-01 -8.05091262e-01
-1.50839761e-01 -6.03086948e-01 9.03229237e-01 9.75939445e-03
-1.18795693e+00 -1.65912271e+00 8.83426964e-01 -7.27025047e-02
4.75668877e-01 2.65364170e-01 9.06354845e-01 -8.31466794e-01
1.91493824e-01 3.60438637e-02 2.94831723e-01 1.33217648e-02
5.93115568e-01 5.14213443e-02 -8.18866313e-01 -1.98640510e-01
7.97442794e-02 -2.32877433e-01 1.16840219e+00 1.75129160e-01
1.06998861e+00 -2.29179263e-01 -3.60737383e-01 6.95999190e-02
1.41037452e+00 3.28382581e-01 4.28453684e-01 6.82645500e-01
4.53615665e-01 7.30398118e-01 4.04946089e-01 4.49670345e-01
5.52388430e-01 6.86585784e-01 1.38173066e-02 1.73415110e-01
-6.72676489e-02 -3.08877110e-01 4.36942339e-01 1.48084128e+00
-1.19302355e-01 -2.30001062e-01 -1.05948687e+00 1.13949800e+00
-1.89488900e+00 -1.02527642e+00 5.09182364e-03 1.84246540e+00
1.05501139e+00 1.05612371e-02 9.77194458e-02 1.55475646e-01
7.43260920e-01 5.18004239e-01 -2.11829588e-01 -9.10594285e-01
1.27058715e-01 6.77897036e-01 3.19686770e-01 6.68323815e-01
-8.19958568e-01 1.49675035e+00 7.44999456e+00 3.78029227e-01
-7.40031898e-01 2.48610929e-01 -9.96917952e-03 3.07349682e-01
-8.89615834e-01 -6.98361024e-02 -1.91972494e-01 4.75305706e-01
1.21696985e+00 -4.87769097e-01 2.77842492e-01 5.31213999e-01
1.52122945e-01 9.51673090e-02 -9.98552918e-01 8.13896239e-01
3.32287520e-01 -1.06189585e+00 4.04407918e-01 -1.49558082e-01
7.59885371e-01 -1.06800064e-01 -1.05860867e-01 2.32090130e-01
6.12937033e-01 -8.84723365e-01 4.64451313e-01 8.87495100e-01
4.93751407e-01 -7.71967351e-01 5.08993149e-01 4.54270020e-02
-1.08097947e+00 6.34923279e-02 -2.04568371e-01 -5.37812054e-01
1.96881369e-01 4.36986834e-01 -8.56130719e-01 3.00828636e-01
4.67449635e-01 1.23855412e+00 -7.62118518e-01 4.57780480e-01
-2.74923235e-01 8.82726312e-01 3.13387178e-02 -4.27515894e-01
1.01111926e-01 2.35213619e-03 4.68846112e-01 1.76498461e+00
2.26984024e-01 -2.12894827e-01 -8.91060755e-02 8.22183371e-01
-2.20330488e-02 6.11373603e-01 -1.01504922e+00 -5.77023149e-01
7.91819215e-01 9.30263758e-01 -1.10925007e+00 -4.82507378e-01
-2.65647858e-01 1.10300255e+00 4.37643588e-01 2.24856436e-01
-4.93388683e-01 -3.56530130e-01 7.30905354e-01 -7.77659798e-03
3.09815466e-01 -2.56846935e-01 -4.39576864e-01 -1.14882708e+00
-1.89703047e-01 -4.92434591e-01 4.42873478e-01 -4.91942912e-01
-1.53777111e+00 5.34599960e-01 2.63867319e-01 -7.30978310e-01
-6.10798895e-01 -4.43138748e-01 -5.41020870e-01 3.34959567e-01
-1.37644017e+00 -6.70504093e-01 -2.18938649e-01 3.31945211e-01
7.95652688e-01 -2.41800889e-01 1.05471683e+00 -8.84273127e-02
-2.72187501e-01 3.38580370e-01 -5.57887591e-02 -2.08170205e-01
5.12645543e-01 -1.23085535e+00 7.19302297e-01 5.22525311e-01
5.78959823e-01 9.42340910e-01 7.86410689e-01 -6.34654760e-01
-7.48982191e-01 -8.80176842e-01 1.48220909e+00 -7.01833844e-01
1.00603068e+00 -1.47406623e-01 -9.28588152e-01 8.75620842e-01
6.88314974e-01 -6.32439554e-01 1.07153785e+00 3.80491793e-01
-4.37260717e-01 3.28645617e-01 -8.29658091e-01 5.70696592e-01
1.39559782e+00 -8.62390935e-01 -1.67260194e+00 2.20206827e-01
1.03932190e+00 3.48641187e-01 -7.79326558e-01 1.54313847e-01
2.83413589e-01 -7.30928183e-01 7.13391066e-01 -8.57397437e-01
1.61092907e-01 -3.00705135e-02 -3.26278687e-01 -1.61381733e+00
-4.28545743e-01 -3.12344581e-01 9.59566981e-02 1.28616500e+00
3.56735766e-01 -8.90848696e-01 2.13041425e-01 -6.08480396e-03
1.25419766e-01 -1.78023994e-01 -8.18233192e-01 -5.27642369e-01
4.06931013e-01 -7.91402876e-01 5.26142299e-01 1.51466680e+00
5.43507695e-01 4.45422918e-01 1.01277985e-01 -4.36268061e-01
1.02951527e-01 -5.40548190e-02 1.29970416e-01 -1.37022638e+00
6.13682717e-02 -8.10304463e-01 -6.95148587e-01 -6.18029058e-01
5.67751765e-01 -1.28281546e+00 -1.16548270e-01 -1.30634451e+00
1.52753040e-01 4.92921248e-02 -7.10543096e-01 3.82557899e-01
2.29997048e-03 2.27660388e-01 -3.72986421e-02 2.83337653e-01
-6.76790714e-01 5.47982275e-01 4.43012834e-01 -3.13663855e-02
-3.19226295e-01 -5.77906191e-01 -8.84857714e-01 9.29102480e-01
1.05071926e+00 -4.34031337e-01 -4.98948187e-01 -2.21528322e-01
6.72131598e-01 -8.82912695e-01 6.58391640e-02 -6.18608296e-01
-1.57573745e-01 -2.63231486e-01 -9.75127984e-03 -2.20864043e-01
-1.49825037e-01 -5.62018156e-01 -1.73482855e-04 5.00119448e-01
-8.43171120e-01 4.19180721e-01 1.46980643e-01 5.35769761e-01
-2.60228872e-01 -2.53044248e-01 5.55327058e-01 -2.27835044e-01
-1.25904644e+00 -3.09370458e-01 -8.05004895e-01 6.02603294e-02
1.00090814e+00 -1.48849159e-01 4.26914021e-02 -2.67101943e-01
-9.15194392e-01 -8.53181109e-02 5.48877001e-01 1.02577329e+00
3.83321553e-01 -1.66331875e+00 -3.66131097e-01 5.38494019e-03
3.84029061e-01 -7.34247565e-01 -1.84816539e-01 3.14001828e-01
-2.07409933e-01 4.30208206e-01 -4.16932590e-02 -5.20955384e-01
-1.03690886e+00 6.85178995e-01 2.08706424e-01 -1.08630940e-01
-4.92215455e-01 4.76223618e-01 -1.99673504e-01 -6.55289650e-01
-3.19101363e-02 -5.04980266e-01 -7.54539371e-01 8.22555423e-01
3.10387313e-01 4.58039641e-01 -1.08325314e-02 -9.63724017e-01
-4.44743782e-01 6.19596422e-01 -2.52393484e-02 -2.51021594e-01
1.39251149e+00 -5.08495569e-01 -3.40718031e-01 1.37349224e+00
1.35153627e+00 -1.52273789e-01 -3.87674749e-01 -5.00146925e-01
7.94930696e-01 -2.71290690e-01 -1.26985032e-02 -6.45838439e-01
-6.05420947e-01 6.16087377e-01 7.73512244e-01 6.91606820e-01
8.73272598e-01 3.04489434e-01 6.32001579e-01 5.85750103e-01
3.99060160e-01 -1.41362607e+00 1.56099424e-01 6.90723240e-01
8.26741040e-01 -7.82280207e-01 -1.81327447e-01 3.52635309e-02
-5.83126128e-01 1.05610085e+00 3.60844374e-01 -3.52031022e-01
7.83284187e-01 -1.12030901e-01 1.53352588e-01 -6.59621418e-01
-8.03057611e-01 -4.88597453e-01 1.50556907e-01 4.55353111e-01
8.85158360e-01 3.17038447e-01 -1.07019508e+00 4.22090501e-01
-5.59085071e-01 -6.19149692e-02 4.80877697e-01 1.00797069e+00
-6.16827726e-01 -1.26586831e+00 6.33923784e-02 2.05861688e-01
-1.67298496e-01 -1.04960330e-01 -9.72696364e-01 7.85866737e-01
1.54342428e-01 8.50415766e-01 3.89883488e-01 -4.65669781e-01
5.27843356e-01 7.26936102e-01 2.20221221e-01 -8.00714254e-01
-6.39409959e-01 -1.24813467e-01 6.75134808e-02 -7.31879890e-01
-9.47476625e-01 -9.12831843e-01 -1.34164572e+00 -1.00898549e-01
-6.82381466e-02 -5.81985246e-03 5.82476914e-01 9.73156750e-01
3.05266947e-01 5.82861066e-01 7.64132082e-01 -6.15581453e-01
-2.45539457e-01 -1.32340252e+00 -4.04555827e-01 8.65612864e-01
1.03004918e-01 -8.77579510e-01 -5.07772923e-01 1.96738750e-01] | [10.233344078063965, 8.944576263427734] |
2a968089-ab85-4cab-ad65-2160d31380b2 | an-open-unified-deep-graph-learning-framework | 2301.03424 | null | https://arxiv.org/abs/2301.03424v2 | https://arxiv.org/pdf/2301.03424v2.pdf | An open unified deep graph learning framework for discovering drug leads | Computational discovery of ideal lead compounds is a critical process for modern drug discovery. It comprises multiple stages: hit screening, molecular property prediction, and molecule optimization. Current efforts are disparate, involving the establishment of models for each stage, followed by multi-stage multi-model integration. However, this is non-ideal, as clumsy integration of incompatible models increases research overheads, and may even reduce success rates in drug discovery. Facilitating compatibilities requires establishing inherent model consistencies across lead discovery stages. Towards that effect, we propose an open deep graph learning (DGL) based pipeline: generative adversarial feature subspace enhancement (GAFSE), which first unifies the modeling of these stages into one learning framework. GAFSE also offers standardized modular design and streamlined interfaces for future expansions and community support. GAFSE combines adversarial/generative learning, graph attention network, graph reconstruction network, and optimizes the classification/regression loss, adversarial/generative loss, and reconstruction loss simultaneously. Convergence analysis theoretically guarantees model generalization performance. Exhaustive benchmarking demonstrates that the GAFSE pipeline achieves excellent performance across almost all lead discovery stages, while also providing valuable model interpretability. Hence, we believe this tool will enhance the efficiency and productivity of drug discovery researchers. | ['Wilson Wen Bin Goh', 'JianSheng Wu', 'Chun Ye', 'Jitao Yang', 'Zhen Yang', 'Haifeng Hu', 'Yueming Yin'] | 2022-12-06 | null | null | null | null | ['graph-reconstruction', 'molecular-property-prediction'] | ['graphs', 'miscellaneous'] | [ 1.46242261e-01 -1.66774020e-01 -4.52397048e-01 9.72632095e-02
-8.17892134e-01 -7.97410071e-01 4.11017120e-01 2.86438286e-01
9.05010626e-02 7.55726337e-01 -1.60236910e-01 -9.29125965e-01
-2.09497392e-01 -6.20893121e-01 -7.75173485e-01 -6.15120590e-01
-8.28758180e-02 6.30192876e-01 -2.29335606e-01 -2.60039661e-02
1.63004309e-01 9.32639539e-01 -5.51278234e-01 1.36593908e-01
9.52689648e-01 7.11194813e-01 -1.09747954e-01 4.79466736e-01
3.13098840e-02 7.24655986e-01 -3.01066458e-01 -7.12678075e-01
1.75994545e-01 -2.55900204e-01 -8.62430155e-01 -2.85697699e-01
3.22366536e-01 -1.21043734e-01 -2.20094413e-01 8.26528132e-01
9.59121466e-01 -1.42750710e-01 6.03350878e-01 -1.20836794e+00
-1.02908909e+00 3.98541629e-01 -3.79787743e-01 3.24847107e-03
2.99646676e-01 6.12960577e-01 1.11440647e+00 -1.04571140e+00
6.14444971e-01 1.03626692e+00 8.13789248e-01 5.26711583e-01
-1.45702100e+00 -8.99547100e-01 1.76926672e-01 5.55943064e-02
-1.45943356e+00 -3.59790981e-01 5.91962099e-01 -7.03940630e-01
1.07351565e+00 1.65622056e-01 6.64103627e-01 1.26793456e+00
5.31825423e-01 6.12968862e-01 5.43924809e-01 1.61904126e-01
3.30722034e-01 -8.08255374e-02 8.71009678e-02 7.80339003e-01
2.83067286e-01 1.16725489e-01 -3.31587672e-01 -5.00333309e-01
8.16997528e-01 2.48702794e-01 -2.21498698e-01 -3.76925886e-01
-8.01642060e-01 8.53688300e-01 6.16199970e-01 -9.79765505e-02
-3.40884447e-01 1.13237411e-01 1.95291787e-01 2.02754200e-01
3.27138007e-01 9.43123639e-01 -2.66785860e-01 2.46672213e-01
-7.98525512e-01 3.12894672e-01 7.48991787e-01 9.29556310e-01
5.40807605e-01 1.47038281e-01 -1.24265298e-01 3.68911237e-01
3.96392971e-01 1.14822172e-01 -2.10908763e-02 -4.64864582e-01
5.43314852e-02 6.92743182e-01 -1.55973226e-01 -1.01451707e+00
-6.83857441e-01 -1.10179174e+00 -7.64387846e-01 4.83215600e-02
6.53897151e-02 -9.83264670e-02 -8.52059960e-01 1.71295869e+00
4.79604661e-01 2.37222672e-01 -7.88209885e-02 5.49779236e-01
8.83826017e-01 4.27794337e-01 6.72116578e-01 -4.29934636e-02
8.86548162e-01 -9.63161469e-01 -5.36386251e-01 1.99391916e-02
6.22443318e-01 -7.40997374e-01 9.42454100e-01 4.05927151e-01
-1.02136207e+00 -3.41906339e-01 -1.03423285e+00 -6.70810342e-02
-4.40378428e-01 -7.38205761e-02 1.07252884e+00 6.11558795e-01
-7.81796217e-01 9.41064835e-01 -8.72575223e-01 1.07574336e-01
9.72611785e-01 8.78588438e-01 -3.94079715e-01 -1.74853742e-01
-1.06221008e+00 7.82522798e-01 1.32759154e-01 1.04149163e-01
-1.27038741e+00 -1.35449266e+00 -7.20052361e-01 1.58395991e-02
2.35563219e-01 -1.16567922e+00 7.47569382e-01 -4.57292229e-01
-1.24568272e+00 3.57627690e-01 2.98658628e-02 -4.52991843e-01
4.80022132e-01 -6.02263957e-04 -3.17423642e-01 -2.80766785e-01
3.56883705e-02 4.23040509e-01 6.64623141e-01 -1.00855124e+00
-1.69572141e-03 -4.60439235e-01 -1.98476300e-01 6.30318299e-02
-1.92704216e-01 -1.82943523e-01 -3.78309488e-01 -4.57347929e-01
-3.31579775e-01 -7.72531331e-01 -6.18637204e-01 -1.06238820e-01
-6.95229769e-01 -8.82389843e-02 6.91052675e-01 -6.26425982e-01
1.41826582e+00 -1.90181589e+00 1.95000827e-01 3.25860441e-01
7.42749214e-01 3.73617560e-01 -3.60019028e-01 5.72718501e-01
-3.35387290e-01 5.19519210e-01 -2.29517296e-02 -1.83439538e-01
-1.33577794e-01 -4.52387094e-01 -2.10215613e-01 4.62966651e-01
5.35523176e-01 1.51229739e+00 -9.84499097e-01 -2.33268794e-02
8.65425617e-02 7.52847195e-01 -8.95471454e-01 7.14886039e-02
-3.82119715e-01 6.28601491e-01 -7.84490108e-01 1.09461761e+00
5.91388106e-01 -8.04757953e-01 2.42689669e-01 -2.46007264e-01
1.41106993e-01 1.57405689e-01 -8.52378249e-01 1.63579869e+00
-1.86763242e-01 1.23539139e-02 -3.38038832e-01 -6.28050447e-01
8.02466273e-01 1.19610250e-01 7.80426204e-01 -3.76914263e-01
5.72708845e-02 2.87149936e-01 1.31687090e-01 -1.32873565e-01
-8.71674791e-02 -9.20804292e-02 2.38544375e-01 1.97408587e-01
1.40531644e-01 4.14806567e-02 -9.87281799e-02 2.07467705e-01
1.24665391e+00 2.80486941e-01 2.30604321e-01 -1.47572532e-01
2.08741531e-01 1.09595999e-01 5.43205857e-01 4.80140924e-01
-1.62982970e-01 3.86484236e-01 6.37160778e-01 -5.83436131e-01
-1.04162145e+00 -1.15996122e+00 -2.90824652e-01 8.28586996e-01
-4.73772138e-02 -7.70153642e-01 -5.97721636e-01 -9.91571784e-01
1.37936532e-01 4.34192091e-01 -5.95099330e-01 -5.58050156e-01
-2.08514839e-01 -1.13766742e+00 5.88139534e-01 2.68673480e-01
3.52591649e-02 -6.63768888e-01 3.01507831e-01 5.37525415e-01
4.65402961e-01 -7.02099919e-01 -6.56257093e-01 2.34664217e-01
-4.75441813e-01 -1.36626649e+00 -4.90239084e-01 -7.17512310e-01
4.76243973e-01 3.27342525e-02 1.11059785e+00 2.59043518e-02
-5.95848799e-01 -3.21703628e-02 -1.70904957e-03 -4.27464366e-01
-5.01409173e-01 1.03835218e-01 3.36784939e-03 -1.69072092e-01
2.26210922e-01 -7.60640562e-01 -8.25367093e-01 2.14035749e-01
-7.43010998e-01 -6.23329058e-02 4.97965604e-01 9.56966043e-01
1.02210891e+00 -1.75118029e-01 8.52002025e-01 -1.00596511e+00
6.69378579e-01 -6.04562998e-01 -6.78182006e-01 4.96448010e-01
-1.02452362e+00 3.78068537e-02 9.22476292e-01 -3.65301490e-01
-4.82935518e-01 1.76114470e-01 -5.36227226e-01 -5.57686567e-01
2.71656215e-01 6.13282144e-01 -5.38424730e-01 -5.80578506e-01
8.24064136e-01 2.32666694e-02 6.74716756e-02 -4.43874538e-01
2.54130542e-01 2.76635438e-01 1.56805143e-01 -2.81796843e-01
7.21450984e-01 6.83403984e-02 3.36224824e-01 -5.81169903e-01
-5.03224969e-01 -1.54972136e-01 -2.00277165e-01 2.68975437e-01
6.99630141e-01 -1.03417957e+00 -9.23450291e-01 1.96839750e-01
-1.05526888e+00 -3.27711076e-01 -1.28873646e-01 3.22796106e-01
-2.46641979e-01 2.85614014e-01 -4.95052963e-01 -2.38750353e-01
-7.25703835e-01 -1.70704746e+00 9.91939306e-01 1.76016577e-02
-2.67800510e-01 -1.31716931e+00 1.16785958e-01 4.74113733e-01
4.36612725e-01 6.09460354e-01 1.11874390e+00 -9.78054583e-01
-8.27642918e-01 -2.87096173e-01 -3.62645611e-02 1.56729683e-01
2.49721959e-01 8.94007310e-02 -9.15471435e-01 -3.93066943e-01
-3.99919093e-01 -3.89124453e-01 6.83512211e-01 6.06780231e-01
1.33284318e+00 -3.15901041e-01 -5.19576430e-01 1.10569382e+00
1.32397485e+00 4.19664532e-01 6.69946790e-01 8.13850313e-02
1.21600950e+00 2.43061166e-02 1.24259256e-01 3.52137923e-01
2.11358994e-01 4.54870164e-01 4.70746785e-01 -6.83141172e-01
-8.60297605e-02 -3.39262068e-01 2.44575918e-01 1.59936786e-01
3.18830833e-02 -3.23845327e-01 -9.88586545e-01 -9.04656649e-02
-1.51692140e+00 -8.75401378e-01 -5.58501184e-02 2.13309407e+00
8.79886270e-01 2.80144680e-02 2.16249853e-01 -3.89993846e-01
3.52231681e-01 -1.27998099e-01 -1.04968858e+00 -2.64491707e-01
-1.46767229e-01 6.00176215e-01 4.60276872e-01 5.53853750e-01
-1.08482933e+00 1.20833683e+00 6.59090996e+00 1.06760979e+00
-1.34104443e+00 -9.55563486e-02 1.14323902e+00 6.28413782e-02
-6.68077946e-01 -1.18241347e-02 -8.81775022e-01 2.77426183e-01
8.62569809e-01 -3.14253151e-01 5.16902983e-01 9.25705910e-01
2.75614291e-01 6.45248652e-01 -1.31302309e+00 7.24352539e-01
-2.90764391e-01 -2.10324907e+00 5.53274214e-01 3.08131397e-01
6.83435619e-01 1.48593962e-01 3.73933047e-01 9.14292634e-02
3.97911429e-01 -1.48928308e+00 1.76985011e-01 2.69601762e-01
8.83776307e-01 -9.93216157e-01 3.08592021e-01 -3.89985293e-02
-1.03306699e+00 1.21793747e-01 -1.46179855e-01 3.31151783e-01
-1.04871616e-01 4.17279363e-01 -1.05988467e+00 7.91851401e-01
4.96137887e-02 9.63640630e-01 -6.68357253e-01 1.04843342e+00
6.23221435e-02 4.62211102e-01 9.03824866e-02 1.01258181e-01
1.67634338e-01 -3.56190622e-01 4.69939888e-01 1.05562449e+00
-4.24273610e-02 -3.36563408e-01 4.69832540e-01 1.18273091e+00
-3.67054522e-01 2.52952814e-01 -6.14260197e-01 -4.63120550e-01
4.94365036e-01 1.09048438e+00 -5.08772492e-01 1.64212361e-02
-2.69430995e-01 7.90329695e-01 2.88873732e-01 4.43511248e-01
-1.04373407e+00 -2.82488793e-01 8.60491514e-01 1.05208203e-01
-1.12318277e-01 3.29864994e-02 -4.35488194e-01 -9.75071967e-01
-3.76929879e-01 -1.17867887e+00 3.19551170e-01 -2.32957572e-01
-1.27568352e+00 6.98560476e-01 -6.06353104e-01 -1.01352739e+00
1.65989071e-01 -6.57531500e-01 -6.75513625e-01 1.10584426e+00
-1.42110455e+00 -1.34477937e+00 6.18264936e-02 4.93174851e-01
3.84558469e-01 -3.25413704e-01 9.09061551e-01 5.52963853e-01
-1.18517804e+00 1.00020611e+00 9.60831195e-02 -3.04931104e-01
5.10188282e-01 -1.13971186e+00 6.91727817e-01 6.44153655e-01
5.70041873e-03 9.44387138e-01 4.46587980e-01 -9.39022124e-01
-1.74299121e+00 -1.60989034e+00 4.84534770e-01 -6.37847662e-01
8.70297015e-01 -3.40005636e-01 -8.13683510e-01 5.45619607e-01
-2.70110011e-01 -5.37023209e-02 1.14982700e+00 9.04375017e-02
-4.10490453e-01 8.97672400e-02 -8.60414386e-01 7.64176190e-01
9.38170910e-01 -5.79242289e-01 2.77332425e-01 9.10432816e-01
9.24156547e-01 -2.88453639e-01 -1.33006155e+00 5.12726128e-01
4.47179109e-01 -4.61803794e-01 1.35158825e+00 -1.08031440e+00
4.71003562e-01 -2.86715090e-01 1.36642069e-01 -1.13445950e+00
-6.59836650e-01 -1.17520833e+00 -1.28619105e-01 8.48926365e-01
1.04407084e+00 -4.71757710e-01 8.45240474e-01 7.79439092e-01
-4.21369016e-01 -1.37008476e+00 -5.51844478e-01 -7.66807735e-01
3.84560168e-01 -2.80037165e-01 7.53150105e-01 9.40491140e-01
-1.33091986e-01 7.67933011e-01 -3.63906771e-01 1.61605373e-01
5.89697719e-01 -2.04101987e-02 7.22467840e-01 -1.04078996e+00
-6.46816075e-01 -7.86839247e-01 -3.73538435e-01 -6.68607235e-01
1.20257959e-04 -1.29193354e+00 -7.09224701e-01 -1.56346035e+00
3.05109411e-01 -6.23294890e-01 -3.36043328e-01 6.11291647e-01
-5.11902034e-01 7.16521442e-02 -2.08395272e-01 1.71720043e-01
-4.84021574e-01 5.96993446e-01 1.59050250e+00 -4.22422886e-01
-5.42021573e-01 1.07829198e-01 -1.13153768e+00 3.20932508e-01
7.04999089e-01 -2.19864786e-01 -4.81362909e-01 -8.98057371e-02
1.16477333e-01 -2.98404276e-01 4.18896437e-01 -7.49168634e-01
3.84131297e-02 -2.53574073e-01 6.01161182e-01 -1.93643823e-01
7.46157169e-02 -5.00277340e-01 7.52916336e-01 6.60871446e-01
-1.57108516e-01 -1.35258973e-01 3.42293471e-01 7.22626150e-01
5.13300225e-02 3.94280344e-01 6.53984189e-01 -9.05807316e-03
-2.84722626e-01 1.08852315e+00 2.46008515e-01 -9.82552990e-02
9.91357684e-01 -2.18262389e-01 -2.47662425e-01 6.90268725e-02
-8.78490567e-01 3.07088971e-01 4.67478603e-01 5.04484735e-02
6.19265676e-01 -1.01134861e+00 -6.32373869e-01 2.96983778e-01
1.46930754e-01 1.29793715e-02 3.05493534e-01 8.62707555e-01
-6.48014367e-01 2.40653247e-01 1.50363192e-01 -2.62669265e-01
-1.09497154e+00 7.64172494e-01 5.47227859e-01 -5.52964687e-01
-2.05429018e-01 1.05600023e+00 4.83011901e-01 -2.47250050e-01
2.99586713e-01 1.30013496e-01 6.67342097e-02 -1.58429325e-01
4.67012376e-01 2.98151165e-01 3.04518133e-01 -1.36604607e-01
-3.61336052e-01 3.41180891e-01 -4.98208374e-01 5.67825139e-01
1.45590866e+00 5.02783716e-01 6.38257563e-02 -2.61104912e-01
1.28237486e+00 -1.20998994e-01 -1.30273283e+00 8.83148611e-02
-1.44408256e-01 4.46830057e-02 2.94128060e-01 -9.78449523e-01
-8.31798077e-01 7.66048193e-01 4.07813907e-01 -1.65870696e-01
1.04349649e+00 -7.74660632e-02 7.10904479e-01 3.60618323e-01
1.91028386e-01 -6.15203679e-01 1.60386547e-01 3.09818715e-01
9.45161581e-01 -1.20366764e+00 2.72512108e-01 -5.33834100e-01
-5.10252595e-01 9.53454852e-01 4.93878990e-01 1.66439399e-01
6.62642598e-01 8.27391446e-02 -3.59177947e-01 -5.64719260e-01
-6.97222948e-01 3.28312144e-02 4.79025781e-01 5.87900400e-01
6.05328500e-01 1.13164503e-02 -6.37606382e-02 7.30398893e-01
1.63082331e-01 -1.31199479e-01 -7.25603178e-02 7.32979655e-01
-1.10307679e-01 -1.54211259e+00 -8.85053258e-03 4.75828946e-01
-5.40400267e-01 -6.81497097e-01 -7.30202913e-01 7.12640882e-01
-6.18419275e-02 7.49692917e-01 -5.29134095e-01 -6.40372515e-01
2.97240019e-01 -7.42793530e-02 4.27743733e-01 -5.79683125e-01
-7.99944162e-01 3.85099381e-01 -1.04357310e-01 -5.71110249e-01
2.92160392e-01 -1.73030466e-01 -1.02906466e+00 -4.38592732e-01
-5.06870031e-01 1.64098233e-01 5.76822400e-01 6.19789839e-01
9.85175133e-01 7.37060726e-01 9.63161349e-01 -5.63636959e-01
-5.43550730e-01 -3.56610775e-01 -3.86002719e-01 6.86194226e-02
2.11858720e-01 -4.61597800e-01 -1.14402615e-01 -5.64762764e-02] | [5.079238414764404, 5.826414108276367] |
6c3c5787-dbdb-42ed-bf3c-bcc4fb0b97d5 | cross-domain-adaptation-for-animal-pose | 1908.05806 | null | https://arxiv.org/abs/1908.05806v2 | https://arxiv.org/pdf/1908.05806v2.pdf | Cross-Domain Adaptation for Animal Pose Estimation | In this paper, we are interested in pose estimation of animals. Animals usually exhibit a wide range of variations on poses and there is no available animal pose dataset for training and testing. To address this problem, we build an animal pose dataset to facilitate training and evaluation. Considering the heavy labor needed to label dataset and it is impossible to label data for all concerned animal species, we, therefore, proposed a novel cross-domain adaptation method to transform the animal pose knowledge from labeled animal classes to unlabeled animal classes. We use the modest animal pose dataset to adapt learned knowledge to multiple animals species. Moreover, humans also share skeleton similarities with some animals (especially four-footed mammals). Therefore, the easily available human pose dataset, which is of a much larger scale than our labeled animal dataset, provides important prior knowledge to boost up the performance on animal pose estimation. Experiments show that our proposed method leverages these pieces of prior knowledge well and achieves convincing results on animal pose estimation. | ['Yu-Wing Tai', 'Hao-Shu Fang', 'Xiaoyong Shen', 'Jinkun Cao', 'Hongyang Tang', 'Cewu Lu'] | 2019-08-16 | cross-domain-adaptation-for-animal-pose-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Cao_Cross-Domain_Adaptation_for_Animal_Pose_Estimation_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Cao_Cross-Domain_Adaptation_for_Animal_Pose_Estimation_ICCV_2019_paper.pdf | iccv-2019-10 | ['animal-pose-estimation'] | ['computer-vision'] | [ 2.93071065e-02 -2.13838160e-01 -4.02637005e-01 -7.35974491e-01
-4.97370601e-01 -6.31535113e-01 1.43899813e-01 3.99207510e-02
-4.66469705e-01 8.20363700e-01 -1.53425887e-01 1.87771350e-01
1.46103144e-01 -6.53347135e-01 -1.10065782e+00 -3.04343671e-01
-2.69812614e-01 4.79990691e-01 4.65017110e-01 -2.02371940e-01
-9.06914845e-02 5.88308692e-01 -1.35641909e+00 -4.75226231e-02
6.88507795e-01 9.32631314e-01 3.14121187e-01 1.88553974e-01
3.30358326e-01 3.89915109e-01 -3.55870754e-01 -3.12117010e-01
4.40449208e-01 -3.86667252e-01 -6.72108471e-01 1.45814016e-01
5.13347208e-01 -5.90391099e-01 -1.09286755e-01 7.80328333e-01
3.22190613e-01 1.56647727e-01 6.91114604e-01 -1.21104884e+00
-2.88613379e-01 5.30426383e-01 -9.57086384e-01 8.51596594e-02
2.26151079e-01 4.20703553e-02 9.36415017e-01 -6.80952609e-01
5.92510164e-01 1.23271525e+00 9.38190222e-01 4.80290860e-01
-1.05794287e+00 -9.91965652e-01 1.79985702e-01 9.07849818e-02
-1.33590281e+00 -2.07166299e-01 8.79161417e-01 -4.43112940e-01
1.95991948e-01 -1.28897592e-01 7.55066514e-01 1.41515088e+00
-6.11387892e-03 7.31036007e-01 1.11470091e+00 -6.05508164e-02
1.75392106e-01 3.73368827e-03 -1.80596039e-01 4.56164300e-01
4.84298706e-01 1.09359644e-01 -5.45188487e-01 -2.18150695e-03
9.39397037e-01 1.48967057e-01 -6.69302838e-03 -9.56541419e-01
-1.47347701e+00 8.84334207e-01 8.07083845e-01 -3.16051006e-01
-1.94750234e-01 2.67710447e-01 3.78965884e-01 4.49547451e-03
1.71017066e-01 4.78188038e-01 -8.44490647e-01 1.65152103e-01
-5.49758196e-01 3.84175807e-01 7.66512513e-01 1.31238985e+00
9.18591261e-01 -2.89527148e-01 1.98527694e-01 1.00271380e+00
4.70950782e-01 8.21177900e-01 6.59108013e-02 -6.78342700e-01
5.56265831e-01 4.65470284e-01 9.78855882e-04 -1.09809172e+00
-5.92172921e-01 -5.13006687e-01 -5.26218772e-01 -1.26798183e-01
5.02373815e-01 -1.02737390e-01 -8.95172119e-01 2.01666093e+00
7.32960403e-01 -6.43828139e-02 -1.82517305e-01 9.61892366e-01
9.59597766e-01 3.73217434e-01 3.75043154e-01 1.35583594e-01
1.58012533e+00 -8.10147345e-01 -2.31312856e-01 -6.63358092e-01
4.05638844e-01 -7.05672503e-01 9.32908237e-01 2.53684334e-02
-5.74949563e-01 -4.40938652e-01 -1.28035557e+00 8.71100277e-02
-1.39218345e-01 2.08155051e-01 9.77771759e-01 3.73192698e-01
-9.98485684e-02 5.24163365e-01 -1.00355911e+00 -6.20186567e-01
5.84492326e-01 3.37703347e-01 -6.69605017e-01 -1.47023365e-01
-1.07441187e+00 1.03582537e+00 4.77246910e-01 1.91566527e-01
-8.71917903e-01 -5.63440859e-01 -1.01206577e+00 -5.12684524e-01
5.73365748e-01 -4.91995960e-01 1.34420347e+00 -6.45560384e-01
-1.31132483e+00 9.96907473e-01 4.00929838e-01 -3.99098516e-01
5.77091873e-01 -4.75727916e-01 -2.51573175e-01 1.26202002e-01
2.67698020e-01 1.12560177e+00 7.98194408e-01 -1.34933543e+00
-5.04789770e-01 -6.89888358e-01 8.37573633e-02 1.54800743e-01
-1.26081541e-01 -2.10707068e-01 -5.91025889e-01 -8.60973537e-01
3.66902858e-01 -1.39075112e+00 -1.32586062e-01 2.97213644e-01
-1.54126436e-01 1.48125485e-01 7.12575436e-01 -5.78080833e-01
9.17587042e-01 -1.97354138e+00 8.33766088e-02 1.53033674e-01
-1.04013972e-01 -1.92340031e-01 -2.19906509e-01 3.58278006e-01
-1.99035816e-02 -3.86902243e-01 -4.19502914e-01 -2.39941161e-02
-1.02688611e-01 5.49684703e-01 -9.95532423e-02 7.56694973e-01
2.89816141e-01 8.14721763e-01 -8.08773041e-01 -8.88627112e-01
2.36740685e-03 2.30675340e-01 -9.79508460e-01 2.67050236e-01
-1.82117671e-01 8.52758825e-01 -8.22732866e-01 9.84530389e-01
9.23578382e-01 -6.97358102e-02 2.48773307e-01 -5.00350356e-01
3.35476339e-01 -9.74280536e-02 -1.11855292e+00 1.90978575e+00
-4.31808621e-01 3.08187716e-02 -5.50394207e-02 -8.47907960e-01
7.73489535e-01 -9.85460654e-02 6.63422585e-01 -2.67858326e-01
3.23680013e-01 2.74097979e-01 1.29249781e-01 -4.55088764e-01
1.53337792e-01 -2.60160387e-01 -4.17516559e-01 1.47258133e-01
2.07755387e-01 -4.37175065e-01 2.42710128e-01 -2.34599948e-01
8.54694963e-01 7.53161728e-01 5.45403123e-01 -2.85383046e-01
2.71144837e-01 1.65695921e-01 8.70261312e-01 2.78963268e-01
-3.90306264e-01 5.90448320e-01 2.05880627e-01 -4.01904702e-01
-9.82494831e-01 -1.18829715e+00 -4.35464472e-01 1.47342908e+00
4.51728523e-01 -2.60896206e-01 -7.82455266e-01 -7.83655465e-01
2.68432647e-01 7.76607767e-02 -6.71091318e-01 -2.44957745e-01
-8.51024210e-01 -6.23929203e-01 4.35984641e-01 9.09572780e-01
6.36207521e-01 -1.07269228e+00 -7.84793735e-01 1.24451235e-01
-1.96598962e-01 -1.29674637e+00 -5.99623859e-01 3.13903719e-01
-9.00763333e-01 -1.20606101e+00 -5.08926392e-01 -9.37047064e-01
6.83855712e-01 8.82826671e-02 1.01688468e+00 -2.57682979e-01
-4.20984328e-01 2.80380677e-02 -4.73557264e-01 -5.78652740e-01
-2.14379176e-01 2.25033000e-01 2.72919863e-01 -3.75646770e-01
2.67362148e-01 -6.01433933e-01 -7.65710354e-01 7.57861197e-01
-7.44410276e-01 -6.05190843e-02 9.12558556e-01 7.08387792e-01
6.45635784e-01 -2.03950703e-01 6.18917227e-01 -9.40567732e-01
-1.75287813e-01 -4.80445117e-01 -6.12376809e-01 1.78743042e-02
-6.84019737e-03 8.76491964e-02 4.21829373e-01 -7.64169455e-01
-8.27058434e-01 5.16754329e-01 -2.26264875e-02 -4.93807904e-02
-2.54103452e-01 3.67412359e-01 -4.38096762e-01 -1.68359086e-01
6.43191874e-01 -3.91911119e-01 2.88614761e-02 -8.12071919e-01
2.35932231e-01 4.12002683e-01 7.61425972e-01 -8.83272588e-01
1.11718106e+00 3.15347224e-01 2.00856015e-01 -4.96546000e-01
-9.15972114e-01 -4.11148846e-01 -8.01865935e-01 -5.13219535e-02
9.41226542e-01 -1.24117613e+00 -5.85117757e-01 4.48173761e-01
-8.71940672e-01 -6.05711378e-02 -6.22406676e-02 8.62333059e-01
-7.21557319e-01 3.03267002e-01 -4.89502579e-01 -2.48549685e-01
-9.25371423e-02 -1.17358601e+00 1.25974441e+00 -7.34147802e-02
-3.34614098e-01 -5.54529965e-01 -1.70906540e-02 5.10215700e-01
4.97396328e-02 5.33255696e-01 6.23598397e-01 -5.58967650e-01
-5.03461480e-01 -2.98220575e-01 -1.34003311e-01 2.15130121e-01
1.67529553e-01 -2.45003134e-01 -6.26640379e-01 -5.34526050e-01
-9.41791460e-02 -7.23159254e-01 5.14410436e-01 1.19459875e-01
1.16140401e+00 1.58536255e-01 -3.49740744e-01 6.33752286e-01
1.02848816e+00 -1.86469987e-01 1.77014351e-01 3.65464956e-01
7.46313334e-01 8.29279423e-01 1.09755731e+00 5.20160735e-01
4.25720751e-01 9.22041535e-01 3.77923459e-01 4.80581046e-04
-1.15456671e-01 -6.56780481e-01 1.76966950e-01 7.62061894e-01
-3.66344899e-02 -2.07574596e-03 -1.05853760e+00 3.09934974e-01
-1.63494408e+00 -6.09717488e-01 2.22838849e-01 2.08623528e+00
1.07620168e+00 -7.55911022e-02 2.43281141e-01 -2.50870764e-01
7.66430914e-01 -5.76355830e-02 -7.39164889e-01 4.40209001e-01
3.20310235e-01 1.67223588e-02 7.78277457e-01 -2.82380193e-01
-1.50178981e+00 7.95269251e-01 6.89688158e+00 7.69137442e-01
-9.80062544e-01 -2.02649802e-01 1.25757337e-01 1.03395455e-01
2.00398058e-01 -1.39859036e-01 -8.32703471e-01 4.40529138e-01
4.87103224e-01 1.13887765e-01 2.15826370e-02 1.19882262e+00
-1.61317915e-01 -1.87642239e-02 -1.46368194e+00 9.90087152e-01
-8.72432888e-02 -5.48602939e-01 -1.86721846e-01 -1.72820669e-02
6.47893846e-01 -1.13664135e-01 -8.60525519e-02 4.52686995e-01
3.01258177e-01 -8.68593693e-01 8.80312920e-01 -4.22106236e-02
7.36313999e-01 -8.26443911e-01 7.45839834e-01 4.96725082e-01
-1.37239468e+00 9.93297771e-02 -4.16339070e-01 1.39195636e-01
3.89484130e-02 1.15843989e-01 -6.66800559e-01 4.03537124e-01
8.42877448e-01 8.24621499e-01 -8.85095298e-01 1.07773542e+00
-1.73910379e-01 3.55162144e-01 -6.15805924e-01 2.58592546e-01
-1.02875955e-01 -5.02260663e-02 2.38362521e-01 8.24350893e-01
2.06855372e-01 2.51880870e-03 6.17726386e-01 5.50139666e-01
-2.80409962e-01 1.76533848e-01 -3.87510806e-01 7.37192333e-02
4.74098325e-01 1.23640740e+00 -8.79957974e-01 1.10163420e-01
-4.64420676e-01 6.89618289e-01 2.21403673e-01 -7.34841824e-02
-1.34607518e+00 -7.49499649e-02 6.35490775e-01 1.89584076e-01
4.60004389e-01 -1.92164257e-01 4.57282178e-02 -1.18098843e+00
4.19773832e-02 -1.03758979e+00 3.22420329e-01 -4.10367072e-01
-1.58994269e+00 2.73652315e-01 5.53314567e-01 -1.75746870e+00
-2.26537600e-01 -6.17310643e-01 7.85446465e-02 2.81498104e-01
-1.25298047e+00 -1.63780928e+00 -4.49450523e-01 3.69973809e-01
4.68194574e-01 -4.28391667e-03 5.83688259e-01 4.33262408e-01
-2.91143358e-01 6.94382429e-01 -1.64713398e-01 2.31150717e-01
1.04287708e+00 -7.93525338e-01 1.74115017e-01 5.07034242e-01
-6.03854954e-02 7.96806514e-01 7.55888522e-01 -7.63670623e-01
-1.59475875e+00 -1.19352734e+00 1.26383137e-02 -5.79189003e-01
6.59395993e-01 -3.92289668e-01 -7.08708644e-01 8.08629930e-01
-4.48115945e-01 3.65602612e-01 7.86296487e-01 2.17464995e-02
-3.55348617e-01 -5.82714789e-02 -1.15892613e+00 4.05305207e-01
1.51228273e+00 -3.22412223e-01 -7.62077749e-01 2.62925953e-01
5.40848136e-01 -7.13086307e-01 -9.90221918e-01 1.12703729e+00
7.64910519e-01 -4.21748281e-01 1.16924739e+00 -6.29231632e-01
6.25281870e-01 -5.17950118e-01 -3.98308456e-01 -1.10054803e+00
-1.06673323e-01 2.70276159e-01 4.98747267e-02 1.19865561e+00
3.09533596e-01 -3.50547433e-01 8.55207980e-01 3.32171023e-01
5.51285632e-02 -6.73799634e-01 -5.91690242e-01 -9.65171993e-01
1.38543904e-01 -9.76571217e-02 7.11135149e-01 8.23507547e-01
-5.16950428e-01 3.82977754e-01 -5.43306708e-01 2.54246444e-01
7.52212465e-01 6.26825929e-01 1.21697128e+00 -1.32806778e+00
-4.00903434e-01 1.59822449e-01 -7.77097225e-01 -1.21240354e+00
2.28341058e-01 -7.65291154e-01 4.41816211e-01 -1.07296348e+00
4.76373196e-01 -5.33646166e-01 -1.11682825e-01 6.44131660e-01
-2.11391017e-01 6.26935601e-01 6.79155672e-03 3.11709344e-01
-5.71101069e-01 4.69682634e-01 1.50821304e+00 -8.94069090e-04
1.43186286e-01 1.34529397e-01 -5.61204970e-01 1.10004544e+00
4.81827408e-01 -5.40728271e-01 -4.16835099e-01 -3.70270729e-01
-4.17887494e-02 -8.60893354e-02 3.70854050e-01 -1.13734877e+00
-1.22504972e-01 -4.05898899e-01 6.33605480e-01 -7.82630742e-01
3.27388704e-01 -1.17996597e+00 3.43397260e-01 5.14567196e-01
-1.45648867e-01 -3.34697574e-01 1.87483400e-01 8.78290176e-01
-1.69572219e-01 2.42083743e-02 1.14739382e+00 -1.92570627e-01
-9.60604727e-01 6.21808171e-01 1.73093781e-01 2.45763570e-01
1.28987825e+00 -1.44628242e-01 9.69160423e-02 -8.40199739e-02
-5.10987580e-01 3.42417181e-01 7.76087582e-01 6.10160768e-01
2.09076241e-01 -1.48645830e+00 -5.81458926e-01 2.20130071e-01
7.26405740e-01 2.70128042e-01 9.84170660e-02 6.40834272e-01
-8.56440604e-01 6.22380003e-02 -6.12451255e-01 -8.23268294e-01
-1.29095078e+00 6.59864783e-01 -1.77294854e-02 -1.01947458e-02
-4.67302054e-01 7.98350096e-01 5.18248141e-01 -8.91498327e-01
1.06864519e-01 -4.76777375e-01 -1.53067201e-01 1.03208184e-01
1.21729314e-01 1.95711553e-01 -9.49112400e-02 -8.74293268e-01
-5.60976028e-01 9.07163620e-01 -1.82048697e-02 3.08489144e-01
1.60071492e+00 -1.16653547e-01 1.02879807e-01 3.80140513e-01
1.33782554e+00 8.34566653e-02 -1.41902220e+00 -3.21021736e-01
2.84716655e-02 -5.55479705e-01 -4.70149785e-01 -4.62975055e-01
-9.29620266e-01 6.58502400e-01 5.28133154e-01 -4.94214118e-01
8.69296551e-01 2.80721188e-01 9.80252624e-01 6.20927513e-01
9.96658921e-01 -1.15677464e+00 2.51126885e-01 3.36773872e-01
8.78627896e-01 -1.70423949e+00 4.65305120e-01 -7.76627839e-01
-7.74154603e-01 6.96616530e-01 9.96852756e-01 -5.96763194e-02
4.14341420e-01 1.86508179e-01 3.60744558e-02 -1.61869064e-01
-2.62526929e-01 -1.89891636e-01 3.68071824e-01 5.99295080e-01
5.13074636e-01 4.97262403e-02 -2.77376115e-01 5.57382584e-01
-5.18778622e-01 -6.77449256e-02 -3.21930163e-02 1.23347270e+00
-4.78835106e-01 -1.21557760e+00 -4.95444894e-01 2.61953622e-01
-3.76630247e-01 3.13138694e-01 -2.54102498e-01 1.00457335e+00
3.75107318e-01 5.46454549e-01 -3.13098609e-01 -1.01869591e-01
4.89923000e-01 -1.83530405e-01 8.59608769e-01 -8.44449997e-01
-1.36818454e-01 -7.54114240e-02 -4.56003211e-02 -5.26380002e-01
-5.29484987e-01 -4.31455165e-01 -1.35071635e+00 -3.23590904e-01
-5.04527152e-01 -1.01948112e-01 5.93174934e-01 6.95003152e-01
-7.34719634e-02 2.90257961e-01 5.33381343e-01 -1.10365331e+00
-7.86384702e-01 -1.01886511e+00 -7.52255499e-01 7.78923392e-01
2.82554701e-02 -1.19927096e+00 -1.21057518e-01 2.90165305e-01] | [7.588891983032227, -0.9202432036399841] |
356784fe-242b-418f-a525-e476cbbed9c9 | metrabs-metric-scale-truncation-robust | 2007.07227 | null | https://arxiv.org/abs/2007.07227v2 | https://arxiv.org/pdf/2007.07227v2.pdf | MeTRAbs: Metric-Scale Truncation-Robust Heatmaps for Absolute 3D Human Pose Estimation | Heatmap representations have formed the basis of human pose estimation systems for many years, and their extension to 3D has been a fruitful line of recent research. This includes 2.5D volumetric heatmaps, whose X and Y axes correspond to image space and Z to metric depth around the subject. To obtain metric-scale predictions, 2.5D methods need a separate post-processing step to resolve scale ambiguity. Further, they cannot localize body joints outside the image boundaries, leading to incomplete estimates for truncated images. To address these limitations, we propose metric-scale truncation-robust (MeTRo) volumetric heatmaps, whose dimensions are all defined in metric 3D space, instead of being aligned with image space. This reinterpretation of heatmap dimensions allows us to directly estimate complete, metric-scale poses without test-time knowledge of distance or relying on anthropometric heuristics, such as bone lengths. To further demonstrate the utility our representation, we present a differentiable combination of our 3D metric-scale heatmaps with 2D image-space ones to estimate absolute 3D pose (our MeTRAbs architecture). We find that supervision via absolute pose loss is crucial for accurate non-root-relative localization. Using a ResNet-50 backbone without further learned layers, we obtain state-of-the-art results on Human3.6M, MPI-INF-3DHP and MuPoTS-3D. Our code will be made publicly available to facilitate further research. | ['István Sárándi', 'Timm Linder', 'Bastian Leibe', 'Kai O. Arras'] | 2020-07-12 | null | null | null | null | ['3d-absolute-human-pose-estimation'] | ['computer-vision'] | [-9.38255191e-02 4.37722951e-01 -2.08375275e-01 -5.38321316e-01
-7.67141640e-01 -4.93619412e-01 2.77038485e-01 -6.27125055e-02
-3.86304915e-01 4.55783337e-01 2.50647515e-01 1.04747750e-01
-2.86874212e-02 -6.06649280e-01 -8.11188936e-01 -2.46250704e-01
-4.18048799e-01 9.60067570e-01 1.02931961e-01 -3.29777271e-01
-9.49791670e-02 3.65056545e-01 -1.21843135e+00 -2.68290728e-01
4.66789335e-01 1.09810030e+00 -3.35377604e-01 5.64494193e-01
3.94304067e-01 -6.19955510e-02 -5.05906105e-01 -4.45509464e-01
4.45671082e-01 -2.89481252e-01 -7.43971288e-01 1.21151023e-01
8.42341185e-01 -4.28701311e-01 -3.10821563e-01 6.15897596e-01
7.07895279e-01 -1.47430643e-01 6.90034628e-01 -1.33720040e+00
-4.66385096e-01 2.39668369e-01 -8.01037967e-01 -2.84906179e-01
5.25061429e-01 2.35001128e-02 7.62166619e-01 -9.32200670e-01
7.23293781e-01 1.38442767e+00 1.16308534e+00 4.93602544e-01
-1.23173678e+00 -5.93644917e-01 -1.28384262e-01 -4.09873426e-02
-1.62389755e+00 -7.94003978e-02 9.51005518e-01 -4.34779108e-01
7.55302310e-01 2.70003676e-01 1.00096929e+00 1.07104564e+00
5.40332496e-01 6.42390430e-01 1.04117227e+00 -2.54635334e-01
2.03674942e-01 -4.26720172e-01 -2.86425292e-01 1.02571797e+00
2.42658317e-01 2.64192820e-02 -5.54403305e-01 1.03293613e-01
1.15192640e+00 -1.27681583e-01 -4.57452089e-02 -1.13010252e+00
-1.33708870e+00 8.51138890e-01 9.87465501e-01 -4.97727580e-02
-1.99847713e-01 4.72052634e-01 2.64968812e-01 -1.28769241e-02
7.26798832e-01 4.50739652e-01 -5.53718925e-01 -2.42292047e-01
-1.01371002e+00 5.87680697e-01 4.69868213e-01 9.40764189e-01
6.42877460e-01 -3.05492789e-01 5.15191406e-02 7.14441776e-01
2.63007492e-01 7.30580866e-01 3.53024542e-01 -1.21589077e+00
6.08837903e-01 7.63360202e-01 -4.37555201e-02 -1.30462980e+00
-1.10604048e+00 -4.32539701e-01 -8.26991379e-01 2.05252931e-01
7.29874969e-01 -4.96973656e-02 -1.12146223e+00 1.82823026e+00
5.92114091e-01 -2.53027856e-01 -5.24358928e-01 1.43728292e+00
4.46016520e-01 1.16501234e-01 -1.98218554e-01 4.21311378e-01
1.49767709e+00 -7.33090222e-01 -6.00469224e-02 -2.94992834e-01
6.33808911e-01 -3.94124240e-01 1.14135516e+00 2.13286296e-01
-1.20146120e+00 -5.51099002e-01 -1.17222357e+00 -4.05406356e-01
-3.57410669e-01 -4.95628687e-03 7.20334709e-01 8.20215344e-01
-1.06213868e+00 6.58933461e-01 -1.20062673e+00 -5.19986629e-01
2.51008749e-01 3.87338787e-01 -5.80453873e-01 1.82448983e-01
-1.15564108e+00 1.00780714e+00 -5.94220608e-02 2.74797171e-01
-6.19979739e-01 -6.41800582e-01 -1.15830588e+00 -3.08501720e-01
2.56930590e-01 -9.65566337e-01 1.05183554e+00 -2.13236377e-01
-1.28742063e+00 1.21244812e+00 2.84149021e-01 -3.29134732e-01
9.98292029e-01 -5.21567822e-01 1.14171892e-01 3.03794265e-01
1.71288967e-01 1.06143475e+00 7.50803411e-01 -9.77948129e-01
-9.13913473e-02 -9.60309982e-01 3.67103815e-02 4.39239264e-01
-7.08650500e-02 -3.87791127e-01 -7.38714397e-01 -7.13919401e-01
5.99318147e-01 -1.29376602e+00 -2.31896043e-01 6.24018550e-01
-6.31342828e-01 7.78300166e-02 3.45940799e-01 -7.94577837e-01
8.67036998e-01 -1.78407264e+00 6.84176862e-01 3.94777924e-01
3.67600739e-01 -3.37463945e-01 -2.54977904e-02 7.14724464e-03
1.46777526e-01 4.10431139e-02 -4.07990098e-01 -6.31055355e-01
1.12904221e-01 9.59591642e-02 2.71433353e-01 8.81016970e-01
1.13561131e-01 1.04390812e+00 -7.22599983e-01 -7.15609610e-01
3.88484925e-01 7.43596196e-01 -7.09796131e-01 -7.17595816e-02
7.14794174e-02 5.17918646e-01 -3.33812505e-01 8.58741283e-01
6.05399668e-01 -1.56067684e-01 -7.30999047e-03 -4.02725309e-01
2.04913467e-01 -4.34035286e-02 -1.05622482e+00 2.37533450e+00
-4.40016896e-01 2.95005202e-01 1.82197038e-02 -7.17406929e-01
8.19145024e-01 6.90129492e-03 8.98191929e-01 -6.84670985e-01
2.79973835e-01 7.22828284e-02 -2.64622331e-01 1.93741303e-02
4.76002783e-01 6.88890880e-03 -4.59004343e-01 2.60537386e-01
3.66264842e-02 -3.94904226e-01 -2.54907664e-02 -5.76273762e-02
8.75156820e-01 5.48398197e-01 7.77554139e-02 -1.03743650e-01
1.97726846e-01 1.28893629e-02 4.29445624e-01 3.11192244e-01
-3.07694227e-01 1.15047801e+00 4.84160572e-01 -6.16111934e-01
-1.25652850e+00 -1.49763548e+00 -3.79929602e-01 8.29469562e-01
2.29947045e-01 -5.77823043e-01 -9.87931907e-01 -7.09130764e-01
3.94835234e-01 1.77585214e-01 -1.02305377e+00 -2.40836844e-01
-9.06716526e-01 -6.18854821e-01 6.84777021e-01 7.84853935e-01
3.44417214e-01 -5.42508006e-01 -1.02194715e+00 6.62123561e-02
-2.87550062e-01 -7.82672584e-01 -5.80384076e-01 1.39701366e-01
-8.47972274e-01 -1.22734797e+00 -1.25424409e+00 -4.76393044e-01
6.57423437e-01 -2.33410329e-01 1.19046235e+00 -1.53195322e-01
-5.16361833e-01 4.66041923e-01 -2.98540831e-01 -1.15188800e-01
6.66429801e-03 2.83570200e-01 3.81460577e-01 -4.26812887e-01
2.12692209e-02 -7.20920265e-01 -1.00599086e+00 7.29889572e-01
-3.45028818e-01 2.00197309e-01 5.44329762e-01 5.49122572e-01
8.66502941e-01 -5.21622181e-01 7.33334050e-02 -3.99919957e-01
2.34430566e-01 -1.11836381e-01 -4.11028743e-01 3.60854566e-02
-3.73084277e-01 9.46160853e-02 2.12745845e-01 -2.63020515e-01
-3.37337881e-01 3.25473070e-01 -1.98922887e-01 -5.28173745e-01
6.37795329e-02 2.79661894e-01 2.93258745e-02 -2.23733231e-01
8.22250128e-01 -9.27827284e-02 2.57776171e-01 -6.66918159e-01
5.77893734e-01 2.74023116e-01 7.56295502e-01 -7.76857555e-01
7.96147108e-01 4.79207397e-01 3.51261139e-01 -5.30905306e-01
-6.16021097e-01 -2.04882115e-01 -1.06110609e+00 -2.77654558e-01
1.12935436e+00 -9.02274370e-01 -7.99208581e-01 3.25836927e-01
-9.45488393e-01 -3.30765873e-01 -2.00653240e-01 4.55296129e-01
-9.92314160e-01 2.37865746e-01 -6.54624462e-01 -4.83252972e-01
-4.30400848e-01 -1.11504543e+00 1.79475498e+00 -2.24550098e-01
-6.07450962e-01 -6.44008934e-01 2.06926540e-01 4.81870741e-01
1.52187899e-01 7.41149306e-01 5.81821561e-01 -9.07405838e-02
-1.43405735e-01 -3.81229281e-01 -1.43627107e-01 5.21734245e-02
-1.85323656e-02 -4.69032496e-01 -7.31818974e-01 -4.55501795e-01
-3.39627355e-01 -5.38253307e-01 6.01804137e-01 6.33884430e-01
1.18830085e+00 -4.81513813e-02 -3.05166185e-01 9.71769810e-01
9.75105464e-01 -5.25308490e-01 4.78735775e-01 3.77686352e-01
8.73846471e-01 6.42967343e-01 5.85123241e-01 4.64737147e-01
7.49805212e-01 1.12010229e+00 4.96356905e-01 -2.87155002e-01
-3.73533577e-01 -5.04783869e-01 9.44281742e-02 7.32910335e-01
-3.31485063e-01 3.06097955e-01 -1.04453683e+00 2.43469566e-01
-1.50973129e+00 -4.33751911e-01 1.82517752e-01 2.29716802e+00
9.02233601e-01 3.05750847e-01 5.49631655e-01 4.45311740e-02
4.14978594e-01 1.60608724e-01 -7.50684440e-01 -6.69001564e-02
1.50190011e-01 1.46441489e-01 6.72493398e-01 4.63097215e-01
-1.26234877e+00 7.26010084e-01 6.04782677e+00 4.96347219e-01
-1.01920557e+00 2.21434962e-02 5.14538705e-01 -3.35035473e-01
9.04610679e-02 -2.99109429e-01 -5.90797007e-01 3.53096157e-01
6.55412793e-01 3.14892203e-01 2.25515559e-01 9.93913770e-01
-7.47433752e-02 -5.34893721e-02 -1.30029309e+00 1.17959881e+00
1.34801894e-01 -8.79539311e-01 -3.44839305e-01 1.90255105e-01
3.40863705e-01 4.64768410e-02 7.36217722e-02 3.14102679e-01
-1.92539040e-02 -1.03371811e+00 1.00813246e+00 3.54386955e-01
1.14671206e+00 -7.93155849e-01 4.08023238e-01 1.07423402e-01
-1.40438640e+00 3.46054107e-01 -4.80401486e-01 -2.79430095e-02
2.78396040e-01 4.90745932e-01 -9.24857974e-01 5.82325161e-01
8.89633894e-01 5.50460219e-01 -8.99901569e-01 8.62331271e-01
-1.23926423e-01 4.81195152e-02 -7.26240158e-01 1.18860446e-01
-1.35731086e-01 5.63170649e-02 3.51144284e-01 8.93466473e-01
3.32465142e-01 -3.13725084e-01 2.79200166e-01 8.24325144e-01
4.87003615e-03 9.24108841e-04 -3.09275061e-01 4.19859201e-01
2.01639608e-01 1.09627092e+00 -8.61441970e-01 -6.19069859e-03
-1.04866689e-02 1.15848053e+00 2.41047904e-01 5.05902506e-02
-1.07077479e+00 -3.46074462e-01 7.87274539e-01 4.01439011e-01
-3.66460271e-02 -6.45978153e-01 -4.95368510e-01 -1.13192749e+00
2.56690979e-01 -4.75809395e-01 2.68135309e-01 -7.68362522e-01
-1.10007989e+00 4.58584279e-01 3.62850189e-01 -1.21376896e+00
-5.25207639e-01 -5.92999697e-01 3.87217887e-02 6.06375337e-01
-8.68074536e-01 -1.40292513e+00 -3.06546599e-01 4.40214097e-01
2.54061639e-01 3.76905054e-01 8.51822913e-01 2.14204594e-01
-1.92097068e-01 9.21113849e-01 -3.59329253e-01 4.39364374e-01
7.58611560e-01 -1.39170384e+00 8.59449685e-01 3.77728969e-01
2.12160766e-01 4.29105848e-01 7.16083884e-01 -7.27256775e-01
-1.51092815e+00 -9.00848985e-01 6.59476340e-01 -9.78248596e-01
3.00947726e-01 -8.59154761e-01 -3.94109726e-01 7.18579412e-01
-5.10347307e-01 2.21772999e-01 2.95182109e-01 3.68686438e-01
-5.04990101e-01 -1.56841472e-01 -1.19924700e+00 5.24164081e-01
1.53719604e+00 -3.25415701e-01 -5.98270416e-01 3.36109698e-01
7.88413525e-01 -8.98853183e-01 -1.22280538e+00 5.58023155e-01
9.80032265e-01 -8.59892011e-01 1.34824824e+00 -2.78061152e-01
2.61504769e-01 -2.93614775e-01 -1.91945270e-01 -1.23020363e+00
-1.33243963e-01 -1.80788442e-01 -2.00274467e-01 6.54318273e-01
1.95498973e-01 -2.43562222e-01 1.27288151e+00 5.85305750e-01
-1.25515938e-01 -1.05968320e+00 -1.29388773e+00 -8.78318787e-01
2.82754809e-01 -5.16099751e-01 5.00196815e-01 6.27815008e-01
4.95703109e-02 5.01486696e-02 -5.08820534e-01 2.78916559e-03
8.76267374e-01 6.58393800e-02 9.68704224e-01 -1.25800729e+00
-1.88724622e-01 -4.25306171e-01 -8.98029923e-01 -1.17013121e+00
-1.21998169e-01 -7.40562916e-01 1.19918734e-01 -1.30812824e+00
2.97743678e-02 -4.19714630e-01 -6.59552738e-02 6.61981821e-01
2.08686233e-01 9.08629239e-01 2.53341109e-01 4.41748798e-02
-5.61186850e-01 7.22310543e-01 1.31213474e+00 -9.62745398e-02
-1.29555240e-01 -2.57113010e-01 -3.19980979e-01 6.78978860e-01
7.18704224e-01 -3.00873905e-01 -3.37788731e-01 -4.43832487e-01
2.67072290e-01 3.51159014e-02 6.04912639e-01 -1.35648584e+00
-8.05450454e-02 1.42138571e-01 1.03948677e+00 -7.32034206e-01
7.89128542e-01 -7.03381896e-01 3.22421677e-02 5.75312793e-01
-2.69797832e-01 3.48018140e-01 -1.18457615e-01 3.51995081e-01
1.41673088e-01 4.79422122e-01 6.02941751e-01 -1.40594468e-01
-4.10760969e-01 4.36890453e-01 3.02163512e-01 1.84963182e-01
8.82676303e-01 -3.55028778e-01 9.46675763e-02 -3.89806032e-01
-7.83701062e-01 3.00536871e-01 8.81163955e-01 5.98135412e-01
6.86253011e-01 -1.52474821e+00 -7.37057865e-01 2.20833629e-01
2.13251874e-01 2.67728448e-01 1.99552536e-01 1.00455427e+00
-8.51942062e-01 4.68762755e-01 -4.15112883e-01 -1.02885199e+00
-9.27196085e-01 4.32506561e-01 3.97207677e-01 -1.12306848e-01
-8.28690290e-01 7.92926311e-01 6.46201819e-02 -8.68263423e-01
3.76229078e-01 -6.68777168e-01 2.85374135e-01 -1.94399059e-01
3.34041148e-01 3.95199209e-01 1.89316586e-01 -8.47738385e-01
-8.00905108e-01 1.09826303e+00 3.26878339e-01 -2.30567276e-01
1.16103756e+00 -2.11954921e-01 2.73218364e-01 3.11815441e-01
1.45982099e+00 -2.59093493e-01 -1.44437528e+00 7.50387833e-02
-2.42877305e-01 -3.86683971e-01 -2.81263679e-01 -8.62465680e-01
-1.17137480e+00 9.28780079e-01 9.07466412e-01 -3.04992527e-01
8.79095793e-01 2.80837536e-01 8.82713318e-01 1.23439111e-01
7.04505920e-01 -1.10263932e+00 9.28478986e-02 2.76254594e-01
1.19221139e+00 -1.19107187e+00 2.78919458e-01 -3.13978702e-01
-3.39388043e-01 9.10111248e-01 6.56678259e-01 -2.51797736e-01
5.90435326e-01 -3.14920284e-02 6.15640692e-02 -3.07080656e-01
-1.11605860e-01 -4.34009545e-02 7.82268822e-01 7.66906917e-01
5.49757302e-01 3.19346488e-01 -2.71487892e-01 4.10030186e-01
-7.92879820e-01 -3.30526263e-01 -7.59678036e-02 8.26520622e-01
-2.43670419e-01 -9.98082399e-01 -6.47572935e-01 1.45256057e-01
-1.95312247e-01 3.48990113e-01 -5.04323661e-01 1.05344319e+00
1.20022796e-01 5.11011660e-01 1.38403177e-01 -6.41380072e-01
5.46723545e-01 -2.22443547e-02 9.00472522e-01 -3.46977174e-01
-1.14341773e-01 5.28880805e-02 1.71078797e-02 -1.07034945e+00
-2.06502408e-01 -4.82026756e-01 -1.39572763e+00 -3.88722122e-01
8.62317607e-02 -2.26037577e-01 9.85913515e-01 5.78695416e-01
3.78866106e-01 1.86025143e-01 2.17378333e-01 -1.51877797e+00
-5.67109883e-01 -9.30701077e-01 -5.08756340e-01 5.79658985e-01
1.83812097e-01 -1.08220792e+00 2.81856600e-02 -2.73512751e-01] | [6.979128837585449, -0.9774619340896606] |
1877b7e2-e69a-459c-8c15-52b4fff5f2fa | language-based-colorization-of-scene-sketches | null | null | https://sketchyscene.github.io/SketchySceneColorization/ | http://mo-haoran.com/files/SIGA19/SketchColorization_paper_SA2019.pdf | Language-based Colorization of Scene Sketches | Being natural, touchless, and fun-embracing, language-based inputs have been demonstrated effective for various tasks from image generation to literacy education for children. This paper for the first time presents a language-based system for interactive colorization of scene sketches, based on semantic comprehension. The proposed system is built upon deep neural networks trained on a large-scale repository of scene sketches and cartoonstyle color images with text descriptions. Given a scene sketch, our system allows users, via language-based instructions, to interactively localize and colorize specific foreground object instances to meet various colorization requirements in a progressive way. We demonstrate the effectiveness of our approach via comprehensive experimental results including alternative studies, comparison with the state-of-the-art methods, and generalization user studies. Given the unique characteristics of language-based inputs, we envision a combination of our interface with a traditional scribble-based interface for a practical multimodal colorization system, benefiting various applications. | ['Ruofei Du', 'Hongbo Fu', 'Changqing Zou', 'Haoran Mo', 'Chengying Gao'] | 2019-11-17 | null | null | null | transactions-on-graphics-2019-11 | ['sketch'] | ['computer-vision'] | [ 3.90524089e-01 -3.38620603e-01 1.85047507e-01 -3.66189510e-01
-2.57208377e-01 -7.22544789e-01 6.96303427e-01 2.57534329e-02
-3.00291181e-01 2.61812478e-01 -1.19877726e-01 -5.55575848e-01
1.65464640e-01 -9.13522840e-01 -5.94618917e-01 -2.31971651e-01
3.05223882e-01 2.85105139e-01 1.18705556e-01 -3.12602282e-01
7.52681851e-01 7.22486079e-01 -1.77808154e+00 6.70647919e-01
1.04388177e+00 7.13416219e-01 5.12524307e-01 8.11756372e-01
-1.12390387e+00 5.42749465e-01 -5.81828296e-01 -4.43715543e-01
-4.92744446e-02 -2.89683968e-01 -3.76861274e-01 -1.48892961e-02
8.40465128e-01 -5.14128625e-01 3.08753736e-02 9.86591280e-01
2.39704385e-01 8.96634161e-02 6.72157466e-01 -1.32252789e+00
-1.35863662e+00 4.93282914e-01 -5.75334072e-01 -6.28187120e-01
8.15214396e-01 1.47954702e-01 7.14968443e-01 -1.21700740e+00
5.32727361e-01 1.46352923e+00 2.51400143e-01 9.18057561e-01
-1.13458800e+00 -8.96503150e-01 5.56439221e-01 -1.13194548e-02
-1.12015891e+00 -2.41537485e-03 1.11947453e+00 -3.36934447e-01
6.41007960e-01 3.98146808e-01 1.09038699e+00 9.41339850e-01
-3.70079011e-01 1.16830647e+00 1.19476593e+00 -8.21675897e-01
2.61407375e-01 3.14245373e-01 -3.32872570e-02 1.11226225e+00
9.56884772e-02 -3.45514655e-01 -6.01379335e-01 1.44694418e-01
1.22913790e+00 2.23929703e-01 -2.01148987e-01 -5.83989799e-01
-1.16036165e+00 3.10304850e-01 4.97006088e-01 3.19956481e-01
-1.54182792e-01 5.28101265e-01 1.17988147e-01 4.52053733e-02
3.07129949e-01 5.98620176e-01 -1.88314050e-01 -1.10912189e-01
-8.90718639e-01 2.34006643e-01 7.37396419e-01 1.25820267e+00
5.33295512e-01 9.31924134e-02 -3.93538743e-01 8.40216875e-01
2.61460304e-01 6.78015828e-01 -7.13032112e-02 -7.67419100e-01
3.96367520e-01 8.42599988e-01 1.97394058e-01 -7.77095616e-01
-1.92830220e-01 3.04134160e-01 -7.42710590e-01 7.44707406e-01
5.50344467e-01 5.09273931e-02 -1.16993225e+00 1.64451683e+00
3.07239369e-02 -1.01994127e-01 -1.08219348e-01 9.91752326e-01
1.09488201e+00 7.27649331e-01 5.05248308e-01 4.71017033e-01
1.41872203e+00 -9.35171902e-01 -7.17770517e-01 6.32171556e-02
1.36532217e-01 -7.44797587e-01 2.06812739e+00 8.05756986e-01
-1.08050001e+00 -6.60271406e-01 -1.06683886e+00 -3.18252951e-01
-9.00009632e-01 5.93616486e-01 9.51181233e-01 8.21868896e-01
-1.17934537e+00 4.23669517e-01 -4.85748559e-01 -7.58892596e-01
4.45947051e-01 -2.27490161e-02 -1.98268279e-01 -8.29175413e-02
-7.70565033e-01 6.96458519e-01 1.83127463e-01 8.78826976e-02
-4.92876559e-01 -8.03991258e-01 -6.78960919e-01 1.58707917e-01
1.58337161e-01 -6.40644312e-01 1.11713529e+00 -1.17552996e+00
-1.92654860e+00 7.74355471e-01 6.71175867e-02 2.46060669e-01
6.26044810e-01 -5.92239618e-01 -5.46320900e-02 4.07053649e-01
-3.07631552e-01 1.07796371e+00 7.63959646e-01 -1.75925124e+00
-4.25924987e-01 7.37278163e-03 4.02752429e-01 3.60781223e-01
-5.81485987e-01 -3.93798836e-02 -8.65222812e-01 -7.79974103e-01
-1.07095623e-02 -6.02482736e-01 -8.73719901e-02 7.73976564e-01
-4.12801236e-01 -2.06654623e-01 9.17633533e-01 -4.83599991e-01
1.03639555e+00 -2.06647372e+00 5.83343543e-02 2.84948885e-01
1.41123757e-01 2.37130746e-01 -3.90836328e-01 7.43678272e-01
1.19142763e-01 2.92424083e-01 -1.70601219e-01 -4.11561310e-01
1.78696498e-01 -1.87070563e-01 -3.17264855e-01 -1.70889154e-01
3.28221202e-01 9.70589519e-01 -9.63504791e-01 -4.30179626e-01
7.56726682e-01 6.17305100e-01 -3.98613930e-01 5.30623257e-01
-5.14379740e-01 2.40528315e-01 -2.80786157e-01 8.08496118e-01
7.88837314e-01 5.23673743e-02 2.17651978e-01 -4.02746588e-01
-1.55220091e-01 -2.76594549e-01 -1.07488418e+00 2.20136666e+00
-9.37500715e-01 8.74310970e-01 -6.12553842e-02 -3.90972644e-01
1.01162577e+00 3.26626562e-02 -1.40251219e-01 -7.35881686e-01
-1.47578672e-01 -5.16909398e-02 -4.31521475e-01 -5.65573096e-01
7.87547231e-01 2.06643507e-01 -5.76213039e-02 7.21723378e-01
-1.64431378e-01 -5.74325383e-01 2.22280532e-01 4.97257024e-01
4.84692097e-01 8.49989831e-01 -2.39613112e-02 -2.35223204e-01
3.27848732e-01 -2.09372982e-01 -4.99481291e-01 8.80096257e-01
2.25511029e-01 7.84591734e-01 3.48750889e-01 -4.45752561e-01
-1.04928529e+00 -1.31427336e+00 3.76225621e-01 1.50096154e+00
4.04568315e-01 -2.35737443e-01 -9.16643918e-01 -3.60618681e-01
-1.32559255e-01 9.54384744e-01 -5.77385068e-01 4.16560948e-01
-4.89441216e-01 -6.99171647e-02 3.03014636e-01 6.58618689e-01
5.99128246e-01 -1.40364707e+00 -1.02743888e+00 -6.86919019e-02
4.67789859e-01 -8.13428581e-01 -4.01165843e-01 -2.55431354e-01
-6.63815141e-01 -8.72127354e-01 -1.16717923e+00 -9.28694248e-01
1.05674243e+00 5.07840931e-01 1.15854669e+00 4.84984428e-01
-5.00920236e-01 8.42948556e-01 -4.06863421e-01 -3.79711837e-01
-2.43573368e-01 -1.74876645e-01 -4.10868883e-01 2.25123595e-02
7.82836750e-02 -5.47492862e-01 -9.04506743e-01 -2.48699903e-01
-1.14853060e+00 9.07159328e-01 5.64954638e-01 5.46527863e-01
3.83745134e-02 -5.66212356e-01 3.58250618e-01 -1.06863236e+00
1.04305363e+00 1.31015822e-01 -5.63617945e-01 8.57623219e-01
-2.89407164e-01 1.76075339e-01 6.56984329e-01 -6.73571050e-01
-1.43518376e+00 1.43408775e-01 1.17860593e-01 -3.33943903e-01
-4.20839936e-01 1.38259068e-01 -1.14651643e-01 -2.06668139e-01
3.73777598e-01 2.55380839e-01 -3.41870934e-01 -4.91039872e-01
1.33904815e+00 6.55881643e-01 7.67679214e-01 -1.13579071e+00
8.24690759e-01 3.37739408e-01 -2.95966238e-01 -8.01383793e-01
-1.23204269e-01 -5.95205501e-02 -8.13741863e-01 -6.29839659e-01
9.68170762e-01 -5.36852598e-01 -9.72233832e-01 4.80107486e-01
-1.51302814e+00 -4.22523618e-01 8.17613155e-02 -6.15084358e-02
-2.82843500e-01 2.03100562e-01 -4.29764867e-01 -1.13704991e+00
-5.60863018e-01 -9.57395852e-01 1.17336142e+00 6.74033523e-01
-7.36749247e-02 -8.25721920e-01 -3.08825821e-01 -6.96149543e-02
5.06493032e-01 3.79131347e-01 1.44527078e+00 -1.04200423e-01
-8.68107140e-01 -2.05545500e-01 -7.69887626e-01 -2.06947014e-01
1.65721148e-01 5.18911481e-01 -1.00202692e+00 1.68476775e-02
-1.03363407e+00 -4.74788755e-01 6.33680999e-01 -9.16667953e-02
1.55350173e+00 9.93275736e-03 -2.36922994e-01 5.73109925e-01
1.66673458e+00 5.29153407e-01 4.97993588e-01 1.83842063e-01
9.55735683e-01 4.13678199e-01 2.77958959e-01 3.88375372e-01
3.77321839e-01 3.62824529e-01 1.70000836e-01 -6.90706015e-01
-3.20990860e-01 -6.15002215e-01 -1.05744608e-01 4.32687730e-01
-1.20786488e-01 -3.14926654e-01 -8.18201900e-01 3.62569988e-01
-1.52983129e+00 -5.99298894e-01 9.02292281e-02 2.08977866e+00
7.82088935e-01 -1.81783605e-02 -4.60777283e-02 -2.13805541e-01
5.48124790e-01 -2.98304390e-02 -4.21359420e-01 -7.88380206e-01
1.59738883e-01 5.93106091e-01 6.36895597e-02 3.02468807e-01
-8.67553890e-01 1.26381409e+00 6.08476830e+00 5.87333500e-01
-1.32377434e+00 -2.11771026e-01 4.78144348e-01 1.57500893e-01
-7.23108947e-01 -2.41313130e-01 -4.22840901e-02 -1.72332991e-02
4.37146565e-03 1.11739822e-01 7.79648840e-01 6.58821583e-01
8.06129277e-02 -2.96033591e-01 -1.39662576e+00 1.22120929e+00
1.39566749e-01 -1.48964703e+00 6.16863132e-01 -7.73690224e-01
6.13310635e-01 -6.89659059e-01 3.39514434e-01 1.74696460e-01
1.99521467e-01 -1.00591660e+00 1.07795787e+00 6.85822368e-01
1.33803236e+00 -5.99078000e-01 -1.22130312e-01 -3.90933305e-02
-1.29441702e+00 2.71641076e-01 5.51695228e-02 -1.90004930e-01
5.07022021e-03 -9.03541520e-02 -6.64890468e-01 3.25930804e-01
5.64144135e-01 4.26805079e-01 -8.16791236e-01 8.67643476e-01
-3.96374345e-01 2.45231032e-01 -1.22019418e-01 -7.78073251e-01
1.43768191e-01 -2.59474337e-01 -1.70336515e-02 1.68221235e+00
3.36690694e-01 2.94056594e-01 1.05541356e-01 1.36313415e+00
-1.60326362e-02 4.45046395e-01 -6.97043180e-01 -3.31853092e-01
5.17331243e-01 1.40594375e+00 -1.07752657e+00 -5.15463352e-01
-5.31183839e-01 1.42685306e+00 3.68388116e-01 8.29521298e-01
-6.07328951e-01 -9.56978500e-01 3.14595789e-01 -7.68753812e-02
-3.76520641e-02 -5.20996273e-01 -6.89682722e-01 -1.06583023e+00
-1.86583474e-01 -6.49222195e-01 -9.18991417e-02 -1.42475116e+00
-1.14944303e+00 5.84903479e-01 1.24762163e-01 -1.02902830e+00
2.30948716e-01 -9.88304257e-01 -1.07385027e+00 8.88350904e-01
-1.28438878e+00 -1.56760287e+00 -6.42517030e-01 3.50817144e-01
8.18944097e-01 -2.25408345e-01 9.14286077e-01 1.44180685e-01
-2.88527489e-01 5.18505216e-01 -2.43252113e-01 1.43921122e-01
5.13825476e-01 -1.57305229e+00 6.98490918e-01 7.31773317e-01
4.17485982e-01 8.76975060e-01 3.71961445e-01 -5.45557201e-01
-1.79482961e+00 -5.58447897e-01 2.14422151e-01 -1.93901777e-01
3.40448976e-01 -8.94767821e-01 -7.85321712e-01 1.04749188e-01
7.61474907e-01 -5.39889395e-01 4.99448508e-01 1.96176558e-03
-5.82284510e-01 -9.29692611e-02 -9.28117096e-01 1.18316603e+00
1.06422079e+00 -6.59771502e-01 -4.97003436e-01 1.58440694e-01
5.76214910e-01 -4.15200323e-01 -2.52624631e-01 2.56628674e-02
1.04976165e+00 -8.14810336e-01 1.04173207e+00 -5.78803718e-01
6.01848185e-01 -4.08038586e-01 9.09002870e-02 -1.06041515e+00
-1.59427464e-01 -5.86328089e-01 4.59732980e-01 1.23583400e+00
3.54596168e-01 -1.32684961e-01 6.91819370e-01 9.73437726e-01
7.52370954e-02 -4.74231690e-01 -4.57271934e-01 -8.39493126e-02
-4.24423069e-02 -5.01051545e-01 7.04345107e-01 6.95938051e-01
4.41060327e-02 4.46286723e-02 -2.27310598e-01 1.19421668e-01
4.80079949e-01 5.51812172e-01 1.00259280e+00 -9.95501280e-01
-5.55178486e-02 -8.59721661e-01 1.43173829e-01 -1.03411853e+00
-5.55490218e-02 -7.95791566e-01 6.55753464e-02 -1.89722598e+00
2.14814931e-01 -5.05748630e-01 -1.37857884e-01 4.70721871e-01
-3.20407450e-01 3.89239281e-01 8.77914786e-01 -2.87777722e-01
-6.44539595e-01 2.64474779e-01 1.34511650e+00 -3.21228713e-01
-4.50680524e-01 -4.86480892e-01 -6.11963034e-01 6.69039607e-01
6.09187543e-01 1.89821795e-01 -6.63630366e-01 -5.79405487e-01
2.50843197e-01 -2.60257311e-02 3.60719681e-01 -1.03020215e+00
2.28110924e-01 -4.98657227e-01 6.98355079e-01 -7.29242206e-01
3.29094827e-01 -8.49228263e-01 -3.33805606e-02 4.29914415e-01
-6.68138683e-01 1.45351395e-01 4.05206800e-01 3.02357763e-01
2.62884945e-01 -5.54528609e-02 4.47210729e-01 -2.68794924e-01
-1.12411642e+00 -6.79353476e-02 -3.57021838e-01 -3.67940217e-01
7.72249281e-01 -3.35097879e-01 -2.83606976e-01 -5.89937091e-01
-3.53244424e-01 -6.53598234e-02 5.36327779e-01 6.93955183e-01
1.16615903e+00 -1.39175987e+00 -3.76824617e-01 4.43529129e-01
3.89547288e-01 -2.19388649e-01 6.30946979e-02 -1.12257227e-01
-1.08741772e+00 2.10390791e-01 -4.93111819e-01 -4.66622025e-01
-1.13141227e+00 6.65403366e-01 2.00540293e-02 4.21921492e-01
-6.78973317e-01 8.64812493e-01 5.32276988e-01 -3.56108636e-01
5.00622988e-01 -8.17962408e-01 -1.12602577e-01 -3.30564797e-01
5.18502712e-01 2.77013443e-02 -4.24346149e-01 2.82972485e-01
-8.62190947e-02 8.70546043e-01 2.46256322e-01 -4.63084608e-01
8.61870408e-01 6.83155954e-02 -3.34981978e-02 5.20818174e-01
6.97067320e-01 -4.56143618e-02 -1.31463587e+00 9.31850076e-03
-1.51249424e-01 -6.31883264e-01 -3.40930581e-01 -1.30550897e+00
-9.06446457e-01 1.24905360e+00 7.21578479e-01 2.19244268e-02
1.27716291e+00 -2.79930979e-01 4.35793042e-01 6.42752647e-01
3.33256096e-01 -9.36148942e-01 4.83326793e-01 7.19290748e-02
1.15085077e+00 -1.16963816e+00 -7.45199397e-02 -2.83850521e-01
-7.04327583e-01 1.51656199e+00 9.82167363e-01 -7.50453472e-02
1.38951898e-01 2.66953558e-01 3.33554059e-01 -1.11038953e-01
-5.60283899e-01 -1.47339806e-01 5.61161041e-01 7.29763091e-01
6.97313488e-01 2.73335218e-01 -9.93535221e-02 3.10260475e-01
1.44596040e-01 3.50001343e-02 3.78864050e-01 7.89885581e-01
-4.30142045e-01 -1.15293705e+00 -3.60283315e-01 1.39719158e-01
1.86229035e-01 -4.55588073e-01 -5.79280436e-01 9.39603567e-01
1.95762403e-02 6.36826575e-01 1.06742032e-01 -6.90487102e-02
3.64377737e-01 2.01286629e-01 8.32966566e-01 -4.79307353e-01
-5.46819150e-01 -1.16823554e-01 -1.66290060e-01 -4.59583253e-01
-3.43374431e-01 -1.89812288e-01 -1.31953073e+00 -2.00683977e-02
1.80865034e-01 -2.41487801e-01 1.08167434e+00 5.59710085e-01
1.10207565e-01 4.59856719e-01 2.50046611e-01 -1.25940800e+00
1.89815864e-01 -8.29431891e-01 -2.61998653e-01 6.30516887e-01
1.04932144e-01 -4.63827968e-01 3.45156252e-01 1.49676442e-01] | [11.437841415405273, -0.7394251823425293] |
cadb6362-e340-422b-98c9-258f237ef06e | virtual-node-tuning-for-few-shot-node | 2306.06063 | null | https://arxiv.org/abs/2306.06063v1 | https://arxiv.org/pdf/2306.06063v1.pdf | Virtual Node Tuning for Few-shot Node Classification | Few-shot Node Classification (FSNC) is a challenge in graph representation learning where only a few labeled nodes per class are available for training. To tackle this issue, meta-learning has been proposed to transfer structural knowledge from base classes with abundant labels to target novel classes. However, existing solutions become ineffective or inapplicable when base classes have no or limited labeled nodes. To address this challenge, we propose an innovative method dubbed Virtual Node Tuning (VNT). Our approach utilizes a pretrained graph transformer as the encoder and injects virtual nodes as soft prompts in the embedding space, which can be optimized with few-shot labels in novel classes to modulate node embeddings for each specific FSNC task. A unique feature of VNT is that, by incorporating a Graph-based Pseudo Prompt Evolution (GPPE) module, VNT-GPPE can handle scenarios with sparse labels in base classes. Experimental results on four datasets demonstrate the superiority of the proposed approach in addressing FSNC with unlabeled or sparsely labeled base classes, outperforming existing state-of-the-art methods and even fully supervised baselines. | ['Huan Liu', 'Kaize Ding', 'Ruocheng Guo', 'Zhen Tan'] | 2023-06-09 | null | null | null | null | ['meta-learning', 'graph-representation-learning'] | ['methodology', 'methodology'] | [ 6.08952284e-01 5.90281367e-01 -6.88291848e-01 -2.76123852e-01
-3.45792860e-01 -4.46438849e-01 7.31613934e-01 2.49687359e-01
-5.97729422e-02 5.99634409e-01 1.39033556e-01 -7.52806067e-02
-1.54284276e-02 -1.06991088e+00 -5.44987857e-01 -7.98262656e-01
9.20913666e-02 3.96923661e-01 3.06670219e-01 -3.65487278e-01
-1.04844324e-01 3.52229804e-01 -1.50381470e+00 5.93477450e-02
8.92547607e-01 7.64795542e-01 2.12449625e-01 3.92738581e-01
-4.84544784e-01 7.71727741e-01 -4.37119067e-01 -6.50050163e-01
1.44934267e-01 -4.06302005e-01 -6.53285861e-01 -4.19847071e-02
3.68979484e-01 2.41360590e-02 -7.10149169e-01 1.02328885e+00
5.72678804e-01 3.57360244e-01 7.56624460e-01 -1.64298773e+00
-1.05704570e+00 9.07323241e-01 -5.91124713e-01 1.93511128e-01
-9.17835087e-02 9.96547192e-02 1.38625228e+00 -8.19496870e-01
9.62439001e-01 1.19511163e+00 6.11241281e-01 1.01759768e+00
-1.27782261e+00 -6.78624094e-01 3.54987890e-01 3.71605486e-01
-1.34316826e+00 -4.47408587e-01 1.16583550e+00 -2.44665444e-01
7.28674293e-01 -1.04338028e-01 5.36458731e-01 1.32302666e+00
-2.27907956e-01 7.24590898e-01 6.51091516e-01 -4.43680048e-01
2.69394934e-01 3.56069058e-01 1.52528867e-01 7.02680767e-01
2.03302711e-01 2.71786433e-02 -4.71856564e-01 -5.46222702e-02
3.68513077e-01 2.08616525e-01 -2.65899599e-01 -6.13766432e-01
-9.07934785e-01 1.03947806e+00 8.45337689e-01 4.31772441e-01
-3.48959953e-01 1.96233943e-01 6.35011613e-01 3.36248636e-01
7.17917204e-01 3.36434156e-01 -3.53419870e-01 1.84970886e-01
-5.86186230e-01 -2.90434211e-01 6.34266615e-01 1.03585100e+00
9.45381284e-01 3.97244871e-01 -6.41376495e-01 8.73551667e-01
3.65231968e-02 1.96097925e-01 5.21888375e-01 -5.37179589e-01
4.53851253e-01 8.52897406e-01 -5.64037383e-01 -9.69888270e-01
-2.69242376e-01 -1.08235133e+00 -8.57887387e-01 -2.20885381e-01
-3.24273765e-01 1.33193498e-02 -1.14196646e+00 1.98255253e+00
6.52832925e-01 7.91769385e-01 1.77536830e-01 4.82130826e-01
1.18526709e+00 8.33901525e-01 2.73982018e-01 -3.09549004e-01
1.00329161e+00 -1.33623755e+00 -7.06481755e-01 -2.76579648e-01
7.24065840e-01 -1.38686001e-01 1.12042046e+00 -2.36667469e-01
-5.59569716e-01 -5.35199881e-01 -9.64752614e-01 1.26682550e-01
-5.34589648e-01 -2.11315215e-01 5.80995858e-01 7.82274187e-01
-1.18834078e+00 5.60390532e-01 -5.53705454e-01 -3.27099234e-01
7.56073117e-01 2.63067722e-01 -2.56845295e-01 -3.83674681e-01
-1.29981005e+00 6.55999482e-01 5.66399872e-01 -1.42408967e-01
-1.21676505e+00 -1.00173748e+00 -1.08211017e+00 4.62054014e-01
5.67649424e-01 -4.72912580e-01 8.61325562e-01 -9.02627230e-01
-1.48563266e+00 6.53082192e-01 -8.31963196e-02 -1.87632665e-01
-5.90461902e-02 6.76590621e-01 -4.64808553e-01 3.14370722e-01
1.54372185e-01 8.35123003e-01 1.14685905e+00 -1.46639502e+00
-4.44075882e-01 -2.85585344e-01 1.69278860e-01 2.09843680e-01
-1.12166238e+00 -4.87403274e-01 -3.15575093e-01 -8.02500248e-01
-7.58697540e-02 -8.48346591e-01 -1.68469071e-01 -7.58436620e-02
-4.53613549e-01 -5.77631772e-01 1.06461239e+00 -2.79040545e-01
1.26764679e+00 -2.08136868e+00 4.96083856e-01 1.47339985e-01
4.29511279e-01 6.11895621e-01 -6.34975791e-01 5.42537451e-01
-7.69320801e-02 1.00909241e-01 -1.19810224e-01 -4.32394534e-01
8.34105164e-02 3.77919912e-01 -9.23648477e-02 1.44018814e-01
3.30756009e-01 1.12538135e+00 -1.33452868e+00 -4.75267440e-01
5.89761846e-02 6.67726219e-01 -4.76476133e-01 3.12388510e-01
-3.04661423e-01 4.19720173e-01 -3.79085928e-01 8.18560004e-01
5.63868344e-01 -6.22640491e-01 2.55700976e-01 -3.07159036e-01
4.93575454e-01 -3.16243656e-02 -8.14197719e-01 1.69022691e+00
-5.41760206e-01 2.61591822e-01 -3.97974737e-02 -1.41927052e+00
1.01984775e+00 2.55392879e-01 3.22209418e-01 -5.21453321e-01
2.10957915e-01 5.24256676e-02 -1.55118480e-01 -4.11426812e-01
2.75580496e-01 -1.45025656e-01 1.56217337e-01 4.11696404e-01
6.77310586e-01 1.28885582e-01 2.52728373e-01 5.96924484e-01
1.47614062e+00 -1.86182305e-01 1.78829342e-01 1.49465233e-01
5.18141210e-01 -2.63838261e-01 7.58417487e-01 5.61608136e-01
-3.91604334e-01 3.27486336e-01 4.93536890e-01 -9.75628570e-02
-7.40240991e-01 -8.58527362e-01 2.06938893e-01 1.56471670e+00
1.45098507e-01 -6.39737725e-01 -4.93419707e-01 -1.24266851e+00
-3.27301398e-02 8.93627763e-01 -8.96258116e-01 -7.83358514e-01
-3.43588650e-01 -6.26661897e-01 2.43643299e-01 5.39719880e-01
2.59327501e-01 -1.17255950e+00 -1.12944962e-02 1.93634406e-01
1.02591105e-01 -1.24789715e+00 -3.69191647e-01 2.50805259e-01
-8.78167808e-01 -9.86018419e-01 -7.05867887e-01 -1.03105402e+00
9.01756406e-01 5.56203306e-01 1.02435493e+00 3.59730154e-01
-2.75516331e-01 5.74458361e-01 -8.85941863e-01 1.95472920e-03
-4.93919313e-01 5.12588024e-01 -2.58994848e-01 4.22324389e-01
1.71380356e-01 -8.03351760e-01 -5.16119957e-01 -4.31309417e-02
-8.84647429e-01 -9.55942925e-03 4.60776150e-01 1.02825928e+00
4.33581710e-01 -7.51942769e-02 1.00846350e+00 -1.47365522e+00
4.20312643e-01 -9.31092501e-01 -2.68687844e-01 5.00666857e-01
-8.97998273e-01 1.92535166e-02 9.89463389e-01 -4.99911517e-01
-9.77878928e-01 -5.15189767e-02 1.14897527e-02 -7.55196869e-01
2.52614439e-01 5.83605528e-01 -5.04873842e-02 -3.45171213e-01
6.34863198e-01 1.62119806e-01 -6.66290671e-02 -2.52035320e-01
6.48257196e-01 4.98506367e-01 3.20084155e-01 -3.32878441e-01
1.10349154e+00 3.31397980e-01 9.42318365e-02 -5.55961549e-01
-9.90348816e-01 -3.62801611e-01 -6.11037970e-01 -3.71405542e-01
5.56199253e-01 -9.38676238e-01 -1.14424042e-01 1.88983083e-01
-7.78289020e-01 -4.11376625e-01 -6.35091484e-01 2.16890760e-02
-1.86864793e-01 1.34072259e-01 -4.27867502e-01 -5.54094017e-01
-5.33621192e-01 -1.06038010e+00 9.37949479e-01 4.01543468e-01
3.02772731e-01 -1.42372966e+00 4.26737368e-02 2.02370197e-01
6.65050328e-01 2.12758154e-01 1.16391075e+00 -8.49675179e-01
-4.55105156e-01 -3.34394485e-01 -2.95910478e-01 2.22125351e-01
3.31242979e-01 -1.89029053e-01 -1.10260546e+00 -7.17606246e-01
-4.81542140e-01 -7.36529708e-01 1.05304635e+00 4.59285714e-02
1.29533041e+00 -1.17415853e-01 -7.32516110e-01 6.61102116e-01
1.44528103e+00 -1.67567536e-01 2.16444418e-01 2.12334953e-02
1.07968831e+00 5.09996533e-01 5.37896931e-01 4.24979270e-01
6.26368761e-01 7.06881762e-01 7.12876379e-01 7.21083209e-02
-7.14964509e-01 -5.06281137e-01 1.66391686e-01 9.70374286e-01
2.79430211e-01 -6.23164058e-01 -7.83978701e-01 6.44369423e-01
-1.61621571e+00 -6.88555717e-01 3.62897485e-01 1.72522044e+00
7.70763576e-01 -1.92116946e-02 -3.11671942e-01 1.94114655e-01
1.09283090e+00 6.07217133e-01 -8.07370245e-01 2.72815358e-02
9.97680500e-02 4.01114285e-01 1.93279266e-01 2.35567197e-01
-9.64818716e-01 1.17378163e+00 4.75613976e+00 9.65112090e-01
-1.09705424e+00 6.13978982e-01 4.07519013e-01 2.29153797e-01
-6.53664172e-01 1.37628227e-01 -8.76484573e-01 3.58656496e-01
9.92113233e-01 -1.15989052e-01 3.79944593e-01 9.33127344e-01
-2.58330762e-01 6.30903482e-01 -1.01948512e+00 8.45566213e-01
4.36335623e-01 -1.52994251e+00 1.96456224e-01 -1.70283273e-01
1.11516774e+00 4.25478965e-02 1.36090338e-01 8.12061489e-01
3.73753041e-01 -7.46929169e-01 2.29327559e-01 8.85316283e-02
1.04424524e+00 -4.59837049e-01 4.44096684e-01 2.58083284e-01
-1.43877065e+00 -3.79801691e-01 -7.06643701e-01 3.36120337e-01
8.41583535e-02 4.29566294e-01 -9.92590725e-01 6.24119759e-01
3.63471121e-01 1.17136991e+00 -8.38058412e-01 8.79168451e-01
-4.02854264e-01 7.89651275e-01 4.23303396e-02 -8.44535008e-02
2.44254172e-01 7.68157020e-02 4.94452417e-01 7.75467336e-01
3.97054374e-01 1.63194925e-01 2.73409218e-01 6.45807624e-01
-6.04956806e-01 9.80882794e-02 -5.55421650e-01 -3.70296597e-01
8.94490421e-01 1.57437873e+00 -8.47208500e-01 -3.31462204e-01
-4.84240621e-01 9.72752631e-01 7.94142485e-01 3.94764602e-01
-6.21735752e-01 -4.25638825e-01 2.34037787e-01 -7.16620609e-02
5.29819489e-01 3.64292055e-01 1.25837967e-01 -1.25105643e+00
-2.99055666e-01 -6.24730170e-01 5.70075572e-01 -4.67542827e-01
-1.39917231e+00 6.98296845e-01 -1.20574266e-01 -1.31200552e+00
-3.25208306e-02 -3.17624211e-01 -8.51397753e-01 3.38191777e-01
-1.86205578e+00 -1.62271547e+00 -5.10156989e-01 6.17311656e-01
6.07224166e-01 -5.65359294e-01 8.78003955e-01 4.15337414e-01
-7.74318814e-01 1.05006540e+00 3.60975750e-02 -1.40025496e-01
5.54894209e-01 -1.15091515e+00 3.38685393e-01 6.79644108e-01
2.43782893e-01 1.67852372e-01 5.19345343e-01 -6.32096529e-01
-1.38412333e+00 -1.62830913e+00 7.37816930e-01 -1.08925933e-02
5.78286290e-01 -5.01868844e-01 -1.00258017e+00 7.31280267e-01
1.96796358e-01 6.72950745e-01 8.29261422e-01 2.41017379e-02
-5.41501105e-01 -3.03712875e-01 -1.06627512e+00 3.49166781e-01
1.36653054e+00 -5.22350430e-01 -2.92624503e-01 7.20217109e-01
1.24871397e+00 -1.39435783e-01 -7.99314380e-01 5.87287903e-01
2.41341591e-02 -4.89283592e-01 9.34841871e-01 -6.25607431e-01
2.45289162e-01 5.32985479e-02 2.15143673e-02 -1.80376613e+00
-4.60110903e-01 -3.46736878e-01 -5.02541780e-01 1.55639458e+00
4.65613782e-01 -6.03925288e-01 1.01814604e+00 1.75429031e-01
-5.43572128e-01 -8.15451860e-01 -7.68987536e-01 -7.09477484e-01
-2.23866343e-01 -3.73040363e-02 6.22509897e-01 1.25682974e+00
-1.62204042e-01 5.49214542e-01 -3.82869422e-01 9.39305276e-02
6.62411988e-01 9.81979370e-02 5.73401332e-01 -1.47747612e+00
-3.38742971e-01 -1.31869271e-01 -5.18946946e-01 -5.72401226e-01
8.55620682e-01 -1.67838931e+00 -1.60050273e-01 -1.63623512e+00
2.38613889e-01 -8.28710318e-01 -6.17868364e-01 8.18917453e-01
-3.58337879e-01 2.38709569e-01 2.00921908e-01 -1.79926232e-02
-6.91516280e-01 1.09766233e+00 1.31467879e+00 -5.03297627e-01
-9.24233049e-02 -3.68849598e-02 -6.98812127e-01 2.75585890e-01
6.42724276e-01 -7.28530943e-01 -9.26370203e-01 -3.79206598e-01
1.95736170e-01 -1.10606506e-01 1.39235243e-01 -8.73680532e-01
3.53189439e-01 6.45979634e-03 3.05981166e-03 -4.29839879e-01
3.91519606e-01 -8.39395106e-01 -6.34285584e-02 4.81815249e-01
-4.97750670e-01 -3.09471607e-01 -1.88522667e-01 1.01445484e+00
-6.44805748e-03 -6.03329659e-01 7.89663613e-01 9.07816961e-02
-1.02639008e+00 7.73890018e-01 2.40022883e-01 3.59637946e-01
1.27820361e+00 2.28053275e-02 -5.88135302e-01 -1.89020425e-01
-7.59481668e-01 4.09341216e-01 2.63142884e-01 5.64852715e-01
8.12734127e-01 -1.55399716e+00 -5.74341893e-01 3.05018187e-01
4.98814523e-01 -7.03534111e-02 4.97393161e-01 5.11869073e-01
-7.97769129e-02 1.33966813e-02 -1.34869039e-01 -4.55663979e-01
-9.15344119e-01 7.51888752e-01 7.37332180e-02 -3.14780861e-01
-5.89702129e-01 1.31250322e+00 -5.15579246e-02 -8.92044485e-01
4.02151644e-01 3.37847531e-01 -6.08365238e-01 3.35273385e-01
2.55321205e-01 2.75650263e-01 6.31749853e-02 -7.00514436e-01
-3.66983302e-02 4.37068075e-01 -3.12550455e-01 5.57402074e-01
1.48817706e+00 -1.70228660e-01 -3.90316993e-02 4.04187262e-01
1.38245618e+00 -4.65314269e-01 -1.22793257e+00 -6.57900751e-01
-2.03459382e-01 -3.48922074e-01 1.31090760e-01 -4.02389258e-01
-1.60052478e+00 9.38418627e-01 6.42964303e-01 1.07679300e-01
1.03336596e+00 1.74153477e-01 8.63804936e-01 3.29027534e-01
2.89642006e-01 -8.21342647e-01 4.45036709e-01 2.80417502e-01
5.53427458e-01 -1.33962178e+00 -2.88066208e-01 -7.34289646e-01
-6.22366011e-01 8.66891682e-01 9.04892206e-01 2.47555450e-02
8.33229065e-01 -2.80826032e-01 -2.57841498e-01 -3.09993207e-01
-1.00163245e+00 -2.16943145e-01 9.33363363e-02 7.58176982e-01
1.04496591e-01 -2.60167234e-02 -7.76247531e-02 5.13675511e-01
1.23182856e-01 -2.62962580e-01 5.17676950e-01 9.03072000e-01
-3.78069133e-01 -1.26154876e+00 2.72946715e-01 7.98851669e-01
1.65591091e-02 -1.67618230e-01 -1.23897716e-01 5.69094062e-01
-1.37272654e-02 8.42103243e-01 -1.20505236e-01 -6.42701685e-01
2.17434898e-01 2.04453513e-01 2.67761230e-01 -1.20782995e+00
-5.71896017e-01 -3.93179923e-01 -1.54188454e-01 -2.98202634e-01
-3.93175304e-01 -2.66346633e-01 -1.07125449e+00 9.61560905e-02
-5.44860780e-01 2.79952943e-01 4.25789356e-01 7.06879675e-01
4.34500307e-01 9.43561733e-01 9.15929496e-01 -8.03074241e-01
-8.02240729e-01 -8.51473510e-01 -6.75642252e-01 4.30169284e-01
2.32786745e-01 -9.08099532e-01 -5.75864911e-01 -2.40837112e-01] | [7.429309844970703, 6.171928405761719] |
9b6633a6-535a-45c6-9a08-d67b8e3bb023 | unsupervised-face-recognition-using-unlabeled | 2211.07371 | null | https://arxiv.org/abs/2211.07371v1 | https://arxiv.org/pdf/2211.07371v1.pdf | Unsupervised Face Recognition using Unlabeled Synthetic Data | Over the past years, the main research innovations in face recognition focused on training deep neural networks on large-scale identity-labeled datasets using variations of multi-class classification losses. However, many of these datasets are retreated by their creators due to increased privacy and ethical concerns. Very recently, privacy-friendly synthetic data has been proposed as an alternative to privacy-sensitive authentic data to comply with privacy regulations and to ensure the continuity of face recognition research. In this paper, we propose an unsupervised face recognition model based on unlabeled synthetic data (USynthFace). Our proposed USynthFace learns to maximize the similarity between two augmented images of the same synthetic instance. We enable this by a large set of geometric and color transformations in addition to GAN-based augmentation that contributes to the USynthFace model training. We also conduct numerous empirical studies on different components of our USynthFace. With the proposed set of augmentation operations, we proved the effectiveness of our USynthFace in achieving relatively high recognition accuracies using unlabeled synthetic data. | ['Naser Damer', 'Arjan Kuijper', 'Meiling Fang', 'Marcel Klemt', 'Fadi Boutros'] | 2022-11-14 | null | null | null | null | ['unsupervised-face-recognition'] | ['computer-vision'] | [ 4.88006026e-01 3.01083446e-01 7.25353928e-03 -1.12596738e+00
-4.83256370e-01 -3.59909654e-01 5.47148585e-01 -6.15194261e-01
-4.06829298e-01 1.00642300e+00 -6.48508370e-02 8.57523456e-02
1.71321422e-01 -9.11641300e-01 -7.50736415e-01 -6.81755364e-01
3.25626075e-01 2.49307036e-01 -6.41341090e-01 1.47698045e-01
3.24619934e-02 7.10079134e-01 -1.61239624e+00 2.59588003e-01
7.63322115e-01 1.22150111e+00 -4.83494073e-01 7.39446357e-02
-1.06561475e-01 4.74038422e-01 -5.44466317e-01 -1.00439227e+00
9.25519764e-01 -5.89759707e-01 -6.62851036e-01 2.33010158e-01
7.38652706e-01 -5.65326989e-01 -4.58616585e-01 9.98024523e-01
5.08077145e-01 1.56777967e-02 4.21593040e-01 -1.90198803e+00
-1.01653600e+00 4.35425222e-01 -5.88771343e-01 -4.03375030e-01
2.56773457e-03 2.69779586e-03 4.94811714e-01 -1.03567493e+00
6.10556066e-01 1.16285944e+00 6.81546688e-01 9.75827217e-01
-1.09113681e+00 -1.15477908e+00 -1.80898860e-01 8.17521438e-02
-1.60643792e+00 -8.56912792e-01 6.95618868e-01 -1.18742965e-01
3.37416828e-01 3.57158959e-01 2.46239066e-01 1.33447170e+00
-2.91895449e-01 5.83402336e-01 1.20809007e+00 -5.54144800e-01
2.34739140e-01 5.61697721e-01 -1.56767398e-01 5.32871246e-01
4.43479359e-01 2.08882719e-01 -5.11860609e-01 -2.56753832e-01
8.26490104e-01 1.90840334e-01 -2.62102395e-01 -5.07575750e-01
-6.42298162e-01 8.93963933e-01 1.87564701e-01 -1.56176323e-02
-1.33738831e-01 -1.19805060e-01 2.83229440e-01 4.82020855e-01
4.36820954e-01 3.95514488e-01 -3.33249480e-01 3.77458841e-01
-6.86191320e-01 -5.60840704e-02 7.19268560e-01 1.11073041e+00
8.37000728e-01 2.23760396e-01 -1.79196671e-01 9.86449242e-01
1.34238988e-01 4.27180469e-01 4.81402010e-01 -8.51681828e-01
2.19657108e-01 5.19217491e-01 -6.73994329e-03 -8.83217514e-01
1.35607645e-01 -4.06339735e-01 -1.02150118e+00 2.05208972e-01
3.67474914e-01 -3.37416172e-01 -9.43549931e-01 2.06854057e+00
2.95477003e-01 1.66999310e-01 1.50484189e-01 6.34062588e-01
5.41272998e-01 4.45428550e-01 1.82614759e-01 -1.01932362e-01
8.58791590e-01 -7.37263441e-01 -8.05521786e-01 6.61834255e-02
5.71020663e-01 -4.71690744e-01 9.57342327e-01 2.84100205e-01
-7.40776956e-01 -3.69284093e-01 -9.36743140e-01 1.13310881e-01
-4.23565477e-01 2.98523813e-01 6.76835537e-01 1.38128662e+00
-1.15993905e+00 4.84940529e-01 -5.40603161e-01 -3.49670589e-01
1.12593949e+00 5.68054616e-01 -8.47727120e-01 -3.68937463e-01
-1.04375577e+00 4.35821116e-01 3.11052650e-01 2.41275311e-01
-8.98690224e-01 -5.35076857e-01 -9.14842188e-01 3.51576954e-02
4.67999488e-01 -2.97926784e-01 8.93638790e-01 -1.36319494e+00
-1.35753572e+00 1.16788721e+00 1.97369009e-02 -3.59992087e-01
6.49146497e-01 -9.88928229e-02 -5.78357935e-01 -9.90161523e-02
-4.07892652e-02 7.30006099e-01 1.01002657e+00 -1.34248364e+00
-2.50334948e-01 -7.78466165e-01 -2.88694322e-01 -2.01306805e-01
-7.83875704e-01 -4.40357104e-02 -3.17744583e-01 -8.46229494e-01
-1.10686265e-01 -9.63641226e-01 8.36951062e-02 3.73349220e-01
-5.10564566e-01 8.08734819e-02 1.17603862e+00 -5.90250194e-01
5.80708563e-01 -2.24129415e+00 -2.80222863e-01 2.48335496e-01
5.22615574e-02 6.50568008e-01 -4.24472928e-01 2.53810823e-01
-1.98064059e-01 3.23477507e-01 -4.89943177e-01 -5.55765629e-01
-8.06408748e-02 1.01798654e-01 -4.52155769e-01 4.86412913e-01
4.52240586e-01 9.43938434e-01 -4.27055657e-01 -2.03381062e-01
-7.49739856e-02 5.72196901e-01 -4.12131548e-01 3.37814987e-01
1.63469426e-02 4.42200065e-01 -4.66602951e-01 8.42436194e-01
1.23499310e+00 -3.34310792e-02 1.38827354e-01 2.81549692e-02
4.02916819e-01 -5.03976583e-01 -7.81553328e-01 1.47605932e+00
-2.67613977e-01 5.71541786e-01 3.74643728e-02 -1.04628754e+00
1.11076808e+00 3.25561047e-01 3.94102365e-01 -6.00217164e-01
3.56068969e-01 6.25021607e-02 -3.32585543e-01 -2.09805652e-01
2.59525925e-01 -1.84587151e-01 3.32738340e-01 6.10707104e-01
1.05812900e-01 3.30516398e-01 -4.09840256e-01 1.43094236e-04
6.60709023e-01 1.77785665e-01 3.87630016e-02 -2.59463582e-02
6.64331138e-01 -3.59218478e-01 7.26162314e-01 5.73630810e-01
-3.35392088e-01 6.34261727e-01 1.44868955e-01 -4.14295465e-01
-9.72908676e-01 -8.03065538e-01 -2.11675569e-01 7.13940978e-01
-2.07979307e-01 -2.91341823e-02 -1.05273914e+00 -1.01272166e+00
8.39292705e-02 6.59663022e-01 -8.25571537e-01 -3.58650506e-01
-3.20478618e-01 -6.55924559e-01 9.86711025e-01 4.60191727e-01
1.09144640e+00 -1.03552556e+00 -1.17162190e-01 -2.25897655e-01
1.79707840e-01 -1.14209187e+00 -5.23518860e-01 -1.77950427e-01
-5.33328712e-01 -1.00840175e+00 -7.35108137e-01 -7.52775669e-01
1.10452890e+00 9.65323746e-02 5.69118202e-01 6.75103441e-02
-4.72459316e-01 2.78343230e-01 -2.74084836e-01 -5.80565453e-01
-3.16229016e-01 -1.31988302e-01 1.89246312e-01 6.43860877e-01
5.16796172e-01 -5.39824307e-01 -4.38735604e-01 4.61706072e-01
-1.13122582e+00 -2.48782143e-01 5.26901543e-01 9.83921409e-01
3.37248743e-01 -1.50841445e-01 9.24425125e-01 -1.17026162e+00
4.86636668e-01 -2.99489886e-01 -6.42472625e-01 4.29946154e-01
-7.21666634e-01 -1.59407146e-02 7.58862555e-01 -4.32376564e-01
-1.19437504e+00 2.29965001e-01 -1.40151247e-01 -7.21780121e-01
-2.24687636e-01 1.80106480e-02 -6.55969381e-01 -5.31760991e-01
4.21930313e-01 2.58445859e-01 3.78418177e-01 -2.44431630e-01
3.02147567e-01 9.43789482e-01 4.31460321e-01 -3.58208120e-01
8.83261263e-01 5.23599744e-01 5.30576184e-02 -7.28057504e-01
-7.35209584e-01 1.27791017e-01 -4.25819814e-01 3.92979309e-02
5.90479374e-01 -7.66292393e-01 -8.07187140e-01 9.69298363e-01
-7.55187511e-01 4.07388769e-02 -3.14928234e-01 3.94810498e-01
-2.87991285e-01 3.36257100e-01 -4.42016274e-01 -9.96628106e-01
-4.18430448e-01 -9.87058818e-01 8.06574881e-01 2.78701276e-01
1.24393627e-01 -5.76184928e-01 -2.45523348e-01 5.14485121e-01
5.33349693e-01 4.72781837e-01 6.15355790e-01 -9.65098798e-01
-5.51762640e-01 -4.07977581e-01 -3.06420058e-01 7.32421458e-01
5.40291250e-01 -1.94548815e-01 -1.39545941e+00 -6.13865376e-01
2.62127697e-01 -7.57093966e-01 5.98324239e-01 9.90021378e-02
1.58024037e+00 -6.94737136e-01 -2.05734774e-01 9.38511968e-01
1.32287514e+00 5.49383759e-01 8.84514809e-01 1.05759822e-01
6.63023889e-01 8.85675132e-01 5.01864016e-01 4.34385955e-01
-7.82098174e-02 4.38537270e-01 4.47001576e-01 -1.47259027e-01
2.93471515e-01 -3.57643276e-01 1.00482024e-01 1.25882462e-01
2.32028008e-01 -3.89316767e-01 -6.01986527e-01 2.68800050e-01
-1.38242567e+00 -9.18490648e-01 3.93505752e-01 2.43750548e+00
6.61956489e-01 -3.72134417e-01 -1.75406992e-01 -4.89384830e-02
9.37247992e-01 -8.77259597e-02 -8.63877058e-01 -1.70827672e-01
-2.33253151e-01 4.48676497e-01 5.03664553e-01 9.15835574e-02
-1.05661368e+00 9.02827322e-01 6.20869780e+00 6.66291118e-01
-1.30749226e+00 -5.76571142e-03 1.20243418e+00 1.20272659e-01
-2.98245162e-01 -2.79546738e-01 -7.44340897e-01 2.33445540e-01
6.48573041e-01 -2.47331291e-01 4.57269400e-01 1.05296195e+00
-1.58464938e-01 4.10236210e-01 -1.19027221e+00 1.19968104e+00
3.21181834e-01 -1.30160165e+00 3.32567453e-01 3.46458733e-01
6.89604878e-01 -4.74752516e-01 3.53812128e-01 2.19725817e-01
2.06713900e-01 -1.44887102e+00 3.44014972e-01 2.15153217e-01
1.22947812e+00 -8.71594191e-01 6.27968252e-01 8.41036811e-02
-6.80730045e-01 4.61338263e-04 -3.33770424e-01 2.61676997e-01
-3.49433035e-01 1.93806201e-01 -9.43347335e-01 5.64392090e-01
4.50009733e-01 4.15407062e-01 -4.78593528e-01 5.73696733e-01
-4.82622124e-02 5.15772939e-01 -1.82127953e-01 2.75129139e-01
-4.81507704e-02 -4.31956202e-01 7.32130408e-02 5.33749998e-01
4.79386091e-01 1.78318202e-01 -3.45911026e-01 1.13300323e+00
-7.65517950e-01 1.33461595e-01 -7.66802311e-01 -4.78167713e-01
6.70376062e-01 1.12205100e+00 -2.30666041e-01 -6.90157637e-02
-2.24520832e-01 1.12237668e+00 2.79747844e-01 2.89270878e-01
-6.90944016e-01 -3.58971864e-01 7.31616974e-01 1.44631505e-01
1.01741046e-01 1.45598546e-01 -2.50601172e-01 -1.15128469e+00
1.67874873e-01 -1.00846028e+00 3.47843736e-01 -4.50439870e-01
-1.34704328e+00 1.01003695e+00 -3.20365161e-01 -1.11677277e+00
-1.21386588e-01 -4.28628474e-01 -4.16039526e-01 9.79166687e-01
-1.38996935e+00 -1.26756227e+00 -4.01658893e-01 8.25659811e-01
1.82133630e-01 -6.45328879e-01 1.08050549e+00 3.42218459e-01
-8.05397987e-01 1.25568175e+00 2.80438423e-01 4.96661842e-01
8.45337689e-01 -6.40657425e-01 5.48928201e-01 7.95381725e-01
5.84584177e-02 8.64335895e-01 2.17810586e-01 -5.82501292e-01
-1.37058842e+00 -1.51103306e+00 5.27893424e-01 -1.35531723e-01
-9.96003821e-02 -5.77324390e-01 -9.59068358e-01 1.00681114e+00
9.15489197e-02 2.96158940e-01 1.02598369e+00 -2.50073880e-01
-6.59694374e-01 -4.29325491e-01 -1.89536369e+00 3.81763309e-01
1.04564893e+00 -6.01942658e-01 -1.55657604e-01 2.47968614e-01
4.74094421e-01 1.19611599e-01 -7.09388137e-01 5.05933762e-01
6.77179158e-01 -7.63314545e-01 7.30113149e-01 -8.30961406e-01
3.34165186e-01 -9.16845798e-02 -1.88776955e-01 -1.05955267e+00
1.21763006e-01 -6.28715992e-01 3.10775965e-01 1.49980044e+00
2.73202598e-01 -9.77915287e-01 1.31902170e+00 9.89149809e-01
2.00577572e-01 -5.08488655e-01 -7.47441232e-01 -7.75320351e-01
-3.73412967e-02 8.09950456e-02 9.64103520e-01 1.31610942e+00
-5.33752143e-01 -1.12483047e-01 -8.11741054e-01 1.29981160e-01
9.96046484e-01 -1.36877596e-01 8.59330535e-01 -1.15873384e+00
8.34348947e-02 1.08665474e-01 -4.10233468e-01 -4.62495893e-01
3.88715297e-01 -8.70010734e-01 -2.62325436e-01 -8.45745742e-01
1.76521510e-01 -5.83705664e-01 -2.67681777e-01 7.92457879e-01
5.96914664e-02 5.42440414e-01 -1.64733939e-02 7.59474784e-02
8.97637308e-02 9.16777492e-01 1.14898729e+00 -1.83905557e-01
3.44397575e-02 -3.94557975e-02 -9.12201047e-01 3.64637703e-01
8.30927432e-01 -4.55158204e-01 -7.92515278e-01 -3.84931177e-01
-4.50239897e-01 -3.65925021e-02 2.63346583e-01 -8.71510684e-01
-4.40980159e-02 -1.90493733e-01 6.19097650e-01 -1.39021844e-01
4.60661769e-01 -1.01025796e+00 5.70363581e-01 2.54512638e-01
-5.77336013e-01 -2.91729361e-01 2.53314197e-01 4.37933028e-01
-2.85506487e-01 9.72833857e-02 9.82455850e-01 -2.42446959e-02
-5.13365149e-01 7.14305103e-01 2.24913478e-01 -1.70546696e-01
1.39565277e+00 -4.41747814e-01 -2.89005309e-01 -5.19517720e-01
-5.92428267e-01 1.54152429e-02 6.23161554e-01 4.58114952e-01
8.23037207e-01 -1.49860251e+00 -7.23888576e-01 7.92999864e-01
1.78510174e-01 -2.62234241e-01 2.28589848e-01 1.34592041e-01
-3.24219167e-01 3.15916032e-01 -7.05377221e-01 -1.83452263e-01
-1.31251860e+00 6.19959891e-01 3.00589472e-01 1.15948424e-01
-2.63886541e-01 9.72848296e-01 3.62012267e-01 -6.80285513e-01
3.99319023e-01 2.52878726e-01 1.75483063e-01 -2.32105821e-01
6.18478358e-01 9.69119295e-02 3.42285149e-02 -6.92321002e-01
-3.12979400e-01 1.65235832e-01 -4.00748163e-01 -3.59897465e-02
1.30882597e+00 -3.16531770e-02 -1.12513557e-01 -2.30178684e-01
1.35313463e+00 -6.69943467e-02 -1.16350949e+00 -1.66136399e-01
-1.84624329e-01 -9.81932640e-01 -2.47967407e-01 -7.70132720e-01
-1.58513081e+00 6.23801351e-01 7.24795938e-01 -2.17254564e-01
1.30548835e+00 -5.68236351e-01 7.28936434e-01 4.33780819e-01
6.11067057e-01 -7.25470662e-01 -1.75768241e-01 -6.12555351e-03
8.02633643e-01 -1.50791192e+00 -1.41780198e-01 -6.79863036e-01
-6.77096307e-01 7.98194647e-01 8.18986773e-01 2.52715886e-01
4.99294072e-01 1.34199128e-01 2.50056922e-01 1.72096342e-01
-2.70546168e-01 3.77850801e-01 -2.93121114e-02 8.40021849e-01
1.87524751e-01 -1.00280538e-01 -1.72281310e-01 5.46271265e-01
-1.01187013e-01 2.86602169e-01 3.43047082e-01 1.09085274e+00
1.54377431e-01 -1.62984669e+00 -3.45124722e-01 5.69008768e-01
-3.87444258e-01 2.65759025e-02 -6.50223732e-01 7.99228668e-01
4.61251214e-02 8.60624671e-01 -1.27211422e-01 -5.80698490e-01
1.09133296e-01 1.82487994e-01 3.79054248e-01 -4.06909287e-01
-4.31859225e-01 -4.97247756e-01 -9.23991427e-02 -4.45380151e-01
-2.89690703e-01 -5.31367421e-01 -9.14840102e-01 -3.98956001e-01
-2.31020361e-01 2.55167335e-01 7.16008306e-01 6.80774987e-01
6.31070495e-01 4.84588556e-03 1.10777056e+00 -3.01557511e-01
-7.97059536e-01 -8.72141242e-01 -8.25145423e-01 5.84147990e-01
1.48847744e-01 -4.36784685e-01 -3.58162135e-01 3.64651047e-02] | [12.823098182678223, 0.6797406077384949] |
61de76af-827e-4f15-8c5c-c14aeaa3dbec | stochastic-unrolled-federated-learning | 2305.15371 | null | https://arxiv.org/abs/2305.15371v1 | https://arxiv.org/pdf/2305.15371v1.pdf | Stochastic Unrolled Federated Learning | Algorithm unrolling has emerged as a learning-based optimization paradigm that unfolds truncated iterative algorithms in trainable neural-network optimizers. We introduce Stochastic UnRolled Federated learning (SURF), a method that expands algorithm unrolling to a federated learning scenario. Our proposed method tackles two challenges of this expansion, namely the need to feed whole datasets to the unrolled optimizers to find a descent direction and the decentralized nature of federated learning. We circumvent the former challenge by feeding stochastic mini-batches to each unrolled layer and imposing descent constraints to mitigate the randomness induced by using mini-batches. We address the latter challenge by unfolding the distributed gradient descent (DGD) algorithm in a graph neural network (GNN)-based unrolled architecture, which preserves the decentralized nature of training in federated learning. We theoretically prove that our proposed unrolled optimizer converges to a near-optimal region infinitely often. Through extensive numerical experiments, we also demonstrate the effectiveness of the proposed framework in collaborative training of image classifiers. | ['Alejandro Ribeiro', 'Navid Naderializadeh', 'Samar Hadou'] | 2023-05-24 | null | null | null | null | ['unrolling'] | ['computer-vision'] | [-2.20642343e-01 3.69661063e-01 -9.71928462e-02 -3.53265673e-01
-6.52237415e-01 -8.00289869e-01 1.77329212e-01 -1.29052298e-02
-5.34152031e-01 6.29362881e-01 2.56360974e-03 -6.58120513e-01
-4.47923034e-01 -6.36711180e-01 -1.27418160e+00 -7.20896721e-01
-1.53114438e-01 2.94152886e-01 -4.35702473e-01 3.62574384e-02
-8.11081901e-02 5.59000432e-01 -1.30451286e+00 5.23619279e-02
1.06348264e+00 9.42967534e-01 3.01368330e-02 8.56109917e-01
-3.08995515e-01 1.03915262e+00 -2.87798405e-01 -5.89248538e-01
6.59269810e-01 -3.28499973e-01 -7.64319777e-01 1.70174360e-01
9.40487146e-01 -2.35321999e-01 -2.95382679e-01 1.14345562e+00
3.44590306e-01 2.67942458e-01 2.04268008e-01 -1.28808749e+00
-6.65273249e-01 1.00661933e+00 -3.04779828e-01 3.48388851e-02
-2.76419818e-01 -1.57572217e-02 9.66400623e-01 -9.71738160e-01
5.29355526e-01 8.55755210e-01 8.81532490e-01 6.64067328e-01
-1.08281040e+00 -3.52118701e-01 4.47344273e-01 -1.00402869e-01
-9.86976922e-01 -3.22144508e-01 7.61912584e-01 -4.14622605e-01
6.83537900e-01 4.31134462e-01 5.31936646e-01 5.57873189e-01
4.33500782e-02 8.65046740e-01 7.46838927e-01 -5.70261836e-01
6.23219311e-01 1.93018571e-01 3.31121683e-01 1.48369229e+00
5.21736801e-01 5.36087751e-02 -7.53530383e-01 -3.61118942e-01
4.91208404e-01 2.97261655e-01 -2.97112226e-01 -7.63961911e-01
-8.42041016e-01 9.59863007e-01 7.07660496e-01 3.75732370e-02
-2.65762955e-01 2.65725374e-01 4.75318372e-01 7.25412905e-01
6.69276774e-01 2.32882366e-01 -6.52758300e-01 3.42929870e-01
-9.92241144e-01 9.31276977e-02 1.19542384e+00 8.56269896e-01
1.17215884e+00 3.14211398e-01 -9.63993147e-02 5.66528380e-01
1.35218605e-01 2.53710091e-01 6.40579998e-01 -7.41513073e-01
7.62330115e-01 8.47927153e-01 1.56137776e-02 -8.69359553e-01
-2.79502481e-01 -8.15870821e-01 -9.55170512e-01 1.22165576e-01
4.17753994e-01 -8.39895010e-01 -5.45024157e-01 1.79567015e+00
8.28700423e-01 1.15022548e-01 -8.18877816e-02 1.02397406e+00
3.06790024e-01 2.44498238e-01 -5.69556296e-01 -1.29623428e-01
8.32778037e-01 -1.68106461e+00 -3.45082045e-01 1.81331798e-01
1.02879500e+00 -1.21711768e-01 1.01023805e+00 2.26091623e-01
-8.98237526e-01 -1.83316201e-01 -1.10086226e+00 2.29434341e-01
-2.53546059e-01 3.63387913e-01 6.30023062e-01 9.44683075e-01
-1.26519501e+00 1.07721233e+00 -8.68535459e-01 1.82654019e-02
6.05237663e-01 6.49351180e-01 -4.56111044e-01 -9.93654802e-02
-5.77431321e-01 2.48052478e-01 4.50210385e-02 3.32561612e-01
-1.07875383e+00 -9.00811434e-01 -6.71399891e-01 5.18992357e-02
4.77702320e-01 -1.09382272e+00 1.20769751e+00 -1.18971837e+00
-1.69525909e+00 4.40543205e-01 -9.19082835e-02 -9.13379133e-01
8.14440608e-01 -2.85034150e-01 1.71727702e-01 -7.81607181e-02
-2.44336218e-01 -2.26429954e-01 1.17549706e+00 -8.54666293e-01
-5.47817945e-01 -7.98950613e-01 -5.08776084e-02 2.25911826e-01
-9.48136151e-01 -4.94127631e-01 -8.92227441e-02 -6.07737780e-01
2.26320792e-02 -8.90124381e-01 -6.10452771e-01 -4.64427583e-02
-3.30678225e-01 -1.77440390e-01 9.93284404e-01 -1.13810904e-01
1.04718447e+00 -1.84440351e+00 2.28107825e-01 1.21366188e-01
7.80918419e-01 2.46071175e-01 -3.58476937e-01 5.18076181e-01
1.58796057e-01 -1.47036528e-02 -1.33725777e-01 -8.27279150e-01
6.65145889e-02 3.51311505e-01 -3.27117741e-01 9.57773328e-01
-4.44975257e-01 8.97859097e-01 -9.01640415e-01 -2.96514899e-01
2.16366928e-02 1.19843736e-01 -8.81458938e-01 3.09629232e-01
-3.17842633e-01 2.24604264e-01 -7.00682402e-01 3.85846049e-01
7.19788373e-01 -3.38050276e-01 2.26794153e-01 -5.16747450e-03
-1.24396980e-01 -1.56241745e-01 -1.34701121e+00 1.77771723e+00
-7.83599734e-01 2.34201580e-01 7.01723158e-01 -1.18032181e+00
8.62502694e-01 1.49839923e-01 5.33606768e-01 -7.00583532e-02
1.22553661e-01 2.49125779e-01 -6.07421875e-01 -3.82947266e-01
3.79220545e-01 2.39855513e-01 3.46873432e-01 9.62738931e-01
4.64007348e-01 3.16179365e-01 -1.43586919e-01 3.23540539e-01
1.38158751e+00 -1.29001960e-01 -3.12948786e-02 -4.90523219e-01
6.06845438e-01 -1.41488060e-01 5.75955212e-01 1.24793530e+00
-1.29034579e-01 3.25198144e-01 1.39768139e-01 -9.42745388e-01
-7.29037762e-01 -7.63094723e-01 4.31851983e-01 1.52961373e+00
-1.63056135e-01 -4.49052036e-01 -9.88120556e-01 -1.32418096e+00
2.96151012e-01 2.16497630e-01 -8.31118107e-01 1.78069584e-02
-3.55243027e-01 -5.96059740e-01 4.45727766e-01 -8.00968427e-03
4.10766423e-01 -4.64944869e-01 -6.09389603e-01 1.63460523e-01
2.52007216e-01 -6.10628903e-01 -7.51718938e-01 5.81297696e-01
-1.11282802e+00 -1.20160580e+00 -8.42571378e-01 -9.06211317e-01
1.14935255e+00 3.19444388e-01 9.10790145e-01 6.74792081e-02
-1.52639270e-01 6.36987031e-01 -1.77449197e-01 -1.38937309e-01
-3.69397849e-01 4.22036856e-01 2.47306079e-01 6.65158629e-01
-1.93395004e-01 -5.91730297e-01 -6.91722095e-01 2.05771223e-01
-9.76237833e-01 3.88268568e-02 2.39902422e-01 1.04478312e+00
5.54255366e-01 -1.34018987e-01 5.01530766e-01 -1.35775411e+00
7.34423697e-01 -5.23548603e-01 -9.82542455e-01 5.99278867e-01
-9.48733151e-01 5.68560004e-01 1.34006858e+00 -4.49479014e-01
-7.50003338e-01 5.49698591e-01 3.42733413e-01 -9.70944405e-01
1.93248972e-01 4.15008157e-01 1.42402202e-01 -7.44332552e-01
8.52946877e-01 6.12180941e-02 2.09233135e-01 -6.81650579e-01
8.82422447e-01 5.94469726e-01 4.47160989e-01 -8.12937617e-01
8.95814419e-01 5.06470621e-01 -9.07211751e-03 -5.00888944e-01
-9.28984046e-01 -4.07736033e-01 -2.76249886e-01 -2.70750403e-01
2.55759805e-01 -6.61233068e-01 -9.84217823e-01 2.47771665e-01
-9.52919662e-01 -4.14742112e-01 -6.27770722e-01 4.17877197e-01
-5.28970838e-01 2.54599065e-01 -4.53207731e-01 -7.47856975e-01
-8.88826847e-01 -9.26891804e-01 5.39629281e-01 2.48104751e-01
4.32554156e-01 -1.27469742e+00 4.20002401e-01 3.08986008e-01
4.78531450e-01 3.79510880e-01 6.21944010e-01 -9.25502598e-01
-4.14790690e-01 -3.33478719e-01 2.03284249e-01 3.48968923e-01
2.34340236e-01 -4.60929543e-01 -7.46563077e-01 -7.08770275e-01
5.46702027e-01 -7.12164223e-01 9.30484533e-01 1.81094274e-01
1.19517004e+00 -1.19392967e+00 -8.97148997e-02 1.15425384e+00
1.62668061e+00 -6.31930351e-01 -3.76073658e-01 2.02368632e-01
9.00176704e-01 3.96671355e-01 -4.14858982e-02 7.49573171e-01
2.01192498e-01 1.23941310e-01 7.09989727e-01 -2.78939586e-02
6.56076148e-02 -4.91024703e-01 2.87539601e-01 1.06020439e+00
2.50831038e-01 -1.47782005e-02 -6.82904363e-01 4.66674089e-01
-2.30405426e+00 -5.13475776e-01 1.34405300e-01 2.22547174e+00
7.46024013e-01 -5.24971247e-01 1.93721168e-02 -2.09130362e-01
8.59222651e-01 1.35331854e-01 -1.07818151e+00 -4.72006768e-01
-1.21901967e-01 4.47419286e-02 9.22519743e-01 4.60384697e-01
-9.31002617e-01 7.19697475e-01 5.39111090e+00 7.08457530e-01
-1.16051316e+00 2.82005727e-01 6.24991477e-01 -2.94574916e-01
-3.50165933e-01 3.99456210e-02 -7.16658890e-01 1.18226185e-01
7.31611371e-01 -4.18736666e-01 1.07663023e+00 1.19883835e+00
7.33296871e-02 4.14420903e-01 -1.18009961e+00 1.05556154e+00
1.42437145e-02 -1.81957066e+00 1.05449870e-01 -3.98164876e-02
1.33298469e+00 5.64092040e-01 1.54122353e-01 1.00894282e-02
6.99749053e-01 -8.87171805e-01 6.92467570e-01 2.60389209e-01
5.61323941e-01 -8.34000289e-01 2.88829565e-01 6.04484081e-01
-1.07692981e+00 -5.20673752e-01 -4.26991582e-01 -2.72742454e-02
-4.03238922e-01 7.35386312e-01 -8.19480658e-01 8.43983471e-01
4.91116822e-01 4.06664342e-01 -5.92457950e-01 9.10048366e-01
1.54246151e-01 7.81752646e-01 -5.06724179e-01 -2.31919944e-01
3.70385885e-01 -5.77486694e-01 6.40513301e-01 8.64173949e-01
-1.88889578e-02 -2.93901086e-01 1.06357327e-02 7.25522578e-01
-5.24027288e-01 2.83379853e-01 -7.57384956e-01 -2.87317429e-02
2.47556418e-01 1.48818278e+00 -5.97017944e-01 -5.13205677e-02
-3.86942208e-01 9.21007395e-01 1.08903158e+00 4.31874275e-01
-4.62935120e-01 -4.89180923e-01 3.45871568e-01 -2.19067797e-01
5.71413636e-01 -1.12330884e-01 -2.26255670e-01 -1.38861167e+00
1.16187021e-01 -1.07587051e+00 6.26455903e-01 4.24479432e-02
-1.31374729e+00 7.34063864e-01 -6.76243424e-01 -8.57560575e-01
-1.44871354e-01 -3.44999313e-01 -5.63653111e-01 5.29111266e-01
-1.22679675e+00 -1.00154793e+00 -2.29892135e-01 1.03799725e+00
4.13147211e-01 -4.67770875e-01 7.03688085e-01 1.23741575e-01
-7.16582000e-01 1.04823530e+00 6.62401378e-01 4.59070355e-02
2.74376363e-01 -1.23341382e+00 1.65022239e-01 9.41823006e-01
6.58806741e-01 6.38218462e-01 4.02715534e-01 -3.53217095e-01
-2.15125251e+00 -1.57534146e+00 4.83564287e-01 -7.36948252e-02
7.46880949e-01 -4.45425123e-01 -5.21420956e-01 7.93092310e-01
-6.03113659e-02 5.02464235e-01 4.93569016e-01 2.30129302e-01
-4.73351330e-01 -4.72693712e-01 -1.16369450e+00 4.67724860e-01
1.00201201e+00 -4.19049710e-01 -1.98257759e-01 8.21191669e-01
1.00948131e+00 -4.24806476e-01 -5.36287189e-01 1.49879381e-01
2.98162490e-01 -9.13612127e-01 4.96662468e-01 -9.28689063e-01
-2.16657594e-02 -2.64031556e-03 -6.29935265e-02 -1.33877146e+00
1.18646882e-01 -1.48844671e+00 -6.13225758e-01 8.73426139e-01
3.46352667e-01 -1.14885104e+00 1.42692745e+00 5.58661997e-01
-1.71691343e-01 -1.20779359e+00 -1.12339818e+00 -7.59983957e-01
1.49040535e-01 -1.95758015e-01 8.13978910e-01 8.42343211e-01
-1.13704681e-01 4.83485647e-02 -3.73132527e-01 9.87176597e-02
1.02463746e+00 3.22110921e-01 8.54555845e-01 -7.70224512e-01
-7.11704433e-01 -1.94802880e-01 3.40023870e-03 -1.06494617e+00
2.74357975e-01 -1.33386540e+00 -2.16277495e-01 -1.18284142e+00
-6.32646233e-02 -5.75117469e-01 -4.36792850e-01 5.65352142e-01
6.90600201e-02 -1.70502171e-01 2.00518250e-01 2.61934280e-01
-7.93033779e-01 7.67210364e-01 9.89100397e-01 -1.35665134e-01
-3.67672086e-01 3.65162760e-01 -5.65747917e-01 5.33222973e-01
6.74133241e-01 -7.26000547e-01 -5.55608273e-01 -8.67850244e-01
2.98803300e-01 7.26664066e-02 3.41150224e-01 -8.94363344e-01
9.40181494e-01 9.12977904e-02 2.60884478e-03 -2.36661717e-01
-3.74999434e-01 -7.62612641e-01 -4.87444848e-02 6.30741298e-01
-5.89452982e-01 2.28045821e-01 -1.05727315e-01 1.01493847e+00
1.26733035e-01 -3.12443972e-01 5.42852759e-01 -3.95535072e-03
-3.12257139e-03 7.01010764e-01 6.79169744e-02 2.04894587e-01
8.66369903e-01 7.56836087e-02 -1.46858618e-01 -1.11551888e-01
-6.33864582e-01 4.78681058e-01 3.24144661e-01 9.68988985e-02
5.54636776e-01 -9.08490419e-01 -5.06800890e-01 4.09196854e-01
-3.33365232e-01 1.84787452e-01 1.77887678e-01 7.45060682e-01
-5.61484277e-01 3.93818557e-01 3.63254517e-01 -4.58024204e-01
-9.85405624e-01 5.36244810e-01 6.96847320e-01 -4.15852994e-01
-6.93485260e-01 1.06115198e+00 -1.52775839e-01 -1.05643511e+00
6.66320920e-01 -1.14008330e-01 3.42830509e-01 -3.68727714e-01
2.45546535e-01 6.36913896e-01 5.50667286e-01 6.05812781e-02
-1.67645916e-01 2.09631771e-01 -2.08161250e-01 9.59928657e-05
1.49124455e+00 -6.82634339e-02 -1.79673985e-01 2.10264444e-01
1.48219621e+00 3.84236872e-02 -1.44666517e+00 -4.80859548e-01
-1.91397339e-01 -1.08051673e-01 2.15543017e-01 -3.47364634e-01
-1.39840877e+00 3.01958710e-01 3.96591783e-01 1.57307476e-01
9.22898471e-01 -3.50693256e-01 8.19936514e-01 9.80861902e-01
4.73009825e-01 -1.01102173e+00 -1.76195845e-01 4.92929995e-01
4.55729991e-01 -9.20882702e-01 -1.31328970e-01 1.73620760e-01
-2.26880208e-01 1.37499678e+00 4.69154924e-01 -4.15103227e-01
9.25698578e-01 2.52321035e-01 -1.96473282e-02 -1.36042699e-01
-8.92047703e-01 3.61147702e-01 5.92182875e-02 2.54423887e-01
-7.75778070e-02 -5.91477491e-02 -1.71749279e-01 6.08884931e-01
-2.33385101e-01 1.19460545e-01 1.55944631e-01 8.80087912e-01
-2.27824062e-01 -8.61803353e-01 -1.73885047e-01 5.63800454e-01
-3.72003943e-01 2.67540179e-02 -3.22005421e-01 2.71562219e-01
-9.09458622e-02 9.65287626e-01 -1.73716202e-01 -4.66328621e-01
-1.20522626e-01 -4.89692055e-02 4.30584580e-01 -4.51278478e-01
-1.07958007e+00 -5.15215576e-01 -3.28416288e-01 -6.96320057e-01
-3.84066664e-02 -1.08533956e-01 -1.08865595e+00 -3.56386542e-01
-5.51315129e-01 7.42826819e-01 9.77356374e-01 9.20170844e-01
8.41876447e-01 2.74016440e-01 1.24466264e+00 -6.21983707e-01
-1.46114731e+00 -4.21977550e-01 -7.13008165e-01 -5.19800410e-02
7.55135119e-01 -3.58819887e-02 -7.03731954e-01 -1.53902873e-01] | [6.126292705535889, 5.303414821624756] |
d0febb90-1068-4d5f-b4ae-0b4c9cb71840 | universal-battery-performance-and-degradation | 2008.01527 | null | https://arxiv.org/abs/2008.01527v2 | https://arxiv.org/pdf/2008.01527v2.pdf | Universal Battery Performance and Degradation Model for Electric Aircraft | Development of Urban Air Mobility (UAM) concepts has been primarily focused on electric vertical takeoff and landing aircraft (eVTOLs), small aircraft which can land and takeoff vertically, and which are powered by rechargeable (typically lithium-ion) batteries. Design, analysis, and operation of eVTOLs requires fast and accurate prediction of Li-ion battery performance throughout the lifetime of the battery. eVTOL battery performance modeling must be particularly accurate at high discharge rates to ensure accurate simulation of the high power takeoff and landing portions of the flight. In this work, we generate a battery performance and thermal behavior dataset specific to eVTOL duty cycles. We use this dataset to develop a battery performance and degradation model (Cellfit) which employs physics-informed machine learning in the form of Universal Ordinary Differential Equations (U-ODE's) combined with an electrochemical cell model and degradation models which include solid electrolyte interphase (SEI) growth, lithium plating, and charge loss. We show that Cellfit with U-ODE's is better able to predict battery degradation than a mechanistic battery degradation model. We show that the improved accuracy of the degradation model improves the accuracy of the performance model. We believe that Cellfit will prove to be a valuable tool for eVTOL designers. | ['Venkatasubramanian Viswanathan', 'Evan Frank', 'William L. Fredericks', 'Alexander Bills', 'Matthew Guttenberg', 'Devin Charles', 'Shashank Sripad'] | 2020-07-06 | null | null | null | null | ['physics-informed-machine-learning'] | ['graphs'] | [-4.44368809e-01 -8.80446792e-01 -4.88737971e-01 2.55667478e-01
-2.87904531e-01 -5.17238975e-01 4.29637522e-01 9.17340349e-03
-7.10652769e-02 1.09392285e+00 -2.50946909e-01 -1.15520370e+00
-2.57515937e-01 -7.01042831e-01 -8.38559568e-01 -7.01855481e-01
-5.35880588e-03 7.63000607e-01 2.83939838e-01 -5.22968709e-01
1.19020917e-01 9.98943627e-01 -1.73170161e+00 -2.91585494e-02
1.03311241e+00 1.31343448e+00 -2.07630481e-04 6.39543235e-01
2.08417863e-01 7.04144657e-01 -2.88398862e-02 2.81581461e-01
-1.21417440e-01 -1.40445545e-01 -3.20964754e-01 -7.21303642e-01
-3.51950079e-01 -1.62019461e-01 -3.25842619e-01 2.09907755e-01
3.86514783e-01 2.28675991e-01 1.26834929e+00 -1.81917655e+00
-5.14622405e-02 -1.25573128e-01 3.08793426e-01 3.12837690e-01
-1.60478845e-01 2.06640899e-01 5.34027696e-01 -1.06458271e+00
5.95022626e-02 6.10212326e-01 1.04654753e+00 5.52399278e-01
-7.67208159e-01 -8.18917871e-01 -3.60971808e-01 1.65496752e-01
-1.43685400e+00 -4.19254154e-01 3.09634626e-01 -5.66904604e-01
1.60501945e+00 7.27489054e-01 1.32124317e+00 4.04376805e-01
1.18961716e+00 6.16083682e-01 9.19322431e-01 -4.41542603e-02
5.77930748e-01 2.23687425e-01 5.34415059e-02 1.56697527e-01
8.40415001e-01 2.65437305e-01 -3.35501969e-01 -5.83572201e-02
1.36326686e-01 -9.20987055e-02 -1.34810954e-01 -3.66368115e-01
-2.73851901e-01 4.47819941e-02 1.68385580e-01 8.40284228e-02
-2.16938987e-01 5.70457101e-01 1.38760820e-01 -1.34221390e-01
2.55635023e-01 2.26270556e-01 -2.77203172e-01 -5.48282325e-01
-1.05253160e+00 6.74282849e-01 1.03236783e+00 9.62106884e-01
2.93133080e-01 4.49506700e-01 1.10912018e-01 4.01312351e-01
5.10368824e-01 9.80835736e-01 2.13708252e-01 -6.02946937e-01
-3.81187387e-02 2.91229993e-01 7.95110583e-01 -2.29836609e-02
-4.28525418e-01 -4.90037262e-01 -3.14434618e-01 1.87434480e-01
-3.56680512e-01 -2.18422815e-01 -8.69562149e-01 1.00703430e+00
-6.22249544e-02 -2.33048536e-02 4.10074145e-01 5.20577669e-01
6.66041195e-01 9.78733301e-01 3.04190487e-01 -1.23557925e-01
5.14571905e-01 -5.35654068e-01 -5.81161678e-01 -9.97909456e-02
9.81979966e-01 1.61057413e-01 7.99672961e-01 2.37324506e-01
-1.52991819e+00 -1.63199738e-01 -1.56709898e+00 -5.18710129e-02
-6.50543451e-01 -2.07201123e-01 3.72998744e-01 8.29188764e-01
-9.91978467e-01 7.29025245e-01 -1.02159727e+00 -1.37454048e-01
2.61399657e-01 7.22491801e-01 8.05844665e-02 2.34695878e-02
-1.29563487e+00 1.40329778e+00 -2.01746538e-01 2.27360860e-01
-1.31368947e+00 -1.14104617e+00 -6.19937956e-01 8.61073583e-02
-3.75127226e-01 -6.90469503e-01 1.42453945e+00 1.95678044e-02
-1.06156242e+00 1.52275026e-01 -2.30406031e-01 -5.17125487e-01
3.55674565e-01 -5.75922318e-02 -5.92465281e-01 -2.46661723e-01
-2.77026713e-01 3.60874474e-01 -9.68591422e-02 -1.50575304e+00
-5.80448925e-01 6.50877580e-02 -5.05944014e-01 1.68248847e-01
-4.08130616e-01 -5.79622090e-01 1.74408987e-01 2.18476310e-01
-2.55034357e-01 -1.37315702e+00 1.14312083e-01 -2.18924955e-01
1.85907811e-01 -8.29502717e-02 7.84534872e-01 -7.72783816e-01
1.44398916e+00 -1.79277360e+00 -3.05994153e-02 2.99960136e-01
-4.67146516e-01 2.01852009e-01 4.40280974e-01 8.80125165e-01
1.01203628e-01 3.80441815e-01 -3.67285788e-01 -3.37628275e-01
7.92453736e-02 4.33737159e-01 -2.68142015e-01 4.46432441e-01
-1.73307478e-01 1.11356509e+00 -7.24797666e-01 -4.82431352e-02
-2.37455927e-02 3.56531471e-01 -3.53125870e-01 -2.32819542e-01
-2.22531945e-01 1.57191813e-01 -7.53593370e-02 9.86563325e-01
6.49583697e-01 3.91962528e-01 -9.19623896e-02 -1.44711316e-01
-8.04583371e-01 2.42134407e-01 -2.87914485e-01 1.03415561e+00
-7.05585420e-01 5.57500660e-01 1.53938562e-01 -5.27093053e-01
7.96827793e-01 1.46310613e-01 7.86321044e-01 -1.06462073e+00
2.63655841e-01 8.91477168e-01 -1.62727922e-01 -5.87938547e-01
6.17151260e-01 -9.44683671e-01 6.22024834e-02 -3.88270505e-02
-3.59325528e-01 -6.18405640e-01 2.09633354e-02 -8.63943771e-02
1.07494736e+00 -2.60312110e-03 -7.16323972e-01 -8.23575020e-01
3.39766294e-01 5.05075037e-01 4.58513051e-01 1.18352041e-01
8.27327743e-02 -4.67719994e-02 2.01153189e-01 1.29303530e-01
-1.28892362e+00 -9.97366309e-01 -5.22995472e-01 3.08472037e-01
8.16807747e-01 -2.16979399e-01 -4.80558366e-01 3.62266392e-01
7.99155235e-01 1.36455452e+00 -1.84588492e-01 -1.03072882e+00
-2.66529858e-01 -1.07420397e+00 6.07842326e-01 1.21584773e+00
2.98349231e-01 -2.53792137e-01 -5.02184212e-01 1.99432820e-01
1.19197980e-01 -6.43344223e-01 1.66836623e-02 5.51187217e-01
-9.98292804e-01 -7.90424943e-01 -1.89073727e-01 -4.43202496e-01
2.28992015e-01 1.18975483e-01 8.51690888e-01 3.37229341e-01
-6.68099299e-02 8.36256385e-01 5.82188182e-02 -8.35460126e-01
-5.94211578e-01 -8.67899731e-02 6.09254897e-01 -4.84634757e-01
1.79471999e-01 -3.74918550e-01 -7.05916405e-01 5.60715079e-01
-8.12964022e-01 2.66417842e-02 3.24956030e-01 3.64318192e-01
6.89761102e-01 5.67577183e-01 8.77514780e-01 -1.27751440e-01
5.78420520e-01 -9.52147067e-01 -3.14465255e-01 2.01289892e-01
-1.21047437e+00 -2.41377354e-01 2.24240229e-01 -8.41847956e-02
-9.85952079e-01 -4.22229618e-01 -5.93934417e-01 -2.90547311e-01
5.04413664e-01 7.22796977e-01 -4.57442522e-01 -2.66981691e-01
1.54775739e-01 2.09655002e-01 2.94266999e-01 -4.81561571e-01
-5.47622085e-01 1.05819762e+00 6.60402060e-01 -6.73659205e-01
5.07155001e-01 4.48343784e-01 5.84948123e-01 -9.88175392e-01
2.31038500e-02 -2.15862185e-01 -2.17931569e-01 -6.21890366e-01
4.61546570e-01 -1.14934790e+00 -8.57322752e-01 6.20219469e-01
-2.51528174e-01 -1.06209648e+00 -1.90705016e-01 3.21139038e-01
-4.33132023e-01 -1.66075870e-01 -4.60509360e-01 -1.48263836e+00
-3.72747689e-01 -9.83042121e-01 7.01162577e-01 2.14420378e-01
-2.03413710e-01 -1.20915461e+00 1.16301917e-01 3.07905644e-01
6.63719893e-01 2.54918069e-01 1.12927735e+00 -2.26383880e-02
-5.40245533e-01 -5.03980219e-01 3.04243505e-01 5.20343006e-01
-1.37121782e-01 -4.10255045e-02 -7.10387230e-01 -6.82748675e-01
-1.49724871e-01 2.25277498e-01 7.91829586e-01 3.65906447e-01
9.29468215e-01 2.83239096e-01 -1.10541630e+00 4.14907008e-01
1.63491905e+00 6.61154389e-01 8.53841424e-01 2.69496083e-01
5.00538409e-01 2.95778904e-02 8.56794000e-01 2.23556608e-01
4.99700457e-01 3.64188015e-01 5.98441601e-01 2.83378780e-01
6.82070702e-02 -3.24467063e-01 6.99695706e-01 7.01295018e-01
-4.58256006e-02 -7.71829128e-01 -1.07322669e+00 7.18910336e-01
-1.41468799e+00 -5.82827151e-01 -5.81420481e-01 2.34706473e+00
6.37215197e-01 3.71600419e-01 5.73239103e-02 4.60840553e-01
7.92880058e-02 -7.59931564e-01 -8.98980081e-01 -7.13582456e-01
-1.07725263e-01 -4.35427055e-02 1.04607797e+00 4.58951980e-01
-1.98173553e-01 2.29859546e-01 6.95095682e+00 4.84987706e-01
-1.14127553e+00 2.46237949e-01 2.98940867e-01 -4.19848263e-01
-9.95949149e-01 1.50724024e-01 -1.11798477e+00 6.69494390e-01
1.68908107e+00 -3.23274344e-01 3.03456306e-01 4.45050985e-01
3.21343690e-01 -6.16285264e-01 -1.08379066e+00 7.31544375e-01
-1.00910261e-01 -1.30966520e+00 -6.79095984e-02 2.38280952e-01
5.13971567e-01 -6.33613393e-02 -4.52303179e-02 4.67986941e-01
-5.57399333e-01 -9.03011382e-01 1.22190952e+00 1.14054847e+00
1.20525789e+00 -9.76378202e-01 6.33639276e-01 4.60594714e-01
-1.17739582e+00 -7.03440845e-01 -1.16484933e-01 -1.37728125e-01
3.95586461e-01 5.20373166e-01 -6.34107172e-01 3.00376445e-01
8.23007762e-01 5.67570865e-01 -4.15106803e-01 1.24544179e+00
7.65645266e-01 5.88767111e-01 -5.45033932e-01 -2.88414240e-01
2.89933234e-02 -3.06363314e-01 3.90906692e-01 9.20112371e-01
8.49525273e-01 1.47034481e-01 -3.26369286e-01 8.08137000e-01
1.92169681e-01 -4.28963035e-01 -7.00778782e-01 -4.35556024e-01
4.73017424e-01 8.94817948e-01 -5.14749944e-01 -1.45157710e-01
-2.68103555e-02 4.33236390e-01 -6.20687962e-01 2.99091607e-01
-9.97144938e-01 -3.51026803e-01 1.07456183e+00 1.19780231e+00
-9.10707191e-02 -5.89688659e-01 -4.44301516e-01 -5.43312967e-01
-2.69996434e-01 2.04165384e-01 -2.80336618e-01 -1.24568272e+00
-9.16095316e-01 -2.27629766e-02 4.96914208e-01 -8.56013954e-01
9.55228657e-02 -1.06657040e+00 -8.68888617e-01 8.71884108e-01
-1.52525067e+00 -1.10012054e+00 -4.73900139e-01 -1.71745628e-01
1.30582780e-01 1.23844706e-01 2.88790494e-01 7.52054453e-01
-6.70348883e-01 2.96176523e-01 7.30377913e-01 -7.94224501e-01
3.42705309e-01 -1.08114457e+00 3.10270905e-01 2.83447772e-01
-1.08291459e+00 2.53014416e-01 9.16687965e-01 -1.49008095e+00
-2.25941634e+00 -1.09921658e+00 5.17888725e-01 -8.56128514e-01
3.23276699e-01 -2.60744631e-01 -8.78105879e-01 3.97953600e-01
-2.14330792e-01 -5.60383499e-01 7.36187816e-01 -5.34260809e-01
8.52165341e-01 -4.93817359e-01 -8.31933737e-01 5.01331210e-01
1.10705960e+00 -4.72226143e-01 5.41637540e-02 1.18887089e-01
2.01974481e-01 -4.92266893e-01 -9.68846798e-01 9.28422272e-01
9.84649658e-01 -4.77815956e-01 8.46053064e-01 -4.81024683e-01
-1.35923952e-01 -3.98483455e-01 -5.42493686e-02 -1.05761039e+00
-2.07681537e-01 -1.32289797e-01 -2.86660761e-01 1.20463812e+00
4.69960600e-01 -6.70686364e-01 2.84262151e-01 1.30108392e+00
-9.70738113e-01 -1.17432106e+00 -1.11864674e+00 -1.18157089e+00
5.85050941e-01 -6.48394823e-01 8.08501601e-01 2.42938306e-02
5.84438108e-02 -3.05590153e-01 -2.97782972e-04 5.28901704e-02
2.34883755e-01 -4.90418822e-01 1.78043917e-01 -1.54741895e+00
5.24599850e-01 -1.27739102e-01 -5.23646586e-02 -5.05869269e-01
3.40181440e-01 -1.00664890e+00 3.50861639e-01 -2.01859236e+00
-9.66578200e-02 -1.08965719e+00 -5.32456398e-01 1.36684805e-01
4.56678182e-01 1.90420598e-01 -2.73965210e-01 2.58340865e-01
-1.10769860e-01 8.56463611e-01 7.76820719e-01 -2.64871091e-01
-2.91146338e-01 6.25845641e-02 -3.69144380e-01 1.47555217e-01
7.33594894e-01 -3.19672674e-01 -5.04742682e-01 1.24473810e-01
8.31969023e-01 7.48276412e-02 7.55052865e-01 -1.62724483e+00
2.68693328e-01 -1.74223498e-01 6.15750313e-01 -9.28930879e-01
5.18567979e-01 -8.82553339e-01 1.02172744e+00 6.63553655e-01
3.82761776e-01 3.06202441e-01 9.67275560e-01 4.70992506e-01
2.33145311e-01 -1.95297241e-01 5.02807975e-01 3.41694713e-01
-4.99392569e-01 1.88046113e-01 -9.97044027e-01 -6.40765786e-01
1.15877473e+00 -5.45737803e-01 -1.48201272e-01 -1.66203380e-01
-3.83328021e-01 5.55095136e-01 1.09408903e+00 3.64083976e-01
5.48820138e-01 -1.41573834e+00 -9.68279317e-03 1.84221208e-01
-1.84970245e-01 -1.62698686e-01 2.95633703e-01 1.01350188e+00
-8.24313223e-01 6.75876856e-01 -1.66773677e-01 -4.43915099e-01
-6.69538081e-01 4.03502524e-01 9.56365943e-01 3.97102863e-01
-1.82924181e-01 3.27704817e-01 -4.63870227e-01 1.42455667e-01
-2.43725389e-01 -6.55408978e-01 2.30366394e-01 -4.22510244e-02
1.38973370e-02 9.23946202e-01 4.57208723e-01 -6.40043259e-01
-5.86302578e-01 4.26853538e-01 8.28840733e-01 -2.08757252e-01
1.04644156e+00 -1.51247293e-01 -2.49122798e-01 9.76901233e-01
8.44266176e-01 -2.08568096e-01 -1.12703598e+00 1.02458704e+00
-2.86472261e-01 9.55786034e-02 4.74807858e-01 -8.36821795e-01
-4.30292279e-01 7.12921739e-01 7.05548584e-01 -7.33211190e-02
8.74195218e-01 -3.65221649e-01 1.26734948e+00 2.63975501e-01
2.88370222e-01 -1.25944304e+00 -2.39231288e-01 2.77904570e-01
8.70608509e-01 -5.02688348e-01 2.88206376e-02 -6.89300224e-02
-3.10723692e-01 7.38673508e-01 7.53954649e-01 2.34951809e-01
1.17893934e+00 5.98658562e-01 -5.41690767e-01 3.08997440e-03
-9.65105414e-01 2.54441142e-01 1.57328501e-01 2.32846633e-01
3.02630831e-02 -1.96518339e-02 -2.42192402e-01 1.06965566e+00
2.16104016e-01 4.22799081e-01 3.85476828e-01 1.39556408e+00
-6.53916180e-01 -1.13120306e+00 -3.99589330e-01 7.28176832e-01
2.49459162e-01 1.82212576e-01 -5.46786934e-02 5.98075509e-01
3.31553012e-01 8.37987006e-01 2.74757832e-01 -5.71269631e-01
6.68047249e-01 5.90275109e-01 4.43383247e-01 -8.60442296e-02
-2.65733272e-01 -4.27986801e-01 2.56920904e-01 -4.56134742e-03
2.69192737e-02 -6.72316551e-01 -1.92246985e+00 -5.73503911e-01
-4.24181432e-01 5.89265347e-01 1.22363329e+00 9.88854289e-01
5.15617430e-01 7.49197006e-01 2.24311799e-01 -1.18698776e+00
-2.94556260e-01 -5.83685458e-01 -1.21162534e+00 -2.90229708e-01
2.66765952e-01 -1.24745393e+00 -4.56785947e-01 -4.25116509e-01] | [6.385476589202881, 2.8084516525268555] |
ab04f1f0-6fee-4eab-aa8f-31cd958edaaa | on-the-diagnostic-of-road-pathway-visibility | 1601.05535 | null | http://arxiv.org/abs/1601.05535v1 | http://arxiv.org/pdf/1601.05535v1.pdf | On the Diagnostic of Road Pathway Visibility | Visibility distance on the road pathway plays a significant role in road
safety and in particular, has a clear impact on the choice of speed limits.
Visibility distance is thus of importance for road engineers and authorities.
While visibility distance criteria are routinely taken into account in road
design, only a few systems exist for estimating it on existing road networks.
Most existing systems comprise a target vehicle followed at a constant distance
by an observer vehicle, which only allows to check if a given, fixed visibility
distance is available. We propose two new approaches that allow estimating the
maximum available visibility distance, involving only one vehicle and based on
different sensor technologies, namely binocular stereovision and 3D range
sensing (LIDAR). The first approach is based on the processing of two views
taken by digital cameras onboard the diagnostic vehicle. The main stages of the
process are: road segmentation, edge registration between the two views, road
profile 3D reconstruction and finally, maximal road visibility distance
estimation. The second approach involves the use of a Terrestrial LIDAR Mobile
Mapping System. The triangulated 3D model of the road and its surroundings
provided by the system is used to simulate targets at different distances,
which allows estimating the maximum geometric visibility distance along the
pathway. These approaches were developed in the context of the SARI-VIZIR
PREDIT project. Both approaches are described, evaluated and compared. Their
pros and cons with respect to vehicle following systems are also discussed. | ['Jean-Philippe Tarel', 'Pierre Charbonnier', 'Francois Goulette'] | 2016-01-21 | null | null | null | null | ['road-segementation'] | ['computer-vision'] | [ 1.66857049e-01 -5.08405603e-02 4.45261374e-02 -3.42362434e-01
-1.87921762e-01 -6.69620097e-01 7.32159078e-01 1.74806148e-01
-7.02817261e-01 5.22905409e-01 -4.96180981e-01 -9.20115769e-01
-4.01491344e-01 -1.36831379e+00 -3.29898089e-01 -4.67830420e-01
2.03867376e-01 7.42737889e-01 7.60117114e-01 -3.57467115e-01
2.72930831e-01 1.28102112e+00 -2.08277702e+00 -6.02752924e-01
9.63976324e-01 6.77234769e-01 5.13862550e-01 7.70931780e-01
-1.69919044e-01 -8.62971768e-02 -2.17574343e-01 -7.21728876e-02
3.64746481e-01 -6.72650104e-03 -4.18962449e-01 2.87338197e-01
2.48164356e-01 -5.01060009e-01 -8.13792571e-02 8.56264591e-01
1.27680466e-01 2.41463110e-01 8.63508046e-01 -1.13378656e+00
4.09884453e-01 -2.18175188e-01 -3.11491191e-01 1.32322654e-01
2.71160901e-01 1.17894068e-01 4.64248806e-01 -7.77503133e-01
4.35751617e-01 9.00208771e-01 3.98412198e-01 -1.85343772e-01
-1.08092761e+00 -2.12518722e-01 -2.82710761e-01 5.81339777e-01
-1.77373254e+00 -4.05764669e-01 5.89106739e-01 -6.06420159e-01
5.69212794e-01 2.62759238e-01 7.15601385e-01 6.75107986e-02
1.36657553e-02 -1.85475782e-01 1.24338281e+00 -5.50925553e-01
1.84129819e-01 5.56036592e-01 2.40526393e-01 3.11172932e-01
4.99114245e-01 5.23955345e-01 2.21958935e-01 3.06033432e-01
3.19538981e-01 -2.39465743e-01 -4.04274583e-01 -5.39193869e-01
-4.64798838e-01 8.46362948e-01 3.80805373e-01 4.11296189e-01
-5.15702546e-01 -3.12476009e-01 6.41559586e-02 1.65261298e-01
2.58742303e-01 -3.78404140e-01 1.48461247e-02 1.10412692e-03
-7.94197023e-01 1.57011241e-01 5.96320510e-01 6.72281206e-01
1.27707160e+00 -1.17341951e-01 3.93562943e-01 4.86908078e-01
6.11315191e-01 8.75499845e-01 -3.74211937e-01 -8.78461957e-01
4.11389560e-01 6.07910156e-01 2.29955778e-01 -1.05930412e+00
-3.17027181e-01 -1.40546754e-01 -1.85611904e-01 1.13692331e+00
5.11208594e-01 -1.13077424e-01 -8.88812482e-01 1.05803370e+00
6.99941874e-01 2.32848208e-02 1.20328061e-01 6.17841423e-01
5.66361725e-01 6.25242114e-01 -2.82624751e-01 -1.49379611e-01
1.23568642e+00 -3.11416477e-01 -5.25941014e-01 -2.22326085e-01
5.16259253e-01 -8.44435811e-01 5.56220055e-01 2.79910624e-01
-8.69393706e-01 -5.58462739e-01 -1.09067631e+00 2.00914428e-01
-7.94920802e-01 1.66336894e-01 -1.42894492e-01 1.00879467e+00
-1.17260182e+00 2.41774440e-01 -2.76467919e-01 -3.87605935e-01
-1.76575601e-01 1.62322655e-01 -3.07223171e-01 -2.72233397e-01
-1.16773796e+00 1.50556028e+00 1.10478535e-01 6.81239665e-01
-7.44538128e-01 -2.90504277e-01 -9.27020490e-01 -4.22941446e-02
3.25902343e-01 -2.17308864e-01 8.06135297e-01 -3.55534554e-01
-1.40592539e+00 9.40595984e-01 -3.68784726e-01 -4.09234881e-01
8.78282666e-01 -8.71042814e-03 -6.93768799e-01 2.07672596e-01
1.99110270e-01 1.45312950e-01 3.79815370e-01 -1.47023189e+00
-9.25749242e-01 -4.62123305e-01 8.90710950e-02 3.67668390e-01
5.26264071e-01 -1.32330924e-01 -4.94437814e-01 6.13048851e-01
-1.14280749e-02 -8.44070733e-01 -2.77718008e-01 2.98567284e-02
-2.64891833e-01 7.99314007e-02 1.03696012e+00 -6.89234018e-01
9.45069849e-01 -1.83878636e+00 -4.46067989e-01 8.56798828e-01
-9.74540040e-02 6.23195410e-01 2.93571413e-01 4.60713208e-01
2.43509784e-01 -2.41485760e-01 -3.50191683e-01 -3.26288305e-02
-2.73162633e-01 2.08843991e-01 2.21704002e-02 6.71765447e-01
-2.42983863e-01 2.19993368e-01 -5.88493586e-01 -5.07940948e-01
1.00094938e+00 8.43944967e-01 1.52595013e-01 3.80818918e-02
1.87037900e-01 4.09586728e-01 -3.69116247e-01 2.03347653e-01
1.10818148e+00 7.56344438e-01 -9.13859997e-03 2.98042838e-02
-1.12280953e+00 1.87311426e-01 -1.49140310e+00 7.10232675e-01
-7.76615977e-01 8.49110901e-01 2.71001905e-01 -7.32414961e-01
1.37206662e+00 2.60749400e-01 9.58787799e-02 -8.28740299e-01
2.28773862e-01 5.42780280e-01 -1.35156721e-01 -5.52366197e-01
7.28449285e-01 1.07203461e-01 4.55395103e-01 1.20742716e-01
-7.93004513e-01 -4.49458092e-01 3.40909749e-01 -1.22909382e-01
4.40228701e-01 -2.44828425e-02 3.61233979e-01 -2.38601770e-02
9.70509768e-01 2.84083724e-01 1.16886139e-01 1.84780866e-01
8.14398229e-02 3.74828458e-01 3.64998490e-01 -1.70348823e-01
-1.03468001e+00 -1.09894919e+00 -5.28669000e-01 9.10263434e-02
6.24040842e-01 3.81908298e-01 -7.10124671e-01 -1.98083282e-01
-2.10462529e-02 8.66670907e-01 -3.96770209e-01 2.09007695e-01
-4.68188971e-01 -1.24615103e-01 7.61101991e-02 3.48557867e-02
5.59744060e-01 -5.66623569e-01 -9.16219234e-01 1.69648454e-02
1.08205535e-01 -1.07611823e+00 1.74843356e-01 -3.56679469e-01
-7.86026478e-01 -1.25465894e+00 -4.00921404e-01 -2.51269341e-01
5.85609257e-01 9.14536655e-01 7.56892502e-01 3.83097380e-02
-1.43031418e-01 4.12504882e-01 -1.54110566e-01 -4.17747766e-01
-7.31788874e-01 -1.17200129e-01 -2.70134330e-01 6.78156689e-02
3.09037060e-01 -3.75611156e-01 -4.17569458e-01 8.68430614e-01
-4.98687625e-01 -3.25127780e-01 4.76901472e-01 -8.92444924e-02
5.81523180e-01 1.94886222e-01 1.70840696e-01 -6.32939339e-01
2.27157161e-01 -4.48573947e-01 -1.35654151e+00 -5.40912226e-02
-7.83532798e-01 -4.62406665e-01 2.47450545e-01 3.08565825e-01
-1.06479692e+00 5.19250631e-02 -3.74470353e-01 -2.04145566e-01
-5.81265390e-01 2.57482201e-01 -7.41490364e-01 -2.28893429e-01
6.02342188e-01 5.61930090e-02 3.25298339e-01 -2.50134021e-01
3.35831821e-01 7.68860936e-01 4.89352435e-01 2.07721218e-01
1.12284541e+00 6.99294031e-01 4.25808966e-01 -1.45640171e+00
-1.13597691e-01 -6.87622190e-01 -8.18813562e-01 -8.86597753e-01
8.72866333e-01 -6.11163199e-01 -6.62403762e-01 3.86373729e-01
-1.13774967e+00 -1.18099250e-01 -1.43092617e-01 7.23935246e-01
-3.32957119e-01 3.33389223e-01 2.15550855e-01 -1.13670039e+00
1.86218079e-02 -1.09313917e+00 5.16399682e-01 2.95193970e-01
1.72361717e-01 -1.21177256e+00 1.69002563e-01 4.51397330e-01
4.31177139e-01 1.94729701e-01 4.90245819e-01 2.98877247e-02
-8.57218385e-01 -3.90367299e-01 -3.00675780e-01 3.76859248e-01
8.96246135e-02 2.54787117e-01 -1.00565815e+00 1.83907494e-01
-1.78399056e-01 4.90694791e-01 6.81507289e-01 6.43979669e-01
2.32611716e-01 3.78574580e-01 -5.43294847e-01 2.91193187e-01
1.70664823e+00 3.87404859e-01 1.02029932e+00 3.61148715e-01
4.21269983e-01 1.11451292e+00 1.24461615e+00 -5.01458496e-02
4.79799509e-01 9.29660976e-01 9.57104445e-01 -2.79705346e-01
-2.00562730e-01 6.36993051e-02 1.91147476e-01 1.51424021e-01
-3.87876958e-01 -1.71293572e-01 -1.00408459e+00 8.20172608e-01
-1.33618295e+00 -1.11113310e+00 -1.04501009e+00 2.81263113e+00
-1.78267032e-01 2.94610351e-01 1.73588514e-01 5.17981946e-01
1.11537528e+00 -7.67150298e-02 -1.03442386e-01 -8.13733935e-01
1.49134189e-01 -1.48768529e-01 1.02546465e+00 1.36571980e+00
-5.43557882e-01 5.14185727e-01 5.46198416e+00 4.65074301e-01
-1.05652761e+00 9.98907089e-02 1.23844771e-02 4.39595103e-01
-4.31329072e-01 3.53440464e-01 -9.78471518e-01 3.03480327e-01
1.08573627e+00 -1.02147378e-01 6.91161528e-02 6.88663781e-01
8.77026320e-01 -8.54200721e-01 -3.35515469e-01 4.82697725e-01
-1.87569991e-01 -1.04880202e+00 -2.93758035e-01 5.58939993e-01
2.47922033e-01 1.78691566e-01 -2.93568701e-01 -2.55725890e-01
-1.01380661e-01 -7.45388031e-01 6.20252252e-01 6.27676845e-01
8.18865895e-01 -9.94658232e-01 8.18357229e-01 4.21907455e-01
-1.59386885e+00 2.19478816e-01 -2.27655634e-01 2.92024910e-02
7.98478961e-01 6.93995059e-01 -1.08735406e+00 8.42521250e-01
2.45680302e-01 1.71670109e-01 -1.97044834e-01 1.37166059e+00
-5.64972699e-01 2.64991939e-01 -3.60411227e-01 2.34679058e-02
2.82758504e-01 -7.82804251e-01 8.50694835e-01 9.09083724e-01
5.56525946e-01 -3.19778085e-01 -1.87139511e-01 7.42827237e-01
4.77050245e-01 4.86128442e-02 -1.10339439e+00 7.54966378e-01
6.55812740e-01 1.23648930e+00 -6.52296007e-01 -1.27186865e-01
-4.52665299e-01 3.82433414e-01 -2.49438062e-01 3.18704069e-01
-8.50737214e-01 -6.10320270e-01 6.62753582e-01 8.04058313e-01
2.28675351e-01 -3.94892097e-01 -1.04470849e-01 -1.39630318e-01
3.18062566e-02 1.71223097e-02 -1.70198619e-01 -7.50584900e-01
-3.49696726e-01 4.38684821e-01 3.48744214e-01 -1.48900175e+00
-1.35964781e-01 -5.98737180e-01 -6.98610663e-01 1.49200535e+00
-1.91361499e+00 -9.02481794e-01 -7.28220999e-01 5.16129255e-01
3.72872829e-01 1.80038586e-01 4.31761175e-01 4.82608646e-01
-3.04030687e-01 -2.73426473e-01 1.32126674e-01 -3.71774316e-01
1.30501643e-01 -9.32002544e-01 2.61398524e-01 1.03640294e+00
-3.31478119e-01 8.89839511e-03 8.49261999e-01 -6.62817359e-01
-8.88434052e-01 -9.43105400e-01 1.25051105e+00 -1.44734293e-01
4.04944927e-01 -1.08860917e-02 -8.43566239e-01 3.09659749e-01
8.16949364e-03 -2.44779766e-01 2.49782652e-01 -3.05796444e-01
2.96395034e-01 -3.87092173e-01 -1.23880255e+00 4.87465948e-01
7.03328371e-01 -4.98742372e-01 -3.60921621e-01 -1.01204656e-01
1.62318170e-01 -1.87409982e-01 -5.34191191e-01 2.78590500e-01
6.60633087e-01 -1.24004674e+00 8.23256433e-01 3.23371232e-01
-1.84513912e-01 -7.41561413e-01 9.34130177e-02 -1.21457303e+00
1.78706139e-01 -3.06930751e-01 5.60868859e-01 1.09564209e+00
6.19953454e-01 -9.95377719e-01 7.46587336e-01 3.71408850e-01
-2.39757463e-01 -3.33024323e-01 -1.06819785e+00 -8.16271842e-01
-3.58513534e-01 -7.04048932e-01 4.40836221e-01 4.49782699e-01
-6.38778806e-01 3.02720189e-01 3.82040488e-03 7.62152374e-01
7.80565381e-01 4.38872240e-02 1.18006194e+00 -1.64457369e+00
3.46110880e-01 -3.71202767e-01 -7.08367169e-01 -7.95506001e-01
-7.20633194e-02 -6.06297791e-01 -1.64022613e-02 -2.01709366e+00
-4.98753428e-01 -6.35326803e-01 4.90539551e-01 -4.54314530e-01
3.22106332e-01 -1.17357383e-02 -2.10690379e-01 1.91893369e-01
2.94388473e-01 2.82490194e-01 8.81017983e-01 2.28461578e-01
-3.42898577e-01 5.55719733e-01 -6.11155219e-02 7.86920547e-01
8.44349444e-01 -2.98460186e-01 -6.73958480e-01 -2.90642887e-01
1.40817806e-01 -3.36340964e-02 4.64824945e-01 -1.11872804e+00
3.28199953e-01 -1.22796670e-01 -2.34734878e-01 -1.11951125e+00
5.75308859e-01 -1.34110975e+00 4.82619911e-01 5.86616755e-01
4.84320313e-01 -2.64967550e-02 6.51506856e-02 4.60332483e-01
-1.94123983e-01 -6.36721253e-01 1.00980484e+00 1.54250607e-01
-8.76175225e-01 1.03798889e-01 -7.82777846e-01 -4.73370552e-01
1.53042459e+00 -9.98769820e-01 -2.40648732e-01 -3.48768681e-01
-7.03393698e-01 3.23246568e-01 6.96781874e-01 -3.60854380e-02
6.94476008e-01 -8.23765755e-01 -7.68559277e-01 1.52724802e-01
1.30264699e-01 -6.21121563e-02 4.23688948e-01 8.74141455e-01
-8.71602476e-01 4.60620791e-01 -2.01214522e-01 -6.79344893e-01
-1.64885545e+00 4.61702287e-01 6.17321610e-01 8.73481333e-02
-4.04326379e-01 2.55618468e-02 -1.72369599e-01 -3.36696059e-01
-2.76459724e-01 -6.97723180e-02 -6.77284420e-01 1.29387632e-01
5.44230521e-01 1.04815161e+00 2.24459410e-01 -1.26255727e+00
-3.26889008e-01 1.28376365e+00 5.22199094e-01 -3.98932695e-01
8.19666803e-01 -7.43428826e-01 1.03426151e-01 5.20484090e-01
8.70661974e-01 4.81041908e-01 -1.00472951e+00 -2.03389507e-02
-1.09684788e-01 -5.84095240e-01 2.49506980e-01 -2.54954845e-01
-8.26021314e-01 1.08550370e+00 8.24475169e-01 5.08732975e-01
9.81474102e-01 -1.28951088e-01 2.63826162e-01 2.20586956e-01
3.76576066e-01 -1.09301388e+00 -8.13411891e-01 4.41940784e-01
5.56215107e-01 -1.09193039e+00 9.03145224e-02 -9.60778534e-01
-3.88278395e-01 1.10511529e+00 2.43450955e-01 2.60747164e-01
7.14038134e-01 1.35675013e-01 1.76594540e-01 -2.78026730e-01
-1.10206850e-01 -9.36181366e-01 -6.37213420e-03 7.59851873e-01
-4.11582924e-02 1.52737573e-01 -5.86318314e-01 -5.24402857e-01
4.82556485e-02 -4.92664725e-02 7.21790016e-01 6.80103481e-01
-1.01642013e+00 -1.05063868e+00 -8.25261712e-01 1.67042628e-01
1.56043649e-01 2.74438709e-01 -1.28080025e-01 1.10521317e+00
4.17162329e-01 1.28641486e+00 2.50165731e-01 -9.38125476e-02
8.85591567e-01 -3.39566022e-01 2.34827176e-01 -4.24330294e-01
-1.42306373e-01 -3.90637726e-01 6.27689242e-01 -2.59790033e-01
-2.53414243e-01 -8.77456844e-01 -1.16081381e+00 -3.65130842e-01
-4.43891346e-01 4.30124253e-01 1.37851691e+00 7.90049136e-01
5.55640087e-02 1.28345400e-01 9.80744600e-01 -9.44490373e-01
6.67367652e-02 -5.89236379e-01 -7.32008100e-01 -3.19619030e-01
2.70379007e-01 -8.79177868e-01 -6.03514969e-01 -3.50072265e-01] | [7.953003883361816, -1.4967306852340698] |
37b1fd7a-d7fa-4c86-a5d2-1bbe8c4cc8ff | character-grounding-and-re-identification-in | null | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/3913_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123500528.pdf | Character Grounding and Re-Identification in Story of Videos and Text Descriptions | We address character grounding and re-identification in multiple story-based videos like movies and associated text descriptions. In order to solve these related tasks in a mutually rewarding way, we propose a model named Character in Story Identification Network (CiSIN). Our method builds two semantically informative representations via joint training of multiple objectives for character grounding, video/text re-identification and gender prediction: Visual Track Embedding from videos and Textual Character Embedding from text context. These two representations are learned to retain rich semantic multimodal information that enables even simple MLPs to achieve the state-of-the-art performance on the target tasks. More specifically, our CiSIN model achieves the best performance in the Fill-in the Characters task of LSMDC 2019 challenges. Moreover, it outperforms previous state-of-the-art models in M-VAD Names dataset as a benchmark of multimodal character grounding and re-identification. | ['Jongseok Kim', 'Youngjae Yu', 'Jiwan Chung', 'Heeseung Yun', 'Gunhee Kim'] | null | null | null | null | eccv-2020-8 | ['gender-prediction'] | ['computer-vision'] | [ 4.27639633e-01 -1.13833562e-01 -5.68475246e-01 -3.75076771e-01
-1.11903560e+00 -6.43777549e-01 8.44899893e-01 2.35956982e-01
-3.60495985e-01 4.30088401e-01 5.03898978e-01 2.90522903e-01
2.64887065e-01 -1.45614937e-01 -9.55851018e-01 -3.55353236e-01
3.00912082e-01 9.35997725e-01 -8.10630172e-02 3.16740751e-01
3.65587950e-01 1.18653663e-01 -1.53510237e+00 8.54949057e-01
5.74803591e-01 1.10320067e+00 2.39092007e-01 9.23434794e-01
-2.40494594e-01 1.06612813e+00 -3.03464234e-01 -1.14578879e+00
-2.42374852e-01 -7.20627680e-02 -1.06831050e+00 9.96410623e-02
1.02908146e+00 -5.38656950e-01 -7.42264688e-01 9.67098951e-01
5.72606802e-01 1.75911114e-01 1.00191832e+00 -1.44494402e+00
-1.08989716e+00 1.33310306e+00 -6.60244286e-01 -7.41723999e-02
6.39703214e-01 -1.88177422e-01 1.20843530e+00 -1.15401387e+00
7.83413827e-01 1.62414122e+00 7.51441836e-01 9.31314409e-01
-1.37041163e+00 -4.33852166e-01 2.98978537e-01 5.69274962e-01
-1.36728847e+00 -6.23662949e-01 4.46950585e-01 -6.57713592e-01
8.53076816e-01 2.88018465e-01 3.23577404e-01 1.92239785e+00
-4.17520404e-01 1.61896229e+00 4.76146758e-01 -4.62044589e-02
-2.56161451e-01 3.54038835e-01 4.83538993e-02 6.82170212e-01
-1.10762753e-01 -5.96673667e-01 -1.01964879e+00 5.25124483e-02
6.46964967e-01 -3.43696028e-01 4.50828491e-04 -2.60564744e-01
-1.67797971e+00 6.95813835e-01 -3.15773368e-01 7.76736513e-02
1.03072852e-01 6.17720068e-01 9.32047606e-01 2.21799631e-02
2.32324481e-01 5.73666930e-01 -1.30947575e-01 -4.28448558e-01
-1.08889949e+00 4.00739282e-01 4.92721707e-01 9.30600107e-01
1.47357613e-01 2.15167373e-01 -9.04626012e-01 1.11084473e+00
3.44291031e-01 5.30654430e-01 5.69978535e-01 -8.12944055e-01
8.36072862e-01 3.17957044e-01 -3.00523508e-02 -8.14890325e-01
-3.06445688e-01 -1.39929190e-01 -6.75449729e-01 -4.39791679e-01
4.88424569e-01 1.21251896e-01 -8.05381417e-01 1.82727039e+00
-3.47290710e-02 7.71866500e-01 8.14569667e-02 6.74586058e-01
1.41885257e+00 6.92556381e-01 2.66370088e-01 2.87837505e-01
1.52350378e+00 -1.17993116e+00 -5.93770266e-01 -3.85162868e-02
3.92994940e-01 -7.81376481e-01 8.52386296e-01 9.14667919e-02
-9.80345845e-01 -7.99703121e-01 -9.34306264e-01 -1.32221758e-01
-1.03633910e-01 8.61249328e-01 2.73830056e-01 6.05135083e-01
-8.65875065e-01 9.15373266e-01 -4.62571502e-01 -6.13667607e-01
3.89289647e-01 1.64205939e-01 -6.64475441e-01 -1.24776391e-02
-9.17317629e-01 6.70859694e-01 7.26367652e-01 -7.14704543e-02
-1.29683173e+00 -8.10604990e-01 -9.79527950e-01 9.45565291e-03
2.85661638e-01 -5.94601214e-01 9.92818713e-01 -8.36727381e-01
-1.56431413e+00 1.55836022e+00 -8.47679153e-02 -3.81933451e-01
5.21611512e-01 -4.71866757e-01 -5.30217290e-01 3.53412151e-01
3.58176321e-01 9.69207287e-01 1.21251881e+00 -1.22966099e+00
-6.08149886e-01 -8.42485130e-02 -3.55929196e-01 1.48875296e-01
-6.44760847e-01 2.55219370e-01 -9.30435002e-01 -9.86117125e-01
-1.54604957e-01 -9.52492356e-01 3.81195307e-01 -3.41564208e-01
-9.37521875e-01 -4.00877148e-01 6.59850597e-01 -1.09411848e+00
1.02903020e+00 -2.02756476e+00 7.42684782e-01 -2.14232519e-01
3.55470449e-01 1.17493860e-01 -5.62299132e-01 2.88679689e-01
-1.71813905e-01 1.77261412e-01 1.89798862e-01 -1.28906906e+00
4.46489751e-01 -1.56572536e-01 -3.80320191e-01 3.35453868e-01
3.51923168e-01 1.02809024e+00 -7.07199275e-01 -8.11656296e-01
2.08248142e-02 3.29422295e-01 -2.40086928e-01 3.90867382e-01
-2.97301501e-01 4.73031938e-01 -1.80126026e-01 1.08096433e+00
5.21446705e-01 -3.80272388e-01 3.50420296e-01 -5.00985920e-01
1.86355665e-01 -2.34090075e-01 -1.14676869e+00 1.82196188e+00
9.93924439e-02 1.02018297e+00 -2.28248194e-01 -1.00656700e+00
7.94728398e-01 3.42900634e-01 4.83451962e-01 -4.55611944e-01
2.73812413e-01 1.63995370e-01 -6.70850635e-01 -6.93098009e-01
1.15850997e+00 3.43979955e-01 -3.69910836e-01 2.55797774e-01
6.70728207e-01 3.75339866e-01 4.04122144e-01 4.25189734e-01
5.96192122e-01 4.52566594e-01 -2.89885044e-01 1.08198933e-01
6.18553877e-01 -3.11955124e-01 4.44803655e-01 7.26709068e-01
-6.35961210e-03 9.10753310e-01 7.70507097e-01 -1.79136932e-01
-1.31877172e+00 -1.04235339e+00 -2.33569831e-01 1.59160697e+00
1.91992104e-01 -6.75135374e-01 -7.65634954e-01 -6.46775067e-01
1.92076489e-01 6.81431293e-01 -8.69879067e-01 1.10377237e-01
-6.21073127e-01 -5.39275885e-01 1.16820812e+00 8.95928919e-01
2.71791667e-01 -9.00264084e-01 1.75288945e-01 3.18885818e-02
-6.04878247e-01 -1.80331159e+00 -5.90091825e-01 -1.73052892e-01
-4.75463212e-01 -9.24717724e-01 -1.30491602e+00 -1.00882709e+00
1.48133025e-01 -1.43168733e-01 1.08038580e+00 -3.03409934e-01
-8.57441574e-02 7.94189215e-01 -4.98998702e-01 8.12635198e-02
-6.57848656e-01 3.44770789e-01 1.14830792e-01 6.59093440e-01
5.44754378e-02 -1.07972436e-01 -1.57979891e-01 2.08553240e-01
-4.71242130e-01 1.38031960e-01 2.52363145e-01 8.74218047e-01
5.95760167e-01 -6.59760296e-01 3.61901313e-01 -6.00569725e-01
3.02660525e-01 -5.67113698e-01 -2.82181263e-01 6.86843634e-01
-1.91705719e-01 2.63412017e-02 3.56658280e-01 -7.07749367e-01
-8.37836683e-01 1.53021455e-01 1.09637501e-02 -8.04969013e-01
-1.10006340e-01 2.90246904e-01 -2.66094625e-01 1.20235480e-01
2.01971814e-01 5.05034685e-01 -4.65101838e-01 -8.95117879e-01
5.31607747e-01 5.66162407e-01 1.13440800e+00 -8.71273279e-01
6.46921754e-01 6.10717423e-02 -1.43814206e-01 -7.53567934e-01
-6.23657942e-01 -3.73566002e-01 -7.78850853e-01 -3.55134726e-01
1.43691278e+00 -1.20107603e+00 -8.30930412e-01 8.65839183e-01
-1.27622926e+00 7.42647238e-03 3.65275778e-02 4.21428710e-01
-6.86275959e-01 8.13497424e-01 -7.61058271e-01 -7.29464293e-01
-2.45408118e-01 -1.13291276e+00 1.28323710e+00 3.07742149e-01
-3.37526679e-01 -7.75417149e-01 1.42518491e-01 6.39782250e-01
-1.07989326e-01 2.88425207e-01 1.08076513e+00 -9.74726558e-01
-4.30266440e-01 -1.33510545e-01 -5.34633517e-01 1.39827743e-01
-4.68275756e-01 2.31437474e-01 -1.16409612e+00 -2.32198969e-01
-9.42527294e-01 -8.24091673e-01 1.39253604e+00 2.72445261e-01
1.18432248e+00 -1.12428159e-01 -3.74566227e-01 9.24025118e-01
1.09702325e+00 -1.70742959e-01 5.66240191e-01 3.68172139e-01
1.41160417e+00 6.92503929e-01 4.38632995e-01 9.06755924e-01
7.00588763e-01 8.41187656e-01 6.21091843e-01 1.10917546e-01
-9.23437551e-02 -6.77923083e-01 3.89700979e-01 4.35171336e-01
-1.72939613e-01 -5.77085972e-01 -7.32987106e-01 6.65355504e-01
-2.33694410e+00 -1.25781631e+00 -1.09716877e-01 1.95350587e+00
5.93254983e-01 -3.51984441e-01 4.08407658e-01 -2.29387254e-01
1.20505214e+00 3.71213615e-01 -5.46297014e-01 -2.41760850e-01
-7.06316710e-01 -5.10419607e-01 4.72496718e-01 -1.01549730e-01
-1.69300723e+00 1.19352555e+00 5.97570038e+00 1.19805253e+00
-6.93103135e-01 4.28597815e-02 8.68907988e-01 -1.20154843e-01
-3.71237367e-01 -2.62451142e-01 -1.16702676e+00 6.35948360e-01
8.39585900e-01 2.34629095e-01 4.69266146e-01 8.88407409e-01
-3.91048253e-01 2.26524800e-01 -1.60647404e+00 1.67294264e+00
6.63113654e-01 -1.75805759e+00 4.00749236e-01 -3.09011072e-01
8.15578282e-01 -2.17603028e-01 3.74935210e-01 3.81571412e-01
-4.70394194e-02 -1.31071377e+00 1.31630695e+00 6.54465616e-01
1.13790894e+00 -6.60248518e-01 5.54047763e-01 -1.60199702e-01
-1.40594876e+00 -1.26697972e-01 -2.55030721e-01 7.46277690e-01
3.73939037e-01 -1.17042489e-01 -7.08420932e-01 3.72099161e-01
6.41986787e-01 1.12318230e+00 -9.10785615e-01 1.05120671e+00
1.35686308e-01 3.18305433e-01 2.70300180e-01 -3.00383214e-02
2.03116983e-01 1.82860374e-01 7.61302352e-01 1.49386907e+00
2.16429815e-01 -5.64027131e-01 1.10542953e-01 7.56207824e-01
-5.47617674e-01 2.14018300e-01 -5.66091277e-02 -7.80712247e-01
3.48308623e-01 1.19786370e+00 -6.04692936e-01 -5.30278802e-01
-1.76556721e-01 1.36082268e+00 4.03155208e-01 2.61350751e-01
-1.10127997e+00 -3.28916907e-02 8.10165226e-01 -1.91519320e-01
4.58293617e-01 6.39002398e-02 -2.98353620e-02 -1.58310509e+00
-2.72131979e-01 -1.02163756e+00 5.00263214e-01 -7.20473170e-01
-1.52505803e+00 6.79372787e-01 -1.48162432e-02 -1.14324582e+00
-3.99230391e-01 -7.96100140e-01 -3.97518307e-01 4.41514939e-01
-1.15145624e+00 -1.82684183e+00 -1.10245995e-01 8.51955295e-01
7.23630548e-01 -8.97278666e-01 7.93487310e-01 5.87516069e-01
-9.60661054e-01 1.29607034e+00 5.24577439e-01 5.68724513e-01
1.06441116e+00 -1.25027180e+00 5.36055028e-01 7.08458960e-01
2.27294549e-01 6.81565255e-02 5.05319893e-01 -8.14991832e-01
-1.44492459e+00 -9.89534736e-01 8.52620721e-01 -6.98245287e-01
7.60355413e-01 -5.65343380e-01 -6.42336130e-01 7.01732516e-01
1.48247540e-01 -4.93313581e-01 8.03972363e-01 8.20004381e-03
-6.68686867e-01 3.40311050e-01 -7.19743133e-01 5.46099424e-01
8.91389430e-01 -8.36514950e-01 -3.36271644e-01 4.03503180e-01
7.51224458e-01 -4.27395254e-01 -8.71066928e-01 1.13539755e-01
8.82452250e-01 -5.56999385e-01 1.30088174e+00 -1.02320242e+00
9.47335303e-01 8.49924311e-02 -7.08177209e-01 -7.06338465e-01
-4.67391521e-01 -6.13271594e-01 -3.69273871e-01 1.99354351e+00
4.50215459e-01 1.40451699e-01 9.17002380e-01 4.49696213e-01
-6.08700477e-02 -4.45867866e-01 -1.02776539e+00 -6.34897649e-01
-1.15617372e-01 -4.66273516e-01 6.10562265e-01 1.10289907e+00
-6.55800328e-02 5.12101829e-01 -1.27961099e+00 -1.09858409e-01
8.41376662e-01 1.32450640e-01 7.09088147e-01 -1.13796258e+00
-4.45520461e-01 -6.19135797e-01 -5.10745525e-01 -1.06958127e+00
7.97297537e-01 -1.15455973e+00 -2.63903111e-01 -1.30021918e+00
8.76552403e-01 -8.77993368e-03 -1.98616862e-01 2.95781940e-01
-1.68679491e-01 5.18077493e-01 7.42011487e-01 2.66251653e-01
-1.22262919e+00 7.87124038e-01 7.64391243e-01 -6.48424506e-01
1.35467976e-01 2.04954785e-03 -5.19549727e-01 6.84224069e-01
3.04113626e-01 -3.84954602e-01 7.78650120e-02 -6.71935081e-01
2.94616580e-01 3.07306796e-01 5.24470091e-01 -8.06789696e-01
4.13148962e-02 -3.00427992e-02 6.60933435e-01 -1.01652956e+00
7.81555414e-01 -2.45892793e-01 -6.94254935e-02 -2.61186492e-02
-6.29846275e-01 1.34378970e-01 1.73888057e-01 6.62636876e-01
-1.59074128e-01 -6.41346157e-01 5.92713177e-01 8.53326870e-04
-1.12876880e+00 3.03989619e-01 -5.67638338e-01 2.66521722e-01
6.87911749e-01 -5.04779994e-01 -5.06112158e-01 -3.70454907e-01
-8.70814860e-01 3.47112805e-01 4.35187727e-01 7.06182301e-01
7.60103285e-01 -1.69588304e+00 -1.13082433e+00 -2.29143888e-01
4.44456458e-01 -9.42411959e-01 6.55957460e-01 7.27436066e-01
-4.07004565e-01 3.57771397e-01 -4.04858112e-01 -6.83506608e-01
-1.59641242e+00 4.49168086e-01 6.35455102e-02 -3.21483105e-01
-3.29786301e-01 1.29969072e+00 9.90851074e-02 -2.97996938e-01
6.27826869e-01 1.04903094e-01 -7.08235741e-01 6.16582632e-01
5.80015063e-01 6.63774610e-01 -4.45511490e-01 -1.08814907e+00
-4.24419016e-01 8.50049019e-01 -7.01530948e-02 -8.26524347e-02
1.26061821e+00 -2.96462148e-01 5.50643802e-02 3.96155328e-01
1.41217983e+00 -1.20501727e-01 -1.31223488e+00 -3.65747452e-01
-2.62633618e-02 -5.47457278e-01 -3.29947889e-01 -6.24010563e-01
-8.35265934e-01 8.62439334e-01 5.83452463e-01 -2.86752760e-01
5.56217313e-01 3.33480805e-01 1.00436902e+00 4.64523733e-01
-3.37806851e-01 -1.49679911e+00 7.05203712e-01 6.39391899e-01
7.45144010e-01 -1.51839089e+00 -1.50099695e-01 -1.47781640e-01
-1.36861396e+00 1.31715977e+00 6.85702503e-01 2.02766746e-01
-2.82196421e-02 -2.24316582e-01 -4.25064772e-01 2.32554272e-01
-4.99005854e-01 -1.85756058e-01 6.90990269e-01 7.52713621e-01
1.71591789e-01 6.44654334e-02 4.02807444e-01 1.10535038e+00
1.44084990e-01 -6.68929875e-01 4.47059065e-01 2.21386403e-01
-1.97673172e-01 -1.00198054e+00 -4.10688609e-01 3.31584215e-01
-6.40616059e-01 -2.89055705e-01 -6.18783116e-01 5.32909632e-01
-2.39172168e-02 6.89080536e-01 1.77473217e-01 -8.53123307e-01
-8.21832567e-02 1.78568512e-01 5.70458055e-01 -3.70349824e-01
-4.56377745e-01 1.30779937e-01 4.89160508e-01 -4.87217665e-01
-3.18614542e-01 -8.41000736e-01 -8.58610034e-01 -5.96617520e-01
3.65933776e-02 -3.70131016e-01 5.32281578e-01 8.61055434e-01
1.76445514e-01 4.36180204e-01 3.51732194e-01 -1.12062478e+00
-3.85256767e-01 -8.41647685e-01 -7.78519809e-01 6.41914487e-01
1.95408240e-02 -6.24700606e-01 1.91669852e-01 3.48995835e-01] | [10.588286399841309, 1.0012539625167847] |
770a00c6-16c0-46c0-a744-8aaa9f4f67c9 | discourse-parsing-of-contentious-non | 2012.04585 | null | https://arxiv.org/abs/2012.04585v1 | https://arxiv.org/pdf/2012.04585v1.pdf | Discourse Parsing of Contentious, Non-Convergent Online Discussions | Online discourse is often perceived as polarized and unproductive. While some conversational discourse parsing frameworks are available, they do not naturally lend themselves to the analysis of contentious and polarizing discussions. Inspired by the Bakhtinian theory of Dialogism, we propose a novel theoretical and computational framework, better suited for non-convergent discussions. We redefine the measure of a successful discussion, and develop a novel discourse annotation schema which reflects a hierarchy of discursive strategies. We consider an array of classification models -- from Logistic Regression to BERT. We also consider various feature types and representations, e.g., LIWC categories, standard embeddings, conversational sequences, and non-conversational discourse markers learnt separately. Given the 31 labels in the tagset, an average F-Score of 0.61 is achieved if we allow a different model for each tag, and 0.526 with a single model. The promising results achieved in annotating discussions according to the proposed schema paves the way for a number of downstream tasks and applications such as early detection of discussion trajectories, active moderation of open discussions, and teacher-assistive bots. Finally, we share the first labeled dataset of contentious non-convergent online discussions. | ['Oren Tsur', 'Yifat Ben-David Kolikant', 'Dina Grossman', 'Tovit Hakak', 'Omri Hadar', 'Stepan Zakharov'] | 2020-12-08 | null | null | null | null | ['discourse-parsing'] | ['natural-language-processing'] | [ 4.67979237e-02 7.34196961e-01 -4.71572459e-01 -3.06777716e-01
-4.58595663e-01 -9.57331002e-01 1.17390597e+00 5.33110499e-01
-2.70572811e-01 7.17582464e-01 9.15540576e-01 -5.03302932e-01
-1.33390740e-01 -6.30419374e-01 -4.68887575e-02 -8.02174866e-01
1.16992705e-01 4.69775051e-01 1.98618814e-01 -3.47462088e-01
5.91439068e-01 8.75570402e-02 -1.45762169e+00 5.07694781e-01
1.03730667e+00 4.66765761e-01 6.68743923e-02 7.54174352e-01
-6.50152862e-01 1.30584502e+00 -9.43982184e-01 -7.36008763e-01
-4.84782606e-01 -5.20867229e-01 -1.30495346e+00 1.80474654e-01
2.43996054e-01 -3.85723114e-02 -3.66544008e-01 7.26647198e-01
4.13088053e-01 3.21583860e-02 7.36239552e-01 -1.19927204e+00
-6.27503395e-01 1.12537348e+00 -1.53184980e-01 4.53946531e-01
5.03790617e-01 -1.38543636e-01 1.20831776e+00 -4.83120888e-01
1.09193242e+00 1.32270706e+00 5.22366345e-01 7.04944789e-01
-9.25719380e-01 -2.54003763e-01 4.57663275e-02 3.09201539e-01
-3.12423587e-01 -3.25578213e-01 7.51804948e-01 -1.06459045e+00
4.48665291e-01 5.96082866e-01 6.37381732e-01 1.56453919e+00
-2.70349532e-01 8.27842534e-01 1.48851526e+00 -4.64592218e-01
-2.60122363e-02 3.25916171e-01 5.94858229e-01 3.89652640e-01
-6.00016974e-02 -4.11617815e-01 -6.31743610e-01 -3.05477262e-01
-1.18408315e-02 -3.16170424e-01 -1.86056450e-01 5.75257316e-02
-1.11194658e+00 1.10059524e+00 1.59958273e-01 9.45610344e-01
3.50119080e-03 -4.04979259e-01 7.73365676e-01 4.66107547e-01
9.84808862e-01 6.31622553e-01 1.58303101e-02 -7.07596540e-01
-4.92985785e-01 2.88531274e-01 1.27106106e+00 5.25216520e-01
5.69827199e-01 -4.89302278e-01 -6.04641557e-01 1.07673585e+00
-3.31963003e-02 -2.39444360e-01 3.85701537e-01 -1.07470834e+00
5.63787997e-01 9.59498465e-01 5.19228838e-02 -1.02105892e+00
-6.45871878e-01 7.79502690e-02 -4.84273195e-01 -2.46475145e-01
9.16973948e-01 -4.53004986e-01 1.26300424e-01 1.67910111e+00
6.92311943e-01 -6.79931119e-02 8.90781581e-02 6.37395799e-01
1.09141898e+00 6.32006288e-01 -4.04663011e-02 -5.85261881e-01
1.46634817e+00 -1.16401207e+00 -1.09845901e+00 1.60928607e-01
8.54599535e-01 -7.83897221e-01 8.50912273e-01 2.18596116e-01
-8.80746186e-01 -7.74681941e-02 -6.22154891e-01 -3.01878631e-01
-3.75423819e-01 -2.61022925e-01 4.74245936e-01 7.31220961e-01
-8.34480226e-01 4.54010844e-01 -5.23656189e-01 -3.70909065e-01
2.47410834e-01 6.45010471e-02 -2.06641182e-01 4.25878793e-01
-9.29645836e-01 1.09507489e+00 1.65212616e-01 -1.17419429e-01
-3.83133531e-01 -5.36389112e-01 -5.19988060e-01 -1.06460646e-01
3.40068579e-01 -8.53131860e-02 1.23530209e+00 -8.21040690e-01
-1.66845345e+00 1.26104259e+00 -8.12220648e-02 -3.12642008e-01
7.37065017e-01 -1.50806174e-01 -1.09183751e-01 1.32854283e-01
2.45624483e-02 2.42844045e-01 5.24973273e-01 -1.04848492e+00
-6.08104110e-01 -1.39408797e-01 4.08735424e-01 1.45767272e-01
-7.95672417e-01 4.67209101e-01 2.21836671e-01 -3.34442735e-01
-1.92280143e-01 -9.54276919e-01 4.09767553e-02 -2.71440089e-01
-5.73585272e-01 -1.08858800e+00 8.65230203e-01 -9.26214755e-01
1.56829667e+00 -1.93702519e+00 4.56194788e-01 -2.89836437e-01
7.89657235e-01 4.26691890e-01 2.46398330e-01 8.28711092e-01
2.31990471e-01 3.56228113e-01 3.37397866e-02 -2.67506033e-01
2.30217278e-01 1.67383924e-01 -8.44777748e-02 4.53718394e-01
-4.20190219e-04 6.96902454e-01 -1.10789621e+00 -4.98122901e-01
1.67998925e-01 2.36121804e-01 -4.75498617e-01 6.62898779e-01
-4.11121458e-01 8.17883551e-01 -6.47895396e-01 1.25654787e-01
3.42936426e-01 -4.11607295e-01 7.37120509e-01 3.39324951e-01
-6.51607394e-01 8.22205663e-01 -6.16464674e-01 1.23560226e+00
-4.76022303e-01 1.34929633e+00 3.52704585e-01 -1.05024803e+00
9.94376957e-01 4.69153851e-01 4.39525217e-01 -3.00339162e-01
4.08876628e-01 1.50770023e-02 2.55522132e-01 -1.12795997e+00
6.77025855e-01 2.57933825e-01 -1.79148629e-01 8.26022387e-01
1.40117645e-01 2.83871979e-01 3.81874144e-01 1.27918169e-01
1.23615193e+00 -2.16383383e-01 4.75493431e-01 -3.24987352e-01
6.57931805e-01 3.92367356e-02 3.21430236e-01 5.33079624e-01
-4.07942861e-01 2.96364337e-01 1.27265728e+00 -3.70109081e-01
-1.03889084e+00 -5.54423094e-01 -2.85966516e-01 1.69239104e+00
-8.60077702e-03 -5.07063806e-01 -9.46343362e-01 -8.05400133e-01
-9.32193249e-02 5.87269485e-01 -4.85056520e-01 5.03695846e-01
-9.65578079e-01 -6.40343428e-01 6.05536520e-01 -1.56410024e-01
3.52235079e-01 -1.17260647e+00 -2.82473773e-01 3.46900523e-01
-5.31660080e-01 -1.07315707e+00 -1.16165601e-01 -1.40827358e-01
-4.79887098e-01 -1.51933467e+00 -3.75305831e-01 -8.61017168e-01
2.05879301e-01 7.94861689e-02 9.82976615e-01 3.25290382e-01
7.28570893e-02 3.72343451e-01 -5.85391879e-01 1.31001780e-02
-1.09945095e+00 2.83392429e-01 -1.78490490e-01 -5.31193390e-02
3.16557765e-01 -5.74337959e-01 -4.48853076e-01 8.62526149e-02
-3.90150607e-01 1.20706409e-01 -1.68165311e-01 1.00078869e+00
-4.27500814e-01 -7.73612857e-01 7.32283235e-01 -1.23409367e+00
1.17851937e+00 -9.28600907e-01 -1.86526164e-01 3.17995518e-01
-5.28999031e-01 -2.53021181e-01 4.54988718e-01 -7.35563695e-01
-1.33089828e+00 -7.10885346e-01 -4.28582817e-01 1.47146463e-01
-2.15758711e-01 2.20877782e-01 3.23194921e-01 3.03940415e-01
8.43612313e-01 -2.71111518e-01 1.49609029e-01 -5.66194355e-01
5.90116382e-01 1.20444036e+00 2.18971521e-01 -6.92057490e-01
4.49186772e-01 9.43017974e-02 -5.42352021e-01 -1.10121906e+00
-1.03178298e+00 -5.83382487e-01 -4.54401582e-01 -5.30145109e-01
1.08610952e+00 -6.35927081e-01 -1.40369678e+00 3.15504134e-01
-1.55381644e+00 -2.81530023e-01 -3.85568812e-02 1.62657693e-01
-4.72357571e-01 6.54228628e-01 -9.47309375e-01 -1.11635637e+00
-2.72900730e-01 -9.16425407e-01 4.06937838e-01 4.16774541e-01
-7.45919585e-01 -1.28805172e+00 2.43465230e-01 9.47895288e-01
2.88172483e-01 5.48452556e-01 9.27309394e-01 -1.43623805e+00
-1.06519304e-01 4.08787459e-01 -3.20547558e-02 3.85978281e-01
-3.90021168e-02 1.08370103e-01 -1.19425523e+00 -4.07718308e-03
4.91668172e-02 -5.48063815e-01 5.62732816e-01 -5.13372608e-02
8.46364856e-01 -8.38308990e-01 -2.09237382e-01 -9.45455506e-02
6.76032126e-01 2.27277558e-02 2.43939012e-01 4.80109453e-01
5.75556815e-01 1.25286126e+00 3.66933614e-01 4.50500637e-01
7.53215373e-01 4.92255121e-01 2.17317715e-01 4.06551003e-01
-4.37453110e-03 -2.66335249e-01 3.74945760e-01 1.31581533e+00
-1.47385910e-01 -6.29304767e-01 -1.11689031e+00 6.97603703e-01
-1.95269060e+00 -1.15412331e+00 -6.94179058e-01 1.43050992e+00
1.02012801e+00 6.26717061e-02 2.72074908e-01 9.20690447e-02
1.02815115e+00 7.09212482e-01 2.82464898e-03 -8.42207491e-01
-5.47736548e-02 1.37152269e-01 -1.69850171e-01 6.86298251e-01
-9.93095100e-01 8.10736060e-01 5.81493521e+00 6.85875714e-01
-9.10964787e-01 6.18484616e-01 5.13408780e-01 4.21773702e-01
-3.68819654e-01 5.58624597e-05 -6.68738544e-01 6.70865417e-01
1.07079864e+00 -3.28427821e-01 2.18795851e-01 8.54588151e-01
1.28684312e-01 -1.07026704e-01 -1.06222486e+00 6.04726017e-01
4.99723181e-02 -1.71189702e+00 -5.08875430e-01 2.89663449e-02
6.72401845e-01 -1.24550380e-01 -2.61138499e-01 5.42711854e-01
2.96817392e-01 -7.71188319e-01 2.97629178e-01 1.24775507e-01
2.59202480e-01 -1.25409499e-01 5.66611767e-01 7.55507410e-01
-4.62044418e-01 -3.14796358e-01 5.29702939e-02 -5.07202923e-01
3.64303365e-02 2.08521605e-01 -1.18570471e+00 3.45172197e-01
2.99921721e-01 6.47696316e-01 -3.23254138e-01 6.53986752e-01
-3.56397480e-01 1.03258944e+00 -9.59645435e-02 -6.58766448e-01
2.23601505e-01 -3.53518128e-01 7.56309807e-01 1.45318270e+00
1.38941064e-01 4.52772044e-02 2.72658225e-02 6.58009708e-01
-2.29824245e-01 1.66623026e-01 -5.09183943e-01 -3.00984502e-01
8.38298738e-01 1.42657661e+00 -4.64431137e-01 -2.65854001e-01
-3.71748865e-01 3.50277901e-01 7.09363580e-01 -3.38323438e-03
-5.68871677e-01 2.21092343e-01 6.63615167e-01 5.88117652e-02
-3.42537090e-02 -1.03854693e-01 -2.30278552e-01 -1.07309937e+00
-1.24179617e-01 -8.87390018e-01 3.14059407e-01 -1.57479495e-01
-1.61799800e+00 6.33722365e-01 1.61355510e-01 -8.55955422e-01
-6.21043742e-01 -4.62731361e-01 -1.05647206e+00 4.86495316e-01
-1.32015777e+00 -9.42386150e-01 -3.64217609e-01 1.13090962e-01
6.98684275e-01 -1.03850238e-01 8.33360672e-01 2.89696634e-01
-6.74966872e-01 2.76757717e-01 1.65088296e-01 2.67697126e-01
7.18912721e-01 -1.36715508e+00 -2.54547931e-02 2.52529860e-01
-1.38451353e-01 3.29980522e-01 9.61919963e-01 -1.80106655e-01
-1.01839626e+00 -5.50360143e-01 1.38988996e+00 -7.14942396e-01
1.22003508e+00 -5.21342993e-01 -1.02572870e+00 5.06707013e-01
8.02408278e-01 -5.71074009e-01 8.81115258e-01 5.94178438e-01
-2.15656608e-01 4.00483727e-01 -9.92611349e-01 5.48656285e-01
1.10720825e+00 -5.78364968e-01 -1.07603443e+00 8.25324237e-01
6.77854061e-01 -5.70990980e-01 -1.10023320e+00 -6.66363463e-02
3.44069898e-01 -1.26878750e+00 6.22741938e-01 -7.25414753e-01
7.48375714e-01 4.81509507e-01 2.35753223e-01 -9.70832765e-01
8.76951367e-02 -1.04262924e+00 -2.90671915e-01 1.75796628e+00
1.38321280e-01 -4.73374575e-01 7.25811839e-01 4.54652011e-01
-3.54065984e-01 -7.15220630e-01 -1.22701705e+00 -2.95718104e-01
4.43172187e-01 -1.30295157e-01 1.89405411e-01 1.48873889e+00
6.50599062e-01 6.58066332e-01 -3.17043394e-01 -2.81309098e-01
6.44891784e-02 2.14287907e-01 9.09823537e-01 -1.50013947e+00
9.32071134e-02 -6.63366199e-01 -4.35981750e-02 -1.25952411e+00
7.06719100e-01 -9.37865257e-01 -1.34083658e-01 -1.45622563e+00
1.35147721e-01 -6.19422495e-01 3.11823994e-01 8.58461857e-02
-1.03195101e-01 -2.69454181e-01 2.14687571e-01 2.31984571e-01
-5.89932919e-01 4.14761215e-01 1.26409423e+00 -8.09233189e-02
-2.72781253e-01 4.96279597e-02 -5.62221885e-01 9.54967976e-01
7.65326023e-01 -5.35267591e-01 -8.30236152e-02 -4.82810959e-02
1.91936776e-01 3.43133390e-01 1.56941786e-01 -3.98287237e-01
2.51625240e-01 -3.18221807e-01 -5.43195963e-01 -3.99018914e-01
3.37917000e-01 -3.24856639e-01 -2.08818167e-01 1.68336168e-01
-7.50090837e-01 -1.83061227e-01 -3.97153020e-01 4.98911560e-01
-2.49607682e-01 -5.42857051e-01 5.42967796e-01 7.37565830e-02
-2.34499291e-01 -2.53262252e-01 -5.68415821e-01 4.92919981e-01
1.02408445e+00 -5.37943207e-02 -1.04895461e+00 -3.31277728e-01
-1.02616239e+00 3.28162253e-01 1.20887369e-01 6.91737950e-01
-1.10618003e-01 -9.33829069e-01 -7.85064340e-01 -4.07744586e-01
-1.18444130e-01 -2.30110511e-01 2.67008752e-01 9.29984987e-01
-2.61207759e-01 1.75516978e-01 -7.68620670e-02 -6.77984953e-01
-1.54530954e+00 8.17214176e-02 -2.55122557e-02 -4.82877761e-01
-6.10276580e-01 5.62881351e-01 -1.04270175e-01 -6.07283950e-01
2.99529850e-01 -1.02122799e-01 -7.67463982e-01 9.53245640e-01
3.56235325e-01 7.72239864e-01 -3.34281474e-01 -7.17806220e-01
-1.73993148e-02 -1.30919488e-02 1.31166756e-01 1.39889657e-01
1.37533164e+00 -2.90760309e-01 -4.34816569e-01 7.80612111e-01
1.05205548e+00 3.36188048e-01 -9.41709638e-01 -2.84079015e-01
6.07343197e-01 -2.82666177e-01 -4.41822618e-01 -4.11364615e-01
-2.17674121e-01 9.26794946e-01 8.89085457e-02 1.26367891e+00
3.89387667e-01 3.30327660e-01 4.42030728e-01 3.11545610e-01
-1.33569255e-01 -1.07272911e+00 2.98181117e-01 8.27765703e-01
8.71637285e-01 -1.25446379e+00 -1.80988401e-01 -5.42146444e-01
-6.61665857e-01 1.41051579e+00 6.13632083e-01 4.91952151e-03
4.85155463e-01 9.20075178e-02 5.47852134e-03 -3.06440860e-01
-1.02407467e+00 -1.26915455e-01 -7.88153708e-02 5.29704452e-01
9.04965222e-01 1.14722341e-01 -1.00801325e+00 5.11911571e-01
-4.10845131e-01 -3.86484087e-01 8.34047675e-01 5.26658356e-01
-6.13249838e-01 -1.18052733e+00 -2.53818274e-01 3.69665682e-01
-6.10943317e-01 4.66283947e-01 -6.72097266e-01 1.01469529e+00
1.36918232e-01 1.34193885e+00 2.28604838e-01 -3.37200820e-01
2.85963379e-02 3.88904393e-01 1.10400513e-01 -7.66146362e-01
-1.30512166e+00 -2.92181909e-01 1.01237786e+00 2.94632632e-02
-1.05875301e+00 -6.99267268e-01 -8.11409414e-01 -9.33912873e-01
-4.18975711e-01 4.60343033e-01 4.29854780e-01 1.11452723e+00
2.28042416e-02 3.38715285e-01 8.43605578e-01 -6.11957788e-01
-5.19038141e-01 -1.35855794e+00 -8.87123570e-02 5.98013818e-01
3.27617794e-01 -7.17339396e-01 -5.75745225e-01 -1.51278213e-01] | [12.349077224731445, 8.12497615814209] |
da655250-d604-4ef9-8491-6b951e6d0027 | disarm-detecting-the-victims-targeted-by-1 | 2205.05738 | null | https://arxiv.org/abs/2205.05738v1 | https://arxiv.org/pdf/2205.05738v1.pdf | DISARM: Detecting the Victims Targeted by Harmful Memes | Internet memes have emerged as an increasingly popular means of communication on the Web. Although typically intended to elicit humour, they have been increasingly used to spread hatred, trolling, and cyberbullying, as well as to target specific individuals, communities, or society on political, socio-cultural, and psychological grounds. While previous work has focused on detecting harmful, hateful, and offensive memes, identifying whom they attack remains a challenging and underexplored area. Here we aim to bridge this gap. In particular, we create a dataset where we annotate each meme with its victim(s) such as the name of the targeted person(s), organization(s), and community(ies). We then propose DISARM (Detecting vIctimS targeted by hARmful Memes), a framework that uses named entity recognition and person identification to detect all entities a meme is referring to, and then, incorporates a novel contextualized multimodal deep neural network to classify whether the meme intends to harm these entities. We perform several systematic experiments on three test setups, corresponding to entities that are (a) all seen while training, (b) not seen as a harmful target on training, and (c) not seen at all on training. The evaluation results show that DISARM significantly outperforms ten unimodal and multimodal systems. Finally, we show that DISARM is interpretable and comparatively more generalizable and that it can reduce the relative error rate for harmful target identification by up to 9 points absolute over several strong multimodal rivals. | ['Tanmoy Chakraborty', 'Preslav Nakov', 'Md. Shad Akhtar', 'Shivam Sharma'] | 2022-05-11 | null | https://aclanthology.org/2022.findings-naacl.118 | https://aclanthology.org/2022.findings-naacl.118.pdf | findings-naacl-2022-7 | ['person-identification'] | ['computer-vision'] | [-5.64264841e-02 8.14654864e-03 2.97614057e-02 1.14958338e-01
-3.95410508e-01 -8.44954848e-01 9.71096277e-01 5.00743508e-01
-4.78305906e-01 7.19600737e-01 3.70320857e-01 -1.03916086e-01
3.56237441e-01 -7.84758627e-01 -3.59545588e-01 -5.48575699e-01
1.67149961e-01 3.55209976e-01 1.09101571e-01 -3.11615616e-01
4.68716621e-01 3.47679466e-01 -1.03788161e+00 2.52655864e-01
7.68280923e-01 6.41567647e-01 -6.26699269e-01 4.15366501e-01
-1.40372710e-02 1.04440176e+00 -1.10978699e+00 -1.04622293e+00
-3.48936945e-01 -4.14358944e-01 -1.05371344e+00 -5.15719168e-02
5.52237689e-01 -1.72181979e-01 -5.19291043e-01 1.14452326e+00
5.99686205e-01 6.33502901e-02 7.51400888e-01 -1.15814817e+00
-8.91287088e-01 5.81786454e-01 -6.52516127e-01 2.87783831e-01
6.20388269e-01 1.59045443e-01 6.44691110e-01 -7.70010948e-01
7.04462647e-01 1.35742176e+00 5.08069217e-01 7.41043389e-01
-1.15819073e+00 -6.83002830e-01 -1.66427508e-01 -2.97095510e-03
-1.37148106e+00 -3.36151898e-01 7.86463797e-01 -6.79976285e-01
3.45594972e-01 3.88260543e-01 2.29291946e-01 1.79322231e+00
-5.93977347e-02 6.48329794e-01 8.70700002e-01 -1.08682655e-01
7.65824392e-02 5.29942691e-01 3.33634824e-01 6.28835082e-01
1.73339814e-01 -2.99626976e-01 -6.09761536e-01 -5.66049814e-01
5.32953404e-02 -2.41330877e-01 -3.91902477e-01 1.12667829e-01
-1.10769796e+00 1.06388664e+00 6.62918568e-01 4.67875212e-01
-3.90228242e-01 -2.42991056e-02 6.18556619e-01 3.75186466e-02
5.47697604e-01 7.01226830e-01 3.34295839e-01 -8.08015317e-02
-5.90320647e-01 1.72772944e-01 1.16587925e+00 2.81060249e-01
5.13929546e-01 -2.09418043e-01 -3.11070442e-01 9.48521495e-01
-1.27977997e-01 6.45511448e-01 1.85101852e-01 -5.62554181e-01
4.99267191e-01 1.05545151e+00 3.49631965e-01 -1.76969123e+00
-5.36847651e-01 -2.97067463e-01 -7.81711459e-01 -3.26532334e-01
3.04504335e-01 -3.96995842e-01 -5.27984083e-01 1.89292943e+00
4.49689507e-01 -9.32975337e-02 -1.35955781e-01 1.02487278e+00
1.14501715e+00 6.79239988e-01 4.87254947e-01 1.09273426e-01
1.34458888e+00 -6.58234835e-01 -6.04092538e-01 -2.28693470e-01
7.05347180e-01 -6.63264275e-01 7.76865840e-01 1.42444326e-02
-8.92891049e-01 3.07461880e-02 -6.47621989e-01 1.63852144e-02
-9.85046029e-01 -1.50874361e-01 1.93233401e-01 1.02353239e+00
-7.23986506e-01 3.12197059e-01 -2.74989635e-01 -8.61306667e-01
3.14421684e-01 1.17458232e-01 -6.86519504e-01 1.96129531e-01
-1.58630371e+00 1.20008171e+00 4.07624304e-01 -4.30186950e-02
-7.19308674e-01 -3.66997600e-01 -8.39919746e-01 1.33547690e-02
3.14908564e-01 -5.40512979e-01 6.30176425e-01 -1.03435814e+00
-9.23176229e-01 1.33183920e+00 1.10578515e-01 -2.31462792e-01
3.29257607e-01 -2.43298620e-01 -8.55343580e-01 2.41295397e-01
2.61024177e-01 5.72161078e-01 8.15456748e-01 -1.46804476e+00
-2.48499528e-01 -5.28644204e-01 3.27914894e-01 1.52574569e-01
-1.08880591e+00 5.20439684e-01 -1.91607818e-01 -5.67388535e-01
-3.67436677e-01 -9.01475370e-01 4.46653813e-01 -4.41771001e-01
-1.22208607e+00 -2.43665338e-01 1.16801512e+00 -8.73556674e-01
1.58315825e+00 -2.10134840e+00 3.54765683e-01 1.80225760e-01
6.25725746e-01 6.03505373e-01 -8.08176845e-02 9.51663852e-01
8.70242193e-02 6.78959727e-01 -2.96704262e-01 -2.83713520e-01
1.21509537e-01 -2.47136921e-01 -1.98985443e-01 6.11988723e-01
1.61162525e-01 7.70083785e-01 -8.48352551e-01 -3.52777541e-01
-1.06863663e-01 5.63471258e-01 -8.60754177e-02 8.34997669e-02
2.54158348e-01 3.10274243e-01 -3.74955267e-01 5.74570954e-01
4.84422624e-01 -2.89552480e-01 1.42879501e-01 -7.44939595e-02
7.11676851e-02 6.91330284e-02 -7.23143399e-01 5.50773263e-01
-9.88078490e-02 9.33249950e-01 2.13627204e-01 -4.54576999e-01
8.79932344e-01 2.40700886e-01 9.41956565e-02 -5.20647168e-01
2.79613256e-01 1.17940158e-01 -3.28145325e-01 -6.96796298e-01
5.60068071e-01 -1.85901910e-01 -4.21656221e-01 6.64526999e-01
-1.72852382e-01 6.85291052e-01 1.84660882e-01 5.07577658e-01
1.33930838e+00 -4.68670815e-01 2.74832606e-01 1.20528266e-01
7.33019173e-01 -8.85200128e-02 1.76152766e-01 5.88431180e-01
-5.57401896e-01 3.63612354e-01 9.15541828e-01 -3.70806307e-01
-8.05202127e-01 -1.02789867e+00 1.16569147e-01 1.40120566e+00
5.29194117e-01 -1.38494775e-01 -7.90672481e-01 -1.11253142e+00
9.17407498e-02 9.16962802e-01 -7.88161457e-01 -3.90478015e-01
-4.03761089e-01 -6.65531635e-01 1.01776111e+00 5.87516241e-02
7.15517938e-01 -1.25001872e+00 -2.77105659e-01 -1.09116346e-01
-5.33635974e-01 -9.53194439e-01 -5.76365113e-01 -4.19427127e-01
-9.80874449e-02 -1.30185425e+00 -7.73766994e-01 -5.32858670e-01
5.76426744e-01 1.19006976e-01 1.01928937e+00 2.63797075e-01
1.84767500e-01 4.75897282e-01 -2.91245490e-01 -7.73593709e-02
-4.82430845e-01 3.87337595e-01 1.38839379e-01 4.44306880e-01
5.72254181e-01 -2.56612599e-01 -4.03549284e-01 4.37631279e-01
-1.03380346e+00 -3.25578034e-01 1.46488696e-01 5.70169985e-01
-5.35677493e-01 -1.79350421e-01 5.79206169e-01 -1.19946098e+00
1.01615524e+00 -1.02167428e+00 2.14265093e-01 3.17976892e-01
1.84089839e-02 -5.24730563e-01 5.83620369e-01 -6.59734368e-01
-8.85053694e-01 -4.44434166e-01 4.23733741e-02 -2.04929352e-01
-4.57569838e-01 5.15457153e-01 -4.66317050e-02 -1.21769384e-01
9.25588191e-01 8.73812381e-03 -1.88163295e-01 -3.41350019e-01
1.23312645e-01 9.81252432e-01 2.86828130e-01 -3.93711537e-01
1.08511293e+00 4.48919624e-01 -2.86547750e-01 -9.75558102e-01
-8.62101614e-01 -5.54928720e-01 -3.39335084e-01 -5.53219736e-01
9.87889528e-01 -5.19172549e-01 -1.12578070e+00 6.04641318e-01
-1.41006696e+00 1.17719218e-01 6.56411350e-01 2.71879397e-02
1.93673462e-01 4.65534031e-01 -8.46131504e-01 -9.10536051e-01
-4.43615347e-01 -5.82635641e-01 7.72605896e-01 2.66427040e-01
-5.86270034e-01 -1.28315234e+00 1.88832864e-01 6.69927061e-01
3.76993686e-01 6.61545515e-01 9.69522417e-01 -1.09134650e+00
3.09937447e-02 -4.78997499e-01 -4.68019128e-01 5.65754361e-02
-6.50954321e-02 -5.16427867e-02 -8.41660678e-01 -2.25784108e-01
-3.29393923e-01 -5.76219320e-01 6.65718138e-01 -4.52804655e-01
4.84939337e-01 -7.14424610e-01 -5.62561214e-01 -1.36167750e-01
1.09845805e+00 -2.72619631e-02 6.79941416e-01 4.64791983e-01
7.65178859e-01 1.01319659e+00 -3.04737724e-02 3.66197914e-01
3.35554689e-01 6.48452520e-01 5.82691669e-01 9.07157362e-03
1.68340594e-01 -3.34798753e-01 6.09065771e-01 4.04100329e-01
3.55814993e-02 -5.77633798e-01 -1.24949467e+00 5.79687953e-01
-1.70243895e+00 -1.31262541e+00 -2.82645464e-01 1.96372080e+00
5.45847416e-01 -1.40102804e-01 4.49608147e-01 -2.88753748e-01
1.32501519e+00 3.52326512e-01 -3.96073848e-01 -4.04787719e-01
-2.31910542e-01 -2.30729833e-01 1.75803214e-01 2.94495940e-01
-1.33705044e+00 9.89786386e-01 5.56214809e+00 6.19859755e-01
-1.17313802e+00 3.26891810e-01 6.49218798e-01 -6.49344251e-02
-1.07167810e-01 -4.28129852e-01 -5.73851109e-01 7.40012467e-01
8.32949400e-01 -5.98489717e-02 3.79424244e-01 5.26651084e-01
-5.05582206e-02 -7.37412497e-02 -8.72464538e-01 9.70345020e-01
5.49900413e-01 -1.04086483e+00 4.94323932e-02 6.74228370e-02
6.11453474e-01 -3.49259377e-01 2.03764200e-01 5.21808207e-01
-5.08966222e-02 -1.00688910e+00 6.86051667e-01 4.61977899e-01
3.82348895e-01 -7.25546658e-01 9.34286356e-01 4.08435822e-01
-5.77632666e-01 -1.59120470e-01 -2.59577893e-02 9.84746888e-02
1.64449230e-01 2.74464935e-01 -4.51782614e-01 1.57430202e-01
5.99904597e-01 3.88971686e-01 -9.17377889e-01 8.93212318e-01
-3.94569844e-01 7.10672975e-01 -1.70136362e-01 -4.28389877e-01
4.05365855e-01 1.26769289e-01 8.91641617e-01 1.32960021e+00
9.96633898e-03 9.03807953e-02 7.32616559e-02 9.68674183e-01
-4.13381785e-01 1.25553787e-01 -8.47719431e-01 -5.86006165e-01
4.96923327e-01 1.41798556e+00 -5.28852522e-01 -5.56601584e-02
-9.24513340e-02 1.20853043e+00 5.22908330e-01 6.00232065e-01
-9.91005063e-01 -5.60826004e-01 3.92218828e-01 1.69593096e-01
-1.13119259e-01 7.98280630e-03 -6.42579570e-02 -1.02651227e+00
-2.07621604e-01 -7.69743443e-01 5.02418697e-01 -7.10288942e-01
-1.68360806e+00 5.88181496e-01 -3.59382957e-01 -7.84676552e-01
8.48129466e-02 -4.69655663e-01 -7.44426489e-01 7.68759549e-01
-8.23182106e-01 -1.27686179e+00 -2.85474122e-01 3.32504809e-01
-1.71787396e-01 -1.95845693e-01 6.91691995e-01 3.95447612e-01
-1.10168576e+00 5.99547029e-01 -2.55225688e-01 7.44737089e-01
6.89076781e-01 -8.45588684e-01 -1.81037813e-01 6.05786979e-01
-2.21258432e-01 8.69394422e-01 7.68797278e-01 -7.48936534e-01
-1.03696012e+00 -9.43545461e-01 1.15628362e+00 -8.75940323e-01
9.67299819e-01 -3.49461436e-01 -8.69119465e-01 5.48172593e-01
4.73435789e-01 -4.92152065e-01 8.34013522e-01 3.12465370e-01
-6.61797702e-01 2.73745775e-01 -1.31927621e+00 7.43822336e-01
8.15774381e-01 -7.14026392e-01 -4.41108525e-01 4.46170658e-01
4.21750128e-01 -1.00069702e-01 -7.75175631e-01 2.10115425e-02
2.69785404e-01 -1.12994695e+00 6.32170439e-01 -1.02545297e+00
8.50746632e-01 8.75735581e-02 8.02943483e-02 -1.18125224e+00
-2.68634021e-01 -5.05589366e-01 -2.32778713e-01 1.61892271e+00
3.78505349e-01 -6.78092062e-01 6.52013958e-01 7.14397073e-01
1.68131724e-01 -5.42713523e-01 -8.71429563e-01 -4.45203662e-01
8.60207081e-02 7.85088092e-02 2.29135156e-01 1.60215151e+00
2.89535373e-01 7.70883381e-01 -6.68763578e-01 2.43052185e-01
6.55345976e-01 -1.89611942e-01 7.16574192e-01 -1.10541463e+00
3.60318393e-01 -6.78297818e-01 -4.38792109e-01 -4.81438249e-01
3.31705540e-01 -7.54375756e-01 -3.29554945e-01 -1.44132996e+00
6.36136711e-01 2.74057481e-02 -8.00080299e-02 6.30520523e-01
-2.28295118e-01 6.12528920e-01 3.88413548e-01 2.70599842e-01
-7.83863425e-01 3.56493056e-01 8.00235093e-01 -3.58474106e-01
-1.28758088e-01 -2.28242084e-01 -7.45924652e-01 6.55688882e-01
6.37473881e-01 -3.70738596e-01 2.23223910e-01 -9.27013084e-02
6.24092937e-01 -4.97163199e-02 7.25772202e-01 -7.95776129e-01
1.43239394e-01 -3.07061393e-02 2.32951149e-01 -2.44266748e-01
3.82219255e-01 -5.70330441e-01 -5.63222505e-02 3.97953033e-01
-3.44699800e-01 -1.62025273e-01 1.20070711e-01 4.73591089e-01
-2.25820467e-01 -3.18676561e-01 8.99077296e-01 8.11957940e-02
-5.60848773e-01 8.96424130e-02 -5.21866262e-01 2.31753364e-01
1.22081220e+00 -3.06642726e-02 -9.64462280e-01 -7.24083006e-01
-7.60345757e-01 2.38108188e-01 4.35247004e-01 5.33883989e-01
5.03210962e-01 -1.25713074e+00 -7.92064428e-01 -3.87146980e-01
2.66181707e-01 -1.05494416e+00 3.60605627e-01 9.18265224e-01
-2.86528647e-01 3.00751120e-01 -4.96217906e-02 -1.84251457e-01
-1.21974158e+00 6.27232969e-01 3.96946430e-01 -2.66679027e-03
-2.77585804e-01 6.74780905e-01 1.50038630e-01 -6.90486789e-01
2.51847237e-01 9.88784790e-01 -5.58209658e-01 4.73062932e-01
5.69661856e-01 7.16486692e-01 -2.93676049e-01 -1.38787901e+00
-4.76613879e-01 1.94564596e-01 -4.29646634e-02 2.26523384e-01
8.76605451e-01 -9.23688859e-02 -5.72968721e-01 4.17555809e-01
1.23319423e+00 1.94189832e-01 -2.83085197e-01 -8.61105993e-02
1.63582578e-01 -3.25457752e-01 -3.37191731e-01 -1.00227523e+00
-9.17336941e-01 8.49217296e-01 3.29830408e-01 9.03987050e-01
6.49153113e-01 1.68615162e-01 1.05975544e+00 2.68508434e-01
1.73091128e-01 -8.36431563e-01 4.16441321e-01 5.80120444e-01
9.96288836e-01 -1.19052434e+00 -3.34332734e-01 -4.37757045e-01
-8.19001973e-01 9.72851455e-01 8.36330771e-01 2.56278157e-01
2.93115646e-01 -4.99651939e-01 2.44540181e-02 -4.97777909e-01
-3.89734358e-01 -1.21326715e-01 3.79558742e-01 4.14328903e-01
3.17953527e-01 3.96738052e-02 -4.04786319e-01 3.79870474e-01
1.17642984e-01 -5.84847808e-01 5.65933883e-01 5.41563511e-01
-5.21037817e-01 -4.74635601e-01 -6.89242899e-01 2.58378774e-01
-5.78474462e-01 5.26983291e-02 -1.29023325e+00 8.54666770e-01
1.17526248e-01 1.09545755e+00 -1.65987372e-01 -6.61012769e-01
2.58284688e-01 2.00031787e-01 -1.05682977e-01 -5.86845279e-01
-1.02831376e+00 -5.57179153e-01 5.05088985e-01 -2.58844256e-01
-3.09888124e-01 -2.93851227e-01 -9.41026330e-01 -9.77613449e-01
-2.44577482e-01 3.70179564e-02 4.19529498e-01 9.13500845e-01
3.74339193e-01 -5.91722876e-02 6.49565458e-01 -6.03349507e-01
-2.52266914e-01 -1.08751535e+00 -4.94693130e-01 9.54790235e-01
5.91455400e-02 -7.66132712e-01 -5.73005378e-01 -3.18564683e-01] | [8.498302459716797, 10.668750762939453] |
c6c5fd91-93bb-43b4-9c91-8fd2ec422cfe | optical-flow-estimation-in-360-circ-videos | 2301.11880 | null | https://arxiv.org/abs/2301.11880v1 | https://arxiv.org/pdf/2301.11880v1.pdf | Optical Flow Estimation in 360$^\circ$ Videos: Dataset, Model and Application | Optical flow estimation has been a long-lasting and fundamental problem in the computer vision community. However, despite the advances of optical flow estimation in perspective videos, the 360$^\circ$ videos counterpart remains in its infancy, primarily due to the shortage of benchmark datasets and the failure to accommodate the omnidirectional nature of 360$^\circ$ videos. We propose the first perceptually realistic 360$^\circ$ filed-of-view video benchmark dataset, namely FLOW360, with 40 different videos and 4,000 video frames. We then conduct comprehensive characteristic analysis and extensive comparisons with existing datasets, manifesting FLOW360's perceptual realism, uniqueness, and diversity. Moreover, we present a novel Siamese representation Learning framework for Omnidirectional Flow (SLOF) estimation, which is trained in a contrastive manner via a hybrid loss that combines siamese contrastive and optical flow losses. By training the model on random rotations of the input omnidirectional frames, our proposed contrastive scheme accommodates the omnidirectional nature of optical flow estimation in 360$^\circ$ videos, resulting in significantly reduced prediction errors. The learning scheme is further proven to be efficient by expanding our siamese learning scheme and omnidirectional optical flow estimation to the egocentric activity recognition task, where the classification accuracy is boosted up to $\sim$26%. To summarize, we study the optical flow estimation in 360$^\circ$ videos problem from perspectives of the benchmark dataset, learning model, and also practical application. The FLOW360 dataset and code are available at https://siamlof.github.io. | ['Yan Yan', 'Gaowen Liu', 'Keshav Bhandari', 'Bin Duan'] | 2023-01-27 | null | null | null | null | ['egocentric-activity-recognition'] | ['computer-vision'] | [-2.10401654e-01 -5.04773498e-01 -1.84526280e-01 -4.61953916e-02
-2.23649636e-01 -5.15329778e-01 3.40911716e-01 -6.37832344e-01
-3.91790360e-01 8.16635132e-01 2.19390407e-01 -2.65638262e-01
-3.07826132e-01 -5.71987510e-01 -6.50893807e-01 -7.17201531e-01
-3.92329603e-01 -2.58022487e-01 -1.16669573e-01 -1.08102672e-01
3.85993183e-01 4.44764465e-01 -1.47074544e+00 -2.04053909e-01
7.54245222e-01 1.19750166e+00 2.01351300e-01 1.01016140e+00
3.52796763e-01 1.37783742e+00 -2.74909824e-01 -4.77584183e-01
6.52427793e-01 -4.08465385e-01 -7.44180501e-01 1.86786931e-02
9.83406901e-01 -6.56191647e-01 -9.85565543e-01 9.27488387e-01
5.09766102e-01 4.45787042e-01 3.10896486e-01 -1.35664523e+00
-5.33745706e-01 -2.39164429e-03 -5.71210861e-01 7.48249352e-01
5.87508857e-01 3.90769809e-01 1.21735382e+00 -1.03871381e+00
8.83125782e-01 1.01372576e+00 2.33691573e-01 4.92475867e-01
-8.56690943e-01 -7.35529423e-01 2.46275380e-01 6.24164522e-01
-1.18323076e+00 -5.32505333e-01 9.17321086e-01 -6.59970284e-01
8.28715563e-01 2.31714658e-02 7.48967946e-01 9.49753463e-01
1.10835835e-01 1.05839360e+00 6.07618868e-01 -2.36661527e-02
1.97578017e-02 -1.97302192e-01 -3.96492571e-01 1.02889013e+00
6.89318478e-02 2.99288690e-01 -6.48032010e-01 3.32402885e-01
9.03635681e-01 -1.08002767e-01 -7.35547841e-01 -7.10229874e-01
-1.29581428e+00 7.12960184e-01 5.34694552e-01 -1.64830387e-01
-5.07410169e-02 2.60703385e-01 5.78802347e-01 2.90659875e-01
2.55731940e-01 3.42136234e-01 -2.61151105e-01 -4.79449034e-01
-7.07824707e-01 2.70085365e-01 4.71143663e-01 1.06811798e+00
7.72334278e-01 6.33616626e-01 1.91301867e-01 7.06227720e-01
2.88633585e-01 6.30414963e-01 4.21389341e-01 -1.61098063e+00
6.64268136e-01 2.45154142e-01 1.91135988e-01 -1.47123885e+00
-2.36693263e-01 -6.37539208e-01 -9.58595991e-01 2.15404794e-01
5.43470204e-01 -1.91288069e-01 -1.01543166e-01 1.90027809e+00
2.46896446e-01 3.76148313e-01 7.48270005e-02 1.12535203e+00
6.77100241e-01 6.53382659e-01 -2.04384238e-01 -3.34466904e-01
8.73965442e-01 -1.26084006e+00 -4.58286375e-01 -6.90039992e-02
6.06253445e-01 -8.65877628e-01 8.93908560e-01 5.25784016e-01
-1.22414744e+00 -7.32198954e-01 -9.08573866e-01 -1.55289039e-01
5.12015680e-03 2.16682982e-02 6.91081107e-01 5.01693249e-01
-1.05300725e+00 3.35998982e-01 -6.03584766e-01 -1.61112785e-01
5.76528549e-01 1.61638960e-01 -4.85972017e-01 -4.34746027e-01
-1.09231150e+00 4.41771895e-01 2.85170712e-02 1.79237351e-01
-9.37793612e-01 -1.14712238e+00 -1.27988648e+00 -1.30386561e-01
4.04120147e-01 -8.16930950e-01 7.10118115e-01 -8.56174886e-01
-1.34178019e+00 6.14940345e-01 -2.46302083e-01 -2.22482249e-01
8.46008480e-01 -2.99759597e-01 -3.64705563e-01 5.94150484e-01
3.07527572e-01 6.35445178e-01 8.64306986e-01 -9.42543864e-01
-8.01946223e-01 -1.02857865e-01 5.01730144e-01 3.03287745e-01
-1.78161949e-01 -2.32909977e-01 -4.28561747e-01 -8.29380155e-01
-2.18364745e-01 -8.80189121e-01 -1.84528790e-02 3.86545211e-01
-6.41768128e-02 1.83144715e-02 8.52883756e-01 -4.81146306e-01
1.21355414e+00 -2.09333706e+00 1.86788976e-01 -2.51203150e-01
3.44755590e-01 4.86014664e-01 -1.61514655e-01 2.24486724e-01
-1.65161952e-01 -2.77359575e-01 -5.44324853e-02 -2.48595417e-01
-2.91711658e-01 6.44601583e-02 -2.51618475e-01 6.94558084e-01
1.21993460e-02 7.99596190e-01 -1.32117426e+00 -5.90920329e-01
7.37270713e-01 5.09571135e-01 -1.08801651e+00 3.59634817e-01
3.73327494e-01 6.91682935e-01 -3.07603687e-01 6.91923380e-01
8.14601302e-01 -2.35844895e-01 -5.59341311e-02 -4.87074137e-01
-1.98407874e-01 -2.13750809e-01 -1.17297661e+00 1.99736059e+00
-6.13990784e-01 9.21419621e-01 1.45907938e-01 -1.35731673e+00
6.49558544e-01 1.58445582e-01 1.10676634e+00 -8.83387208e-01
1.90557335e-02 2.39035979e-01 -2.07632482e-02 -7.27390170e-01
5.03595114e-01 -7.55248824e-03 1.64669961e-01 2.49913037e-01
1.90649867e-01 -8.84783268e-02 7.46343017e-01 3.36360484e-01
9.35738623e-01 5.33112526e-01 2.83035308e-01 -2.48702496e-01
1.11096692e+00 -3.57849211e-01 7.02607989e-01 5.32145083e-01
-8.25333357e-01 6.66398823e-01 3.87021363e-01 -6.92473233e-01
-8.13567221e-01 -1.24533844e+00 -1.42688170e-01 7.47267485e-01
3.93570036e-01 -3.41587812e-01 -4.42984700e-01 -6.30604148e-01
-1.18760698e-01 5.99838756e-02 -4.86300766e-01 -5.83854355e-02
-9.83630240e-01 -5.46213150e-01 2.27891788e-01 4.06820506e-01
8.39735806e-01 -1.02111483e+00 -6.98725700e-01 1.31887838e-01
-5.75208247e-01 -1.52210689e+00 -8.06737483e-01 -6.01769924e-01
-6.95774615e-01 -1.36381507e+00 -7.98734009e-01 -6.39116943e-01
4.42689955e-01 6.55185640e-01 1.11054695e+00 -2.19321176e-01
-5.52450836e-01 6.33613527e-01 -2.32100263e-01 2.33034208e-01
8.56938213e-02 -2.01839164e-01 1.17248133e-01 3.77379417e-01
1.80896986e-02 -7.53067493e-01 -1.41035998e+00 4.80893433e-01
-8.73368859e-01 -1.98367655e-01 3.18544030e-01 8.22604895e-01
3.06214839e-01 -1.74176201e-01 3.65333945e-01 -4.26855713e-01
-2.18636561e-02 -5.74525595e-01 -4.36490864e-01 -3.01854700e-01
-3.61868054e-01 -3.10439438e-01 1.15511286e+00 -5.66152371e-02
-9.98977363e-01 -1.38855100e-01 -1.15071125e-01 -8.68090689e-01
1.60481781e-01 4.19115201e-02 -8.94987136e-02 -2.71994680e-01
4.21209306e-01 3.18788648e-01 6.29154593e-02 -4.46380004e-02
2.74918467e-01 1.99694529e-01 6.21019065e-01 -5.12474656e-01
6.84183121e-01 9.27807748e-01 2.19204247e-01 -1.07915175e+00
-7.31660306e-01 -6.85573578e-01 -3.97064149e-01 -5.64980686e-01
6.69277251e-01 -1.06907785e+00 -1.29088974e+00 5.52745640e-01
-8.56890202e-01 -1.85748532e-01 -3.16732705e-01 9.54204261e-01
-1.05833936e+00 7.32049346e-01 -5.29199481e-01 -4.79427010e-01
-1.45530596e-01 -1.29638529e+00 6.89356863e-01 1.60894722e-01
1.57880321e-01 -1.19853973e+00 1.26241997e-01 6.88212216e-01
3.36129576e-01 2.19475985e-01 5.21895707e-01 3.87472093e-01
-8.31455529e-01 1.61089376e-01 -4.62406069e-01 6.47833288e-01
1.26592398e-01 5.44737466e-03 -8.77619922e-01 -6.01906836e-01
-1.08106650e-01 -3.35497409e-01 7.91312039e-01 6.27479792e-01
1.19288862e+00 -1.57036066e-01 1.14479154e-01 1.11327338e+00
1.61331546e+00 3.66284728e-01 6.33007348e-01 1.76908091e-01
8.41732621e-01 5.99829614e-01 5.63309014e-01 5.38841367e-01
4.96635616e-01 7.02958524e-01 8.53765845e-01 1.04262739e-01
-3.36275250e-01 -3.05507809e-01 4.20797914e-01 9.06901062e-01
-4.92773414e-01 -3.17503273e-01 -4.75618571e-01 6.46155775e-01
-1.59220541e+00 -1.32215166e+00 2.52692461e-01 2.10161400e+00
2.90741473e-01 -2.05797866e-01 2.10985616e-02 2.38137439e-01
4.95253146e-01 7.42273152e-01 -5.65161169e-01 -2.55176455e-01
-1.06686398e-01 3.18994112e-02 3.00122082e-01 7.78589845e-01
-1.19361734e+00 8.40571404e-01 5.34305716e+00 7.35993624e-01
-1.31760323e+00 -2.04908028e-02 5.48596203e-01 -2.95733303e-01
-9.09003168e-02 -1.49009719e-01 -5.39591610e-01 6.39358401e-01
4.64708626e-01 -8.02950338e-02 5.07779658e-01 7.17086911e-01
5.88760912e-01 -9.80135873e-02 -8.28028202e-01 1.47110879e+00
2.91636318e-01 -1.49763906e+00 2.33993158e-01 7.19836876e-02
8.20401669e-01 -1.04158428e-02 2.34262213e-01 1.15270108e-01
-1.22215133e-02 -8.30182016e-01 6.44558847e-01 3.47511351e-01
1.01555359e+00 -5.86856425e-01 3.72330964e-01 -2.90443115e-02
-1.41381657e+00 -4.15042728e-01 -2.72531331e-01 -3.04497957e-01
4.25090045e-01 3.47689241e-01 3.63028771e-03 7.33032465e-01
1.06062412e+00 1.64854372e+00 -1.95351243e-01 8.88017952e-01
7.87637681e-02 2.05031887e-01 1.37099892e-01 2.62645900e-01
2.78923482e-01 -5.98673999e-01 6.89173937e-01 1.08415604e+00
3.65237117e-01 4.25201729e-02 2.11815555e-02 5.85046172e-01
-6.35048598e-02 1.39646351e-01 -7.63882399e-01 3.74066472e-01
-2.04803292e-02 1.17892385e+00 -2.14631513e-01 -1.65896490e-01
-6.78075671e-01 7.84772933e-01 2.45700836e-01 5.95233500e-01
-8.85800660e-01 -4.45502877e-01 1.19368529e+00 2.37028804e-02
2.90841758e-01 -4.08247888e-01 2.17126980e-01 -1.89204061e+00
2.60776669e-01 -6.02256060e-01 3.54155451e-01 -7.93675601e-01
-9.46216464e-01 5.46319187e-01 -1.98790774e-01 -1.71544385e+00
-3.16645116e-01 -8.34892154e-01 -4.26588953e-01 4.55249041e-01
-2.00567865e+00 -6.84708357e-01 -6.91034138e-01 9.03034449e-01
7.81424642e-01 -1.37522683e-01 3.17310482e-01 7.34041035e-01
-5.72180986e-01 5.86223423e-01 1.68208271e-01 3.85977000e-01
6.40673459e-01 -1.03213966e+00 -8.40745270e-02 1.01986098e+00
-3.12757934e-03 4.55506295e-01 4.76266414e-01 1.12628572e-01
-1.56565726e+00 -1.10701835e+00 6.31517291e-01 -2.48908028e-01
7.12268054e-01 -1.46736726e-01 -4.13425028e-01 6.13584340e-01
8.17095190e-02 7.56936729e-01 4.71643150e-01 -6.80774808e-01
-3.04582357e-01 -5.18865287e-01 -1.03615880e+00 6.85625494e-01
1.51624763e+00 -4.56400305e-01 8.00484493e-02 5.46169207e-02
3.55898082e-01 -2.69641370e-01 -8.96220446e-01 4.49197233e-01
8.60170782e-01 -1.40745723e+00 1.37812364e+00 -4.17509437e-01
7.75608122e-01 -3.32996517e-01 -3.60442102e-01 -1.16009784e+00
-1.26067266e-01 -8.92304182e-01 -2.70936042e-01 7.97468722e-01
-6.61316290e-02 -6.49796426e-01 9.65375721e-01 -9.68004856e-03
-1.85388088e-01 -7.27392435e-01 -9.69682455e-01 -6.00662053e-01
1.16114534e-01 -5.82052290e-01 -8.77243280e-02 9.02976692e-01
-1.00709490e-01 -3.10181286e-02 -6.68565750e-01 -1.72900915e-01
8.37617457e-01 1.06263816e-01 8.34418595e-01 -6.70637786e-01
-2.93605953e-01 -4.95319813e-01 -7.12227225e-01 -1.67546260e+00
2.29842916e-01 -7.16699898e-01 -2.23150924e-01 -1.03409755e+00
-7.34860897e-02 -5.02317667e-01 -2.10071683e-01 -2.79486686e-01
-4.87920456e-02 6.61306798e-01 4.02387559e-01 2.35313028e-01
-5.10073185e-01 6.79791927e-01 1.80070806e+00 -1.77164122e-01
1.24531686e-01 -5.87578639e-02 -5.14302790e-01 6.00585282e-01
5.34289002e-01 1.64783016e-01 -8.24204624e-01 -5.39897144e-01
2.33035445e-01 4.11939144e-01 4.69748408e-01 -1.07393360e+00
6.99066967e-02 -2.04056799e-01 1.83222249e-01 -3.16880643e-02
5.45205414e-01 -5.92423141e-01 -2.27084816e-01 5.29602170e-01
-1.62098885e-01 1.83369562e-01 -1.70498848e-01 6.87183082e-01
-5.61540067e-01 2.02606156e-01 9.94078100e-01 -2.37953991e-01
-9.70566690e-01 8.79783452e-01 -2.26000965e-01 5.73456943e-01
1.02391684e+00 -3.99056405e-01 -3.62045884e-01 -5.29477358e-01
-5.98823667e-01 1.06790632e-01 3.29311937e-01 5.71598291e-01
7.09199905e-01 -1.16199231e+00 -6.73297107e-01 4.90682989e-01
1.05277829e-01 -4.23105098e-02 6.38752699e-01 1.00502145e+00
-1.01136494e+00 5.49986005e-01 -4.91823763e-01 -7.90208399e-01
-7.79510140e-01 7.65301645e-01 5.17701387e-01 -9.57296416e-02
-6.77837849e-01 6.36390984e-01 6.64305568e-01 -2.02963725e-01
4.15780284e-02 -9.15777609e-02 -3.53390157e-01 -1.63984686e-01
5.28171062e-01 8.51323068e-01 -2.71784067e-01 -1.01652634e+00
-3.27069730e-01 1.15321445e+00 2.36004159e-01 1.21627890e-01
1.05303335e+00 -5.59159636e-01 2.00798109e-01 -5.71534038e-02
1.74745917e+00 1.28317505e-01 -1.75367904e+00 -1.74152449e-01
-6.70887768e-01 -1.06316435e+00 -2.37276465e-01 -1.38382390e-01
-1.71299982e+00 1.09825909e+00 4.38595146e-01 -1.65996432e-01
1.15601432e+00 -3.07216942e-01 9.24582720e-01 -4.82161194e-02
3.65762919e-01 -6.95659280e-01 3.82274032e-01 3.72174174e-01
6.36912525e-01 -1.47415042e+00 -9.57360864e-02 -5.10609090e-01
-4.33048069e-01 1.06377542e+00 6.99579775e-01 -4.67934668e-01
9.18490291e-01 -9.71446559e-02 1.12964034e-01 6.14903457e-02
-5.72067142e-01 6.61393031e-02 -1.37568917e-02 6.09358490e-01
3.22529942e-01 -3.00192416e-01 -2.49257743e-01 -3.29517275e-01
-2.10926980e-01 1.22338906e-01 7.83717155e-01 7.31628835e-01
4.16407138e-02 -6.26993835e-01 1.18424632e-01 1.10139757e-01
-5.96617758e-01 7.32893199e-02 5.50264001e-01 7.39409447e-01
1.40696257e-01 1.05812800e+00 1.42849475e-01 -1.77223712e-01
2.74010360e-01 -5.84649086e-01 4.99035209e-01 4.95572388e-02
-2.03753896e-02 -2.55976498e-01 -3.13430130e-01 -1.07947922e+00
-8.30419719e-01 -6.25365674e-01 -9.42890942e-01 -6.55424297e-01
2.94063538e-01 1.97780862e-01 2.37982020e-01 6.39059246e-01
3.24224800e-01 2.05971494e-01 1.04987633e+00 -9.93262410e-01
-1.30068287e-01 -5.18809021e-01 -5.83352864e-01 6.80061340e-01
7.37996280e-01 -8.52904618e-01 -7.04239547e-01 2.67288566e-01] | [8.798748016357422, -1.7992980480194092] |
4f2c33fe-b9e3-42ad-bded-5ebca1c2bbe0 | improving-stain-invariance-of-cnns-for | 2304.11445 | null | https://arxiv.org/abs/2304.11445v1 | https://arxiv.org/pdf/2304.11445v1.pdf | Improving Stain Invariance of CNNs for Segmentation by Fusing Channel Attention and Domain-Adversarial Training | Variability in staining protocols, such as different slide preparation techniques, chemicals, and scanner configurations, can result in a diverse set of whole slide images (WSIs). This distribution shift can negatively impact the performance of deep learning models on unseen samples, presenting a significant challenge for developing new computational pathology applications. In this study, we propose a method for improving the generalizability of convolutional neural networks (CNNs) to stain changes in a single-source setting for semantic segmentation. Recent studies indicate that style features mainly exist as covariances in earlier network layers. We design a channel attention mechanism based on these findings that detects stain-specific features and modify the previously proposed stain-invariant training scheme. We reweigh the outputs of earlier layers and pass them to the stain-adversarial training branch. We evaluate our method on multi-center, multi-stain datasets and demonstrate its effectiveness through interpretability analysis. Our approach achieves substantial improvements over baselines and competitive performance compared to other methods, as measured by various evaluation metrics. We also show that combining our method with stain augmentation leads to mutually beneficial results and outperforms other techniques. Overall, our study makes significant contributions to the field of computational pathology. | ['Abdulmotaleb El Saddik', 'Mustaqeem Khan', 'Numan Saeed', 'Kudaibergen Abutalip'] | 2023-04-22 | null | null | null | null | ['whole-slide-images'] | ['computer-vision'] | [ 5.46198130e-01 -1.99241579e-01 6.29858598e-02 -4.78523701e-01
-1.25264800e+00 -9.71327066e-01 2.71183401e-01 1.99164882e-01
-6.32264256e-01 7.04851389e-01 -1.73232764e-01 -4.71093178e-01
2.66112268e-01 -4.69609827e-01 -9.01362479e-01 -1.08637595e+00
2.49273360e-01 3.38671863e-01 3.47641766e-01 -2.79866979e-02
2.02797890e-01 9.93128359e-01 -7.87925363e-01 4.86708045e-01
4.11755562e-01 6.54298484e-01 -3.61678451e-01 1.13928223e+00
2.57779490e-02 5.18560290e-01 -7.43279099e-01 -6.00130081e-01
1.33697048e-01 -2.49132544e-01 -8.70080709e-01 -1.33779660e-01
6.41452551e-01 -3.16341519e-01 -1.71157524e-01 1.00470650e+00
6.99476779e-01 -3.50010365e-01 9.42748785e-01 -1.07338464e+00
-7.06133008e-01 3.28176945e-01 -7.56686807e-01 6.48113966e-01
-3.81668925e-01 5.43285072e-01 9.94450569e-01 -5.63169479e-01
9.44025636e-01 9.66847479e-01 1.03028834e+00 7.66543984e-01
-1.52768874e+00 -6.97324514e-01 -1.54183939e-01 -8.83581713e-02
-1.11558771e+00 -2.82482922e-01 5.68553686e-01 -5.38884759e-01
6.94372296e-01 4.06157106e-01 4.78826970e-01 1.07669210e+00
3.83578360e-01 8.56367171e-01 1.18630934e+00 -1.79980516e-01
6.74375445e-02 -9.45609435e-02 2.89873749e-01 7.15892553e-01
4.20801282e-01 -4.47215348e-01 -1.72030866e-01 -1.53980568e-01
4.34059143e-01 2.02327192e-01 -3.40370089e-01 -6.53114840e-02
-1.09143651e+00 6.34225667e-01 5.30927002e-01 3.82806599e-01
1.68297932e-01 2.69081265e-01 4.72631633e-01 1.57157734e-01
5.69381952e-01 7.16599822e-01 -5.04509687e-01 2.87865460e-01
-8.29574585e-01 1.05990887e-01 5.11563599e-01 2.33897522e-01
6.15755141e-01 -4.46402043e-01 -5.60925901e-01 5.59625626e-01
3.32652219e-02 4.61223662e-01 3.45468193e-01 -5.48789024e-01
8.67701322e-02 6.67131543e-01 -1.09169789e-01 -9.01755393e-01
-8.97296131e-01 -6.79122269e-01 -7.30506301e-01 8.65138993e-02
6.93204701e-01 -1.06408216e-01 -1.26793659e+00 1.66832709e+00
3.68138939e-01 1.68388471e-01 -1.54418677e-01 7.37206161e-01
7.73604393e-01 1.81289613e-01 2.91976869e-01 2.84025609e-01
1.18099904e+00 -8.62828314e-01 -5.05539775e-01 -5.20730205e-02
9.82623279e-01 -6.70478880e-01 1.20049596e+00 2.16133937e-01
-7.79152334e-01 1.67486295e-02 -1.21396637e+00 -3.61653775e-01
-6.34497762e-01 -6.49693385e-02 6.98511362e-01 6.14559710e-01
-1.28082311e+00 6.55863941e-01 -1.12380600e+00 -4.06098098e-01
1.21619582e+00 6.21365964e-01 -2.57027328e-01 -1.05523311e-01
-5.89740753e-01 5.41282654e-01 -7.18251690e-02 1.45523265e-01
-6.25834286e-01 -1.21484351e+00 -5.96925497e-01 -4.20867577e-02
-4.52301130e-02 -5.18542111e-01 1.12644112e+00 -7.10277617e-01
-1.20474088e+00 1.23980665e+00 -1.06232099e-01 -1.77699164e-01
5.35908759e-01 3.86167794e-01 3.96611318e-02 2.56141692e-01
-4.37928922e-02 8.38388383e-01 3.12165767e-01 -1.23516738e+00
-3.91662866e-01 -3.49069208e-01 -2.24587664e-01 -1.99045077e-01
-4.24983442e-01 -1.78457350e-01 -5.40312529e-01 -5.03845155e-01
-2.29301259e-01 -1.08048737e+00 -2.57856816e-01 2.69768655e-01
-7.16385007e-01 2.06321687e-01 7.23296344e-01 -5.27972281e-01
5.95063448e-01 -2.29034877e+00 -2.81353090e-02 3.27133745e-01
5.03443658e-01 3.38237822e-01 -3.95608723e-01 5.40804304e-02
7.39628971e-02 5.59875488e-01 -2.01060280e-01 -5.30853152e-01
-1.07341543e-01 3.89308706e-02 6.95128292e-02 8.37476134e-01
7.16380298e-01 1.33456492e+00 -8.46260786e-01 -6.50638223e-01
-9.00069997e-02 6.74350083e-01 -5.00503242e-01 6.46602213e-02
-5.60428873e-02 5.06994188e-01 -1.88101023e-01 8.99117053e-01
7.71046937e-01 -7.05021918e-01 1.88330501e-01 -2.22542301e-01
4.89691794e-01 -1.36736065e-01 -3.14130455e-01 1.34078181e+00
-2.29283169e-01 8.53660703e-01 5.63683584e-02 -7.43541121e-01
4.58889872e-01 -1.90724909e-01 2.07199767e-01 -2.54410297e-01
4.61152911e-01 2.05856279e-01 3.32275510e-01 -3.77024919e-01
1.45214126e-01 -3.36779386e-01 2.02870905e-01 4.48219627e-01
1.90517277e-01 -6.77035674e-02 2.50561148e-01 1.94568262e-01
1.58591080e+00 -4.36965257e-01 -4.47430536e-02 -2.95097709e-01
3.02492648e-01 1.66796278e-02 4.10756439e-01 7.03476846e-01
-5.52283466e-01 7.55765140e-01 1.06186247e+00 -2.85970181e-01
-1.06930029e+00 -1.01022077e+00 -2.56949931e-01 1.03525758e+00
-9.33709927e-03 3.34080011e-01 -7.41310775e-01 -1.08847189e+00
1.77631229e-01 -1.88505612e-02 -1.33948600e+00 -1.47766545e-01
-4.07420158e-01 -1.33819914e+00 9.47082400e-01 7.13663340e-01
8.13352838e-02 -7.05012381e-01 -2.25139901e-01 -1.11490257e-01
2.61365503e-01 -1.01574218e+00 -5.77186346e-01 4.34680104e-01
-6.75786078e-01 -1.28999901e+00 -6.86877847e-01 -6.90311253e-01
1.06524694e+00 1.96815267e-01 1.04962754e+00 6.10340834e-01
-6.78558588e-01 6.39091954e-02 -2.00780228e-01 -7.15345919e-01
-5.85127890e-01 4.92595613e-01 -6.50921643e-01 1.48759663e-01
4.58108276e-01 -1.95262015e-01 -9.06363130e-01 -1.42705711e-02
-1.30866778e+00 -4.71275635e-02 8.19783032e-01 1.06843412e+00
7.79014707e-01 -3.69437546e-01 5.65746963e-01 -1.53857648e+00
4.58059013e-01 -4.16813195e-01 -3.12364250e-01 2.86275983e-01
-3.34100068e-01 6.92379698e-02 6.65406525e-01 -3.80159318e-01
-8.37902427e-01 -1.23648606e-01 -2.22640142e-01 -1.45456150e-01
-7.15859830e-02 2.72051305e-01 -2.11408809e-01 -5.58052659e-01
6.39228702e-01 -1.91745251e-01 2.51589298e-01 -1.09556457e-02
-4.45591621e-02 3.81660342e-01 4.80953723e-01 -2.41466299e-01
8.73730242e-01 1.04454756e+00 3.69170010e-01 -3.42815280e-01
-9.06047165e-01 -5.79620481e-01 -4.84408051e-01 2.71101370e-02
7.80758440e-01 -4.95257914e-01 -6.48038507e-01 8.17107618e-01
-8.29569280e-01 -7.43050396e-01 1.47502154e-01 3.88241075e-02
-8.90311822e-02 1.82251766e-01 -1.29230070e+00 -2.64840633e-01
-4.62769061e-01 -1.31561255e+00 1.38434803e+00 4.15341765e-01
-2.15727270e-01 -1.31190956e+00 1.95371658e-01 3.76623720e-01
4.77652222e-01 6.79899991e-01 1.04869938e+00 -1.00799572e+00
-4.59583849e-01 -1.79476798e-01 -5.29384792e-01 1.51460752e-01
2.47518599e-01 3.68322939e-01 -1.34382808e+00 -4.52249467e-01
-5.27687967e-01 -4.06945914e-01 1.25146329e+00 3.79239351e-01
1.43034232e+00 2.17668731e-02 -6.37197733e-01 1.13320470e+00
1.62130964e+00 -9.97821465e-02 5.27859509e-01 3.84377480e-01
6.87141776e-01 4.05216783e-01 1.25263348e-01 -2.07211837e-01
1.21162161e-01 1.03877686e-01 3.01093608e-01 -8.46601367e-01
-3.15470904e-01 8.96408856e-02 -1.10742584e-01 3.26434255e-01
2.75602549e-01 -5.16857088e-01 -9.72457826e-01 7.56444335e-01
-1.33632922e+00 -5.69740057e-01 8.38487148e-02 1.51951730e+00
1.16321230e+00 1.24862529e-01 -9.22090635e-02 1.50922043e-02
5.56023300e-01 -8.63593444e-02 -8.75070572e-01 -3.31312060e-01
-3.30450505e-01 5.66391289e-01 7.92013943e-01 3.38508219e-01
-1.10900939e+00 7.01570451e-01 6.74433756e+00 7.65677512e-01
-1.46484196e+00 8.48122910e-02 1.36551201e+00 -1.86573848e-01
-3.68470937e-01 -7.78925598e-01 -7.46437728e-01 3.18960309e-01
7.78400481e-01 2.79683888e-01 1.95765961e-02 3.69768977e-01
-8.56270120e-02 1.79863513e-01 -1.35799587e+00 5.49521148e-01
1.53108433e-01 -1.59714758e+00 -4.29733377e-03 2.98490256e-01
8.90277028e-01 2.34569296e-01 5.55226386e-01 5.21463081e-02
4.55420494e-01 -1.20124209e+00 5.47606349e-02 1.42585993e-01
9.77999628e-01 -7.13395953e-01 1.34315181e+00 -4.51892793e-01
-5.75009286e-01 3.01357388e-01 -1.72188938e-01 4.57946330e-01
-3.53129655e-01 5.92665255e-01 -1.47177994e+00 1.42642885e-01
4.80269611e-01 4.45495278e-01 -9.37091529e-01 9.08698618e-01
5.12330197e-02 8.55880737e-01 -1.90598980e-01 -2.22128898e-01
1.96112767e-01 3.83940279e-01 6.64076358e-02 1.72878230e+00
-9.51134041e-02 -2.51552731e-01 -3.15862536e-01 7.63714373e-01
-4.79005665e-01 1.13800354e-01 -1.93934307e-01 -1.06483810e-01
2.27736503e-01 1.65399802e+00 -1.36782324e+00 -1.27973869e-01
-2.72451043e-01 7.57827103e-01 5.87284446e-01 4.72154737e-01
-8.20567667e-01 -3.67289603e-01 7.81683087e-01 1.82489753e-02
3.22648615e-01 2.77734429e-01 -5.60688078e-01 -8.61717582e-01
-1.55383378e-01 -8.31341147e-01 4.18092668e-01 -2.80092537e-01
-1.60496795e+00 3.73449206e-01 -6.04053617e-01 -7.56545901e-01
3.04568857e-01 -1.07398283e+00 -6.47558808e-01 6.58758163e-01
-1.93630672e+00 -1.36805475e+00 -2.89388210e-01 2.60882676e-01
1.55025825e-01 1.43232793e-01 6.37066782e-01 2.32128218e-01
-7.05870807e-01 1.16806960e+00 2.84557968e-01 4.91724789e-01
1.00332725e+00 -1.47148764e+00 3.81496698e-01 9.43578601e-01
-1.40768901e-01 5.45096576e-01 4.75308686e-01 -4.04877156e-01
-1.30636084e+00 -1.32789969e+00 4.06059533e-01 -7.31207907e-01
9.28661525e-01 -4.39590067e-01 -1.02299917e+00 6.82017744e-01
-6.92399368e-02 3.05479884e-01 1.37945485e+00 5.15446067e-02
-5.48346400e-01 -1.66449863e-02 -1.47012436e+00 7.56485343e-01
7.41275430e-01 -3.70765418e-01 6.60616606e-02 3.49801123e-01
7.35936284e-01 -6.84258878e-01 -7.55604446e-01 2.01738298e-01
4.70620275e-01 -6.90418005e-01 7.48828173e-01 -8.66212070e-01
6.79915011e-01 -1.20808117e-01 1.84025988e-01 -1.30672705e+00
-3.06750745e-01 -2.55543619e-01 3.76822472e-01 1.07432759e+00
7.28514969e-01 -8.30713391e-01 1.05257833e+00 5.28371155e-01
-1.77456573e-01 -1.08598793e+00 -6.49106205e-01 -3.35220069e-01
4.82221484e-01 -9.58519336e-03 6.52560234e-01 7.82166898e-01
-2.56326109e-01 4.85280156e-02 2.56583691e-01 1.82829514e-01
6.01929903e-01 -8.21730271e-02 6.94967270e-01 -9.56979752e-01
-3.90901387e-01 -7.42228329e-01 -5.76349139e-01 -4.30932462e-01
2.82287180e-01 -1.11562359e+00 3.61017473e-02 -1.18189883e+00
7.51026511e-01 -3.91504556e-01 -7.41624951e-01 7.33281553e-01
-6.05960727e-01 1.06556201e+00 7.97826946e-02 7.81129226e-02
-6.78883374e-01 -2.04905257e-01 1.50691915e+00 -5.58793366e-01
4.25534807e-02 -3.64822954e-01 -1.12271297e+00 4.18076068e-01
8.94414544e-01 -4.86474097e-01 -3.08930874e-04 -4.84882474e-01
2.95607448e-01 -5.72673917e-01 3.83931130e-01 -8.22991788e-01
-1.66624144e-03 -1.78590454e-02 7.28630960e-01 -2.93031424e-01
-1.25036854e-02 -4.91990566e-01 -1.82009533e-01 5.27639627e-01
-5.23784459e-01 -3.61505330e-01 4.73218977e-01 5.48956752e-01
1.54440090e-01 6.32611215e-02 8.85472953e-01 1.59332410e-01
1.12713268e-02 3.71786892e-01 -2.00660616e-01 1.20671898e-01
8.42278838e-01 -2.13410437e-01 -8.20147634e-01 8.60222280e-02
-4.43533570e-01 3.03779721e-01 7.39637077e-01 1.20281978e-02
1.27682865e-01 -9.15246010e-01 -6.80333912e-01 6.07060529e-02
1.73777074e-01 2.64526397e-01 3.81043762e-01 9.79083657e-01
-9.55471039e-01 2.25662857e-01 -2.27172628e-01 -7.83200264e-01
-1.32605195e+00 3.16776276e-01 4.71346676e-01 -7.34443128e-01
-1.35584876e-01 1.24617374e+00 4.25214499e-01 -4.87838328e-01
-8.14875588e-02 -5.69242239e-01 6.49149939e-02 -1.96006387e-01
5.71649551e-01 7.14729819e-03 3.52935612e-01 -1.82647631e-01
-3.98155510e-01 3.55625659e-01 -6.01513684e-01 3.09348583e-01
1.44576514e+00 3.11626166e-01 -7.52220824e-02 3.53998601e-01
1.58463442e+00 1.15211137e-01 -1.37162066e+00 1.27836674e-01
-2.53974020e-01 -2.49742433e-01 -8.88397023e-02 -9.30100143e-01
-1.43944371e+00 7.50588238e-01 6.12714708e-01 -4.45336252e-02
9.86619711e-01 1.91836342e-01 8.35448921e-01 2.21822008e-01
-1.40029714e-01 -7.01807559e-01 6.07079603e-02 9.06304941e-02
2.30358541e-01 -1.45686781e+00 -1.33849144e-01 -4.55630481e-01
-1.76846296e-01 1.00635147e+00 6.15590036e-01 -7.95911476e-02
3.47960651e-01 8.01902652e-01 4.64816153e-01 -3.17778975e-01
-7.50378013e-01 8.81403685e-02 1.00328013e-01 7.10095465e-01
5.41949332e-01 -1.32552125e-02 -1.76407829e-01 4.94970024e-01
-1.50045548e-02 -2.46864744e-02 6.23223722e-01 8.38277936e-01
-1.64939240e-02 -9.74257410e-01 -2.39631504e-01 7.33958066e-01
-9.57276285e-01 -1.67447235e-02 -6.04363263e-01 9.62372184e-01
-3.36788893e-02 4.45075870e-01 3.07650238e-01 -2.32441813e-01
-3.87412645e-02 -3.03342026e-02 5.77390850e-01 -5.34736276e-01
-6.74165606e-01 -8.26186687e-02 -3.27939838e-01 -3.61004680e-01
-3.50535721e-01 -6.77910745e-01 -1.32648540e+00 -3.82733673e-01
-2.13919714e-01 -2.42809430e-01 4.10384595e-01 8.57539892e-01
3.68814290e-01 8.87932181e-01 3.56537282e-01 -7.55004108e-01
-3.69947255e-01 -8.05932760e-01 -4.62851584e-01 6.93264842e-01
6.00130200e-01 -4.79591757e-01 -5.51687598e-01 2.76785225e-01] | [15.091075897216797, -2.9744458198547363] |
89edc229-8bb9-4e35-8b3d-aa02db36fe6c | enabling-noninvasive-physical-assault | null | null | https://doi.org/10.1155/2019/8186573 | http://downloads.hindawi.com/journals/wcmc/2019/8186573.pdf | Enabling Noninvasive Physical Assault Monitoring in Smart School with Commercial Wi-Fi Devices | Monitoring physical assault is critical for the prevention of juvenile delinquency and promotion of school harmony. A large portion of assault events, particularly school violence among teenagers, usually happen at indoor secluded places. Pioneering approaches employ always-on-body sensors or cameras in the limited surveillance area, which are privacy-invasive and cannot provide ubiquitous assault monitoring. In this paper, we present Wi-Dog, a noninvasive physical assault monitoring scheme that enables privacy-preserving monitoring in ubiquitous circumstances. Wi-Dog is based on widely deployed commodity Wi-Fi infrastructures. The key intuition is that Wi-Fi signals are easily distorted by human motions, and motion-induced signals could convey informative characteristics, such as intensity, regularity, and continuity. Specifically, to explicitly reveal the substantive properties of physical assault, we innovatively propose a set of signal processing methods for informative components extraction by selecting sensitive antenna pairs and subcarriers. Then a novel signal-complexity-based segmentation method is developed as a location-independent indicator to monitor targeted movement transitions. Finally, holistic analysis is employed based on domain knowledge, and we distinguish the violence process from both local and global perspective using time-frequency features. We implement Wi-Dog on commercial Wi-Fi devices and evaluate it in real indoor environments. Experimental results demonstrate the effectiveness of Wi-Dog which consistently outperforms the advanced abnormal detection methods with a higher true detection rate of 94% and a lower false alarm rate of 8%. | ['Shuo Zhao', 'Qizhen Zhou', 'Jianchun Xing', 'Chenshu Wu', 'and Qiliang Yang'] | 2019-04-01 | null | null | null | wireless-communications-and-mobile-computing | ['rf-based-pose-estimation'] | ['computer-vision'] | [ 4.18439031e-01 -1.88568264e-01 -8.86851490e-01 -9.35066419e-05
-6.02672875e-01 -4.81550723e-01 6.27593324e-02 -6.01304807e-02
-8.88390467e-02 6.63772821e-01 1.14382468e-01 -3.79123002e-01
-6.43739104e-01 -9.91703153e-01 -2.88152575e-01 -8.44488919e-01
-4.22236234e-01 -5.55440247e-01 2.78957784e-01 1.65682048e-01
-9.35049057e-02 3.64160866e-01 -1.18135440e+00 -1.14783920e-01
8.54479611e-01 1.11151314e+00 -5.83708584e-01 5.23798108e-01
4.43983078e-01 4.84351009e-01 -8.56676221e-01 -3.22129071e-01
-8.37469622e-02 -1.37251467e-01 -3.07610571e-01 -1.03115216e-01
1.75175950e-01 -7.65584707e-01 -8.12659502e-01 9.45928097e-01
5.23091435e-01 -4.12924262e-03 4.63409722e-01 -1.47379923e+00
-5.73025644e-02 4.28816348e-01 -9.37125683e-01 7.37458467e-01
9.75416183e-01 -1.06811095e-02 4.25696164e-01 -3.38815510e-01
1.62389606e-01 8.70919585e-01 7.16067553e-01 2.96837658e-01
-1.05342758e+00 -1.44487000e+00 8.18251744e-02 4.01543528e-01
-1.78817475e+00 -4.33443010e-01 1.02644467e+00 -2.19334096e-01
1.37496680e-01 8.65574181e-01 7.71159232e-01 1.35518265e+00
-3.89072150e-02 6.58498824e-01 6.49013102e-01 -2.29501382e-01
-4.42388374e-03 -2.10842401e-01 1.87118098e-01 4.54750806e-01
8.54649901e-01 1.04827985e-01 -7.43718028e-01 -4.87493038e-01
7.09724665e-01 1.12744123e-01 -5.65023005e-01 1.69832602e-01
-8.37624192e-01 3.09549481e-01 -2.83129126e-01 4.82271343e-01
-1.19459353e-01 -7.43599702e-03 7.47681707e-02 1.45862445e-01
5.33751905e-01 -1.09947696e-01 2.17852712e-01 -5.76685667e-01
-8.60652804e-01 1.69160500e-01 3.14677626e-01 8.10277224e-01
-3.09027303e-02 1.28764110e-02 -4.10270602e-01 4.74325120e-01
4.66439039e-01 8.60581934e-01 -9.56658460e-03 -7.52594173e-01
7.39470959e-01 6.53211325e-02 -5.40371723e-02 -1.42868030e+00
-4.98737484e-01 -7.97627270e-01 -9.45163727e-01 -4.08753633e-01
3.70878786e-01 -4.50832367e-01 -5.45109250e-02 1.51184857e+00
8.54904056e-01 1.09127307e+00 -4.79567856e-01 5.29933631e-01
8.11696768e-01 2.35149324e-01 -6.21527294e-03 -5.52179575e-01
1.28632224e+00 2.96752937e-02 -1.02634478e+00 2.41678491e-01
2.29318634e-01 -4.26765919e-01 3.44152093e-01 6.65990412e-01
-9.14222479e-01 -2.61882573e-01 -1.26368773e+00 7.68372059e-01
1.18485652e-01 -9.81414169e-02 6.32248998e-01 1.76341105e+00
-4.66835350e-01 3.60141754e-01 -8.82011235e-01 -5.10535911e-02
7.00773895e-01 2.94183373e-01 -3.64955254e-02 4.06165086e-02
-1.16397274e+00 -1.06365323e-01 -2.02804148e-01 -1.49284542e-01
-2.64115036e-01 -7.74851143e-01 -8.66752982e-01 -1.66615809e-03
2.17322022e-01 2.45004520e-02 8.62008154e-01 -1.18105955e-01
-1.60102987e+00 4.71719772e-01 -1.37112262e-02 -4.07100506e-02
3.59513402e-01 2.48231106e-02 -1.27470064e+00 6.77693188e-01
1.26635134e-01 -3.59986663e-01 8.04593921e-01 -8.15039933e-01
-6.95184290e-01 -4.86255169e-01 -2.04950020e-01 -1.84743375e-01
-7.55500078e-01 2.95588464e-01 -2.23859072e-01 -8.26090813e-01
5.79640329e-01 -4.73557144e-01 9.64305401e-02 -1.84306473e-01
-6.61170542e-01 1.15882255e-01 8.76333952e-01 -5.70274591e-01
1.91689885e+00 -2.28708625e+00 -8.39877784e-01 8.27704191e-01
3.49658221e-01 3.26540232e-01 2.65137225e-01 2.82783359e-01
-1.88048035e-02 1.33646682e-01 1.79983884e-01 1.23610988e-01
-1.07470907e-01 -3.79771888e-01 -2.44906083e-01 1.07613206e+00
-1.94739401e-01 2.87596047e-01 -1.03975427e+00 -3.96824211e-01
3.59792322e-01 2.66040355e-01 -4.34280336e-01 -1.08436659e-01
8.82990241e-01 7.00283170e-01 -1.08127022e+00 1.00831854e+00
1.10520399e+00 4.56783921e-01 5.38976006e-02 1.43020526e-01
-2.10830733e-01 2.14977756e-01 -1.41388571e+00 1.27817416e+00
1.51206702e-02 4.67936754e-01 4.39478904e-01 -1.15493155e+00
6.61771238e-01 7.67433822e-01 9.18332279e-01 -6.46959305e-01
2.24320456e-01 -1.79821670e-01 -3.60917658e-01 -9.29487348e-01
3.94520760e-02 4.39386189e-01 -3.23379397e-01 4.61678594e-01
-1.27957538e-01 3.85398984e-01 -2.53267109e-01 -1.43641695e-01
1.49321997e+00 -2.88864106e-01 8.98356289e-02 -6.97565973e-02
4.02472258e-01 -4.23367977e-01 7.26283371e-01 8.96635413e-01
-4.56394345e-01 1.85160294e-01 1.64271565e-03 -6.94378454e-04
2.06168130e-01 -1.39167368e+00 -6.34695232e-01 7.22447515e-01
5.78166425e-01 -2.01088607e-01 -6.63156390e-01 -5.93857765e-01
-1.28200620e-01 6.15569472e-01 -9.82969105e-02 -4.99873966e-01
-5.55834234e-01 -8.18718910e-01 1.08006430e+00 1.35931298e-01
7.68928230e-01 -1.27984375e-01 -4.04463917e-01 3.30323666e-01
-3.24492961e-01 -1.20153689e+00 -3.80218297e-01 -5.62883377e-01
-5.44463158e-01 -9.86748457e-01 -3.76634955e-01 -3.55743527e-01
3.67425770e-01 8.23590994e-01 3.73330683e-01 3.10832839e-02
-3.48024458e-01 6.96551979e-01 -1.73531890e-01 -5.03506005e-01
3.46770227e-01 -1.04981311e-01 3.03721845e-01 5.02581239e-01
4.64698553e-01 -1.05257654e+00 -8.05508673e-01 3.91382337e-01
-4.81089026e-01 -5.78721702e-01 -1.08919404e-02 1.21617272e-01
1.39708677e-02 5.74385524e-01 6.16731763e-01 -2.12527812e-01
5.61432064e-01 -9.47984099e-01 -3.30932617e-01 -1.44037351e-01
-3.06900893e-03 -7.65673816e-01 1.86819106e-01 -7.49034166e-01
-1.01353562e+00 -3.19825798e-01 -2.76938945e-01 -6.87895715e-02
-5.66864967e-01 7.66151994e-02 -6.08411789e-01 -3.30802947e-01
3.73444676e-01 1.26924634e-01 -6.42803609e-01 -3.40765446e-01
-1.81904614e-01 8.72930169e-01 8.36360455e-01 -6.29277289e-01
9.94095922e-01 6.18826210e-01 2.32434019e-01 -1.35260975e+00
-3.57810140e-01 -8.19825470e-01 -1.80123255e-01 -6.49149597e-01
6.78626657e-01 -9.37812567e-01 -1.20167148e+00 6.87028646e-01
-8.53920460e-01 2.34282210e-01 2.05735192e-01 1.06516385e+00
4.80894297e-02 5.02238274e-01 -5.30011117e-01 -1.18026829e+00
-1.80607036e-01 -5.75744450e-01 9.95611846e-01 4.49898034e-01
-3.43440920e-01 -6.40518010e-01 3.56097668e-02 5.28058589e-01
1.44950360e-01 6.83844984e-01 3.37044895e-01 -3.69818598e-01
-3.95501018e-01 -5.25282323e-01 1.31144539e-01 -3.45628470e-01
1.18730381e-01 -1.35322750e-01 -1.10923636e+00 -1.29352026e-02
1.43754795e-01 3.96711051e-01 2.43082061e-01 6.43924057e-01
1.64129937e+00 -3.45054775e-01 -8.14810097e-01 7.37355590e-01
9.27718580e-01 5.21111786e-01 5.86169541e-01 -9.10252035e-02
5.18113494e-01 3.81065488e-01 7.08758652e-01 7.87226439e-01
3.12378705e-01 7.48812854e-01 3.78350466e-01 8.02831575e-02
2.37791836e-01 -3.31317544e-01 2.27929264e-01 4.12719876e-01
-2.11861864e-01 -4.18143332e-01 -5.66944242e-01 4.61300224e-01
-1.52490401e+00 -1.29761446e+00 -5.08507133e-01 2.57040834e+00
5.16253650e-01 2.20390037e-01 3.19551855e-01 7.80605912e-01
8.67902339e-01 1.30756781e-01 -1.59270883e-01 -7.46458620e-02
3.86510670e-01 5.28787196e-01 7.42515266e-01 2.47926086e-01
-1.39464402e+00 2.10258693e-01 5.81675816e+00 1.41046417e+00
-8.84272814e-01 3.58266354e-01 5.31454921e-01 -2.50311404e-01
-2.74459422e-01 -6.72089458e-01 -6.42998397e-01 6.65181398e-01
6.42936468e-01 4.30330075e-02 -8.52591451e-03 3.83045405e-01
6.10677421e-01 -2.08584070e-01 -7.12384224e-01 1.05798197e+00
-2.16583282e-01 -9.61937666e-01 -7.59771705e-01 3.28736991e-01
2.86132723e-01 -7.75389194e-01 1.20323978e-01 -1.65342674e-01
-7.35971853e-02 -7.56565988e-01 5.10296881e-01 5.34056783e-01
8.83352995e-01 -1.09546888e+00 1.19200192e-01 4.17091399e-01
-1.50653398e+00 5.80009371e-02 2.05739513e-01 -3.82536769e-01
-1.43124452e-02 8.43203783e-01 -1.82360008e-01 6.41571581e-01
6.41110957e-01 4.69080657e-01 2.09358320e-01 1.28602004e+00
-3.60754222e-01 1.41673958e+00 -6.57584846e-01 -3.83259989e-02
-1.45697981e-01 -1.38256878e-01 8.74707580e-01 1.10142589e+00
6.21752441e-01 7.17732608e-01 1.43626973e-01 4.88531858e-01
2.61417806e-01 -2.01430216e-01 -8.56269419e-01 5.91722608e-01
9.75359797e-01 1.08256340e+00 -4.45690572e-01 2.07734883e-01
-5.02673030e-01 4.89098489e-01 -7.29685485e-01 5.50296366e-01
-1.15133309e+00 -6.19940400e-01 1.13927507e+00 4.19919819e-01
1.36831747e-02 -3.32504183e-01 -3.97933334e-01 -9.96689916e-01
7.36478493e-02 -4.58039314e-01 3.20733905e-01 -4.50326242e-02
-6.70976341e-01 -2.41144091e-01 2.13716909e-01 -1.67102015e+00
-6.76214695e-02 5.14304675e-02 -1.22237778e+00 2.68782079e-01
-9.74415660e-01 -8.69240105e-01 -1.64747939e-01 9.64734197e-01
1.00875862e-01 2.68185526e-01 6.20582461e-01 8.56754482e-01
-1.18196630e+00 1.15331948e+00 -3.68386470e-02 4.23992515e-01
7.68607557e-02 -5.33861458e-01 7.01998472e-02 1.15369248e+00
2.55337227e-02 3.81895840e-01 5.22302210e-01 -6.35873199e-01
-1.43089294e+00 -1.17467475e+00 7.37070382e-01 -2.31697746e-02
5.91636896e-01 -5.09172618e-01 -4.44153249e-01 2.73257464e-01
-1.37566805e-01 1.25363871e-01 9.82119977e-01 -8.78894851e-02
2.76797891e-01 -3.22969288e-01 -1.59678125e+00 5.90489745e-01
1.51314759e+00 -1.91100091e-01 -1.91738844e-01 2.42228493e-01
3.23781520e-01 -3.41104269e-01 -9.09803629e-01 3.50553304e-01
7.43520856e-01 -8.26993704e-01 1.43990767e+00 -1.48257747e-01
-2.16599569e-01 -4.78228815e-02 1.26691118e-01 -6.19365573e-01
-3.05942893e-01 -1.00248706e+00 -2.98860013e-01 1.72805369e+00
-2.59307772e-02 -8.21957469e-01 7.42129266e-01 4.59212184e-01
3.12279195e-01 -5.73962271e-01 -1.47544408e+00 -8.01258743e-01
-5.79356611e-01 -1.15366685e+00 9.44983721e-01 1.07484591e+00
6.06609941e-01 -3.25005800e-01 -4.14528906e-01 8.40106547e-01
9.15231228e-01 -2.29292840e-01 4.35886979e-01 -1.03143120e+00
-2.42916197e-01 -3.31202805e-01 -8.19712937e-01 -1.01375639e+00
-2.21466526e-01 -1.81171983e-01 -3.77205789e-01 -8.54184270e-01
-3.12622577e-01 -3.23605508e-01 -2.93146491e-01 1.82302326e-01
2.44019013e-02 4.63935465e-01 -5.45702577e-01 -3.78504068e-01
-5.16441524e-01 3.02045316e-01 8.60559404e-01 -3.19232315e-01
-3.67242306e-01 8.95236552e-01 -4.02750134e-01 1.11473024e+00
7.13315427e-01 -4.26708132e-01 -5.14686882e-01 3.57985422e-02
-1.56786799e-01 4.00406122e-01 3.65216732e-01 -1.48270965e+00
8.35996643e-02 -4.45292473e-01 4.83368754e-01 -4.51804727e-01
3.39340717e-01 -9.55232680e-01 8.86452198e-02 4.92338687e-01
1.59318015e-01 -6.53443038e-01 -8.02077167e-03 4.83459383e-01
2.61902325e-02 8.29870626e-02 2.71405935e-01 6.98212504e-01
-2.68097401e-01 4.99719858e-01 -8.28126609e-01 -9.16222408e-02
1.26227736e+00 -4.35568035e-01 -2.89681077e-01 -6.85838580e-01
-3.24010789e-01 1.02052152e-01 -3.48047853e-01 3.34575206e-01
5.83157539e-01 -1.49311352e+00 -3.91346723e-01 5.07165849e-01
-3.11860330e-02 -5.03761530e-01 4.22597706e-01 1.00595129e+00
4.58613820e-02 2.77812392e-01 2.02406511e-01 -6.09295845e-01
-1.43157959e+00 -1.29442597e-02 -1.36533007e-02 -1.83865335e-02
-5.48331499e-01 7.74109066e-01 -1.20738216e-01 2.40210876e-01
4.31809753e-01 -5.57967961e-01 -4.01870370e-01 -1.44676074e-01
9.05601025e-01 8.91279459e-01 -3.27523462e-02 -5.38074970e-01
-4.69002873e-01 6.86426044e-01 6.99785531e-01 -1.78004459e-01
1.00603485e+00 -4.70639586e-01 4.81745958e-01 -1.60727471e-01
7.50603080e-01 6.98003590e-01 -9.36724603e-01 -8.23908970e-02
-2.84286588e-01 -9.63659406e-01 1.96731433e-01 -1.01932317e-01
-1.27985024e+00 6.26682639e-01 6.75122201e-01 5.86648822e-01
1.43350434e+00 -1.55181319e-01 1.08550882e+00 -2.10827395e-01
7.54920363e-01 -7.71834433e-01 -2.93990791e-01 -1.84786066e-01
1.45599589e-01 -6.59894407e-01 1.28039569e-01 -9.17663872e-01
1.49562612e-01 7.50485599e-01 4.12291139e-01 7.03617632e-02
8.84438157e-01 3.36553901e-01 -4.75629002e-01 -1.90025941e-01
2.02684123e-02 -1.09947652e-01 4.03060883e-01 9.68975127e-01
-4.35864851e-02 3.15887541e-01 -4.23566401e-01 1.38234198e+00
-1.17052384e-01 -6.48547113e-02 2.21348926e-01 1.00980997e+00
-5.24800122e-01 -8.72235239e-01 -8.39413941e-01 5.21757305e-01
-9.22529101e-01 3.63872737e-01 -4.54783402e-02 5.67473233e-01
6.07282341e-01 1.76882041e+00 -6.72211125e-02 -6.68725729e-01
4.16309595e-01 -4.06966686e-01 3.91183287e-01 -2.56229371e-01
-2.80143619e-01 1.08161800e-01 2.82482266e-01 -8.51271987e-01
-2.24684641e-01 -8.54046464e-01 -9.32732046e-01 -5.12361765e-01
-2.86776811e-01 -9.28479731e-02 4.32414860e-01 1.09501219e+00
9.40866396e-02 4.56775248e-01 1.06836879e+00 -4.67495710e-01
1.02636674e-02 -5.25681555e-01 -1.03999615e+00 -8.40253830e-02
5.29147804e-01 -7.23787069e-01 -1.98474631e-01 -5.86938322e-01] | [6.705839157104492, 0.7117884159088135] |
3c30d517-5de3-45d7-a8cb-f5d6f6db76d0 | automatic-face-understanding-recognizing | 2102.08941 | null | https://arxiv.org/abs/2102.08941v1 | https://arxiv.org/pdf/2102.08941v1.pdf | Automatic Face Understanding: Recognizing Families in Photos | We built the largest database for kinship recognition. The data were labeled using a novel clustering algorithm that used label proposals as side information to guide more accurate clusters. Great savings in time and human input was had. Statistically, FIW shows enormous gains over its predecessors. We have several benchmarks in kinship verification, family classification, tri-subject verification, and large-scale search and retrieval. We also trained CNNs on FIW and deployed the model on the renowned KinWild I and II to gain SOTA. Most recently, we further augmented FIW with MM. Now, video dynamics, audio, and text captions can be used in the decision making of kinship recognition systems. We expect FIW will significantly impact research and reality. Additionally, we tackled the classic problem of facial landmark localization. A majority of these networks have objectives based on L1 or L2 norms, which inherit several disadvantages. The locations of landmarks are determined from generated heatmaps from which predicted landmark locations get penalized without accounting for the spread: a high scatter corresponds to low confidence and vice-versa. To address this, we introduced an objective that penalizes for low confidence. Another issue is a dependency on labeled data, which is expensive to collect and susceptible to error. We addressed both issues by proposing an adversarial training framework that leverages unlabeled data to improve model performance. Our method claims SOTA on renowned benchmarks. Furthermore, our model is robust with a reduced size: 1/8 the number of channels is comparable to SOTA in real-time on a CPU. Finally, we built BFW to serve as a proxy to measure bias across ethnicity and gender subgroups, allowing us to characterize FR performances per subgroup. We show performances are non-optimal when a single threshold is used to determine whether sample pairs are genuine. | ['Joseph P Robinson'] | 2021-01-10 | null | null | null | null | ['face-alignment'] | ['computer-vision'] | [ 2.69648787e-02 2.26744175e-01 -2.34768942e-01 -8.17593575e-01
-9.31792796e-01 -5.43862283e-01 3.90158892e-01 -5.06131388e-02
-4.82132405e-01 5.92550218e-01 8.21642876e-02 -2.93553583e-02
-1.02258965e-01 -7.82620907e-01 -7.25010514e-01 -6.48147523e-01
-4.08464134e-01 4.49186504e-01 -1.93698019e-01 7.72855384e-03
-9.83260646e-02 5.45055032e-01 -1.57733667e+00 3.05463642e-01
6.53003454e-01 1.14360380e+00 -3.97812665e-01 4.39674437e-01
3.43847722e-01 4.72124636e-01 -5.66721320e-01 -7.45928168e-01
4.30333018e-01 -2.28762180e-01 -6.16705537e-01 -3.56412351e-01
8.86150956e-01 -2.38273144e-01 -2.18623802e-01 8.76168430e-01
7.56043196e-01 -1.35052353e-01 7.25928068e-01 -1.60706913e+00
-6.21878684e-01 8.40453804e-01 -6.39954746e-01 -1.48046523e-01
3.37922871e-01 1.00778878e-01 9.05937374e-01 -7.73605943e-01
6.29466593e-01 1.29003811e+00 1.09207952e+00 6.92089319e-01
-1.38195086e+00 -1.10613263e+00 6.18615299e-02 2.01044858e-01
-1.86176145e+00 -7.42062986e-01 5.44212878e-01 -3.73682171e-01
3.89748901e-01 3.57305706e-01 4.58177269e-01 1.29057229e+00
-3.76407325e-01 5.74989736e-01 9.13592935e-01 -3.31624389e-01
1.20343737e-01 2.04039097e-01 7.37350881e-02 7.79797018e-01
-7.74393789e-03 2.14491248e-01 -7.27637231e-01 -2.31486991e-01
4.82765585e-01 -2.33315706e-01 -3.09106827e-01 -2.37579271e-01
-9.35209632e-01 9.25483286e-01 4.24394935e-01 1.71888158e-01
1.03284173e-01 1.00745521e-01 2.37192661e-01 3.70776206e-01
4.22496349e-01 3.46159935e-01 -2.44288921e-01 -1.54394224e-01
-1.25604868e+00 6.46295846e-02 6.97997451e-01 6.73634291e-01
6.94101512e-01 -2.72160500e-01 -2.58245140e-01 8.75797868e-01
2.23990336e-01 6.73093915e-01 4.52266425e-01 -1.02875948e+00
4.85700995e-01 4.35401231e-01 -1.27886266e-01 -1.37755370e+00
-6.21503413e-01 -2.23311186e-01 -1.01517701e+00 8.97063464e-02
8.53495657e-01 -1.23925552e-01 -9.77372348e-01 2.07604241e+00
1.26290977e-01 3.51452529e-01 -3.11269104e-01 9.79684114e-01
7.15844393e-01 3.36017400e-01 -1.64752245e-01 1.71604276e-01
1.15539193e+00 -4.51889008e-01 -3.65839034e-01 1.25194743e-01
7.05856621e-01 -6.32012606e-01 9.09703076e-01 2.53224134e-01
-8.41345251e-01 -5.12447536e-01 -8.84122193e-01 1.65051848e-01
-5.23951948e-01 2.78835326e-01 8.17643046e-01 1.18215144e+00
-1.38334417e+00 7.98342645e-01 -7.19523072e-01 -4.06011701e-01
8.16255867e-01 7.30609536e-01 -6.80280447e-01 6.93429708e-02
-1.23910272e+00 5.85545838e-01 3.57527323e-02 4.02596980e-01
-6.35873497e-01 -7.54185498e-01 -9.47954834e-01 8.15612450e-02
1.71002131e-02 -1.52462259e-01 6.83118999e-01 -1.12242234e+00
-1.39086390e+00 1.14185834e+00 -6.24861978e-02 -4.52088475e-01
5.22852361e-01 2.15878114e-01 -6.23459458e-01 3.31213102e-02
6.67147264e-02 9.30540025e-01 8.44470382e-01 -9.38763261e-01
-3.21069956e-01 -4.63296711e-01 -2.19918296e-01 -3.55189681e-01
-5.37524819e-01 -5.92232831e-02 -5.51714182e-01 -5.55791259e-01
1.13999303e-02 -1.10495210e+00 1.75328165e-01 1.33610889e-01
-5.23845255e-01 1.91534031e-02 5.44371665e-01 -5.89405358e-01
1.13402712e+00 -2.30089784e+00 -1.94145188e-01 7.54399836e-01
2.90195346e-01 2.37006664e-01 -3.75956357e-01 6.21220805e-02
-3.54220718e-01 2.96035230e-01 1.66555103e-02 -3.72583389e-01
1.04988158e-01 -1.83868363e-01 -1.89074308e-01 6.78617597e-01
2.62989640e-01 9.00968552e-01 -5.66799343e-01 -5.18940032e-01
-2.38478348e-01 4.42569941e-01 -7.29763567e-01 -1.48935895e-02
2.38402829e-01 1.24614581e-01 5.68333119e-02 9.09781218e-01
8.27354550e-01 -1.25112742e-01 1.74429521e-01 -3.12824100e-01
2.21817181e-01 -9.80235115e-02 -1.23276079e+00 1.49423492e+00
-2.84133554e-01 8.42168093e-01 1.91723824e-01 -1.05338061e+00
1.04519045e+00 9.50978845e-02 4.22615618e-01 -4.82937247e-01
2.33830094e-01 6.14624210e-02 4.53492738e-02 -2.13162914e-01
3.88547957e-01 3.14723700e-01 -4.29864191e-02 5.08979261e-01
1.16631515e-01 2.94775128e-01 -7.47900009e-02 1.05179548e-01
1.08208978e+00 -3.75098526e-03 -2.68751353e-01 -2.05876574e-01
2.44224206e-01 -2.66756445e-01 6.24432027e-01 8.76165211e-01
-2.73239017e-01 9.71889973e-01 7.09744394e-01 -4.57052022e-01
-7.35937357e-01 -1.11047113e+00 -3.67517501e-01 1.00232673e+00
-6.80966377e-02 -4.38343287e-01 -7.96871960e-01 -8.18445802e-01
2.03985095e-01 3.01579952e-01 -9.02351260e-01 -2.50541061e-01
-4.44000483e-01 -8.87397528e-01 1.08711684e+00 4.13380474e-01
3.75766248e-01 -7.86479712e-01 -1.15478739e-01 -1.74014792e-01
-2.14905292e-02 -8.40327024e-01 -6.22974694e-01 3.84715907e-02
-3.87373894e-01 -1.27755463e+00 -8.93880665e-01 -6.60389125e-01
7.76018977e-01 -1.42533883e-01 1.05890846e+00 -8.12859908e-02
-4.14686829e-01 3.70390862e-01 -2.19833300e-01 -2.43426695e-01
-1.57907739e-01 3.14366639e-01 2.97112852e-01 2.59732693e-01
5.32879293e-01 -6.06046796e-01 -4.79875535e-01 7.03650653e-01
-5.47810912e-01 -3.31627250e-01 3.28166902e-01 9.32652533e-01
1.36818603e-01 -3.94715428e-01 6.25850677e-01 -8.57210815e-01
3.56058717e-01 -3.62214208e-01 -6.38766110e-01 3.57179016e-01
-5.17615199e-01 5.35962731e-02 3.55550736e-01 -6.06302202e-01
-6.57888293e-01 1.65305272e-01 -2.68034302e-02 -4.83762026e-01
-1.25494361e-01 1.25900909e-01 -1.48410112e-01 -2.18612507e-01
6.74308836e-01 -2.24931523e-01 2.49870852e-01 -2.91605860e-01
3.17299932e-01 7.96384275e-01 4.69926536e-01 -6.23732984e-01
7.05304742e-01 4.95586723e-01 -6.20881915e-02 -5.52318037e-01
-5.91572285e-01 -1.04535654e-01 -6.40694439e-01 -1.85869053e-01
5.81208169e-01 -9.03279066e-01 -1.41479278e+00 4.66583669e-01
-9.91249621e-01 -3.09712946e-01 -1.36783510e-01 4.43776488e-01
-2.69021571e-01 2.70658910e-01 -5.85782111e-01 -7.01104105e-01
-2.74299026e-01 -9.17725503e-01 1.03367245e+00 1.65035456e-01
-4.38631386e-01 -7.67692029e-01 -1.78776130e-01 3.50175947e-01
4.17896509e-01 2.52755463e-01 6.30782127e-01 -7.35207558e-01
-3.19961905e-01 -3.62430185e-01 -4.36693162e-01 2.11123347e-01
-5.63447289e-02 1.82172418e-01 -1.33756292e+00 -3.33365738e-01
-5.84629416e-01 -4.85080183e-01 9.05620337e-01 3.01665157e-01
1.41822112e+00 -1.91575974e-01 -4.64828283e-01 8.42856705e-01
8.79477501e-01 -4.03291872e-03 5.63165069e-01 6.80159330e-02
6.97282016e-01 7.72152901e-01 5.18093050e-01 2.71195799e-01
2.83458889e-01 8.84791970e-01 2.49539822e-01 -2.87666351e-01
-6.67043105e-02 -2.32196927e-01 3.73728126e-01 3.42969060e-01
-1.42231449e-01 -2.07904130e-01 -1.06497550e+00 4.71503496e-01
-1.79282963e+00 -1.07442963e+00 1.62364051e-01 2.38668251e+00
8.35009396e-01 1.69014111e-02 2.81297803e-01 -1.87274311e-02
8.00907671e-01 -7.72631317e-02 -3.32737774e-01 -9.70497206e-02
-1.77652031e-01 3.35629374e-01 6.79601848e-01 4.00592744e-01
-1.11502898e+00 8.04207563e-01 6.67237520e+00 9.60833251e-01
-1.36036730e+00 -2.01675687e-02 1.02113831e+00 -4.39535677e-01
3.24039608e-02 -3.90796602e-01 -9.67844427e-01 6.60932064e-01
9.48151529e-01 2.23697692e-01 3.66760790e-01 8.10654938e-01
1.39184043e-01 -5.79602532e-02 -1.28147662e+00 1.31285954e+00
2.59010732e-01 -1.24303091e+00 -3.35081607e-01 2.56304353e-01
5.23168385e-01 6.19809963e-02 2.53444314e-01 1.13246486e-01
3.86777073e-01 -1.37110901e+00 6.17628157e-01 4.70982194e-01
1.11583996e+00 -8.59488904e-01 8.96069348e-01 -6.92420527e-02
-9.77870643e-01 -1.83067787e-02 -3.59879076e-01 1.42424732e-01
-9.25247744e-02 5.71163118e-01 -9.10808682e-01 3.55977327e-01
6.80099010e-01 6.29130125e-01 -7.01301634e-01 7.13436186e-01
3.24943736e-02 5.69010079e-01 -7.30842829e-01 1.00687481e-01
-2.30118423e-03 -8.23125541e-02 7.19107613e-02 1.15476048e+00
4.40440655e-01 -2.41905645e-01 -5.68397045e-02 8.57041419e-01
-2.70774812e-01 3.18296663e-02 -6.70217037e-01 2.24393129e-01
5.77721298e-01 1.35556293e+00 -6.15959406e-01 -8.18521753e-02
-2.18481883e-01 8.24421167e-01 4.11795825e-01 2.69380450e-01
-9.73593235e-01 -3.43803495e-01 7.66678393e-01 3.40432256e-01
1.55145526e-01 1.22612759e-01 -1.34289935e-01 -1.18108618e+00
-6.73841089e-02 -7.63784468e-01 4.78072584e-01 -4.16087508e-01
-1.40694892e+00 6.33317888e-01 -1.58073947e-01 -1.12576449e+00
-2.57046729e-01 -5.79842389e-01 -4.19646233e-01 7.91555226e-01
-1.25672007e+00 -1.15184891e+00 -3.59860897e-01 6.69605255e-01
-1.99423939e-01 -3.50370884e-01 9.49905276e-01 7.92631328e-01
-8.10284078e-01 1.40292597e+00 2.24905624e-03 7.37450898e-01
1.17859423e+00 -9.40576375e-01 2.42885724e-01 6.49675846e-01
4.00066465e-01 5.70140779e-01 2.47187242e-01 -3.81929904e-01
-9.82833564e-01 -1.08115220e+00 8.58529508e-01 -4.82013702e-01
5.27487576e-01 -7.45407939e-01 -5.97321212e-01 5.77040672e-01
-3.57777297e-01 2.27206603e-01 8.33505332e-01 5.47112346e-01
-6.05338573e-01 -4.30084795e-01 -1.34485424e+00 4.35823917e-01
1.09096193e+00 -5.14657140e-01 7.33446255e-02 2.44795203e-01
3.93213481e-01 -2.66816944e-01 -8.97638917e-01 2.46293381e-01
8.79280388e-01 -1.09625852e+00 9.03985560e-01 -4.60901052e-01
3.25747669e-01 -1.57312661e-01 -7.11320639e-02 -1.11975968e+00
-1.69142395e-01 -5.48683882e-01 2.51015961e-01 1.67090249e+00
6.70182645e-01 -7.34474301e-01 1.21026051e+00 1.01098597e+00
3.35070610e-01 -6.06940567e-01 -1.05012274e+00 -8.51508677e-01
-1.64580107e-01 -5.17737150e-01 7.92460859e-01 1.21920609e+00
1.16262034e-01 6.72659054e-02 -4.95516777e-01 8.12944025e-02
6.61432803e-01 8.64078663e-03 7.85754263e-01 -1.19926119e+00
-2.56321549e-01 -4.49701965e-01 -7.77987480e-01 -6.08286858e-01
3.94008189e-01 -1.10423791e+00 -1.66993350e-01 -5.58201849e-01
9.47641656e-02 -8.39749157e-01 -1.28914520e-01 8.85734022e-01
1.27248868e-01 7.80380964e-01 2.51824290e-01 5.95230907e-02
-3.67611259e-01 3.05237055e-01 6.36763155e-01 -4.01054978e-01
-2.24403366e-01 4.22084481e-02 -5.28683186e-01 6.77022398e-01
8.06652725e-01 -5.45808196e-01 -8.93772095e-02 -2.05900818e-01
1.30037889e-01 -1.86346471e-01 5.26960015e-01 -1.04248822e+00
2.46070921e-01 2.63479114e-01 5.23625374e-01 -4.28921193e-01
5.13037562e-01 -8.72150481e-01 1.54196247e-01 1.16601497e-01
-3.89702439e-01 -2.30475172e-01 3.37740369e-02 2.87919849e-01
-2.72418380e-01 -1.05903819e-02 8.07640612e-01 2.68240541e-01
-3.18957716e-01 4.09692109e-01 -6.95346370e-02 -5.50838299e-02
9.77753520e-01 -1.45009309e-01 -2.55300671e-01 -5.54818392e-01
-7.66885221e-01 2.96389759e-01 5.05193651e-01 3.57032418e-01
3.27117383e-01 -1.47371805e+00 -6.70239091e-01 5.25442898e-01
1.22912452e-01 -3.80180955e-01 6.82619289e-02 8.44435930e-01
-3.76445532e-01 2.50081599e-01 -1.87055811e-01 -7.72183716e-01
-1.51204240e+00 4.08000797e-01 3.78064930e-01 -2.26752330e-02
-1.13192029e-01 1.00513041e+00 -7.07918629e-02 -5.89795947e-01
5.33274770e-01 -1.66336864e-01 -8.96592513e-02 4.73934948e-01
6.28951430e-01 3.09848189e-01 5.12496158e-02 -6.67980313e-01
-5.23437917e-01 6.86213493e-01 -6.35891408e-02 -1.35292839e-02
1.31287062e+00 3.83657217e-02 -1.06137089e-01 1.25175133e-01
1.31477678e+00 1.86145455e-02 -1.02224290e+00 1.27078742e-02
2.73238868e-02 -4.91836846e-01 -1.64814293e-01 -7.38694727e-01
-1.39015913e+00 7.57816076e-01 8.33186448e-01 1.02896191e-01
9.95688975e-01 -8.58164951e-02 4.71365958e-01 2.54652977e-01
2.66811639e-01 -1.03535378e+00 -2.57708222e-01 2.48369291e-01
4.82600242e-01 -1.45443213e+00 -1.45826429e-01 -4.03744370e-01
-4.66145098e-01 1.06102502e+00 6.21333659e-01 1.98254794e-01
7.97021925e-01 4.26830798e-01 2.01280728e-01 -1.07088573e-01
-3.38431716e-01 -1.83366761e-02 3.54218990e-01 8.79109561e-01
4.09573436e-01 7.08634034e-02 4.31693532e-02 7.48567939e-01
-3.85112762e-01 9.34573263e-03 1.79638654e-01 3.59861642e-01
7.23627284e-02 -1.17655683e+00 -5.16047120e-01 5.64374745e-01
-4.42502648e-01 -6.88600689e-02 -3.35848927e-01 6.94499075e-01
3.15214097e-01 8.68876994e-01 2.92139590e-01 -6.30885541e-01
1.51474267e-01 6.32919222e-02 3.22829545e-01 -3.12019914e-01
-6.04568005e-01 -3.67994606e-01 1.36323303e-01 -8.78628910e-01
-3.79468977e-01 -5.45518339e-01 -8.16897452e-01 -6.22858465e-01
-3.48621607e-01 1.26465052e-01 6.12788200e-01 5.42202592e-01
5.54407954e-01 -4.38078865e-02 6.78057611e-01 -8.89396846e-01
-3.58545095e-01 -8.29616845e-01 -7.46271551e-01 4.73285407e-01
1.42602131e-01 -5.85101545e-01 -3.80658120e-01 -2.56518554e-02] | [13.34774112701416, 0.8785491585731506] |
a184e15a-6718-4d29-b9da-2a50c701e87d | lightweight-learning-from-label-proportions | 2306.12461 | null | https://arxiv.org/abs/2306.12461v1 | https://arxiv.org/pdf/2306.12461v1.pdf | Lightweight learning from label proportions on satellite imagery | This work addresses the challenge of producing chip level predictions on satellite imagery when only label proportions at a coarser spatial geometry are available, typically from statistical or aggregated data from administrative divisions (such as municipalities or communes). This kind of tabular data is usually widely available in many regions of the world and application areas and, thus, its exploitation may contribute to leverage the endemic scarcity of fine grained labelled data in Earth Observation (EO). This can be framed as a Learning from Label Proportions (LLP) problem setup. LLP applied to EO data is still an emerging field and performing comparative studies in applied scenarios remains a challenge due to the lack of standardized datasets. In this work, first, we show how simple deep learning and probabilistic methods generally perform better than standard more complex ones, providing a surprising level of finer grained spatial detail when trained with much coarser label proportions. Second, we provide a set of benchmarking datasets enabling comparative LLP applied to EO, providing both fine grained labels and aggregated data according to existing administrative divisions. Finally, we argue how this approach might be valuable when considering on-orbit inference and training. Source code is available at https://github.com/rramosp/llpeo | ['Fabio A. González', 'Raúl Ramos-Pollán'] | 2023-06-21 | null | null | null | null | ['benchmarking', 'benchmarking'] | ['miscellaneous', 'robots'] | [-6.77914992e-02 1.27171930e-02 -4.29659426e-01 -4.03554827e-01
-8.84790480e-01 -6.16013706e-01 9.34739470e-01 2.92092681e-01
-4.18711603e-01 1.15446532e+00 3.16570401e-01 -4.96198535e-01
-3.35793197e-01 -1.07253313e+00 -6.16501272e-01 -9.05277550e-01
-4.04029429e-01 6.69209301e-01 -5.25800735e-02 -2.89175570e-01
-1.14363678e-01 7.70806849e-01 -1.78676677e+00 2.61446893e-01
8.64231288e-01 1.22119308e+00 1.56051487e-01 4.18380141e-01
8.18269253e-02 6.27195239e-01 -2.63127118e-01 3.99884805e-02
4.67928618e-01 1.84997514e-01 -8.81846130e-01 -1.57818943e-01
6.20106518e-01 -2.09046677e-01 2.39338186e-02 9.69014823e-01
4.00035530e-01 1.00452472e-02 8.86598110e-01 -9.61591721e-01
-2.36792881e-02 3.92576754e-01 -2.95929313e-01 2.68300921e-01
-9.05879140e-02 1.33649051e-01 1.17725134e+00 -5.95905960e-01
4.37965870e-01 1.04027212e+00 9.56418335e-01 -1.68525606e-01
-1.49992108e+00 -4.55128819e-01 -2.07399167e-02 1.11271024e-01
-1.73691893e+00 -2.53267169e-01 2.52239376e-01 -9.26843107e-01
6.78676724e-01 4.57938701e-01 3.80914360e-01 7.98177779e-01
1.22971781e-01 3.43898475e-01 1.74421477e+00 -2.44219184e-01
1.82051316e-01 1.26555890e-01 4.45788540e-02 2.74550736e-01
1.98512852e-01 4.69334990e-01 -4.39569429e-02 -1.04980655e-01
3.87207240e-01 4.99687297e-03 -2.43796855e-01 -5.99321835e-02
-1.22192383e+00 9.07096565e-01 8.64515424e-01 3.90269637e-01
-5.59941471e-01 7.65240714e-02 2.85422742e-01 3.25404197e-01
1.10388958e+00 4.48373497e-01 -7.76498973e-01 7.60219693e-02
-1.64740372e+00 4.82790202e-01 6.38981223e-01 5.18382132e-01
1.14348745e+00 -7.42157847e-02 -1.16901562e-01 7.24989057e-01
2.24867478e-01 7.27877796e-01 -1.53911505e-02 -8.83982122e-01
3.60882908e-01 4.36994046e-01 4.30271000e-01 -1.02534544e+00
-8.59785318e-01 -6.63541615e-01 -1.00236452e+00 3.17737520e-01
4.04608816e-01 -2.02773124e-01 -9.14100468e-01 1.46676493e+00
3.54850441e-01 1.04875714e-02 -7.69099370e-02 9.71734703e-01
7.29857206e-01 9.63205814e-01 1.99589238e-01 2.33683199e-01
1.55568230e+00 -4.73022848e-01 -2.86603242e-01 -3.68451267e-01
8.60518813e-01 -3.40867698e-01 8.72339845e-01 1.41632721e-01
-3.74219686e-01 -5.39924085e-01 -6.40001595e-01 1.43318608e-01
-9.60964143e-01 4.06750858e-01 7.05206215e-01 4.79494601e-01
-1.33472598e+00 7.28455663e-01 -7.59820521e-01 -5.14853835e-01
5.15299618e-01 1.75521299e-01 -3.48844945e-01 5.70304133e-02
-1.45621681e+00 1.08694172e+00 5.64744294e-01 2.67620951e-01
-8.05425465e-01 -8.46441567e-01 -8.17012370e-01 2.06865534e-01
2.20933408e-01 -4.36580509e-01 9.57643867e-01 -5.60612977e-01
-8.99076223e-01 1.05873775e+00 1.35343090e-01 -6.85640991e-01
5.44465423e-01 2.10773908e-02 -2.72416294e-01 -1.84221551e-01
3.56102407e-01 9.06952977e-01 5.86026371e-01 -1.27328384e+00
-8.18299413e-01 -3.59113187e-01 2.45824233e-01 -4.38848920e-02
1.18596628e-01 -2.89836321e-02 3.33474755e-01 -6.15191877e-01
-8.53631273e-02 -9.83683288e-01 -2.86780536e-01 -3.15273732e-01
-1.73605785e-01 -2.67872930e-01 5.35680950e-01 -8.39617610e-01
1.16824460e+00 -1.96674871e+00 1.67155221e-01 1.09308928e-01
8.67889449e-02 1.69770092e-01 1.59555048e-01 6.91888332e-01
-1.73615724e-01 4.34663296e-01 -5.92667222e-01 -1.39615104e-01
3.33289415e-01 2.28020191e-01 -5.55191517e-01 6.13562286e-01
3.63365948e-01 6.87311172e-01 -7.99938917e-01 -4.13844168e-01
4.61338967e-01 3.08124751e-01 -2.85179615e-01 -3.10889608e-03
-4.81754422e-01 8.26421857e-01 -2.61420727e-01 8.72549474e-01
8.69817317e-01 -2.05388844e-01 8.46555382e-02 -1.78951919e-01
-5.01763940e-01 3.95967931e-01 -1.10056925e+00 1.15654683e+00
-6.84260070e-01 8.32373679e-01 -5.72940633e-02 -1.20741856e+00
8.80687177e-01 1.56851202e-01 2.81880885e-01 -7.12660432e-01
3.77577990e-02 3.98600012e-01 4.57822531e-03 -2.28848919e-01
6.62225783e-01 -5.05780697e-01 -3.79203647e-01 2.95534432e-01
-6.38378318e-03 -3.87248874e-01 1.64091066e-01 -2.54403293e-01
6.24309003e-01 2.88147449e-01 5.30302048e-01 -7.18623579e-01
3.91763508e-01 4.09923732e-01 3.97102445e-01 8.63513827e-01
-6.94765523e-02 5.29501915e-01 5.85794210e-01 -8.00191641e-01
-1.25881779e+00 -6.27664864e-01 -1.10858178e+00 1.05128253e+00
-3.84653181e-01 -1.75838619e-01 -3.49342227e-01 -4.15453821e-01
3.53517309e-02 5.51695108e-01 -7.65841007e-01 4.92077440e-01
-1.28413036e-01 -1.32156396e+00 6.61689460e-01 2.04479903e-01
5.58790207e-01 -9.87753689e-01 -5.25423527e-01 2.97046881e-02
-1.26355544e-01 -9.81227994e-01 6.02792442e-01 4.76346940e-01
-8.55434597e-01 -7.15872467e-01 -7.07096159e-01 -1.27466634e-01
1.21514991e-01 -1.64236858e-01 1.45580435e+00 -9.79453772e-02
4.20822538e-02 -2.60408223e-01 -4.92073774e-01 -1.09335579e-01
-1.97784454e-01 5.80361307e-01 3.88472229e-02 -7.58946612e-02
2.99355268e-01 -8.05362463e-01 -4.45605814e-01 3.72715503e-01
-9.18168545e-01 2.05354230e-03 6.32404983e-01 7.40701497e-01
5.06259501e-01 1.86181709e-01 3.51810813e-01 -9.13489163e-01
-2.50037946e-02 -9.74513769e-01 -9.40353036e-01 -1.15141504e-01
-3.21093947e-01 1.53914660e-01 5.37040472e-01 2.13665545e-01
-8.08869183e-01 -1.63536578e-01 -3.90848845e-01 9.05440077e-02
-8.39443684e-01 1.06500959e+00 1.08315654e-01 8.60916823e-02
8.85915577e-01 5.57857268e-02 -4.00630981e-01 -7.50737786e-01
1.97166353e-01 1.09437096e+00 3.57263684e-01 -7.38589346e-01
7.55431652e-01 7.50070989e-01 1.52905405e-01 -7.38212943e-01
-1.18283510e+00 -5.81820488e-01 -8.76139462e-01 -1.02110282e-01
7.12785006e-01 -1.22143316e+00 -2.38975361e-01 2.10953534e-01
-7.42146611e-01 -9.32982802e-01 -2.26585180e-01 3.96981031e-01
-4.82039869e-01 1.45360865e-02 -2.79476255e-01 -7.57145643e-01
-1.31080240e-01 -1.09635198e+00 1.55504942e+00 8.02550241e-02
-8.12781528e-02 -1.12104464e+00 2.99113721e-01 6.74244687e-02
5.59513748e-01 6.03929877e-01 7.30421424e-01 -3.53533953e-01
-4.95600849e-01 -2.22832724e-01 -5.75253963e-01 1.68841392e-01
7.78707266e-02 -7.97781721e-02 -1.19566190e+00 -2.91905999e-01
-3.08153868e-01 -3.89891982e-01 1.29885268e+00 4.70262408e-01
1.06297004e+00 -3.72115016e-01 -3.20419401e-01 7.10845113e-01
1.57564175e+00 -4.13182586e-01 5.49510896e-01 7.83797145e-01
5.55688202e-01 9.86463606e-01 7.39455879e-01 5.34827709e-01
4.75582063e-01 8.91880095e-01 6.81748748e-01 -1.97991714e-01
-3.82864252e-02 1.65499344e-01 -4.91613001e-02 1.97303087e-01
-3.99210662e-01 4.05427971e-04 -1.48070693e+00 7.51580656e-01
-1.83582819e+00 -9.82961476e-01 -4.46611822e-01 2.18755507e+00
7.51625955e-01 -1.60738811e-01 2.55643260e-02 4.80490141e-02
4.88831937e-01 6.66723073e-01 -1.71094000e-01 -4.44918908e-02
-3.46036673e-01 1.58557445e-01 1.05505645e+00 5.83798349e-01
-1.61505628e+00 8.79139245e-01 5.49700212e+00 1.06210184e+00
-1.30162549e+00 3.72999042e-01 8.03787112e-01 1.82374418e-01
-2.60938965e-02 9.93264169e-02 -1.03802097e+00 7.50934839e-01
1.45926476e+00 4.73238230e-01 2.68858314e-01 7.80401707e-01
4.60242450e-01 -3.30436558e-01 -7.86701858e-01 7.28251636e-01
-4.67452347e-01 -1.53400207e+00 -2.87962079e-01 5.00883341e-01
9.28047061e-01 7.69670963e-01 -7.66344145e-02 4.50626701e-01
4.91324842e-01 -1.26801705e+00 7.47645199e-01 6.74443901e-01
9.44732130e-01 -5.04876792e-01 1.24894488e+00 4.62649018e-01
-1.11711776e+00 -7.00324178e-02 -5.01519620e-01 -3.24452072e-01
6.62971884e-02 8.53303194e-01 -5.44813514e-01 1.00581992e+00
1.13246047e+00 8.22737098e-01 -6.32653415e-01 1.13830757e+00
-2.41820514e-01 6.68013752e-01 -7.17104256e-01 4.85038638e-01
6.39090002e-01 -1.63375840e-01 2.84059793e-01 1.33901262e+00
5.90442598e-01 -5.03021069e-02 1.06200360e-01 6.64630651e-01
1.15337484e-02 -6.36660531e-02 -8.07441890e-01 1.70984507e-01
3.66407484e-01 1.49990344e+00 -5.67078769e-01 -2.77841300e-01
-2.80902773e-01 2.89218992e-01 3.60748500e-01 2.08877578e-01
-6.62534475e-01 1.76554963e-01 6.68250084e-01 3.35782588e-01
1.57197848e-01 -2.61901408e-01 -2.05902234e-01 -1.11605358e+00
-3.53594780e-01 -7.82480240e-01 4.09619063e-01 -7.80658543e-01
-1.28773463e+00 6.48702681e-01 3.28503847e-01 -1.53962672e+00
-2.06695586e-01 -8.84860873e-01 -4.54709560e-01 9.01049852e-01
-1.88678169e+00 -1.23463762e+00 -4.72737074e-01 -9.65251476e-02
1.35046259e-01 1.09601766e-01 7.95196295e-01 3.87639135e-01
-3.49403977e-01 -5.03492728e-02 5.48327565e-01 -1.12323932e-01
5.03709793e-01 -1.45427310e+00 1.83947250e-01 5.52593052e-01
1.84105724e-01 7.90571794e-02 7.48933971e-01 -3.61056149e-01
-5.95489621e-01 -1.43936443e+00 1.04636836e+00 -5.10512471e-01
9.32996273e-01 -3.10632974e-01 -9.59435046e-01 7.43926585e-01
1.87365208e-02 2.88633972e-01 5.91743767e-01 3.60834748e-01
-6.32138327e-02 -2.15166852e-01 -1.03652775e+00 2.17494354e-01
5.72595477e-01 -5.70685327e-01 -5.14258265e-01 6.67479813e-01
3.51614594e-01 -3.03761572e-01 -1.27927160e+00 7.18267083e-01
3.77086163e-01 -1.07167113e+00 8.05382192e-01 -2.70034671e-01
5.82530022e-01 -6.03348553e-01 -5.16426444e-01 -1.58359110e+00
-3.04525852e-01 5.79504035e-02 3.42893511e-01 1.07312012e+00
4.39071059e-01 -7.71438599e-01 5.42930424e-01 1.58954605e-01
-2.02488437e-01 -5.97969353e-01 -1.10885787e+00 -7.85330653e-01
4.72690314e-01 -5.11555851e-01 7.88928032e-01 1.21343541e+00
-5.12395561e-01 -6.24285601e-02 -4.25610304e-01 5.93222082e-01
3.92363667e-01 3.20125639e-01 6.77675247e-01 -1.67376828e+00
-1.21924253e-02 -4.55861270e-01 -3.60707611e-01 -5.12040377e-01
2.64299870e-01 -9.87667799e-01 7.23715033e-03 -1.61356795e+00
-8.48900974e-02 -9.75395679e-01 -1.26029521e-01 6.29994690e-01
1.32807359e-01 6.63432777e-01 4.49420847e-02 5.54867864e-01
-2.96629667e-01 5.51325142e-01 5.70810497e-01 -2.26764470e-01
1.73017338e-01 -9.83872786e-02 -2.04116836e-01 7.32363880e-01
1.00577724e+00 -7.58421659e-01 3.82108212e-01 -3.19864839e-01
5.43229818e-01 -8.81921723e-02 7.40667880e-01 -1.28390932e+00
-6.95895404e-02 -1.19187430e-01 3.05062413e-01 -8.93884659e-01
1.62831739e-01 -9.24727976e-01 4.25629675e-01 2.70400792e-01
1.37949660e-01 -4.32647586e-01 3.64552766e-01 1.77891776e-01
-5.17908394e-01 -2.17660293e-01 7.61940897e-01 -1.78939492e-01
-9.28551853e-01 3.15760404e-01 -3.49439532e-01 -2.61181921e-01
6.06200337e-01 -9.34767909e-03 -3.89406681e-01 -9.37772170e-02
-8.32242727e-01 3.28326941e-01 6.06311858e-01 -6.09288141e-02
-3.51640731e-01 -1.20552385e+00 -9.76263285e-01 -6.62190542e-02
2.98156083e-01 2.25678921e-01 4.48365629e-01 1.10318637e+00
-7.11268246e-01 9.30191100e-01 -2.72294670e-01 -8.00115108e-01
-6.58408821e-01 1.71583191e-01 5.65670133e-01 -6.44895673e-01
-4.26738262e-01 4.78468269e-01 1.75521359e-01 -1.07698405e+00
-3.43374878e-01 -4.59115505e-01 -3.97222877e-01 7.24878967e-01
3.03686976e-01 1.68578893e-01 2.70828068e-01 -9.43601847e-01
-3.32821757e-01 4.71536160e-01 3.91734451e-01 1.18358776e-01
1.58948195e+00 -2.64247864e-01 -1.79966569e-01 6.21227384e-01
8.58777881e-01 -1.69265911e-01 -1.42376244e+00 -2.40921095e-01
3.06151628e-01 -4.51967359e-01 4.23133463e-01 -7.10789382e-01
-8.41463745e-01 1.05625522e+00 7.27056205e-01 5.26029348e-01
8.88904810e-01 2.74942458e-01 1.61414281e-01 3.50658983e-01
5.13522863e-01 -8.08284819e-01 -7.63214290e-01 4.66894954e-01
8.74093473e-01 -1.53492081e+00 1.43554315e-01 1.91148440e-03
-3.19530100e-01 8.10236275e-01 -2.09690034e-02 -1.49839148e-01
7.12754726e-01 3.48451547e-02 -1.67727768e-01 -3.86335492e-01
-3.97560030e-01 -8.19530845e-01 3.52028906e-01 4.23473805e-01
4.31006581e-01 4.96282876e-01 -2.06183523e-01 3.61383051e-01
-4.42153007e-01 -2.69299140e-03 3.85884255e-01 5.60243964e-01
-5.70374370e-01 -1.01865733e+00 -7.29897618e-01 8.01654518e-01
-5.95318794e-01 -5.91600418e-01 1.39383137e-01 9.34672296e-01
4.07954216e-01 6.27419174e-01 3.46175164e-01 -3.92645746e-02
-1.63807198e-02 1.72083871e-03 5.43477498e-02 -7.61338174e-01
-3.73028457e-01 -2.74529815e-01 3.04337710e-01 -4.23569113e-01
-6.83219075e-01 -7.79283106e-01 -6.21017873e-01 -5.01393557e-01
1.30192280e-01 2.39730313e-01 7.18201160e-01 1.03665042e+00
2.22783372e-01 3.52827162e-01 4.71990615e-01 -1.56239927e+00
-2.12456390e-01 -1.45513451e+00 -8.22718084e-01 6.23269342e-02
5.72053552e-01 -1.03216815e+00 -5.68655789e-01 -1.28248975e-01] | [9.518356323242188, -1.4895392656326294] |
568540a5-a9a5-4885-b1f0-217423ef10c8 | eeg-based-emotion-recognition-using | 1907.07835 | null | https://arxiv.org/abs/1907.07835v4 | https://arxiv.org/pdf/1907.07835v4.pdf | EEG-Based Emotion Recognition Using Regularized Graph Neural Networks | Electroencephalography (EEG) measures the neuronal activities in different brain regions via electrodes. Many existing studies on EEG-based emotion recognition do not fully exploit the topology of EEG channels. In this paper, we propose a regularized graph neural network (RGNN) for EEG-based emotion recognition. RGNN considers the biological topology among different brain regions to capture both local and global relations among different EEG channels. Specifically, we model the inter-channel relations in EEG signals via an adjacency matrix in a graph neural network where the connection and sparseness of the adjacency matrix are inspired by neuroscience theories of human brain organization. In addition, we propose two regularizers, namely node-wise domain adversarial training (NodeDAT) and emotion-aware distribution learning (EmotionDL), to better handle cross-subject EEG variations and noisy labels, respectively. Extensive experiments on two public datasets, SEED and SEED-IV, demonstrate the superior performance of our model than state-of-the-art models in most experimental settings. Moreover, ablation studies show that the proposed adjacency matrix and two regularizers contribute consistent and significant gain to the performance of our RGNN model. Finally, investigations on the neuronal activities reveal important brain regions and inter-channel relations for EEG-based emotion recognition. | ['Peixiang Zhong', 'Di Wang', 'Chunyan Miao'] | 2019-07-18 | null | null | null | null | ['eeg-emotion-recognition'] | ['miscellaneous'] | [ 4.15785238e-02 -1.50121465e-01 3.86531383e-01 -5.35015881e-01
1.17924139e-01 -3.74738336e-01 3.33465412e-02 -9.79201943e-02
-1.39594868e-01 9.22585607e-01 1.92737937e-01 1.44038033e-02
-4.08611149e-01 -5.49847245e-01 -8.35144043e-01 -7.52335310e-01
-6.67515278e-01 -1.68102533e-01 -5.10134220e-01 -6.18361607e-02
1.96479866e-03 5.12066722e-01 -7.92654812e-01 9.05143693e-02
1.07779706e+00 1.30338442e+00 -2.57398158e-01 2.76887000e-01
1.55235097e-01 5.96189857e-01 -8.62205982e-01 -3.66839886e-01
4.85958345e-02 -6.12390459e-01 -5.63660264e-01 5.47794113e-03
-1.54356942e-01 5.16516387e-01 -8.39039981e-01 1.43133473e+00
7.08314776e-01 8.66212845e-02 8.26136649e-01 -1.67384732e+00
-8.80732000e-01 5.91126800e-01 -5.96875012e-01 5.10784566e-01
2.74449140e-01 -8.69789869e-02 7.53889859e-01 -5.42851508e-01
3.09339583e-01 6.55069649e-01 6.47235513e-01 2.81410277e-01
-1.29177785e+00 -1.27428722e+00 3.35373193e-01 5.35311282e-01
-1.74477172e+00 -1.59687027e-02 1.32800186e+00 -3.01188260e-01
9.04502094e-01 2.54923701e-01 1.31098127e+00 1.65450931e+00
9.02143359e-01 3.40269327e-01 1.46506703e+00 1.29005998e-01
3.62142801e-01 -2.89639622e-01 2.99713314e-01 4.48165387e-01
1.32585242e-01 -1.74836472e-01 -7.59645760e-01 -3.41841161e-01
8.32408249e-01 -6.40389547e-02 -7.20459819e-01 -2.13760406e-01
-1.10541177e+00 4.56165552e-01 7.39020050e-01 4.17062223e-01
-7.01929271e-01 1.48005858e-01 6.48909926e-01 5.32789707e-01
6.59560800e-01 6.56796336e-01 -3.91950220e-01 1.60653397e-01
-6.08771145e-01 -4.75336760e-01 8.41316164e-01 8.87352288e-01
6.58025324e-01 4.37123626e-01 -6.70272708e-02 8.38708818e-01
9.51839685e-02 3.15967977e-01 5.62611282e-01 -3.49447668e-01
2.20078841e-01 7.06025183e-01 -6.20026350e-01 -1.67501283e+00
-8.09398770e-01 -7.63022482e-01 -1.72597766e+00 -4.07748371e-01
-1.40755937e-01 -5.05771279e-01 -6.58301234e-01 1.97845340e+00
-2.54216403e-01 1.01916051e+00 -1.26997933e-01 8.30596745e-01
9.59441602e-01 3.54300648e-01 2.29349747e-01 -3.07098299e-01
1.12955797e+00 -5.24551630e-01 -9.06856179e-01 -3.39273036e-01
1.16721354e-01 2.33653244e-02 8.05754602e-01 4.18434560e-01
-6.92166686e-01 -1.52023405e-01 -9.66196597e-01 5.92384160e-01
-5.43681264e-01 -2.40203053e-01 1.01117539e+00 6.25147045e-01
-1.18733692e+00 1.88938960e-01 -7.07041204e-01 5.11081098e-03
7.48890638e-01 6.77207053e-01 -6.57102287e-01 1.35717243e-01
-1.54336929e+00 5.72110415e-01 1.05910018e-01 3.39369923e-01
-6.91089153e-01 -6.64953351e-01 -6.99391782e-01 2.22245798e-01
-2.80400030e-02 -6.37310147e-01 8.70531499e-02 -1.14232707e+00
-1.30896938e+00 5.15207827e-01 -6.67882990e-03 -2.66985685e-01
-1.11478955e-01 3.16400856e-01 -8.46350193e-01 3.33952308e-01
-3.04313004e-01 4.68917012e-01 8.51908922e-01 -1.05794036e+00
2.63145506e-01 -6.07623279e-01 -3.29562902e-01 3.90093029e-02
-6.99023128e-01 -1.78108335e-01 -3.14391136e-01 -8.87538731e-01
1.53461546e-01 -7.73468852e-01 -2.93180104e-02 -5.22236288e-01
-6.69294596e-01 -9.01503488e-02 4.34187710e-01 -6.54554844e-01
1.12825596e+00 -2.13352895e+00 2.97466546e-01 8.57934117e-01
5.34125030e-01 -2.69525796e-01 -5.00094473e-01 1.38187230e-01
-6.16463542e-01 5.28181791e-02 -3.48679185e-01 1.39900416e-01
-3.95888388e-02 1.31889015e-01 -3.18018012e-02 7.30182171e-01
3.33736986e-02 1.31099844e+00 -7.30259120e-01 -3.43441926e-02
-5.48276752e-02 7.43975043e-01 -2.67744839e-01 1.16535760e-01
5.45451224e-01 7.27935791e-01 -4.19068456e-01 5.30207574e-01
7.28062391e-01 -2.22833604e-01 3.01025271e-01 -4.41365749e-01
5.75798273e-01 6.41238093e-02 -8.97912323e-01 1.90399945e+00
-3.20266068e-01 7.48617113e-01 4.20651622e-02 -1.47366560e+00
1.05820107e+00 4.08081412e-01 6.89864516e-01 -7.89917707e-01
5.31372666e-01 -2.67603099e-01 4.24582899e-01 -4.08634514e-01
-1.20798431e-01 2.33603626e-01 -4.03495245e-02 4.68056262e-01
4.05306280e-01 1.50804088e-01 -2.66649187e-01 2.17575997e-01
1.38372266e+00 -4.99906272e-01 2.44093955e-01 -6.29475296e-01
2.74559855e-01 -7.98425972e-01 6.45112276e-01 6.57772005e-01
-3.38679165e-01 3.58096868e-01 8.50061357e-01 -3.12241256e-01
-4.05629486e-01 -8.54814231e-01 -2.91993469e-01 7.47544765e-01
3.25484276e-01 -3.07761014e-01 -9.44825292e-01 -6.60890341e-01
-2.00623423e-01 4.36170846e-01 -1.07210410e+00 -6.87939763e-01
1.24522382e-02 -1.28535795e+00 8.80301058e-01 5.63158751e-01
5.78675985e-01 -1.29879272e+00 -2.74666429e-01 9.61413309e-02
-2.67123461e-01 -1.09047103e+00 -3.72398257e-01 4.76363808e-01
-5.76129079e-01 -9.76449311e-01 -4.34804529e-01 -7.41896749e-01
8.74440730e-01 -2.47878551e-01 1.07786381e+00 -2.99701076e-02
-3.76134664e-01 6.38577163e-01 -4.13633943e-01 -3.88641655e-01
3.29184830e-01 -1.67053957e-02 2.71547705e-01 4.54571754e-01
4.22722220e-01 -1.44350159e+00 -6.65539086e-01 3.22187722e-01
-8.60026658e-01 -2.69319892e-01 5.51099777e-01 9.32210028e-01
5.72651744e-01 3.06787819e-01 1.02658236e+00 -7.82967031e-01
1.23510194e+00 -7.42958128e-01 1.31063978e-03 3.36285323e-01
-4.40962553e-01 -2.29597077e-01 6.52689755e-01 -6.59647644e-01
-5.71417034e-01 -1.21645793e-01 -8.28567594e-02 -4.95371580e-01
-2.25731462e-01 7.74322510e-01 -4.49519426e-01 -5.59606731e-01
4.46026832e-01 5.35198808e-01 -4.11928326e-01 7.41317645e-02
2.03456968e-01 4.66295302e-01 4.96329010e-01 -4.75966722e-01
1.47010162e-01 1.39395595e-01 -2.95909657e-03 -6.43195689e-01
-3.02747160e-01 -1.96649328e-01 -5.64115226e-01 -2.92047739e-01
7.30190039e-01 -9.20380831e-01 -9.75316286e-01 6.85902059e-01
-1.07781088e+00 -2.86250859e-01 -2.55904607e-02 5.25887311e-01
-2.98535168e-01 1.68546379e-01 -7.65269816e-01 -5.04176199e-01
-5.91269970e-01 -1.01618969e+00 4.70005512e-01 6.41905889e-02
-1.22841336e-01 -1.25685751e+00 -1.66090459e-01 -2.82544792e-01
3.62278044e-01 3.81131679e-01 1.00884461e+00 -7.31534064e-01
-2.66386513e-02 -1.68819770e-01 -3.06373894e-01 3.59326214e-01
1.83722615e-01 -5.15554786e-01 -8.73830080e-01 -7.02617988e-02
1.55545682e-01 -2.72745788e-01 6.65334284e-01 5.77680707e-01
1.72504818e+00 -1.89856708e-01 -2.92676061e-01 1.01631141e+00
1.12031877e+00 2.85704255e-01 9.90120769e-01 -1.23271748e-01
8.67078006e-01 4.17372465e-01 -4.05169398e-01 5.32449245e-01
3.50454956e-01 1.91495076e-01 4.39725488e-01 -2.36290127e-01
2.69273400e-01 1.46140054e-01 7.55064264e-02 1.37144256e+00
-2.74968833e-01 -4.40371007e-01 -9.52638984e-01 3.46127778e-01
-1.67347860e+00 -6.60947561e-01 -1.21242493e-01 1.69497907e+00
5.72085500e-01 -1.94800034e-01 -2.50221699e-01 1.84939783e-02
5.60469866e-01 1.04633920e-01 -7.88752139e-01 -2.27143466e-01
-5.06202281e-01 5.70127964e-01 2.23481610e-01 -9.13100690e-02
-8.51628125e-01 5.96081376e-01 6.59860134e+00 6.48253322e-01
-1.21679175e+00 3.53167325e-01 7.78667569e-01 -4.37539555e-02
-2.89941847e-01 -6.59522474e-01 1.35910690e-01 6.51696503e-01
8.36985469e-01 -2.79322982e-01 1.08207572e+00 3.38913172e-01
-1.23780213e-01 2.98553884e-01 -9.42156136e-01 1.50865543e+00
4.05784100e-01 -1.02195048e+00 -9.78946909e-02 7.97994733e-02
8.32368135e-01 2.03660145e-01 5.38001594e-04 2.27710515e-01
1.06283925e-01 -1.35046458e+00 3.02730829e-01 8.33413720e-01
8.58043134e-01 -1.02833712e+00 8.40981424e-01 8.98801833e-02
-1.18286848e+00 3.00754756e-02 -5.25770724e-01 -9.73536223e-02
-2.28847504e-01 8.08380127e-01 -7.97567144e-02 6.48031116e-01
8.64263713e-01 1.13291299e+00 -6.99752569e-01 8.95635307e-01
-5.14542580e-01 8.27162385e-01 -9.18274373e-02 -9.64466110e-02
-5.97643144e-02 -5.30539095e-01 4.38913733e-01 1.36544788e+00
1.71555981e-01 3.95794898e-01 -1.93566337e-01 1.17302048e+00
-6.02540374e-01 1.45276010e-01 -8.05963576e-01 -7.59267733e-02
3.93819064e-01 1.54174328e+00 -8.32533121e-01 2.38843635e-02
-3.55069458e-01 1.30053377e+00 6.55609250e-01 8.90221059e-01
-8.10933053e-01 -6.89946949e-01 7.68413424e-01 -4.70252156e-01
-2.05146626e-01 -2.95934472e-02 -5.17718017e-01 -1.31069171e+00
-6.86618611e-02 -8.54417264e-01 8.17930996e-02 -8.53856623e-01
-1.79786670e+00 9.74687934e-01 -3.84210110e-01 -8.38583589e-01
3.17039788e-01 -4.54420388e-01 -8.46351445e-01 8.00298631e-01
-1.20293927e+00 -7.95526385e-01 -4.29934353e-01 1.22928846e+00
-1.05557948e-01 -2.38046780e-01 9.91831362e-01 3.84923726e-01
-7.47679651e-01 7.99897075e-01 -5.79171740e-02 4.29497302e-01
4.47998852e-01 -1.02103412e+00 1.50568888e-01 6.72089100e-01
4.04402018e-01 8.47032487e-01 8.93849358e-02 -5.18480659e-01
-1.56654871e+00 -1.18846214e+00 1.97391972e-01 -1.47716194e-01
9.80670094e-01 -8.92472565e-01 -1.15856743e+00 6.46504045e-01
5.77418983e-01 3.73349994e-01 9.53404963e-01 4.24722463e-01
-3.60438079e-01 -2.68806845e-01 -9.34064627e-01 3.93299669e-01
1.27388752e+00 -8.38629186e-01 -4.48257089e-01 4.24705625e-01
3.09202611e-01 -1.07014582e-01 -1.10812330e+00 2.89100617e-01
5.01806915e-01 -6.29519284e-01 8.35254729e-01 -6.60318911e-01
2.78952301e-01 1.07853897e-01 -8.31229426e-03 -2.18793869e+00
-5.81838191e-01 -5.29573202e-01 4.17707264e-02 1.01800370e+00
3.60541642e-01 -9.98532414e-01 3.54724318e-01 6.06361330e-01
-2.18869910e-01 -8.33195031e-01 -1.04629874e+00 -6.83266819e-01
-6.42679632e-02 -7.18885362e-01 8.05848479e-01 1.38105178e+00
5.98078549e-01 4.92438674e-01 -4.27593976e-01 2.28831664e-01
4.07663435e-01 -1.60395861e-01 1.82352260e-01 -1.13753617e+00
-1.95993520e-02 -5.27438045e-01 -8.85233879e-01 -5.17585933e-01
9.37349617e-01 -1.29542649e+00 -1.21884793e-01 -1.43585408e+00
3.19060922e-01 -1.87938154e-01 -1.10085285e+00 6.04126394e-01
-1.15537003e-01 5.22238314e-01 -1.99268579e-01 -2.67046779e-01
-7.30544209e-01 8.38839650e-01 1.17191815e+00 -3.68340969e-01
-2.01986656e-01 -4.47602093e-01 -7.72041023e-01 7.22845972e-01
8.96915674e-01 -6.03592277e-01 -5.76325834e-01 -3.27107936e-01
3.67290765e-01 -4.46288055e-03 4.48886424e-01 -1.10523701e+00
4.78451788e-01 1.96959093e-01 7.27370739e-01 2.16951951e-01
1.90575033e-01 -9.98983085e-01 2.78231651e-01 6.83549047e-02
-3.22512001e-01 1.72327235e-01 3.37159306e-01 8.53752315e-01
-2.95470685e-01 3.64473730e-01 3.85225505e-01 1.12648793e-01
-3.93368363e-01 7.10309029e-01 -6.63925827e-01 2.52036154e-01
9.49748695e-01 2.05265209e-02 -3.42605859e-01 -5.69947720e-01
-9.41532612e-01 3.51013355e-02 -1.37386352e-01 2.69202173e-01
8.69392931e-01 -1.45128477e+00 -5.40198743e-01 5.68154812e-01
4.63246219e-02 -6.64077520e-01 4.27269369e-01 1.22721589e+00
7.50528052e-02 8.90913978e-02 -7.07579792e-01 -3.46406311e-01
-8.44888151e-01 3.47977161e-01 4.48280811e-01 -4.98694703e-02
-4.77987200e-01 1.05650985e+00 4.52747613e-01 -3.52365255e-01
2.34180003e-01 -1.69502765e-01 -3.03019732e-01 6.68474706e-03
2.58208364e-01 2.30236232e-01 1.17285669e-01 -5.40083289e-01
-6.34777188e-01 2.54657775e-01 3.68099451e-01 2.47936711e-01
1.52483261e+00 -2.79871393e-02 -6.44024491e-01 4.34039831e-01
1.24889946e+00 -2.20643833e-01 -9.09044385e-01 -4.58220718e-03
-4.37253654e-01 -1.27717912e-01 2.05621079e-01 -1.07657528e+00
-1.67379475e+00 1.00323665e+00 5.08702457e-01 2.84243226e-01
1.66086245e+00 -2.02932760e-01 4.42230940e-01 4.46300715e-01
6.93067729e-01 -8.77808392e-01 4.51174378e-02 4.12031353e-01
1.06163800e+00 -7.40271747e-01 -1.78768203e-01 -3.41550142e-01
-6.74827874e-01 8.17346632e-01 5.75066864e-01 -2.92427063e-01
1.11013484e+00 4.30946708e-01 -2.08136454e-01 -5.53542614e-01
-5.55340171e-01 3.38006824e-01 6.80162787e-01 7.89272666e-01
2.13818073e-01 3.49088520e-01 -2.12988764e-01 1.16537166e+00
-1.99350968e-01 -1.66005090e-01 2.69929320e-01 3.09950590e-01
1.95476785e-01 -7.09054649e-01 5.31465113e-02 9.16242003e-01
-5.17521977e-01 -4.69932407e-01 -7.42512345e-01 6.15807176e-01
3.38198454e-03 1.00198638e+00 2.61822827e-02 -9.14455295e-01
3.64162892e-01 2.36796454e-01 5.60214400e-01 -3.00964773e-01
-8.40945721e-01 1.58782266e-02 -3.06761086e-01 -6.43228292e-01
-5.03306448e-01 -2.50661880e-01 -1.25189364e+00 -3.87177207e-02
-2.57797241e-01 2.31315255e-01 5.22214651e-01 7.85964549e-01
8.47430110e-01 1.06555188e+00 6.14692509e-01 -7.30813503e-01
3.12058628e-01 -1.16468930e+00 -1.34486282e+00 5.41824281e-01
-2.65517123e-02 -7.30341077e-01 -3.24424863e-01 -8.32634568e-02] | [13.091193199157715, 3.4941980838775635] |
9582203d-debb-4b2d-891e-c5a01e50b8c4 | coarse-to-fine-multi-label-image | 2012.13662 | null | https://arxiv.org/abs/2012.13662v1 | https://arxiv.org/pdf/2012.13662v1.pdf | Coarse to Fine: Multi-label Image Classification with Global/Local Attention | In our daily life, the scenes around us are always with multiple labels especially in a smart city, i.e., recognizing the information of city operation to response and control. Great efforts have been made by using Deep Neural Networks to recognize multi-label images. Since multi-label image classification is very complicated, people seek to use the attention mechanism to guide the classification process. However, conventional attention-based methods always analyzed images directly and aggressively. It is difficult for them to well understand complicated scenes. In this paper, we propose a global/local attention method that can recognize an image from coarse to fine by mimicking how human-beings observe images. Specifically, our global/local attention method first concentrates on the whole image, and then focuses on local specific objects in the image. We also propose a joint max-margin objective function, which enforces that the minimum score of positive labels should be larger than the maximum score of negative labels horizontally and vertically. This function can further improve our multi-label image classification method. We evaluate the effectiveness of our method on two popular multi-label image datasets (i.e., Pascal VOC and MS-COCO). Our experimental results show that our method outperforms state-of-the-art methods. | ['Baochuan Fu', 'Qiming Fu', 'Zhengtian Wu', 'Victor S. Sheng', 'Fuyuan Hu', 'Fan Lyu'] | 2020-12-26 | null | null | null | null | ['multi-label-image-classification'] | ['computer-vision'] | [ 2.40060747e-01 -3.67052257e-01 -1.60640553e-01 -5.58945417e-01
-6.62216246e-01 -4.36631680e-01 2.98148632e-01 8.86284746e-03
-3.46680045e-01 4.06781971e-01 -9.77898240e-02 -1.38127774e-01
1.66728467e-01 -6.45601749e-01 -5.20389795e-01 -7.92727351e-01
7.85129786e-01 3.46051872e-01 1.31524265e-01 3.16762067e-02
2.32785910e-01 3.21487308e-01 -1.46041298e+00 3.17463964e-01
7.69778430e-01 1.22555971e+00 3.72800291e-01 3.99391741e-01
4.68806513e-02 1.23792124e+00 -5.78751445e-01 -2.25863442e-01
-6.05483614e-02 -2.66257614e-01 -1.04883194e+00 6.16379142e-01
5.49528241e-01 -3.08810771e-01 -8.24816152e-02 1.49620628e+00
3.15575719e-01 2.24484265e-01 6.99372530e-01 -1.18436182e+00
-1.08442962e+00 9.50203985e-02 -8.94641757e-01 3.00874710e-01
8.75089765e-02 1.56883806e-01 1.20960701e+00 -8.43396842e-01
1.87107772e-01 1.33658910e+00 2.76069939e-01 3.28618169e-01
-8.87339592e-01 -7.24680901e-01 5.20493865e-01 4.24614459e-01
-1.48306668e+00 -1.48324132e-01 7.18241632e-01 -4.58472848e-01
4.28402513e-01 4.08266604e-01 3.41586590e-01 7.67706752e-01
1.04370996e-01 1.08905411e+00 1.30916715e+00 -3.80635440e-01
-2.15615809e-01 2.24323332e-01 2.75355995e-01 7.31789351e-01
2.65675858e-02 -2.88007140e-01 7.62379989e-02 2.17454895e-01
6.80417180e-01 4.27284718e-01 -2.41757497e-01 4.35479358e-02
-1.23737001e+00 8.79083157e-01 4.69220638e-01 4.17233735e-01
-3.89280975e-01 2.38827735e-01 2.42478043e-01 -1.30112663e-01
3.66375446e-01 1.38113171e-01 -3.69574517e-01 2.47730762e-01
-4.84457403e-01 -1.94500268e-01 1.78320706e-01 8.49531710e-01
9.83201802e-01 -3.91706795e-01 -3.39549124e-01 1.04914498e+00
4.24768955e-01 4.64341342e-01 5.05460978e-01 -7.98601627e-01
3.55361700e-01 8.66657257e-01 7.72414804e-02 -1.24400187e+00
-4.97914016e-01 -3.70076954e-01 -1.07794881e+00 6.07085563e-02
2.50883311e-01 4.35627177e-02 -8.43226969e-01 1.34265208e+00
4.12129283e-01 7.99921080e-02 -1.94142699e-01 1.10544276e+00
9.12666023e-01 7.18162477e-01 2.75528222e-01 -3.75373736e-02
1.49429643e+00 -1.56169033e+00 -7.97644496e-01 -5.31439185e-01
6.58841133e-01 -7.01689005e-01 1.38336074e+00 2.70989805e-01
-7.06519365e-01 -6.75593317e-01 -8.62915993e-01 -1.59246773e-01
-3.78273785e-01 4.73883331e-01 3.97053063e-01 3.84705216e-01
-6.31119430e-01 1.48899227e-01 -4.21349108e-01 -2.71504492e-01
4.28790182e-01 1.79176182e-01 -2.53456473e-01 -2.89939344e-01
-1.11498809e+00 7.50663996e-01 3.50182444e-01 1.32339224e-01
-8.35674763e-01 -6.69651628e-02 -7.69080400e-01 2.89067268e-01
4.12423611e-01 -2.33168989e-01 1.34515953e+00 -1.34665608e+00
-1.19034600e+00 1.14876020e+00 -1.97802186e-01 1.49410874e-01
1.93502724e-01 -9.35188234e-02 -5.38220882e-01 8.37968737e-02
4.57317352e-01 7.99477816e-01 6.20960355e-01 -1.52288592e+00
-9.92794454e-01 -2.07518101e-01 5.16139746e-01 2.88294762e-01
-4.51077878e-01 2.59700060e-01 -5.13647079e-01 -5.60082316e-01
6.37725294e-02 -8.94607425e-01 -2.01084048e-01 -9.76934060e-02
-4.77664888e-01 -5.01276016e-01 9.42692101e-01 -4.45827544e-01
1.20272970e+00 -2.29675937e+00 -1.86656237e-01 -4.76405695e-02
2.39673406e-01 2.71971405e-01 -1.88251346e-01 -6.80649206e-02
-1.08992808e-01 2.91071534e-01 1.53360851e-02 -3.32895696e-01
-4.78885546e-02 2.90521115e-01 -5.91703877e-03 5.41729391e-01
1.07320473e-01 9.29780722e-01 -8.64409685e-01 -7.90672898e-01
2.06057101e-01 2.84151137e-01 -2.23103538e-01 2.67937779e-01
-4.61790822e-02 5.83600104e-01 -5.88003278e-01 9.33077395e-01
5.34767270e-01 -1.10590076e+00 -7.73186088e-02 -2.36564189e-01
-2.33756080e-02 -1.34624809e-01 -1.14649546e+00 1.41862559e+00
-6.92455351e-01 5.23556292e-01 -1.38885928e-02 -1.07357359e+00
7.00618029e-01 2.72162646e-01 3.23815286e-01 -8.96256804e-01
4.83255118e-01 3.07817683e-02 -3.48875701e-01 -7.00706184e-01
3.67083967e-01 6.09209575e-02 -1.90111548e-01 5.52588701e-01
-3.86571407e-01 2.04569981e-01 2.13868976e-01 -1.76051911e-03
6.00066066e-01 -3.86787832e-01 4.06617254e-01 -2.23362446e-01
8.84939611e-01 -1.79789290e-01 7.95205534e-01 5.58724999e-01
-6.09810591e-01 5.38634479e-01 3.75860214e-01 -6.54102802e-01
-7.08710968e-01 -5.13861954e-01 2.02358942e-02 1.49936485e+00
6.92904711e-01 1.04528666e-02 -5.92488766e-01 -1.03302312e+00
-2.69029737e-01 4.14018571e-01 -8.14878047e-01 8.92066490e-03
-3.89566213e-01 -5.58888197e-01 2.87348807e-01 5.57194710e-01
9.42496181e-01 -1.28759646e+00 -4.90437508e-01 1.66292489e-02
-4.37098622e-01 -1.10599685e+00 -8.59800696e-01 3.11533604e-02
-2.78410196e-01 -1.03273690e+00 -6.65353894e-01 -1.09533191e+00
8.67109060e-01 7.42630184e-01 1.00411832e+00 4.04772878e-01
-1.77558661e-02 7.55630285e-02 -4.78937358e-01 -1.96657479e-01
3.07119191e-02 6.52397946e-02 -2.66530216e-01 5.35823166e-01
3.48850846e-01 -1.06524751e-01 -6.93098426e-01 7.60949433e-01
-1.03659296e+00 2.82604158e-01 7.39683747e-01 9.50358987e-01
7.88061738e-01 2.14446545e-01 4.67193455e-01 -8.72552395e-01
4.37516928e-01 -4.18614000e-01 -4.30609703e-01 6.91736877e-01
-4.52098906e-01 -1.41575634e-01 6.28889561e-01 -5.81045628e-01
-7.70222306e-01 2.09350809e-01 -3.96128148e-02 -5.10133862e-01
-4.53625798e-01 3.59516263e-01 -3.57128233e-01 -2.13092208e-01
1.20407395e-01 2.16543615e-01 -4.41892862e-01 -2.35611439e-01
2.89776474e-01 1.02703905e+00 2.66378850e-01 -4.12989199e-01
3.12819362e-01 5.35289645e-01 -1.19076125e-01 -4.74126458e-01
-1.46707046e+00 -8.24437976e-01 -4.49206650e-01 -5.94385564e-01
1.26688766e+00 -7.33647108e-01 -1.04237866e+00 6.32346451e-01
-1.19707155e+00 -2.64271408e-01 2.69660175e-01 1.90278575e-01
-2.63162196e-01 3.62869024e-01 -6.45420492e-01 -5.01760721e-01
-1.72460645e-01 -1.55662322e+00 1.45408046e+00 5.07008433e-01
5.50568327e-02 -1.10534632e+00 -1.62761837e-01 6.87129140e-01
3.81825924e-01 7.25570023e-02 7.15676546e-01 -4.13971543e-01
-6.30781412e-01 2.41541211e-03 -9.47568357e-01 3.35026115e-01
2.68363297e-01 -2.26567894e-01 -1.02194691e+00 -2.31676251e-01
-1.32152308e-02 -6.63690865e-01 7.37375975e-01 1.52654007e-01
1.59808087e+00 -3.36212873e-01 -3.72413754e-01 4.17301655e-01
1.48565936e+00 3.62945318e-01 5.17212391e-01 5.61415911e-01
1.00694013e+00 4.92270499e-01 7.36057341e-01 2.46899858e-01
6.82002425e-01 7.02314973e-01 7.06212044e-01 -3.94537657e-01
7.14206323e-02 2.09952176e-01 -2.94261966e-02 7.45110452e-01
1.58285290e-01 -4.97097671e-01 -9.46762562e-01 5.12229025e-01
-2.02496219e+00 -8.20893347e-01 -2.09137112e-01 1.75684738e+00
5.64887285e-01 8.98288097e-03 -3.26321095e-01 4.70139012e-02
9.73183155e-01 2.72260576e-01 -5.99748909e-01 -1.13714516e-01
-4.62674499e-02 -2.01805562e-01 4.83497083e-01 3.36755186e-01
-1.67840874e+00 8.61157358e-01 5.69429111e+00 1.09874773e+00
-1.32275188e+00 2.70853907e-01 1.06763601e+00 8.37414861e-02
6.44733682e-02 -3.61857414e-01 -8.67260695e-01 6.39399052e-01
4.52853799e-01 1.73684478e-01 3.06436062e-01 1.07637048e+00
-5.47214746e-02 -1.21955477e-01 -8.21256399e-01 1.15589428e+00
2.06090093e-01 -9.67604399e-01 -5.83162941e-02 3.21130380e-02
1.04293442e+00 -1.82787210e-01 2.24541068e-01 2.71652043e-01
4.51740801e-01 -1.10356855e+00 7.95358062e-01 4.72467005e-01
7.84233928e-01 -6.28134906e-01 7.93125212e-01 4.82476354e-01
-1.43998718e+00 -2.73737729e-01 -1.68863758e-01 8.44434835e-03
1.26784265e-01 5.56451261e-01 -2.38180622e-01 3.44646811e-01
6.35951757e-01 8.76094878e-01 -7.15933919e-01 8.55306268e-01
-4.53171432e-01 3.62083197e-01 5.55090699e-03 -9.28803906e-02
5.82873821e-01 -2.04176664e-01 -1.97998539e-01 1.09090698e+00
1.49565607e-01 3.31839710e-01 6.48241162e-01 5.22621751e-01
-1.46808922e-01 2.76294738e-01 -4.02850270e-01 1.89993978e-01
2.18775094e-01 1.54190361e+00 -8.32622707e-01 -4.94663596e-01
-8.20799112e-01 1.06680727e+00 6.07509971e-01 3.57292682e-01
-1.04233575e+00 -2.87434369e-01 4.79123831e-01 -2.18748450e-01
2.61268616e-01 4.85613057e-03 -3.87092233e-01 -1.06819749e+00
-1.56210929e-01 -8.40313256e-01 5.40169537e-01 -1.01962137e+00
-1.23234200e+00 5.92796803e-01 -5.62440038e-01 -1.24969435e+00
3.51416528e-01 -5.34158826e-01 -6.15147471e-01 6.39542639e-01
-1.50256348e+00 -1.34239364e+00 -6.53758407e-01 4.89630878e-01
7.99978614e-01 2.26728439e-01 5.65003514e-01 7.75388479e-01
-9.29895878e-01 6.00677192e-01 9.36877206e-02 3.84134293e-01
6.75764441e-01 -1.05531204e+00 -1.17826886e-01 6.37706280e-01
1.46356881e-01 3.51542681e-01 1.71082318e-01 -2.69998223e-01
-6.87992811e-01 -1.29835784e+00 8.24815035e-01 -2.32284427e-01
5.72491169e-01 -2.01291181e-02 -7.95541644e-01 8.29790711e-01
3.49358618e-01 1.71291739e-01 6.04989529e-01 -1.07229128e-01
-2.59072304e-01 -1.93125129e-01 -9.21846509e-01 3.37337077e-01
7.05998182e-01 -5.11251807e-01 -2.14391053e-01 7.02157557e-01
6.34166002e-01 -2.21551448e-01 -6.81696892e-01 5.11231661e-01
4.30019259e-01 -7.71641016e-01 8.18332911e-01 -3.70954990e-01
5.51543951e-01 -4.66944277e-01 -2.37642542e-01 -1.27974772e+00
-7.52652407e-01 2.15655133e-01 3.11579108e-01 1.16898513e+00
3.15736562e-01 -6.31145358e-01 5.50240874e-01 4.17052716e-01
-1.66728675e-01 -9.98717546e-01 -7.46633768e-01 -4.90166754e-01
-1.31549910e-01 2.69107576e-02 6.34291708e-01 1.30227017e+00
-4.08273101e-01 4.31093752e-01 -4.61111456e-01 3.91538531e-01
4.21713352e-01 3.89218450e-01 5.22067130e-01 -1.16697049e+00
-1.65054709e-01 -7.03583062e-01 -2.42086038e-01 -1.29282534e+00
2.36168265e-01 -5.93640924e-01 1.55870393e-01 -1.62337303e+00
6.22124732e-01 -6.17802978e-01 -5.83654642e-01 6.43689692e-01
-5.08035898e-01 3.82517785e-01 8.25355649e-02 4.72486198e-01
-1.21869922e+00 5.01545250e-01 1.62259054e+00 -6.58293724e-01
3.48587155e-01 -2.12044075e-01 -7.19852090e-01 7.98678339e-01
8.44895005e-01 -4.92076516e-01 -9.46183354e-02 -6.64474666e-01
1.94540367e-01 -1.49653688e-01 2.05782413e-01 -9.53697085e-01
2.82212049e-01 -5.28855324e-01 1.99567214e-01 -5.61547995e-01
2.39599317e-01 -9.87084627e-01 -6.40741810e-02 3.56542021e-01
-3.66539419e-01 1.40418112e-01 -1.82929114e-01 4.63468701e-01
-4.17224407e-01 -1.67638361e-01 1.00800705e+00 -2.51681924e-01
-9.76478517e-01 3.89281601e-01 -2.95916855e-01 -6.67542443e-02
1.31481910e+00 1.41815394e-01 -5.94203293e-01 -3.28995585e-01
-6.64231539e-01 6.36662245e-01 3.88264328e-01 4.02871788e-01
4.14533079e-01 -1.71900797e+00 -5.50622940e-01 9.20915231e-02
4.38003778e-01 3.94240431e-02 4.01340693e-01 7.85719395e-01
-5.13505459e-01 5.19620538e-01 -9.37178284e-02 -6.64254606e-01
-1.51717877e+00 8.21824610e-01 5.02425909e-01 -5.72746515e-01
-1.97437197e-01 9.05914426e-01 7.68281698e-01 -4.10885543e-01
2.58494824e-01 -2.62119144e-01 -5.36267638e-01 9.84797552e-02
6.47287548e-01 1.34706557e-01 -1.85446769e-01 -9.73513484e-01
-2.98998773e-01 9.57138538e-01 -3.52219701e-01 2.72885770e-01
8.81133735e-01 -3.07624459e-01 -3.45498145e-01 5.19683421e-01
1.41552007e+00 -3.44205648e-01 -1.06858730e+00 -3.66758645e-01
-2.92206913e-01 -5.69471240e-01 1.01553716e-01 -7.35485613e-01
-1.42086697e+00 9.24240470e-01 6.68667495e-01 4.38374460e-01
1.29072917e+00 1.25929847e-01 8.39075923e-01 2.91447371e-01
3.51552129e-01 -1.16487527e+00 4.87380594e-01 5.22348821e-01
5.88353336e-01 -1.76862514e+00 -1.79610610e-01 -4.04120624e-01
-7.46661603e-01 7.53138185e-01 8.89271498e-01 2.12198690e-01
5.64259648e-01 -2.35777780e-01 3.91895235e-01 -4.66626048e-01
-4.80201781e-01 -3.81874442e-01 3.01546693e-01 4.27419394e-02
2.36126870e-01 1.84340864e-01 -1.58574164e-01 3.85755301e-01
4.89070833e-01 -2.93886781e-01 3.59479994e-01 8.59361768e-01
-7.58345842e-01 -9.54168618e-01 -6.08262420e-01 3.49169910e-01
-5.80229878e-01 3.03151198e-02 -2.49516502e-01 4.72045541e-01
4.54235196e-01 1.26083219e+00 9.75893587e-02 -4.30406511e-01
1.87829345e-01 -2.25086272e-01 4.60854769e-02 -5.48726380e-01
-4.03495312e-01 -1.44794375e-01 -2.02702612e-01 -3.84501189e-01
-5.69862127e-01 -5.45677722e-01 -1.11660361e+00 -1.81873053e-01
-5.60774386e-01 -4.30317670e-02 5.43511629e-01 1.04365373e+00
6.70475140e-02 7.39292264e-01 9.02567625e-01 -6.26161814e-01
-1.60084188e-01 -1.07616842e+00 -5.91259956e-01 7.97329664e-01
2.29605347e-01 -7.81590223e-01 -3.23648840e-01 1.31315127e-01] | [9.80652904510498, 3.95959734916687] |
e65ce18f-8873-471e-98cc-5e39fbcf1698 | a-semantics-based-approach-to-disclosure | null | null | https://aclanthology.org/2020.findings-emnlp.312 | https://aclanthology.org/2020.findings-emnlp.312.pdf | A Semantics-based Approach to Disclosure Classification in User-Generated Online Content | As users engage in public discourse, the rate of voluntarily disclosed personal information has seen a steep increase. So-called self-disclosure can result in a number of privacy concerns. Users are often unaware of the sheer amount of personal information they share across online forums, commentaries, and social networks, as well as the power of modern AI to synthesize and gain insights from this data. This paper presents an approach to detect emotional and informational self-disclosure in natural language. We hypothesize that identifying frame semantics can meaningfully support this task. Specifically, we use Semantic Role Labeling to identify the lexical units and their semantic roles that signal self-disclosure. Experimental results on Reddit data show the performance gain of our method when compared to standard text classification methods based on BiLSTM, and BERT. In addition to improved performance, our approach provides insights into the drivers of disclosure behaviors. | ['Sarah Rajtmajer', 'Anna Squicciarini', 'Chandan Akiti'] | 2020-11-01 | null | null | null | findings-of-the-association-for-computational | ['semantic-role-labeling'] | ['natural-language-processing'] | [ 1.95813403e-01 7.68372238e-01 -7.24201798e-01 -7.25286722e-01
-6.49818122e-01 -6.39449537e-01 7.66967952e-01 5.30153930e-01
-2.09050596e-01 8.03413093e-01 1.14896488e+00 1.07472152e-01
4.77217346e-01 -4.14477557e-01 -7.14316219e-02 2.33555343e-02
1.03382722e-01 -7.87704065e-02 -3.36064905e-01 -3.54060322e-01
1.62865594e-01 -1.22651316e-01 -1.43156838e+00 7.09177017e-01
6.11219823e-01 1.17961442e+00 -4.04462248e-01 1.19059443e-01
-4.74098206e-01 1.07509375e+00 -6.70524240e-01 -8.04895401e-01
-2.37760410e-01 -4.10749137e-01 -9.59627986e-01 1.37461066e-01
2.80646324e-01 -3.91754746e-01 -9.64407027e-02 1.08713484e+00
2.47813731e-01 7.96277002e-02 2.22764552e-01 -1.29485929e+00
-9.20663953e-01 9.57011998e-01 -2.92231053e-01 3.02073896e-01
7.75287986e-01 -2.95121014e-01 1.52180779e+00 -3.86620790e-01
9.99549329e-01 1.52055216e+00 5.06834626e-01 8.82265151e-01
-1.27160215e+00 -6.10739768e-01 2.07378238e-01 -1.75259829e-01
-7.75871038e-01 -9.76808488e-01 1.08589005e+00 -5.76359391e-01
7.82984018e-01 4.24662232e-01 6.38578057e-01 1.73094845e+00
-1.27250016e-01 9.95671809e-01 1.01965249e+00 -1.81509346e-01
1.83142245e-01 6.67945087e-01 4.06016231e-01 6.18520141e-01
4.06671882e-01 -3.14946353e-01 -1.06203973e+00 -5.20056129e-01
-7.91446194e-02 -1.71989784e-01 -2.53045976e-01 1.26165226e-01
-9.48294342e-01 9.76718068e-01 9.35697258e-02 6.61622822e-01
-1.59879699e-01 -2.85364669e-02 8.03337455e-01 3.44168305e-01
1.13030052e+00 7.23548770e-01 -8.06526169e-02 -5.02094090e-01
-5.90672314e-01 2.15244725e-01 1.13794088e+00 6.05580449e-01
5.33450842e-01 -3.26654106e-01 -2.51784265e-01 7.19776690e-01
2.52460420e-01 -1.46189362e-01 9.18391466e-01 -1.07479417e+00
4.64476258e-01 8.47030401e-01 3.27714324e-01 -1.39687276e+00
-1.59676284e-01 -1.63299337e-01 -4.64154541e-01 -3.09109658e-01
2.32408211e-01 -2.87517518e-01 -7.16783032e-02 1.89113319e+00
1.86303079e-01 -4.63152409e-01 -8.59807059e-03 6.53914034e-01
5.29174626e-01 2.74024785e-01 2.51045346e-01 -2.90890396e-01
1.57038593e+00 -2.68940419e-01 -1.41252995e+00 -5.14908969e-01
8.11458468e-01 -3.25051844e-01 9.06467736e-01 -1.02565391e-02
-7.06375599e-01 2.58674193e-02 -1.10187912e+00 -5.18453062e-01
-2.45379597e-01 -2.10744098e-01 7.57657290e-01 9.03208017e-01
-6.73007250e-01 7.64624238e-01 -5.32007635e-01 -6.16120636e-01
9.28838015e-01 -2.56042689e-01 -2.64096439e-01 5.67375243e-01
-1.44374502e+00 6.11792266e-01 2.86226064e-01 -1.93990394e-01
8.43223408e-02 -6.37828052e-01 -1.06053662e+00 1.07549295e-01
6.42557323e-01 -3.62999260e-01 1.49081302e+00 -1.45284200e+00
-1.30864501e+00 1.32138288e+00 -4.97173548e-01 -7.12956607e-01
5.26242077e-01 -3.65571171e-01 -4.66333985e-01 -5.11461571e-02
2.53882736e-01 5.64750612e-01 8.63203168e-01 -8.29306245e-01
-4.95753884e-01 -5.31542003e-01 6.31462410e-02 2.95647979e-01
-7.80960441e-01 2.52011806e-01 4.65019256e-01 -6.79263890e-01
-1.69309393e-01 -5.48709989e-01 3.48676443e-01 2.72686452e-01
-5.40732741e-01 -5.52072763e-01 9.06270146e-01 -8.35576594e-01
1.31178272e+00 -2.40985036e+00 -2.88207501e-01 -1.71926931e-01
6.10777855e-01 2.03439295e-02 5.20457625e-01 4.57122892e-01
1.33259743e-01 9.45183039e-01 -1.86278783e-02 -5.08809626e-01
1.86267555e-01 5.20731322e-02 -6.61848843e-01 3.78539741e-01
1.46249294e-01 9.10141945e-01 -8.66571248e-01 -3.23334605e-01
-3.04088205e-01 1.99827671e-01 -3.25974733e-01 9.31598097e-02
-5.76166272e-01 1.38003841e-01 -6.90418303e-01 4.59001154e-01
1.07293285e-01 -5.53279340e-01 3.86057079e-01 6.65088072e-02
-1.40520204e-02 1.13562250e+00 -4.89269227e-01 1.38232017e+00
2.16520224e-02 1.01276124e+00 3.96269172e-01 -6.63969934e-01
7.91706681e-01 3.33144128e-01 3.41338038e-01 -6.12460017e-01
3.30297858e-01 -2.08553448e-01 -1.30987599e-01 -4.66336548e-01
5.40096462e-01 -3.59925717e-01 -3.14230412e-01 9.73533273e-01
-4.09507632e-01 4.92599308e-01 -1.63367212e-01 2.80145556e-01
7.37101555e-01 -1.62218317e-01 7.24162877e-01 -3.58523905e-01
3.41642499e-01 -3.92238162e-02 6.39463961e-01 3.38508755e-01
-6.17468715e-01 -4.47000787e-02 9.58099544e-01 -3.36171925e-01
-6.39502287e-01 -3.73852342e-01 7.49030262e-02 1.17578638e+00
9.74904299e-02 -5.97405970e-01 -7.52281308e-01 -5.01271069e-01
2.72171915e-01 9.91638184e-01 -7.48593807e-01 -2.70929128e-01
-2.30860665e-01 -4.52964157e-01 5.18132448e-01 9.02750045e-02
7.13988721e-01 -9.92866576e-01 -8.43555450e-01 1.35993108e-01
-5.76105475e-01 -1.22623742e+00 -6.54539943e-01 -3.42337608e-01
-6.75636530e-01 -9.36526954e-01 -6.26289239e-03 -3.37908179e-01
2.38243148e-01 2.46960431e-01 9.82793570e-01 -5.52399829e-02
-1.19699210e-01 3.81945431e-01 -1.22405365e-01 -5.08546829e-01
-6.71130121e-01 1.23107217e-01 -4.13431833e-03 6.57593548e-01
7.68053591e-01 -5.20876288e-01 -3.53448451e-01 -4.95114177e-02
-7.44159520e-01 7.89195299e-02 -1.48731768e-01 4.09312576e-01
-1.33522943e-01 -3.13101321e-01 6.52628958e-01 -1.50044870e+00
1.36818337e+00 -9.34191227e-01 7.94044882e-03 7.68574625e-02
-9.13394511e-01 1.17542461e-01 3.08745831e-01 -1.31585404e-01
-1.36063564e+00 -3.47306013e-01 4.14512813e-01 1.19890764e-01
-1.11379780e-01 4.78148818e-01 -1.96508378e-01 4.87889677e-01
7.61328995e-01 -2.54828990e-01 3.69850069e-01 -3.89466912e-01
4.10053223e-01 8.80436540e-01 2.25508794e-01 -4.02712613e-01
5.60487807e-01 7.24377930e-01 -6.93358004e-01 -9.43572044e-01
-1.36913168e+00 -2.78686166e-01 -1.26112014e-01 -1.39174908e-01
9.10191357e-01 -9.17533934e-01 -1.00775635e+00 3.15615684e-01
-1.15416169e+00 1.22144111e-01 -2.70501852e-01 -3.33696194e-02
-1.57646343e-01 5.63220799e-01 -7.49189913e-01 -1.08034062e+00
-4.74203706e-01 -2.83630759e-01 7.40925968e-01 -1.08748591e-02
-1.24060547e+00 -1.11066163e+00 -1.69658527e-01 8.02040815e-01
3.77662092e-01 4.66222525e-01 8.84417474e-01 -1.06252110e+00
-3.15938026e-01 -1.49697080e-01 -5.23962267e-02 1.65503681e-01
5.10230303e-01 -4.76989150e-01 -1.25819528e+00 -5.44178337e-02
3.63469213e-01 -6.26224995e-01 5.96351922e-01 -1.16669320e-01
1.07791948e+00 -1.11820686e+00 -3.40637475e-01 1.48868054e-01
8.82943511e-01 4.60335389e-02 2.41981670e-01 2.02033952e-01
4.94167477e-01 1.14236760e+00 2.82339633e-01 7.92338252e-01
6.20761812e-01 3.93206805e-01 1.70130685e-01 3.97767633e-01
3.41186911e-01 -6.29167616e-01 5.33412993e-01 1.81055427e-01
4.60190743e-01 -3.25844914e-01 -6.65669799e-01 3.15932065e-01
-1.85696495e+00 -1.24234688e+00 2.06775069e-01 1.85753644e+00
9.97026861e-01 3.17862988e-01 1.93774268e-01 -1.40926074e-02
9.37390864e-01 7.28809714e-01 -8.81631076e-01 -5.69212437e-01
-2.37343818e-01 -3.30403745e-01 4.00140613e-01 5.70202589e-01
-9.80818093e-01 8.18346679e-01 6.21940756e+00 1.76728204e-01
-1.11706388e+00 2.36379787e-01 1.04865861e+00 -2.13429675e-01
-7.18431890e-01 -2.39037663e-01 -7.65690207e-01 6.02128267e-01
9.82057691e-01 -8.01291466e-01 4.00989056e-01 7.96832740e-01
1.70605138e-01 -6.29409850e-02 -1.44768178e+00 1.06546080e+00
3.66953969e-01 -1.40380347e+00 -2.26843745e-01 1.80366173e-01
2.23742008e-01 -3.75248373e-01 6.86254054e-02 -4.15427312e-02
2.20330179e-01 -7.96076834e-01 9.72262084e-01 1.21090338e-01
5.35442829e-01 -2.67409831e-01 3.82157296e-01 4.93507683e-01
-4.54650402e-01 -9.63123813e-02 1.36007771e-01 -4.88615930e-01
1.46911085e-01 5.25023818e-01 -1.11337090e+00 -1.62162185e-01
5.08020759e-01 1.19166744e+00 -4.35291231e-01 1.61330961e-02
-3.22858274e-01 6.30975068e-01 -7.59519860e-02 -4.64655310e-01
-1.59920037e-01 -9.04417858e-02 6.84213281e-01 9.56374347e-01
5.63435517e-02 1.41573086e-01 -1.24067798e-01 1.35921252e+00
-5.42462051e-01 -1.07337497e-01 -7.43349552e-01 -1.08301735e+00
5.82392871e-01 1.05636406e+00 -3.71845514e-01 -2.92684644e-01
-3.81543577e-01 1.13524210e+00 5.18099666e-01 2.41232038e-01
-2.79857635e-01 1.06678225e-01 1.13614619e+00 5.18067956e-01
-2.21051991e-01 1.42595572e-02 -4.72938120e-01 -1.39253294e+00
4.37105566e-01 -6.86756730e-01 5.48042238e-01 -6.65296972e-01
-1.41069460e+00 2.22459406e-01 -1.85474157e-01 -5.86391091e-01
-5.57224810e-01 -2.73660898e-01 -4.35548246e-01 6.29069090e-01
-1.14798021e+00 -7.96755552e-01 -1.23381026e-01 1.46944076e-01
4.26055580e-01 -7.28243515e-02 8.34416032e-01 3.39181162e-02
-4.82628942e-01 5.82872570e-01 -3.10535938e-01 3.37223083e-01
7.21826494e-01 -1.02775180e+00 7.06910253e-01 6.17078841e-01
2.82451753e-02 8.13026190e-01 7.87512064e-01 -8.99186313e-01
-1.24755251e+00 -8.32585514e-01 1.38379312e+00 -7.48412549e-01
8.49812269e-01 -8.15116644e-01 -1.06440377e+00 1.05059505e+00
2.93214470e-01 -4.78360653e-01 1.26916754e+00 3.26027185e-01
-7.50975192e-01 5.68008758e-02 -1.43670428e+00 5.78688443e-01
1.29079890e+00 -1.15514493e+00 -9.47342515e-01 3.62643659e-01
9.53386843e-01 -1.86329722e-01 -5.48093379e-01 -3.62697840e-01
5.42125940e-01 -1.01155317e+00 5.11101246e-01 -8.07453930e-01
5.52952707e-01 3.78302246e-01 4.99592163e-03 -1.04188347e+00
-1.42362922e-01 -1.24504638e+00 -2.05202669e-01 1.63832617e+00
3.45030725e-01 -8.54863703e-01 8.04502606e-01 1.54221380e+00
2.54983157e-01 -2.20805526e-01 -8.84547114e-01 -1.96080163e-01
-3.50750208e-01 -2.29547322e-01 3.92526388e-01 1.47565377e+00
6.01780534e-01 8.61039460e-01 -3.02288920e-01 -3.71261328e-01
6.73088431e-01 4.84058559e-02 1.86982736e-01 -1.55942500e+00
1.10022724e-02 -4.72352177e-01 -9.18543413e-02 -6.12787068e-01
6.34035051e-01 -1.02015781e+00 -5.94790041e-01 -1.25023186e+00
1.88432872e-01 -3.31084356e-02 1.15337737e-01 7.15083778e-01
7.34693259e-02 -1.96488008e-01 6.09429106e-02 2.59220958e-01
-6.01754665e-01 5.23143291e-01 6.80766940e-01 -1.25409499e-01
-3.43265861e-01 -2.56967217e-01 -1.58183455e+00 6.61617219e-01
9.23864603e-01 -3.06955785e-01 -3.84307384e-01 3.72784324e-02
7.76759326e-01 -7.62284547e-02 2.54191011e-01 -4.27322268e-01
-1.62945539e-01 2.60051396e-02 1.44241169e-01 -1.18307851e-01
3.41439694e-01 -5.89472353e-01 -1.48648575e-01 3.05536568e-01
-1.14818084e+00 -2.11496443e-01 2.16032658e-02 8.19043279e-01
-2.06651524e-01 5.15750889e-03 5.24740458e-01 -2.62965292e-01
-6.02910757e-01 -4.84236553e-02 -6.94233716e-01 4.20363158e-01
8.77795160e-01 -2.81920612e-01 -4.89619046e-01 -8.84227037e-01
-6.00939810e-01 2.54464507e-01 4.01498109e-01 1.05056036e+00
2.68806547e-01 -1.32766318e+00 -5.10464251e-01 1.22831054e-01
2.02154100e-01 -4.38354492e-01 -2.37814113e-02 2.10340917e-01
2.34162405e-01 3.76545191e-01 -1.40145734e-01 -5.42963296e-02
-1.34286928e+00 3.45560402e-01 1.64151549e-01 1.77175432e-01
-6.29740119e-01 7.23480821e-01 1.50508452e-02 9.71140116e-02
2.55223572e-01 -3.42093199e-01 -3.86385322e-01 7.08002985e-01
6.91588759e-01 2.60316342e-01 -2.52031624e-01 -6.38335347e-01
-2.97341138e-01 -4.00688171e-01 -2.60158002e-01 -3.18743438e-01
1.07501602e+00 -3.30344379e-01 -4.19169933e-01 7.91520178e-01
1.52033532e+00 1.94051526e-02 -9.08538818e-01 -3.99465322e-01
6.04307055e-01 -5.52283168e-01 -1.17662214e-01 -7.16192007e-01
-5.55521846e-01 5.72394669e-01 1.15700617e-01 6.09758198e-01
4.84754562e-01 1.97639570e-01 1.00898778e+00 3.64607245e-01
1.41607359e-01 -1.38413906e+00 3.02536786e-02 1.47450939e-01
9.23875511e-01 -1.39283407e+00 -6.00712858e-02 -6.29754961e-01
-8.96789193e-01 8.70457888e-01 5.87153912e-01 3.24945599e-01
5.06487131e-01 -3.08786258e-02 2.09455732e-02 -2.00087771e-01
-1.00383532e+00 2.37841085e-01 2.30267078e-01 3.25069606e-01
8.99907649e-01 1.19574763e-01 -3.93408865e-01 8.83886218e-01
-5.05015373e-01 -1.28391981e-01 7.07938015e-01 7.92740822e-01
-4.81745750e-01 -1.02816308e+00 -1.17369154e-02 5.27210474e-01
-9.37348187e-01 6.77530542e-02 -1.21324980e+00 2.45054826e-01
-3.62419754e-01 1.02648795e+00 1.09901361e-01 -3.08877856e-01
-1.09746762e-01 5.70434153e-01 -1.36361927e-01 -7.88935125e-01
-7.65199959e-01 -2.70283073e-01 8.48422706e-01 -8.93036306e-01
-6.09136641e-01 -1.04417050e+00 -1.20940912e+00 -2.74230182e-01
3.04939091e-01 2.07685590e-01 5.13828874e-01 9.46619749e-01
8.13930154e-01 4.82872166e-02 3.70286405e-01 2.93243099e-02
-6.79651439e-01 -8.30014169e-01 -2.55400985e-01 8.32484245e-01
6.52250767e-01 -2.87681222e-01 -5.46788216e-01 2.80334428e-02] | [8.799736022949219, 10.087385177612305] |
67b53ef2-32e8-4b62-bf26-932a75528f76 | beyond-appearance-a-semantic-controllable | 2303.17602 | null | https://arxiv.org/abs/2303.17602v1 | https://arxiv.org/pdf/2303.17602v1.pdf | Beyond Appearance: a Semantic Controllable Self-Supervised Learning Framework for Human-Centric Visual Tasks | Human-centric visual tasks have attracted increasing research attention due to their widespread applications. In this paper, we aim to learn a general human representation from massive unlabeled human images which can benefit downstream human-centric tasks to the maximum extent. We call this method SOLIDER, a Semantic cOntrollable seLf-supervIseD lEaRning framework. Unlike the existing self-supervised learning methods, prior knowledge from human images is utilized in SOLIDER to build pseudo semantic labels and import more semantic information into the learned representation. Meanwhile, we note that different downstream tasks always require different ratios of semantic information and appearance information. For example, human parsing requires more semantic information, while person re-identification needs more appearance information for identification purpose. So a single learned representation cannot fit for all requirements. To solve this problem, SOLIDER introduces a conditional network with a semantic controller. After the model is trained, users can send values to the controller to produce representations with different ratios of semantic information, which can fit different needs of downstream tasks. Finally, SOLIDER is verified on six downstream human-centric visual tasks. It outperforms state of the arts and builds new baselines for these tasks. The code is released in https://github.com/tinyvision/SOLIDER. | ['Xiuyu Sun', 'Rong Jin', 'Fan Wang', 'Yaohua Wang', 'Hao Luo', 'Jian Jia', 'Xianzhe Xu', 'Weihua Chen'] | 2023-03-30 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Chen_Beyond_Appearance_A_Semantic_Controllable_Self-Supervised_Learning_Framework_for_Human-Centric_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Chen_Beyond_Appearance_A_Semantic_Controllable_Self-Supervised_Learning_Framework_for_Human-Centric_CVPR_2023_paper.pdf | cvpr-2023-1 | ['person-re-identification', 'pedestrian-attribute-recognition', 'pedestrian-detection', 'human-parsing', 'person-search'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [-2.25142129e-02 2.04938471e-01 -2.15387121e-01 -8.17608476e-01
-2.82277822e-01 -3.91355604e-01 3.93703222e-01 -2.69043177e-01
-4.84244704e-01 4.27286506e-01 2.90922999e-01 5.09748869e-02
5.04999816e-01 -5.19713283e-01 -5.84976554e-01 -4.30392236e-01
4.70472634e-01 4.64530766e-01 2.41512731e-01 -1.25180647e-01
-1.90964893e-01 -4.64504324e-02 -1.43979955e+00 4.84144241e-01
6.33911490e-01 8.82573903e-01 5.26484430e-01 3.26328009e-01
-3.49267200e-02 6.43105268e-01 -5.61097264e-01 -4.31376278e-01
4.41588014e-01 -4.28804278e-01 -7.00673163e-01 3.49334657e-01
4.89511758e-01 -4.22701538e-01 -3.79949808e-01 1.17258954e+00
5.44681847e-01 1.90745935e-01 5.69885492e-01 -1.58399618e+00
-1.06621683e+00 4.94896591e-01 -6.41022205e-01 -5.09557277e-02
1.74629882e-01 6.46923423e-01 8.41126800e-01 -8.07705104e-01
4.62038428e-01 1.47304547e+00 3.83996904e-01 1.06980205e+00
-9.10985291e-01 -7.51501024e-01 5.65386236e-01 2.09495053e-01
-1.10165226e+00 -4.22809005e-01 6.49476051e-01 -4.13453668e-01
5.40421426e-01 6.98926672e-02 5.36099374e-01 1.49347091e+00
-4.10168827e-01 1.13569522e+00 1.20165515e+00 -2.37553924e-01
3.83890308e-02 3.56991857e-01 2.74298400e-01 7.78696358e-01
3.34554791e-01 1.85673442e-02 -4.88306046e-01 3.49689871e-01
8.78583193e-01 4.81348395e-01 -2.10208118e-01 -3.39011222e-01
-1.20840561e+00 6.22081101e-01 7.84078479e-01 1.73702329e-01
-3.66239622e-02 2.15740740e-01 3.96089971e-01 2.35245079e-01
2.82352477e-01 1.13466270e-01 -5.78117192e-01 2.53679663e-01
-5.56089282e-01 -6.38102368e-02 3.45277995e-01 1.29185271e+00
8.57757092e-01 -6.28992766e-02 -3.12101811e-01 8.40496302e-01
5.83412766e-01 5.89265883e-01 6.84035480e-01 -8.62130404e-01
5.58600843e-01 8.47733140e-01 7.54587799e-02 -7.20468879e-01
-4.22064126e-01 -4.44089353e-01 -9.43797112e-01 1.66582763e-01
6.07446492e-01 -2.81915367e-01 -1.19093955e+00 1.90357184e+00
2.63390154e-01 3.56970467e-02 3.72691220e-03 1.33565485e+00
1.07291329e+00 5.62370181e-01 4.52258736e-01 2.58043826e-01
1.59207869e+00 -1.47581100e+00 -5.49927413e-01 -6.96825147e-01
4.75653440e-01 -3.94975871e-01 1.32324874e+00 2.02064782e-01
-7.11666703e-01 -9.32114303e-01 -9.09159064e-01 -3.36337835e-01
-4.67386901e-01 5.25385082e-01 5.63471675e-01 4.97357339e-01
-1.08061349e+00 2.24420071e-01 -6.54964924e-01 -5.12884319e-01
4.12197322e-01 1.10543601e-01 -4.98000562e-01 -2.24527746e-01
-1.15893722e+00 6.07256353e-01 4.50324565e-01 9.51337516e-02
-7.53697753e-01 -3.61630052e-01 -1.07099819e+00 -7.03663682e-04
4.33649778e-01 -9.26208556e-01 1.38344097e+00 -1.34779537e+00
-1.19318306e+00 1.15700626e+00 -1.35919765e-01 -1.95870429e-01
7.18433380e-01 -2.63527781e-01 -3.23443919e-01 1.97385371e-01
4.08140212e-01 1.01247334e+00 1.06486166e+00 -1.36899829e+00
-5.68173885e-01 -5.36055744e-01 1.11304559e-01 2.42586300e-01
-4.14869189e-01 -7.19659552e-02 -9.52970028e-01 -7.37945676e-01
-1.28404424e-01 -1.06308198e+00 -2.98890173e-01 2.54878521e-01
-5.60289443e-01 -3.47461194e-01 9.16447818e-01 -6.94698751e-01
6.97990596e-01 -2.31347227e+00 1.05881080e-01 2.31767148e-02
4.84370768e-01 3.37477684e-01 -3.42055738e-01 -5.29009383e-03
3.04443240e-02 -7.97839239e-02 -1.68594599e-01 -6.94844544e-01
6.35602549e-02 3.84736527e-03 7.43829310e-02 2.30806544e-01
2.55634878e-02 1.24312115e+00 -9.48293388e-01 -6.15766168e-01
2.94600308e-01 4.27950889e-01 -2.55777001e-01 4.17842776e-01
-1.79853112e-01 6.19613945e-01 -6.47550106e-01 6.74885035e-01
6.51265681e-01 -6.51426554e-01 1.61176156e-02 -2.83241302e-01
2.70873785e-01 -2.91813433e-01 -9.22586918e-01 1.73144817e+00
-4.05161768e-01 4.05435503e-01 9.01676994e-03 -1.10897517e+00
1.01811695e+00 5.76861128e-02 2.16699168e-01 -9.04552579e-01
1.76703319e-01 -1.21459685e-01 -2.99997270e-01 -4.85955656e-01
1.07890435e-01 -7.75259212e-02 -1.95928916e-01 4.10111576e-01
1.66175112e-01 3.94901454e-01 -5.82655966e-02 3.87277633e-01
7.72018373e-01 2.89376616e-01 1.51380479e-01 -3.51742245e-02
4.83887434e-01 -1.21668447e-02 7.70982623e-01 5.27395606e-01
-6.04986131e-01 8.45530272e-01 1.37755588e-01 -4.56842244e-01
-8.09499264e-01 -1.13437378e+00 3.51599902e-01 1.40722418e+00
5.22321284e-01 -3.48092914e-01 -8.04893017e-01 -9.75325465e-01
5.30511811e-02 4.31399792e-01 -5.92316508e-01 -2.26941973e-01
-3.70551378e-01 -3.58812362e-01 2.85491973e-01 7.61602581e-01
1.00997961e+00 -1.38803363e+00 -6.36152923e-01 -1.10750325e-01
-4.06885982e-01 -1.22498953e+00 -6.85432374e-01 -1.32157668e-01
-5.65123439e-01 -1.27880073e+00 -9.39488471e-01 -1.02621615e+00
1.05709291e+00 6.66887879e-01 9.43578959e-01 8.48096609e-02
-5.01127541e-01 6.22094631e-01 -6.31363153e-01 -2.50250161e-01
-9.41266045e-02 3.60346437e-02 -4.90560941e-02 2.29623631e-01
6.09298289e-01 -1.20801076e-01 -9.16823983e-01 3.85187775e-01
-6.91178143e-01 4.40236449e-01 6.59518123e-01 7.57481873e-01
4.10690188e-01 -1.50939390e-01 5.84499538e-01 -1.05202973e+00
3.74727279e-01 -4.25630450e-01 -3.29588592e-01 3.62277597e-01
-3.80630404e-01 9.32086110e-02 7.31571496e-01 -4.13478881e-01
-1.33570552e+00 3.78241509e-01 2.01196596e-01 -5.48905075e-01
-3.28035742e-01 7.57346824e-02 -5.06385028e-01 4.33691382e-01
6.38169527e-01 1.94632813e-01 5.76240011e-02 -5.38839698e-01
6.31562114e-01 8.23408902e-01 6.63652360e-01 -5.54397285e-01
8.67539942e-01 4.48073566e-01 -5.51374018e-01 -3.80038053e-01
-1.16403520e+00 -6.84224010e-01 -6.15738451e-01 -1.82097182e-01
1.09631681e+00 -1.23924839e+00 -5.89998603e-01 7.13796854e-01
-1.13907361e+00 -6.58421576e-01 -2.10363805e-01 1.84880227e-01
-4.92256016e-01 2.98029006e-01 -6.03290975e-01 -7.13755071e-01
-3.98645818e-01 -1.07829666e+00 1.17737806e+00 5.19747674e-01
-6.86250329e-02 -8.80280912e-01 -4.56915140e-01 7.57332563e-01
2.89005995e-01 7.09754974e-03 3.60359758e-01 -3.98303539e-01
-4.41759795e-01 -6.44820370e-03 -6.97530031e-01 4.96336341e-01
2.96116084e-01 -5.08971274e-01 -1.03393984e+00 -3.79068673e-01
-1.75809249e-01 -5.78033388e-01 1.16014194e+00 2.20377669e-01
1.28879356e+00 -4.07707632e-01 -4.57223564e-01 6.63418472e-01
1.22849774e+00 -7.99653158e-02 5.17541409e-01 4.37273979e-01
1.04702652e+00 9.39748287e-01 7.23976493e-01 3.21518481e-01
6.90477967e-01 6.18711352e-01 1.79843143e-01 -4.35943335e-01
-4.23305660e-01 -5.53582013e-01 4.32659954e-01 3.21471572e-01
-8.82314239e-03 -1.36174053e-01 -7.89275765e-01 3.57846409e-01
-2.22935629e+00 -8.72425497e-01 -4.38731685e-02 1.94034910e+00
7.64955640e-01 -1.62230119e-01 2.27837116e-01 -3.19406897e-01
1.07617319e+00 6.53050467e-02 -9.13828731e-01 1.65747240e-01
7.07915276e-02 -2.45592311e-01 4.39551413e-01 1.46516979e-01
-1.20099139e+00 1.38689125e+00 5.21970701e+00 5.33593774e-01
-9.81612802e-01 3.72417748e-01 7.95943677e-01 3.40792648e-02
-4.63246889e-02 -7.24360123e-02 -8.91287029e-01 7.62624741e-01
3.29652518e-01 -1.49879353e-02 1.44771487e-01 1.10907972e+00
1.92143470e-01 1.13926977e-01 -1.04345894e+00 1.43894839e+00
1.88030109e-01 -8.14493775e-01 1.29531518e-01 -1.91824004e-01
6.04223907e-01 -8.71836245e-02 4.33228649e-02 4.35324490e-01
5.29094696e-01 -9.27960515e-01 6.98926032e-01 4.91960645e-01
1.00695682e+00 -4.81850892e-01 6.69478357e-01 3.49263668e-01
-1.34289455e+00 -2.05258369e-01 -4.14262682e-01 -4.47665192e-02
2.55424201e-01 5.53175926e-01 -3.81268293e-01 2.11968645e-01
9.06156719e-01 8.86402845e-01 -8.06885362e-01 7.32414901e-01
-6.72405124e-01 2.95290083e-01 6.49025217e-02 2.47097522e-01
7.46936873e-02 -1.20055594e-01 -2.72210445e-02 1.23316658e+00
9.45141241e-02 6.39918298e-02 5.31091750e-01 8.24369073e-01
-1.06645525e-01 4.70568687e-02 -4.58514065e-01 1.44002169e-01
4.47599411e-01 1.40405273e+00 -5.62619865e-01 -5.68690419e-01
-5.83970726e-01 1.56584275e+00 4.64517474e-01 6.11989856e-01
-7.91566133e-01 -1.56565607e-01 4.71712679e-01 2.47335866e-01
7.68552795e-02 -2.00956658e-01 -7.47641250e-02 -1.51878810e+00
3.31288464e-02 -8.44507873e-01 5.70007801e-01 -9.15395439e-01
-1.63064671e+00 5.03589153e-01 -1.55113816e-01 -1.18039322e+00
-1.38562649e-01 -7.54018426e-01 -4.78913814e-01 6.99211657e-01
-1.54691410e+00 -1.61899292e+00 -8.81019711e-01 9.78243887e-01
7.62341261e-01 -3.28413516e-01 5.31374872e-01 1.78687483e-01
-6.97053015e-01 7.69549966e-01 -5.43983817e-01 4.83264238e-01
1.09378672e+00 -1.30030000e+00 5.75715125e-01 8.26440990e-01
-4.40209433e-02 6.36415124e-01 4.18364078e-01 -7.15992570e-01
-1.10188925e+00 -1.41364181e+00 7.50731766e-01 -3.44402283e-01
3.47884357e-01 -3.24752688e-01 -7.95439124e-01 9.91072357e-01
1.71190258e-02 3.11331570e-01 5.85019469e-01 7.36226737e-02
-7.59414315e-01 -1.22486807e-01 -1.11133611e+00 6.56463683e-01
1.52801085e+00 -3.63571346e-01 -5.04041076e-01 5.36866844e-01
8.56029391e-01 -1.41787052e-01 -3.33993375e-01 2.36009240e-01
2.78284669e-01 -8.98917973e-01 1.01596332e+00 -5.11551917e-01
2.55856305e-01 -5.43426514e-01 1.25566408e-01 -1.31253970e+00
-4.10818607e-01 -3.22833508e-01 -1.06598534e-01 1.21862912e+00
2.43392661e-01 -7.41126835e-01 8.42171252e-01 8.94325137e-01
7.67863297e-04 -3.32983255e-01 -5.14430583e-01 -9.15506363e-01
-2.08580673e-01 -1.08826034e-01 4.66408163e-01 1.05847764e+00
3.88309471e-02 6.61158383e-01 -6.71745479e-01 3.16304751e-02
9.17833388e-01 3.06378659e-02 9.48114634e-01 -1.20347142e+00
-3.95373970e-01 -2.65325695e-01 -3.46935034e-01 -1.32530856e+00
2.42331117e-01 -9.80383098e-01 1.31260902e-02 -1.75687373e+00
6.27248108e-01 -6.70696199e-01 -2.76324898e-01 1.04046082e+00
-4.42484081e-01 3.16194445e-01 5.28170645e-01 3.19557369e-01
-9.30844367e-01 6.12538755e-01 1.29778421e+00 -3.80049616e-01
-7.22796023e-02 -8.85270908e-02 -1.07913756e+00 6.82180405e-01
1.25092149e+00 -1.44460008e-01 -6.53563619e-01 -6.26483679e-01
-8.17408413e-02 -2.46743545e-01 6.54000342e-01 -9.77687001e-01
2.80952126e-01 -1.74606800e-01 6.79518521e-01 -2.54399270e-01
2.90569872e-01 -8.18564892e-01 -2.15056747e-01 3.85679901e-01
-3.37572217e-01 -1.30009085e-01 -1.82882190e-01 5.79158902e-01
-1.91365540e-01 2.67731812e-04 7.95012951e-01 -4.82006788e-01
-1.23369277e+00 5.99064708e-01 7.80808255e-02 2.44078021e-02
1.12333953e+00 -1.97923109e-01 -4.46761489e-01 -4.90120262e-01
-8.97586763e-01 5.83405495e-01 5.17588377e-01 8.38946640e-01
7.55883515e-01 -1.34375441e+00 -7.35287368e-01 1.24896266e-01
5.46300530e-01 1.21623777e-01 3.84254992e-01 4.54599023e-01
-1.64416298e-01 1.51604936e-01 -3.43318820e-01 -5.49905062e-01
-1.21572185e+00 7.99446046e-01 1.43620253e-01 2.72931959e-02
-7.86162078e-01 8.17482948e-01 6.99432135e-01 -5.76229513e-01
2.01377824e-01 3.60772200e-02 -2.29774326e-01 -1.30417705e-01
7.33795583e-01 9.69041064e-02 -4.58804220e-01 -6.92129731e-01
-3.03663641e-01 4.88684803e-01 -5.65587990e-02 -4.15669829e-02
9.82169867e-01 -3.12488377e-01 2.86264509e-01 2.37605706e-01
1.27742064e+00 -3.11677158e-01 -1.65891767e+00 -4.13843215e-01
-2.29212299e-01 -4.74919617e-01 -2.45298728e-01 -8.53012145e-01
-1.51305354e+00 9.33739960e-01 5.60657203e-01 -2.50948340e-01
1.06299841e+00 3.56959760e-01 8.52452695e-01 2.89108157e-01
6.66146517e-01 -1.29734218e+00 4.95459169e-01 2.43866891e-01
8.10462356e-01 -1.73336446e+00 -1.75426081e-01 -6.62202835e-01
-1.14809370e+00 8.30028534e-01 1.20399129e+00 4.67499718e-02
3.39978307e-01 2.52123596e-03 3.95770103e-01 -1.09270819e-01
-4.35681760e-01 -6.16741002e-01 2.31884316e-01 1.00321507e+00
3.09988290e-01 3.03228378e-01 -1.12311237e-01 8.46165717e-01
-1.21013612e-01 -1.20488815e-01 1.64553165e-01 6.00177288e-01
-5.23758829e-01 -1.22378397e+00 -2.86375076e-01 2.97322571e-01
1.19918346e-01 5.01263276e-05 -4.54695255e-01 6.06831908e-01
3.46859783e-01 1.06447256e+00 -4.35889252e-02 -3.45195085e-01
1.93007663e-01 -1.18843488e-01 2.41924062e-01 -9.45680022e-01
-2.33920306e-01 -1.69275880e-01 -1.67614631e-02 -7.09410429e-01
-3.65738988e-01 -4.80140865e-01 -1.49061453e+00 9.66091361e-03
7.50908330e-02 -1.00953564e-01 4.39323276e-01 7.94161618e-01
4.29978639e-01 4.76426482e-01 5.00528812e-01 -7.09407508e-01
-3.58909041e-01 -8.76972735e-01 -4.49448735e-01 9.89520550e-01
2.72968616e-02 -7.02425420e-01 -2.48971611e-01 3.77814323e-01] | [10.920047760009766, 1.273616075515747] |
dd47dadc-5cfe-4a21-b949-d65e575ad543 | leveraging-joint-sparsity-in-hierarchical | 2303.16954 | null | https://arxiv.org/abs/2303.16954v1 | https://arxiv.org/pdf/2303.16954v1.pdf | Leveraging joint sparsity in hierarchical Bayesian learning | We present a hierarchical Bayesian learning approach to infer jointly sparse parameter vectors from multiple measurement vectors. Our model uses separate conditionally Gaussian priors for each parameter vector and common gamma-distributed hyper-parameters to enforce joint sparsity. The resulting joint-sparsity-promoting priors are combined with existing Bayesian inference methods to generate a new family of algorithms. Our numerical experiments, which include a multi-coil magnetic resonance imaging application, demonstrate that our new approach consistently outperforms commonly used hierarchical Bayesian methods. | ['Anne Gelb', 'Jan Glaubitz'] | 2023-03-29 | null | null | null | null | ['bayesian-inference'] | ['methodology'] | [ 1.19040072e-01 3.23141180e-02 -4.00779575e-01 -7.73638666e-01
-1.20900357e+00 1.00948289e-02 4.81984973e-01 -1.92550227e-01
-3.24371696e-01 8.85039866e-01 5.12991309e-01 -1.31260991e-01
-3.55653644e-01 -2.81069189e-01 -4.39204067e-01 -9.43251193e-01
-4.01835203e-01 7.99376726e-01 3.54779959e-01 4.91832882e-01
2.58113503e-01 2.41318628e-01 -8.32546651e-01 -6.49789721e-02
4.86497074e-01 6.31791413e-01 3.22739750e-01 5.50722301e-01
5.29289484e-01 5.49997032e-01 -3.12835723e-01 1.13776304e-01
-1.23522459e-02 -2.06178296e-02 -4.27288294e-01 1.08009480e-01
-2.00425950e-03 -5.61787248e-01 -3.45624119e-01 1.09508431e+00
5.80595493e-01 2.65475839e-01 1.05408394e+00 -7.25253880e-01
-3.68589833e-02 8.20600212e-01 -8.65498722e-01 4.08777684e-01
2.91723549e-01 -1.46240607e-01 7.94960797e-01 -8.62554729e-01
4.05723900e-01 1.36638105e+00 8.52718771e-01 -4.39036451e-02
-1.50064766e+00 -7.17568874e-01 5.07742912e-02 7.56902918e-02
-1.86766326e+00 -5.43233812e-01 8.42584431e-01 -6.26592338e-01
6.77970827e-01 -2.77027875e-01 3.46622139e-01 1.19989455e+00
7.82187283e-01 6.81101859e-01 1.38440800e+00 -4.46221471e-01
5.23050725e-01 -1.75166667e-01 2.80124724e-01 6.98757410e-01
4.42038864e-01 2.08986998e-01 -5.39524555e-01 -9.20753419e-01
1.05651236e+00 -1.05541646e-01 -7.68364668e-02 -3.37874740e-01
-1.13057172e+00 1.23746622e+00 -2.62445271e-01 7.65407085e-02
-6.87091529e-01 4.28010792e-01 4.16290052e-02 -4.90170866e-01
2.86142886e-01 -1.35275915e-01 -1.71570256e-01 4.87812646e-02
-1.14599693e+00 3.79719943e-01 8.77778351e-01 8.58422935e-01
6.23552263e-01 3.00621480e-01 -2.87356973e-01 8.68288338e-01
1.13243484e+00 9.07154262e-01 6.98435605e-02 -1.23972297e+00
-2.52836049e-02 -6.32146657e-01 3.39766115e-01 -9.40375626e-01
-5.62317312e-01 -3.32591653e-01 -9.37980831e-01 -2.96955913e-01
1.83640897e-01 -2.56607145e-01 -8.62178802e-01 1.55797565e+00
3.15869242e-01 4.75487709e-01 -4.43613440e-01 9.19029415e-01
6.30309641e-01 5.57074368e-01 2.69213974e-01 -6.20003700e-01
1.37379467e+00 -1.50199994e-01 -1.18467379e+00 -1.31601572e-01
-6.68961033e-02 -1.00872219e+00 3.50962073e-01 6.52214170e-01
-1.09784031e+00 -1.30764976e-01 -9.04385388e-01 5.17023802e-01
4.95349407e-01 -6.78215623e-02 9.06599343e-01 8.25451910e-01
-6.63536489e-01 1.67974234e-01 -1.37935197e+00 2.60229677e-01
2.34216809e-01 2.32121781e-01 3.84423882e-03 -2.12225750e-01
-1.07296550e+00 8.82974923e-01 4.04250741e-01 9.80306864e-02
-1.32085907e+00 -7.11351156e-01 -8.52666318e-01 -1.76846042e-01
1.44123450e-01 -7.85254478e-01 1.38813245e+00 7.54248723e-02
-1.75784922e+00 3.75758171e-01 -4.21898335e-01 -2.59138018e-01
-1.01012968e-01 -1.09168179e-01 -1.89822137e-01 4.48151261e-01
4.30273451e-02 4.34173197e-01 1.33924723e+00 -1.28892565e+00
-3.09111476e-02 -4.41717021e-02 -5.27123213e-01 1.03514768e-01
2.02783704e-01 2.24573538e-01 -6.16126239e-01 -8.71111572e-01
5.48753262e-01 -8.73547554e-01 -7.74234951e-01 -6.56063557e-01
-6.29979849e-01 -5.59597015e-02 1.55154705e-01 -6.44188225e-01
1.12866104e+00 -1.74509466e+00 2.59414166e-01 9.08843994e-01
4.31787759e-01 -7.63718307e-01 2.00204834e-01 7.66749755e-02
-1.04178250e-01 -4.41709280e-01 -5.36203384e-01 -3.59108925e-01
-8.64378884e-02 3.91794771e-01 -9.63973701e-02 9.76882458e-01
-2.82906294e-01 2.68720776e-01 -9.21074390e-01 -7.94220030e-01
2.51580238e-01 6.79087579e-01 -6.95460141e-01 3.85691017e-01
1.42555475e-01 5.64150274e-01 -6.01226866e-01 5.32624781e-01
6.42217219e-01 -3.84305954e-01 7.43444204e-01 -5.08262932e-01
4.27008152e-01 -9.36113298e-02 -1.48544645e+00 1.54135895e+00
3.77356894e-02 1.43982783e-01 3.63684148e-01 -1.15875852e+00
7.50584126e-01 4.80206519e-01 8.48233521e-01 -1.16478883e-01
2.44861081e-01 -7.00216368e-02 -2.93379445e-02 -4.25592154e-01
1.06082805e-01 -7.60294259e-01 -2.10474789e-01 4.01194125e-01
4.92438406e-01 -4.66197580e-01 -2.89182011e-02 4.17016715e-01
9.50468242e-01 -5.41998856e-02 4.85228926e-01 -7.72119820e-01
-3.87822352e-02 -5.57366073e-01 9.37392235e-01 1.25008094e+00
-1.51268780e-01 6.76160634e-01 2.80102611e-01 -1.06023312e-01
-8.78116131e-01 -1.50401056e+00 -8.21238339e-01 9.53327656e-01
-2.55715400e-01 -3.98369938e-01 -4.97889012e-01 -2.98852295e-01
-7.61516541e-02 5.92221916e-01 -3.85800272e-01 2.76705235e-01
-4.05625194e-01 -1.55890560e+00 2.05749288e-01 6.08345330e-01
-1.78139880e-01 -4.75494236e-01 -4.24175829e-01 5.48894942e-01
-2.44637132e-01 -1.23078346e+00 -3.08087081e-01 3.20082277e-01
-1.02427149e+00 -9.30512905e-01 -7.41427243e-01 -1.73197910e-01
5.14552534e-01 -2.43289471e-01 9.78341103e-01 -4.35142905e-01
-4.43415165e-01 6.42998934e-01 -1.17576122e-01 -2.25327894e-01
-3.05755764e-01 -4.44217056e-01 4.97743845e-01 -2.19954550e-02
1.77732989e-01 -7.18459606e-01 -4.70094174e-01 1.05430864e-01
-7.34249175e-01 -3.37219715e-01 5.94673634e-01 9.82140064e-01
7.57142127e-01 -3.81966420e-02 4.48386431e-01 -9.24902022e-01
5.42087734e-01 -7.94898510e-01 -8.78077507e-01 -5.82338087e-02
-4.02846903e-01 2.75845796e-01 -2.68199652e-01 -5.90519011e-01
-1.56083119e+00 1.10157900e-01 -1.59326658e-01 -4.38793510e-01
-2.13371158e-01 7.95250475e-01 1.46799618e-02 -8.62571821e-02
5.89402258e-01 -9.49326530e-02 -1.83909368e-02 -5.49760520e-01
4.67293113e-01 3.62590551e-01 7.65647888e-01 -9.88458931e-01
3.42446953e-01 4.94459718e-01 1.51128218e-01 -8.88658524e-01
-8.03245068e-01 -5.28401613e-01 -5.63853741e-01 -2.43204355e-01
8.31063569e-01 -1.05334389e+00 -5.55925786e-01 3.38305563e-01
-1.00409567e+00 -1.82242811e-01 2.13849887e-01 1.38988984e+00
-7.69255877e-01 6.99137628e-01 -7.38904178e-01 -9.24324453e-01
-2.29115915e-02 -1.42406845e+00 1.10563290e+00 -7.77546540e-02
-3.68681282e-01 -1.21244776e+00 4.85840321e-01 2.45918572e-01
2.27194309e-01 -2.52109200e-01 7.79623568e-01 -4.02919263e-01
-3.14216405e-01 -7.81331584e-02 -1.95882574e-01 6.51925877e-02
-1.15660064e-01 -1.12832747e-01 -6.68821096e-01 -3.94170880e-01
3.91033262e-01 -1.62461087e-01 6.95844173e-01 1.36136281e+00
1.27805448e+00 -5.55722229e-02 -6.94333851e-01 5.31226695e-01
1.27465260e+00 3.27953510e-03 5.23356915e-01 -4.23821211e-01
5.08205891e-01 1.85497642e-01 3.19608241e-01 1.06950271e+00
3.64752382e-01 6.04182303e-01 -8.40617195e-02 1.48212910e-01
1.68423891e-01 1.77545026e-01 9.63792205e-04 1.28574026e+00
1.45338578e-02 1.95886150e-01 -9.41124380e-01 4.92270619e-01
-1.78667033e+00 -8.31173420e-01 -1.18391432e-01 1.97140825e+00
1.12769306e+00 -7.86153302e-02 6.10005967e-02 -3.66291672e-01
6.92618072e-01 -5.72006106e-02 -4.17037368e-01 2.72311658e-01
5.60213029e-02 4.96592432e-01 6.20734632e-01 9.68739212e-01
-1.27883852e+00 5.12905419e-01 9.50720119e+00 9.40442801e-01
-3.59153986e-01 5.05624831e-01 2.50729471e-01 -4.96823192e-02
-4.74816203e-01 1.89481694e-02 -1.07016575e+00 3.97341162e-01
1.00970733e+00 4.87608016e-02 -1.30429643e-03 4.44365352e-01
3.49883467e-01 -5.88187635e-01 -7.32716382e-01 1.20072079e+00
2.03487828e-01 -1.32330656e+00 -3.67208391e-01 8.14001367e-04
7.69645214e-01 2.59311408e-01 -2.38307342e-01 -6.40767440e-02
1.20347631e+00 -9.29760277e-01 3.52334380e-01 1.02256739e+00
3.88984412e-01 -5.09774268e-01 6.70089662e-01 4.23075370e-02
-8.33637655e-01 2.46753976e-01 -4.00606483e-01 1.37606800e-01
7.91907609e-01 1.46811068e+00 -8.07005405e-01 2.80556262e-01
8.21812570e-01 5.08349895e-01 5.31630851e-02 1.31349647e+00
-3.75627160e-01 1.12683237e+00 -8.58858883e-01 2.89442033e-01
-8.63267183e-02 -2.45797440e-01 7.98135757e-01 1.40927494e+00
1.49666416e-02 4.55315083e-01 5.28790057e-01 8.71025085e-01
3.53636026e-01 -2.79684067e-01 -3.95540357e-01 5.17409027e-01
8.18291903e-01 1.17387307e+00 -7.93842316e-01 -5.22386372e-01
-4.16526467e-01 1.78662106e-01 1.53254926e-01 6.38805687e-01
-7.93919325e-01 -1.40360862e-01 2.04972029e-01 -2.73985803e-01
5.20883620e-01 -4.94821459e-01 -2.25930735e-01 -1.38312662e+00
-4.39557880e-01 -7.20793426e-01 5.18591106e-01 -7.29266942e-01
-1.60319674e+00 1.65460691e-01 1.09343088e+00 -7.05154419e-01
-6.10650480e-01 -4.43329424e-01 -3.80219579e-01 8.66912186e-01
-1.12537313e+00 -8.55502546e-01 3.26767057e-01 6.31640196e-01
9.59087536e-02 -2.31662154e-01 8.10671806e-01 1.54879391e-01
-5.10400236e-01 2.77548015e-01 1.88882351e-01 -2.31665909e-01
7.25744486e-01 -1.00025415e+00 -3.39261621e-01 7.26261199e-01
-2.44127795e-01 8.33753765e-01 1.05850017e+00 -9.82480526e-01
-1.20094836e+00 -5.17004013e-01 2.60051399e-01 -3.18642765e-01
9.11797345e-01 -1.97307825e-01 -8.95000637e-01 8.61539364e-01
2.29852870e-01 -1.13818489e-01 1.16327703e+00 4.90429878e-01
-3.91372472e-01 2.74756074e-01 -1.11826575e+00 6.23669811e-02
3.95210415e-01 -3.24048012e-01 -9.23720181e-01 6.06757700e-01
2.67891079e-01 -1.93674490e-01 -1.43428314e+00 7.07325518e-01
6.82314575e-01 -5.68362057e-01 1.22839010e+00 -4.08626229e-01
-2.41063908e-01 -2.69057512e-01 -5.67137003e-01 -1.23000395e+00
-6.25191748e-01 -6.46338224e-01 -5.45410216e-01 6.88379467e-01
7.75034651e-02 -3.95275027e-01 6.65613532e-01 5.12713253e-01
-1.36670902e-01 -4.67247814e-01 -1.15267122e+00 -6.55483544e-01
1.41931608e-01 -8.07522297e-01 8.24777409e-02 8.36143792e-01
6.49143979e-02 3.12215656e-01 -6.09062135e-01 4.68281031e-01
1.76178706e+00 -2.30378076e-01 2.07295194e-01 -1.15341032e+00
-7.26989746e-01 -1.44945383e-01 -1.55697480e-01 -9.77856219e-01
3.48018199e-01 -6.00309789e-01 5.99743128e-01 -1.06107962e+00
9.51035023e-01 -4.69767004e-01 -1.31402120e-01 4.06341076e-01
-3.06794673e-01 1.86516240e-01 -3.23299438e-01 1.77437425e-01
-6.20649755e-01 6.42387807e-01 7.76320398e-01 1.38429269e-01
1.26708016e-01 -1.52499750e-01 -4.41572696e-01 1.02448392e+00
4.20873880e-01 -7.95693099e-01 -2.29162976e-01 -1.60050631e-01
2.68453900e-02 3.49168926e-01 2.43339062e-01 -6.32460058e-01
2.66725093e-01 -2.17504621e-01 4.71013784e-01 -7.70456970e-01
4.43794578e-01 -4.02369827e-01 2.18838617e-01 2.47748688e-01
5.01470380e-02 -3.62673908e-01 5.96810877e-02 7.24092722e-01
-1.09030707e-02 -4.99582022e-01 1.11776698e+00 3.38985659e-02
-2.31631115e-01 1.57624468e-01 -9.77281392e-01 2.60395911e-02
4.12696928e-01 3.31287414e-01 3.16876143e-01 -5.42172730e-01
-1.37132883e+00 1.25839695e-01 -3.14080119e-01 -1.22480690e-01
7.77872562e-01 -1.33151233e+00 -9.52177286e-01 1.76786780e-01
-2.00058758e-01 -1.98096469e-01 4.27178234e-01 1.08995283e+00
4.16390635e-02 2.86268502e-01 -2.37960909e-02 -1.09336650e+00
-8.14166307e-01 1.73922613e-01 5.85991852e-02 -2.97692984e-01
-6.35876775e-01 7.80823946e-01 1.00919187e-01 -5.54639518e-01
2.22818539e-01 -6.75859526e-02 2.25249473e-02 -2.01129362e-01
5.38910151e-01 4.71292138e-01 -2.19162002e-01 -3.77242535e-01
-3.33152205e-01 4.52312857e-01 -6.14341944e-02 -6.87024415e-01
1.39559555e+00 -3.00799012e-01 -2.32596144e-01 5.79787731e-01
9.10902023e-01 -2.44952798e-01 -1.20532095e+00 -4.77142662e-01
-4.73878086e-02 -2.45505497e-01 5.82486212e-01 -5.06341279e-01
-7.84491837e-01 5.69408834e-01 5.00644147e-01 -3.89356762e-01
7.35287666e-01 2.47590780e-01 1.43413126e-01 3.98389816e-01
5.90574324e-01 -9.60202396e-01 8.85555055e-03 4.65251774e-01
6.31249487e-01 -1.18002450e+00 6.76839054e-01 -4.88257736e-01
-4.11064297e-01 8.02556992e-01 9.67218876e-02 -2.74406880e-01
1.43640220e+00 7.75886297e-01 -3.21352005e-01 -3.62293810e-01
-6.09919310e-01 1.25236034e-01 4.16515976e-01 8.12637031e-01
5.24092138e-01 2.61498868e-01 -2.67006695e-01 8.89259517e-01
9.24688652e-02 9.30990502e-02 4.09520626e-01 9.00710404e-01
-3.99303317e-01 -1.01201904e+00 -9.97706711e-01 6.58447027e-01
-5.99828601e-01 -2.18340442e-01 5.92904031e-01 4.70922738e-01
-5.46515703e-01 9.48030233e-01 1.36013508e-01 1.94842622e-01
-2.56768227e-01 1.49062648e-02 9.84663010e-01 -5.77732801e-01
1.69633642e-01 1.10209596e+00 1.31211951e-01 -6.81243718e-01
-6.85126126e-01 -1.17106497e+00 -1.05379164e+00 1.65202525e-02
-3.86049151e-01 1.77657604e-01 6.32615507e-01 1.17825842e+00
-5.88301308e-02 4.54305410e-01 4.71137136e-01 -1.13173378e+00
-7.92177677e-01 -1.23500967e+00 -8.22147906e-01 1.45436704e-01
4.89587523e-02 -1.11003721e+00 -3.55151892e-01 1.68743208e-01] | [6.999424457550049, 3.992400884628296] |
a0b52389-50e9-4787-a349-b537fe24faef | operationalising-representation-in-natural | 2306.08193 | null | https://arxiv.org/abs/2306.08193v1 | https://arxiv.org/pdf/2306.08193v1.pdf | Operationalising Representation in Natural Language Processing | Despite its centrality in the philosophy of cognitive science, there has been little prior philosophical work engaging with the notion of representation in contemporary NLP practice. This paper attempts to fill that lacuna: drawing on ideas from cognitive science, I introduce a framework for evaluating the representational claims made about components of neural NLP models, proposing three criteria with which to evaluate whether a component of a model represents a property and operationalising these criteria using probing classifiers, a popular analysis technique in NLP (and deep learning more broadly). The project of operationalising a philosophically-informed notion of representation should be of interest to both philosophers of science and NLP practitioners. It affords philosophers a novel testing-ground for claims about the nature of representation, and helps NLPers organise the large literature on probing experiments, suggesting novel avenues for empirical research. | ['Jacqueline Harding'] | 2023-06-14 | null | null | null | null | ['philosophy'] | ['miscellaneous'] | [ 4.88345027e-01 7.93746710e-01 -6.39279962e-01 -3.30393851e-01
-5.08375108e-01 -9.25125718e-01 1.10869443e+00 3.79771322e-01
-2.43487313e-01 2.40805298e-01 9.76212144e-01 -1.14676225e+00
-4.10516918e-01 -9.37208235e-01 -6.29988313e-01 -4.19860661e-01
3.62558693e-01 4.36259001e-01 -7.81593993e-02 -7.52099082e-02
1.02798891e+00 6.22641325e-01 -1.28371906e+00 5.85011899e-01
4.75014806e-01 6.95499897e-01 -4.83617820e-02 -4.83354218e-02
-5.58549941e-01 1.28998864e+00 -4.28845078e-01 -5.69857895e-01
-1.18687190e-02 -6.97858453e-01 -1.24815118e+00 -2.40169466e-01
3.01638246e-01 1.96903840e-01 -6.91861063e-02 1.25914311e+00
1.18717566e-01 9.28197429e-02 6.05974257e-01 -1.00218523e+00
-1.23265123e+00 8.85667562e-01 -2.96837300e-01 6.41502142e-01
4.44851398e-01 1.02437668e-01 1.42463887e+00 -5.17360806e-01
5.39038360e-01 1.59971607e+00 7.91015267e-01 3.73413265e-01
-1.65866053e+00 -5.01665890e-01 -4.78646671e-03 3.61807913e-01
-7.96743214e-01 -5.97420573e-01 8.29813004e-01 -7.53047168e-01
1.08426261e+00 2.62482077e-01 9.98252094e-01 1.17851126e+00
6.47053495e-02 8.13754916e-01 1.57855117e+00 -7.71723866e-01
3.37302655e-01 3.31978470e-01 6.43193126e-01 5.03932655e-01
3.84628385e-01 5.35275161e-01 -6.74429595e-01 -4.36783046e-01
8.85804355e-01 -4.73335534e-01 -7.44410753e-02 -2.20990270e-01
-1.08259904e+00 1.17861068e+00 5.13669312e-01 6.99707627e-01
-6.44376278e-01 3.62403214e-01 3.89889300e-01 2.68355757e-01
2.55697429e-01 7.54036784e-01 -3.37237507e-01 -2.89542466e-01
-9.51668620e-01 3.98004442e-01 9.35612321e-01 2.53417045e-01
5.01311719e-01 -8.13638195e-02 1.74124524e-01 5.24374306e-01
7.91617274e-01 -4.91951518e-02 5.42596459e-01 -1.26407123e+00
2.25931510e-01 7.66862035e-01 8.42457563e-02 -1.31250715e+00
-3.47752303e-01 2.03447673e-03 -1.14794850e-01 1.15417339e-01
3.04196179e-01 8.18816796e-02 -4.33450073e-01 1.79413605e+00
-9.91566181e-02 -2.62323946e-01 2.11593315e-01 8.36682558e-01
5.14518559e-01 7.27547526e-01 5.14963686e-01 2.18385141e-02
1.76918864e+00 -4.92819637e-01 -4.75233585e-01 -6.04444146e-01
5.00313818e-01 -3.42673719e-01 1.28140640e+00 2.43275285e-01
-1.22336340e+00 -1.68641031e-01 -1.00538015e+00 -4.75016862e-01
-2.46967465e-01 -3.90675366e-01 1.25678802e+00 8.98398876e-01
-1.05262041e+00 3.65505219e-01 -7.52825260e-01 -4.32636589e-01
6.38305724e-01 -2.26854891e-01 -5.81359258e-03 1.58088416e-01
-1.12616968e+00 1.47792780e+00 7.24248350e-01 1.47811338e-01
-4.52930689e-01 -5.45344591e-01 -6.83174551e-01 1.37265965e-01
1.51640758e-01 -7.66966879e-01 1.18845975e+00 -1.48569489e+00
-1.05985105e+00 1.33435059e+00 -2.71590412e-01 -6.66254997e-01
1.40415654e-02 -2.68403701e-02 -4.30076808e-01 8.77852216e-02
-4.77461405e-02 4.56018180e-01 4.31696415e-01 -1.16913474e+00
-3.97149324e-01 -3.98874462e-01 4.98307347e-01 7.60244504e-02
1.39546946e-01 3.02981019e-01 5.05171537e-01 -4.85500008e-01
6.26672089e-01 -6.41384780e-01 1.55726954e-01 1.20033994e-01
-2.25470841e-01 -5.51205218e-01 -2.16035475e-03 -3.37671548e-01
7.97239065e-01 -2.11112809e+00 -1.25538126e-01 1.01212494e-01
3.69768471e-01 -5.28873093e-02 4.81383316e-02 7.33659983e-01
-2.32031360e-01 4.82630283e-01 -1.19226851e-01 3.53531867e-01
6.32738471e-01 2.07785860e-01 -1.07461917e+00 6.36326671e-01
1.45975962e-01 1.17917812e+00 -8.25609326e-01 -3.33743274e-01
1.24071486e-01 4.32857990e-01 -5.86047411e-01 -4.03552502e-01
-2.54588336e-01 -8.79931897e-02 -5.39878726e-01 6.22667372e-01
3.22766989e-01 -3.09468448e-01 3.35533410e-01 -7.26233795e-02
-3.77021879e-01 1.36191642e+00 -7.09718406e-01 1.28733277e+00
-7.35485330e-02 1.20853329e+00 -1.02204204e-01 -1.28650999e+00
8.82637858e-01 3.57653260e-01 -2.48116150e-01 -8.17868292e-01
3.79860401e-01 9.29962993e-02 5.37608027e-01 -3.95540059e-01
4.10724789e-01 -1.04524779e+00 1.27566010e-01 9.96490657e-01
-1.20682873e-01 2.40953155e-02 -2.49491453e-01 -1.03365473e-01
9.81250226e-01 2.50344336e-01 5.03737628e-01 -7.70365536e-01
2.59012014e-01 2.25109041e-01 3.64383936e-01 7.06326246e-01
-3.00202310e-01 1.22437187e-01 6.33808613e-01 -7.32843995e-01
-1.04145861e+00 -1.14322197e+00 -6.40682399e-01 1.09439385e+00
-2.35976562e-01 -1.75530031e-01 -4.85657543e-01 -1.03782535e-01
-3.53559889e-02 1.33513081e+00 -9.60348308e-01 -7.11717084e-02
-4.11776185e-01 -7.74167478e-01 7.67176628e-01 5.96049666e-01
2.73661762e-01 -1.68793178e+00 -1.26809096e+00 1.65495858e-01
-3.05974543e-01 -4.74542618e-01 7.02516854e-01 4.67506945e-01
-9.75254595e-01 -1.16376138e+00 -1.14201136e-01 -8.75126123e-01
3.65980744e-01 8.19345787e-02 1.39433682e+00 1.90757349e-01
1.60457671e-01 3.49310547e-01 -2.25806102e-01 -7.31604934e-01
-7.46888995e-01 -2.24428982e-01 -2.60033280e-01 -6.59267068e-01
1.35669935e+00 -7.94251919e-01 -2.44131386e-01 -2.14861959e-01
-8.99471939e-01 2.98815221e-02 5.11921465e-01 3.15784484e-01
5.90193532e-02 -4.52575088e-02 6.41736090e-01 -9.11126137e-01
1.09601104e+00 -8.33140790e-01 -2.91966230e-01 2.55300757e-02
-5.27816832e-01 -3.21417116e-02 -1.36733860e-01 -3.17391366e-01
-8.85543227e-01 -7.64761150e-01 -1.42124996e-01 1.89194545e-01
-2.05799013e-01 9.51918960e-01 1.23921931e-02 7.60807618e-02
1.08879328e+00 2.67131776e-01 1.02149345e-01 -3.78440291e-01
6.54870987e-01 1.09294526e-01 3.72594148e-01 -1.08553278e+00
6.16088986e-01 3.40602309e-01 -2.98540771e-01 -8.28929722e-01
-9.59559560e-01 -1.71068117e-01 -4.55271453e-01 2.49199331e-01
5.84262908e-01 -7.31399417e-01 -8.41001272e-01 -4.84438598e-01
-1.14659941e+00 -2.08397448e-01 -5.70707262e-01 3.84594083e-01
-9.28125262e-01 2.83309847e-01 -4.46505815e-01 -8.97278726e-01
-5.35811065e-03 -7.77171552e-01 2.51568466e-01 -1.18936650e-01
-1.01606035e+00 -1.17263246e+00 4.06027108e-01 4.79663223e-01
3.94930661e-01 1.12856790e-01 1.65685856e+00 -9.52657640e-01
-2.96578079e-01 -6.70876577e-02 -3.73974323e-01 -4.62645143e-02
-3.52864444e-01 -3.75453889e-01 -1.29389083e+00 2.74155200e-01
6.83395386e-01 -7.46133089e-01 6.82059944e-01 2.26835042e-01
8.13142300e-01 -7.74293423e-01 -8.00672993e-02 1.63755730e-01
1.55474746e+00 2.37365365e-01 9.07224417e-01 9.08708155e-01
3.51771303e-02 1.05690682e+00 -6.22435063e-02 6.98452741e-02
3.27716798e-01 6.58858940e-02 -2.01373976e-02 4.33511555e-01
1.99598119e-01 -4.32073325e-01 2.64078557e-01 4.33996111e-01
9.54527333e-02 -1.67801797e-01 -1.32734466e+00 8.33962560e-01
-1.79742956e+00 -1.23245633e+00 8.47048014e-02 1.74717116e+00
9.88518834e-01 3.43170971e-01 -5.81788756e-02 2.50142217e-01
5.29912233e-01 5.39203644e-01 -3.12243342e-01 -8.31617475e-01
-1.13116698e-02 1.74120069e-01 -2.26829923e-03 4.52814847e-01
-6.16748691e-01 9.28684831e-01 7.53240585e+00 3.61110568e-01
-6.71973407e-01 1.59876514e-02 2.87516922e-01 2.11135998e-01
-7.50808299e-01 2.99668759e-01 -4.03267652e-01 2.83779681e-01
1.29778135e+00 -2.86694378e-01 5.53064466e-01 6.62259519e-01
2.34841951e-03 -9.08961520e-02 -1.52740192e+00 5.37204623e-01
7.55762383e-02 -1.64576161e+00 1.46106154e-01 3.44659537e-01
2.85461068e-01 3.70113313e-01 -1.61894355e-02 3.26020122e-01
7.80618906e-01 -1.32198691e+00 1.19494200e+00 3.23322624e-01
-8.27111006e-02 -3.51323873e-01 3.90000015e-01 6.25253201e-01
-3.86086702e-01 -6.68679699e-02 -5.65678477e-01 -6.52622283e-01
2.19854370e-01 1.14839606e-01 -4.46479827e-01 -3.27755779e-01
3.53386939e-01 1.58278629e-01 -3.03241700e-01 6.40158117e-01
-5.58099210e-01 9.87745702e-01 -3.36169273e-01 -1.05282098e-01
5.36069214e-01 -1.14231177e-01 4.90915716e-01 1.24143624e+00
-1.93073511e-01 3.11469764e-01 -1.63540378e-01 1.56977391e+00
1.18560381e-01 -1.19701549e-01 -6.06073439e-01 -6.76540494e-01
7.82807350e-01 9.24258530e-01 -9.76590455e-01 -1.38024122e-01
-3.89162749e-01 3.34636390e-01 3.13077092e-01 2.12776750e-01
-6.14693701e-01 1.36139646e-01 5.00735879e-01 2.79715002e-01
1.22480117e-01 -1.07649632e-01 -7.59015501e-01 -9.50025499e-01
-5.04459925e-02 -9.70677733e-01 3.11085861e-02 -8.68872762e-01
-1.41731024e+00 7.91597739e-02 2.15774329e-04 -3.18875492e-01
-3.07998836e-01 -7.98521638e-01 -6.05355918e-01 1.03938377e+00
-1.33320332e+00 -1.11059380e+00 2.99435288e-01 2.49628767e-01
9.30658132e-02 1.52923867e-01 1.16313565e+00 -5.76251924e-01
-5.15263490e-02 4.17276397e-02 -1.78316087e-01 2.21762463e-01
9.24008153e-03 -1.12753820e+00 7.52701581e-01 4.18179095e-01
6.01793766e-01 1.34991384e+00 9.25979018e-01 -3.54837447e-01
-1.48642480e+00 -2.40813375e-01 1.13383877e+00 -9.41747963e-01
1.05198860e+00 3.20162475e-02 -1.13905513e+00 1.13592613e+00
3.75814945e-01 -3.76517504e-01 1.21080983e+00 6.11493230e-01
-8.36485207e-01 7.81292140e-01 -1.33095646e+00 4.67495114e-01
6.68408453e-01 -1.18224037e+00 -1.86543024e+00 1.15724511e-01
4.80008870e-01 1.63508449e-02 -7.02362657e-01 -4.89733480e-02
7.90850580e-01 -9.92495298e-01 1.04338241e+00 -6.86301708e-01
5.20334482e-01 -8.89993235e-02 -4.52233195e-01 -8.55181634e-01
-6.90516770e-01 -2.65037149e-01 -1.24110570e-02 1.18610275e+00
4.13563013e-01 -9.62494135e-01 8.57666790e-01 1.05498135e+00
-4.22275402e-02 -5.73650539e-01 -9.82484221e-01 -3.32909077e-01
8.17056477e-01 -9.15838718e-01 3.47174495e-01 1.36604285e+00
5.36047101e-01 5.96472085e-01 6.22550011e-01 -1.33103102e-01
4.43572283e-01 1.57836229e-01 3.81099284e-01 -1.75865746e+00
-1.95314974e-01 -8.83097589e-01 -3.97482902e-01 -8.19568872e-01
3.37810993e-01 -1.29772258e+00 -3.76886651e-02 -1.83977842e+00
4.04603034e-01 -2.52213240e-01 -2.82570958e-01 5.78682244e-01
6.18904054e-01 1.08634621e-01 5.03064990e-01 5.49087465e-01
-9.98552665e-02 1.96271867e-01 7.49056101e-01 8.77996758e-02
-7.60485157e-02 -6.59060061e-01 -1.72227168e+00 1.30678916e+00
8.74843717e-01 -4.12717342e-01 -4.87221211e-01 -3.79509717e-01
1.10690534e+00 -3.08782458e-01 9.15258706e-01 -6.79701984e-01
1.51256740e-01 -3.52944642e-01 5.01790285e-01 -4.75359440e-01
2.53172696e-01 -8.44817698e-01 9.71635245e-03 3.37931097e-01
-7.97461689e-01 2.50501434e-05 5.72601676e-01 3.79941344e-01
7.70752355e-02 -5.73472202e-01 6.29568994e-01 -5.40791392e-01
-8.02911460e-01 -5.88679850e-01 -6.13702416e-01 5.19494712e-01
4.47112948e-01 -4.87545222e-01 -5.74137509e-01 2.20305741e-01
-6.90690160e-01 -2.21226856e-01 6.65942609e-01 2.44929537e-01
3.96202832e-01 -1.15277100e+00 -4.49376017e-01 6.39793053e-02
1.55614000e-02 -4.39335406e-01 -1.82595655e-01 5.86927831e-01
-5.97196877e-01 7.79727936e-01 -1.49844617e-01 3.28319520e-02
-4.45968062e-01 6.44908845e-01 2.57804543e-01 3.16532642e-01
-1.11805463e+00 8.71065855e-01 3.95053148e-01 -3.44405681e-01
5.49534485e-02 -3.44510496e-01 -3.03596884e-01 4.79049951e-01
7.02481329e-01 2.19862252e-01 -4.17391300e-01 -7.78707802e-01
-3.60596091e-01 9.70298871e-02 1.09996200e-01 -5.73040247e-01
1.38009930e+00 -3.69525477e-02 -4.58659083e-01 9.68743980e-01
8.36225033e-01 -1.55679166e-01 -8.74437571e-01 -2.72990882e-01
4.15614188e-01 -3.90443474e-01 3.67648244e-01 -1.25403702e+00
-3.43559682e-01 1.06204009e+00 2.00685337e-01 5.77736497e-01
4.67757821e-01 3.18961829e-01 3.54342103e-01 5.31964600e-01
3.60863119e-01 -7.60030150e-01 -2.45273784e-01 4.05975878e-01
1.23429239e+00 -6.42748535e-01 2.99630277e-02 1.18431531e-01
-5.63234150e-01 1.15728772e+00 6.13234891e-03 -4.08762634e-01
5.76250196e-01 7.66701400e-02 -6.06253259e-02 -7.44298875e-01
-7.50245392e-01 1.13268338e-01 6.38489798e-02 5.90878963e-01
8.44213247e-01 1.29156828e-01 -5.79982281e-01 7.01889455e-01
-8.87562156e-01 1.86174661e-01 3.49297643e-01 9.26658332e-01
-8.87043238e-01 -6.68217063e-01 -4.41204607e-01 2.63283163e-01
-6.72028363e-01 -4.99608397e-01 -7.87133038e-01 8.18383038e-01
1.44369394e-01 7.25910187e-01 3.80091190e-01 3.98800783e-02
-1.42643198e-01 6.13972306e-01 6.29446745e-01 -6.92804575e-01
-7.82317698e-01 -7.70481303e-02 3.34514618e-01 -4.52568024e-01
-7.30958343e-01 -9.28881705e-01 -1.08980763e+00 -6.42683089e-01
-1.53931886e-01 3.32809985e-01 5.96605778e-01 1.06573367e+00
1.21931992e-01 2.80862898e-01 -1.97588906e-01 -6.53464675e-01
-4.05235410e-01 -9.05662417e-01 -5.23510516e-01 3.13719851e-03
1.96666852e-01 -6.55958891e-01 -5.49696386e-01 -1.50807709e-01] | [9.350739479064941, 6.997243404388428] |
33d4c888-3676-4bce-b072-72171e4261d3 | inductive-relation-prediction-using-analogy | null | null | https://openreview.net/forum?id=PTRo58zPt3P | https://openreview.net/pdf?id=PTRo58zPt3P | Inductive Relation Prediction Using Analogy Subgraph Embeddings | Prevailing methods for relation prediction in heterogeneous graphs aim at learning latent representations (i.e., embeddings) of observed nodes and relations, and thus are limited to the transductive setting where the relation types must be known during training. Here, we propose ANalogy SubGraphEmbeddingLearning (GraphANGEL), a novel relation prediction framework that predicts relations5between each node pair based on the subgraphs containing the pair, as well as other (analogy) subgraphs with the same graph patterns. Each graph pattern explicitly represents a specific logical rule, which contributes to an inductive bias that facilitates generalization to unseen relations and leads to more explainable predictive models. Moreover, our method also removes the limited neighborhood constraint of graph neural networks. Our model consistently outperforms existing models on heterogeneous graph based recommendation as well as knowledge graph completion. We also empirically demonstrate our model’s capability in generalizing to new relations while producing explainable heat maps of attention scores across the discovered logic. | ['Quan Gan', 'Yong Yu', 'David Wipf', 'Zheng Zhang', 'Weinan Zhang', 'Kounianhua Du', 'Yangkun Wang', 'Jiarui Jin'] | 2021-09-29 | null | null | null | iclr-2022-4 | ['inductive-relation-prediction'] | ['graphs'] | [ 2.50295579e-01 1.20428753e+00 -7.91505039e-01 -3.44591022e-01
9.83091816e-02 -4.57113296e-01 8.48637462e-01 5.05079389e-01
5.13197482e-01 6.74392998e-01 5.66896796e-01 -5.20648897e-01
-3.61143112e-01 -1.56521237e+00 -9.96350646e-01 -3.43946815e-01
-2.69923896e-01 8.78774822e-01 2.31501851e-02 -3.51727724e-01
-3.31156582e-01 2.18259007e-01 -1.09707677e+00 3.42090786e-01
9.44693387e-01 6.24761879e-01 -2.57468224e-01 3.80348176e-01
9.60279703e-02 1.29692590e+00 2.59780940e-02 -9.42255497e-01
-5.83892129e-02 -4.53036487e-01 -1.19116461e+00 -3.49982530e-01
3.56418878e-01 -1.59082741e-01 -9.73222673e-01 8.65058959e-01
-3.14664453e-01 2.31925547e-01 8.59179437e-01 -1.32443917e+00
-1.70663106e+00 1.26488960e+00 -7.10968524e-02 8.66082981e-02
6.45523012e-01 -2.59572774e-01 1.86938560e+00 -6.59618020e-01
1.00530577e+00 1.19848382e+00 6.95868075e-01 4.44631100e-01
-1.56007302e+00 -3.78618896e-01 2.95891374e-01 5.76815724e-01
-1.17652178e+00 -1.37330696e-01 1.00266552e+00 -2.51987278e-01
1.47103679e+00 2.51813412e-01 9.51191843e-01 1.26809692e+00
3.05147529e-01 4.95323300e-01 4.92489427e-01 -4.18138683e-01
1.69705093e-01 2.07377151e-02 3.60781610e-01 8.72564375e-01
6.12141192e-01 -1.00403063e-01 -5.35255611e-01 -6.54048696e-02
6.76729321e-01 1.45141378e-01 -3.83594096e-01 -6.87813818e-01
-1.04422784e+00 7.92006910e-01 1.03155887e+00 3.02848786e-01
-5.04045188e-01 2.55514860e-01 -3.75876315e-02 3.92765135e-01
6.87489390e-01 7.89102435e-01 -5.89077711e-01 5.18175483e-01
-4.02077883e-02 2.14129016e-02 1.01718855e+00 1.25360024e+00
1.06945276e+00 -1.90487847e-01 -7.31062293e-02 3.92193377e-01
4.42150891e-01 2.49861091e-01 1.20080628e-01 -5.99860251e-01
4.86135125e-01 1.31947315e+00 -3.72722000e-01 -1.39778054e+00
-2.76571214e-01 -5.30456185e-01 -7.64731228e-01 -5.91154337e-01
4.59834710e-02 1.76922306e-01 -5.71771562e-01 1.85618007e+00
2.19904914e-01 3.82387459e-01 2.28844196e-01 4.28758949e-01
1.10632873e+00 4.51792538e-01 1.62851885e-01 -5.56079708e-02
1.11701131e+00 -9.75821793e-01 -6.48744285e-01 -3.18091899e-01
9.65785801e-01 6.76240399e-02 1.01462138e+00 -2.87236124e-01
-8.42179239e-01 -3.24951500e-01 -8.90077353e-01 -2.49728262e-01
-5.78686595e-01 -5.85773647e-01 1.38929689e+00 2.29517892e-01
-1.28224766e+00 8.26470435e-01 -5.93302727e-01 -7.48345196e-01
5.62418342e-01 4.80031550e-01 -5.71899593e-01 -2.03693494e-01
-1.51597202e+00 9.79070723e-01 5.49334824e-01 1.14299051e-01
-6.58456385e-01 -7.44313955e-01 -1.17841160e+00 5.44949591e-01
6.23326421e-01 -9.86439228e-01 4.96388853e-01 -7.65572011e-01
-1.13253224e+00 7.18628109e-01 -1.48392618e-01 -5.48388243e-01
-4.15396392e-01 4.63926345e-02 -6.47345185e-01 2.33059675e-02
-1.61809921e-01 5.06375194e-01 3.42975318e-01 -1.26371777e+00
-5.86850569e-02 -2.63289154e-01 5.92843592e-01 1.84405386e-01
-6.69902563e-01 -6.65596187e-01 -1.67932406e-01 -3.24152410e-01
3.03030729e-01 -9.20745671e-01 3.28710303e-02 -5.36915481e-01
-8.71503890e-01 -6.31897449e-01 4.68035161e-01 -3.65124375e-01
1.08463883e+00 -1.80840719e+00 5.36002815e-01 4.83680993e-01
8.36446881e-01 -1.89104035e-01 -2.88172781e-01 6.43667758e-01
-3.71761143e-01 3.05286020e-01 3.03102314e-01 2.24773034e-01
1.33293748e-01 5.21833718e-01 -4.35066342e-01 1.10294022e-01
3.14196348e-01 1.73563612e+00 -1.27024615e+00 -3.21075022e-01
-1.50765821e-01 2.67351955e-01 -7.93195784e-01 2.41757810e-01
-4.99740094e-01 2.13423371e-01 -4.10301954e-01 4.60096896e-01
1.89110488e-01 -9.03786302e-01 9.60689127e-01 -3.53299290e-01
7.08652020e-01 7.17652798e-01 -5.37037373e-01 1.49467921e+00
-5.18421412e-01 4.97207075e-01 -7.71844208e-01 -1.08529949e+00
1.04698360e+00 3.23393762e-01 1.52528033e-01 -2.16382533e-01
-1.19302459e-02 -2.07800657e-01 2.68662006e-01 -4.50896084e-01
3.61334354e-01 4.41239551e-02 2.11634770e-01 5.65987408e-01
4.33094352e-01 -4.37058712e-04 -5.38201742e-02 8.38122249e-01
1.48980236e+00 3.29853266e-01 4.56304252e-01 -2.73845702e-01
1.83914006e-01 -8.09782222e-02 4.02500212e-01 6.68392301e-01
2.34896243e-01 -1.76550120e-01 9.06693995e-01 -5.77146709e-01
-7.50733554e-01 -1.22537816e+00 2.02572241e-01 9.64091420e-01
3.60848427e-01 -7.89670467e-01 -1.89125061e-01 -9.68158007e-01
1.73873380e-01 8.51279378e-01 -1.18222845e+00 -7.85868883e-01
-3.90640497e-01 -3.30732167e-01 1.54478535e-01 6.49768293e-01
6.09749556e-02 -1.07248640e+00 5.36900103e-01 -2.21370831e-02
-7.97809884e-02 -1.15839076e+00 -8.81861374e-02 2.37802640e-01
-1.07985151e+00 -1.26500213e+00 3.54050547e-01 -1.05830526e+00
1.07012796e+00 9.59659219e-02 1.51531518e+00 4.07146901e-01
3.15554768e-01 4.09538627e-01 -4.69154865e-01 1.59579456e-01
-3.81575942e-01 3.81803811e-01 3.62744965e-02 -2.11503990e-02
7.22190499e-01 -1.22750139e+00 -3.23004216e-01 -1.72059294e-02
-5.75087726e-01 2.15301648e-01 4.34997946e-01 7.74948180e-01
4.65240568e-01 -6.75224066e-02 5.73660970e-01 -1.60104513e+00
8.01251650e-01 -8.81026208e-01 -2.46550873e-01 6.98754907e-01
-1.09862518e+00 3.51245672e-01 6.70348406e-01 -3.83990437e-01
-9.61277902e-01 -2.55721301e-01 5.24331033e-01 -2.23020270e-01
-8.77495632e-02 7.76458681e-01 -3.39486927e-01 -2.86697261e-02
8.68961811e-01 -1.03487521e-01 -3.89323413e-01 -1.09279022e-01
9.99095559e-01 -9.34886932e-02 2.55738318e-01 -6.18316531e-01
1.03588974e+00 1.82326794e-01 4.02767807e-01 -2.55077153e-01
-1.01984584e+00 3.13038491e-02 -8.45932007e-01 9.59397778e-02
9.04321969e-01 -6.51950777e-01 -8.94214392e-01 -4.42304671e-01
-1.17839742e+00 -3.16947371e-01 -4.71108764e-01 4.56993788e-01
-1.55613318e-01 1.48349747e-01 -1.00896037e+00 -4.07189846e-01
-2.37459987e-01 -5.23901045e-01 5.63603818e-01 1.07806399e-02
-6.57855988e-01 -1.52446139e+00 2.37087235e-01 2.58930713e-01
1.38491824e-01 1.54989287e-01 1.79388487e+00 -9.25471604e-01
-1.02662456e+00 -1.83033183e-01 -2.80389488e-01 -1.92275316e-01
2.45774895e-01 -2.31496707e-01 -8.29528153e-01 1.37228265e-01
-6.64517641e-01 -4.17852879e-01 8.84930849e-01 1.19010545e-01
8.16652060e-01 -6.78724468e-01 -8.03424001e-01 5.97702265e-01
9.98760581e-01 -7.19078183e-02 7.22868145e-01 -3.95158939e-02
1.28558505e+00 6.45591080e-01 -2.72473618e-02 -1.95455745e-01
8.12098980e-01 4.96574491e-01 4.35077369e-01 7.01288357e-02
-1.36106715e-01 -9.59953606e-01 2.59620789e-02 9.10363078e-01
-5.52956045e-01 -4.13015246e-01 -8.21501911e-01 4.13050830e-01
-1.86073947e+00 -9.13703263e-01 -3.59251648e-01 1.72635865e+00
8.27973366e-01 3.86847667e-02 -3.75575036e-01 -1.54348910e-01
6.60345256e-01 1.80883557e-01 -5.69501221e-01 -3.19786966e-01
-1.74048662e-01 5.08864403e-01 6.43205643e-03 8.55830431e-01
-6.52630627e-01 1.33262861e+00 6.07840490e+00 2.46216059e-01
-3.23767424e-01 2.49713957e-02 5.05045056e-01 2.86372483e-01
-1.16783643e+00 4.99118268e-01 -3.46097231e-01 -2.57092386e-01
8.18481445e-01 -2.05415741e-01 8.17183256e-01 7.97429919e-01
-4.86507148e-01 3.65139931e-01 -1.75791121e+00 5.34509361e-01
1.30854454e-02 -1.57545960e+00 6.17999673e-01 2.29167223e-01
9.41082835e-01 -2.32710421e-01 -8.97671357e-02 6.36629105e-01
7.58353889e-01 -1.20736563e+00 1.07771948e-01 5.03127813e-01
5.77942848e-01 -4.33311492e-01 7.99723983e-01 -1.09214127e-01
-1.19141984e+00 7.55247027e-02 -4.50379610e-01 -4.44641113e-01
-2.43141457e-01 4.93147492e-01 -1.12523818e+00 8.18992794e-01
2.17255697e-01 1.16803253e+00 -6.06398642e-01 1.09819286e-01
-1.20485926e+00 6.92435086e-01 1.37453884e-01 -2.41736293e-01
-2.95707703e-01 -2.81236231e-01 2.51050591e-01 5.97110033e-01
6.65544719e-02 2.88350046e-01 -6.94512054e-02 1.25021553e+00
-6.33638442e-01 1.02137931e-01 -1.05924487e+00 -3.43039304e-01
6.69610441e-01 1.27949142e+00 -6.09745383e-01 -2.98117220e-01
-5.25010765e-01 8.79629731e-01 1.03338814e+00 7.82523215e-01
-5.60867310e-01 -1.06237501e-01 4.60675418e-01 6.62831068e-02
1.80082574e-01 1.17544858e-02 -3.09657365e-01 -1.42917013e+00
-1.37432054e-01 -3.42676371e-01 5.34453511e-01 -8.44608486e-01
-1.38253415e+00 7.06031501e-01 -5.88789284e-02 -5.94313681e-01
-1.54758990e-01 -4.32937235e-01 -6.59620047e-01 7.27714002e-01
-1.30284381e+00 -1.57016170e+00 -2.47139692e-01 6.61005497e-01
-2.50106186e-01 -1.58876225e-01 1.35583067e+00 -6.77606687e-02
-4.19506848e-01 6.47371113e-01 -4.34871256e-01 1.58901155e-01
2.37936363e-01 -1.24833012e+00 5.60241938e-01 6.20677233e-01
6.43921971e-01 1.06334412e+00 5.66273212e-01 -9.98605788e-01
-1.62055898e+00 -1.12041104e+00 1.28668392e+00 -8.30714524e-01
1.09445345e+00 -5.68881750e-01 -1.10392392e+00 1.54123807e+00
1.34285256e-01 3.10724854e-01 9.20796096e-01 1.21533108e+00
-8.69145989e-01 -5.66713512e-02 -7.22033441e-01 8.17212462e-01
1.74000776e+00 -8.35852683e-01 -7.25002587e-01 5.43006361e-01
1.12615478e+00 -8.94442871e-02 -1.25419724e+00 3.11288893e-01
3.64937782e-01 -5.49367547e-01 9.53722477e-01 -1.33206272e+00
9.10266161e-01 6.73301741e-02 -6.10453188e-02 -1.46825182e+00
-1.00130165e+00 -5.37098825e-01 -9.29740846e-01 1.10845029e+00
1.07040155e+00 -8.88959169e-01 9.46681499e-01 7.47134566e-01
-7.70416036e-02 -9.23892260e-01 -4.89834577e-01 -5.86890161e-01
-2.04766482e-01 -1.78238854e-01 9.97310400e-01 1.60206914e+00
7.94098139e-01 9.66957271e-01 -2.37541616e-01 4.36634392e-01
4.20774221e-01 5.47060966e-01 6.63341641e-01 -1.50883377e+00
-6.84698403e-01 -1.83343261e-01 -7.14951932e-01 -9.12965298e-01
6.94655955e-01 -1.60193372e+00 -4.57526565e-01 -1.96568608e+00
4.75538313e-01 -3.55519474e-01 -3.86355132e-01 7.33534038e-01
-3.01848173e-01 7.23431185e-02 -3.25212359e-01 1.65206924e-01
-6.66226566e-01 6.18312716e-01 1.42138112e+00 -3.80337656e-01
-1.40237972e-01 -2.89457351e-01 -9.86428440e-01 5.73857248e-01
5.96064687e-01 -5.11292458e-01 -9.56703782e-01 -4.80161250e-01
9.06694353e-01 -1.55456156e-01 3.49065006e-01 -3.32006156e-01
1.37741476e-01 -2.64786452e-01 2.79831976e-01 1.82614818e-01
1.50272802e-01 -7.83705950e-01 3.06187302e-01 2.64494240e-01
-7.07277298e-01 -3.17290604e-01 -2.83579856e-01 1.00540197e+00
5.41812479e-02 2.03468606e-01 2.67313910e-03 5.60288467e-02
-6.24026537e-01 5.12020707e-01 1.39024362e-01 4.30769101e-02
9.11116838e-01 3.82546037e-02 -8.78818214e-01 -6.60428047e-01
-9.98237431e-01 7.08699971e-02 3.25667650e-01 5.09364963e-01
6.61880255e-01 -1.69525790e+00 -4.59833682e-01 4.24204916e-02
4.98648524e-01 -7.35527789e-03 1.54563691e-02 7.48879850e-01
-1.02978066e-01 3.15751433e-01 1.11437574e-01 -9.69073325e-02
-8.70234132e-01 1.03894901e+00 2.71818250e-01 -4.99464422e-01
-8.85454893e-01 1.12464309e+00 3.41176689e-01 -5.07630050e-01
-1.43058792e-01 -6.22331500e-01 -4.10585821e-01 -3.40486228e-01
-4.45129350e-04 4.77468893e-02 -1.73878461e-01 -5.05269825e-01
-3.85181725e-01 2.14345187e-01 -2.11094648e-01 5.29471219e-01
1.50069022e+00 2.60004729e-01 -5.20187795e-01 3.94494355e-01
9.48034942e-01 8.09622332e-02 -7.05857337e-01 -5.69608331e-01
3.00816208e-01 -3.19781065e-01 -1.76927403e-01 -6.35990024e-01
-9.58963335e-01 4.99474198e-01 -3.76829624e-01 4.88300502e-01
6.65078282e-01 7.89117277e-01 6.10958636e-01 7.24833727e-01
3.27060759e-01 -4.40648556e-01 9.19808820e-02 4.99585688e-01
7.19948411e-01 -1.10219717e+00 2.24264175e-01 -9.99468088e-01
-3.60306650e-01 1.08154285e+00 7.09690928e-01 -2.36607894e-01
6.93974733e-01 -1.78079933e-01 -5.15000403e-01 -6.66299522e-01
-1.01589823e+00 -9.51928347e-02 6.88302875e-01 9.95474875e-01
5.75389385e-01 3.71954381e-01 1.47430077e-02 6.66053712e-01
-1.72981396e-01 -2.57126480e-01 2.11523712e-01 4.77724969e-01
-4.31762896e-02 -1.09104002e+00 3.48407060e-01 6.86267138e-01
1.72431380e-01 -5.56148231e-01 -8.14474642e-01 7.41135776e-01
5.17490059e-02 9.39160466e-01 -4.46170904e-02 -9.37904000e-01
1.64820358e-01 8.49364921e-02 7.54120886e-01 -9.92052794e-01
-2.07423121e-01 -6.41756415e-01 4.75397170e-01 -5.26296616e-01
-3.07573885e-01 -1.94310680e-01 -1.33408928e+00 -5.11456728e-01
-4.43713486e-01 3.50303799e-01 -1.57960579e-01 1.13315189e+00
5.30321538e-01 7.29588389e-01 2.84809649e-01 -3.05755585e-01
9.98453647e-02 -1.01615036e+00 -8.41732323e-01 7.92465448e-01
-1.60753086e-01 -7.45625913e-01 -3.40807170e-01 7.72456219e-03] | [8.836387634277344, 7.816189765930176] |
b9884019-a852-4509-a517-1ba99740e482 | navigation-of-micro-robot-swarms-for-targeted | 2306.17598 | null | https://arxiv.org/abs/2306.17598v1 | https://arxiv.org/pdf/2306.17598v1.pdf | Navigation of micro-robot swarms for targeted delivery using reinforcement learning | Micro robotics is quickly emerging to be a promising technological solution to many medical treatments with focus on targeted drug delivery. They are effective when working in swarms whose individual control is mostly infeasible owing to their minute size. Controlling a number of robots with a single controller is thus important and artificial intelligence can help us perform this task successfully. In this work, we use the Reinforcement Learning (RL) algorithms Proximal Policy Optimization (PPO) and Robust Policy Optimization (RPO) to navigate a swarm of 4, 9 and 16 microswimmers under hydrodynamic effects, controlled by their orientation, towards a circular absorbing target. We look at both PPO and RPO performances with limited state information scenarios and also test their robustness for random target location and size. We use curriculum learning to improve upon the performance and demonstrate the same in learning to navigate a swarm of 25 swimmers and steering the swarm to exemplify the manoeuvring capabilities of the RL model. | ['Manoj Varma', 'Akshatha Jagadish'] | 2023-06-30 | null | null | null | null | ['reinforcement-learning-1', 'navigate'] | ['methodology', 'reasoning'] | [ 1.52037758e-02 9.23540369e-02 8.75496417e-02 4.22551811e-01
-4.18754034e-02 -5.83114445e-01 5.87444484e-01 3.32313746e-01
-8.40878963e-01 1.33893704e+00 -3.20378542e-01 -2.95161426e-01
-5.52939355e-01 -4.11360890e-01 -7.65928924e-01 -1.34496188e+00
-5.57625949e-01 7.27429569e-01 1.06096931e-01 -7.22434402e-01
3.01936477e-01 5.90815306e-01 -1.52274430e+00 -5.19743681e-01
1.04540205e+00 4.55627561e-01 9.36545789e-01 1.18459439e+00
4.00669158e-01 7.49436855e-01 -7.32722938e-01 5.80795527e-01
7.57583976e-02 -5.25362790e-02 -4.21794981e-01 -1.49385899e-01
-4.64339256e-01 2.12301582e-01 1.92213580e-01 6.90090418e-01
9.73518729e-01 3.14442843e-01 6.11529887e-01 -8.75531316e-01
-6.19636057e-03 4.43579257e-01 -3.20311755e-01 1.32394657e-01
2.51099199e-01 3.31656039e-01 2.99350709e-01 -8.14342201e-02
7.48953044e-01 1.38377225e+00 5.35857499e-01 7.89777517e-01
-9.56612229e-01 -2.96363086e-01 -2.66768962e-01 -9.92466062e-02
-8.12401474e-01 -6.37040138e-02 2.32398823e-01 -2.03695625e-01
7.50214100e-01 8.35110173e-02 7.95171261e-01 1.17446959e+00
1.02079916e+00 3.61906886e-01 1.20311737e+00 -3.93301576e-01
8.83335412e-01 1.33935302e-01 -6.30951762e-01 4.92410719e-01
6.60320461e-01 4.62740839e-01 -3.50301117e-02 -6.30610883e-02
7.60690153e-01 -2.86267519e-01 -5.06059416e-02 -5.61631143e-01
-1.30369055e+00 1.03567612e+00 5.56297719e-01 3.52861464e-01
-5.32659292e-01 4.93422121e-01 -5.00697047e-02 4.65701401e-01
-1.27895176e-01 1.60363197e+00 -6.94414616e-01 -3.68446201e-01
1.78211089e-03 6.23681307e-01 1.37017059e+00 5.91778517e-01
2.24512607e-01 1.94744214e-01 2.24484533e-01 5.32032907e-01
1.50592208e-01 8.76763523e-01 6.67774737e-01 -1.30754542e+00
-4.01466154e-02 2.82923192e-01 7.13506997e-01 -7.32611239e-01
-9.28531468e-01 -5.26911139e-01 -5.70335269e-01 6.38512492e-01
5.22890925e-01 -1.09558868e+00 -6.52593791e-01 1.42223477e+00
7.35382378e-01 5.69220334e-02 4.71488476e-01 8.68901730e-01
2.12028503e-01 6.28167629e-01 -1.21343583e-01 -3.95487100e-01
9.09036040e-01 -8.04415524e-01 -5.70360780e-01 -1.07985713e-01
9.06847239e-01 -5.10837793e-01 7.00568020e-01 4.76101965e-01
-8.47352028e-01 -1.16463847e-01 -9.66839552e-01 9.53475177e-01
-6.01178527e-01 -3.07239473e-01 5.42682052e-01 4.84968781e-01
-1.12492824e+00 1.06616855e+00 -1.26814330e+00 -7.82078743e-01
2.01607302e-01 8.32110941e-01 -1.16420705e-02 2.14959502e-01
-1.04175842e+00 1.25468338e+00 1.07770003e-01 -1.99331969e-01
-1.04331625e+00 -7.23388910e-01 -4.97729510e-01 -2.22455516e-01
2.96573281e-01 -6.52037442e-01 1.24684274e+00 -3.01446974e-01
-2.37759757e+00 2.68008672e-02 2.72243828e-01 -5.60170949e-01
5.56272388e-01 -9.70477462e-02 2.38489866e-01 1.28416777e-01
1.56880602e-01 6.96794510e-01 6.55060709e-01 -1.23730910e+00
-8.57243359e-01 -2.05718465e-02 -4.82088625e-02 5.55249631e-01
-3.68757904e-01 -1.16814829e-01 7.11957932e-01 -2.45125487e-01
-2.09918737e-01 -1.33872616e+00 -8.61654639e-01 -3.82108629e-01
-2.22741514e-01 -2.13867024e-01 9.56219316e-01 3.05839926e-01
3.79128307e-01 -1.51739371e+00 7.31494308e-01 -3.43364887e-02
-1.98913455e-01 4.30321425e-01 -1.56680465e-01 8.38692367e-01
5.00940204e-01 -5.30791692e-02 1.44653544e-01 -1.08344303e-02
-1.18213929e-01 5.51502705e-01 5.95388971e-02 5.09887993e-01
-4.48210277e-02 6.35978162e-01 -1.36158276e+00 -2.74156511e-01
1.83241040e-01 4.27287847e-01 -5.35093963e-01 1.09152332e-01
-3.90533209e-01 1.00115192e+00 -1.00853801e+00 5.52265465e-01
2.07854271e-01 -9.51908678e-02 2.44736850e-01 6.31957471e-01
-7.44115829e-01 -5.30132532e-01 -1.08068037e+00 1.23694313e+00
-5.15177131e-01 3.82969677e-01 7.04998434e-01 -1.03345788e+00
7.48360455e-01 2.61688501e-01 9.64246154e-01 -6.35835707e-01
3.29513282e-01 3.72579575e-01 5.01222074e-01 -8.70392621e-01
1.91877812e-01 -1.09205090e-01 1.22386597e-01 2.37405792e-01
-2.66556978e-01 -6.91122830e-01 2.37278551e-01 -2.12329924e-01
1.61014616e+00 -8.13891292e-02 -3.34813595e-02 -7.62436867e-01
4.34030473e-01 4.32846308e-01 2.64473289e-01 1.13906193e+00
-3.33870023e-01 2.73452047e-02 3.19999605e-01 -4.99464273e-01
-9.69651163e-01 -8.19864631e-01 -4.51315269e-02 9.91968393e-01
5.34196615e-01 1.37176260e-01 -5.38360119e-01 -8.55841935e-02
3.72901291e-01 3.73527259e-01 -7.11961746e-01 4.96769957e-02
-8.38814557e-01 -9.80527818e-01 3.02464478e-02 -6.80466667e-02
2.24910259e-01 -1.33789515e+00 -1.35381675e+00 5.97660780e-01
5.87921798e-01 -9.78464246e-01 1.87553838e-01 4.62730199e-01
-5.81551492e-01 -1.11422026e+00 -8.38650167e-01 -8.85243475e-01
5.12825191e-01 2.29539543e-01 5.38821936e-01 8.31209421e-02
-3.73614073e-01 5.19097984e-01 -5.79584241e-01 -8.70813608e-01
-8.10100198e-01 2.71678001e-01 6.27186179e-01 -4.58685696e-01
-5.28434098e-01 -6.05291188e-01 -7.09632695e-01 6.48362935e-01
-5.16197324e-01 -2.93875933e-01 7.13923931e-01 8.12286198e-01
2.45145023e-01 2.42449909e-01 6.81741714e-01 -3.91887635e-01
9.42856848e-01 -5.08173347e-01 -7.44446754e-01 -3.90230045e-02
-1.94717214e-01 6.37219250e-02 1.04798281e+00 -9.15435016e-01
-5.15608549e-01 2.46355444e-01 2.40918413e-01 -7.91689306e-02
-1.46450460e-01 2.96513408e-01 5.05357087e-01 -6.95262194e-01
8.32984388e-01 6.64679557e-02 6.32719755e-01 -8.82902220e-02
3.18506733e-02 5.05277812e-01 -2.23029494e-01 -5.34023225e-01
6.60750151e-01 6.04572415e-01 6.39280856e-01 -1.24366868e+00
-3.30824584e-01 -1.76341593e-01 -2.54104644e-01 -2.83688605e-01
8.05693626e-01 -4.38938767e-01 -1.43432283e+00 1.90220967e-01
-7.58341253e-01 -1.09233356e+00 -3.16362709e-01 5.60319483e-01
-8.42363358e-01 -1.82297260e-01 -3.61596853e-01 -8.01456928e-01
-1.53539732e-01 -1.32401276e+00 8.15014720e-01 7.76732862e-01
9.14237648e-02 -1.31003952e+00 3.70757252e-01 -3.71823788e-01
1.11254144e+00 6.58693552e-01 4.30197001e-01 -3.31995040e-01
-9.88151059e-02 -8.67306516e-02 5.19607782e-01 -2.98098117e-01
1.09648079e-01 -1.74758613e-01 -3.13940763e-01 -8.47009301e-01
5.94431609e-02 -4.59847212e-01 2.74126589e-01 8.00064325e-01
4.52378273e-01 -3.86586100e-01 -7.91979671e-01 2.52055258e-01
1.24718618e+00 4.97750670e-01 6.87674806e-02 8.29074383e-01
3.97388786e-01 7.15230346e-01 8.23910236e-01 6.18859649e-01
1.39705122e-01 6.30079985e-01 8.08049500e-01 1.87695742e-01
2.25924700e-01 2.29753047e-01 3.23871374e-01 4.03324693e-01
-1.53818175e-01 -2.18818203e-01 -9.28184569e-01 1.22147337e-01
-1.83823538e+00 -7.74291515e-01 9.67658237e-02 1.98488128e+00
5.12247682e-01 -1.68643653e-01 6.71858490e-02 -2.02467650e-01
5.52443802e-01 -4.08952266e-01 -7.23376811e-01 -4.62073535e-01
-3.00988611e-02 5.85854501e-02 8.95340025e-01 6.11853898e-01
-9.60003316e-01 8.97091627e-01 6.09915924e+00 4.46890205e-01
-1.51043928e+00 -2.67247647e-01 1.60974130e-01 -1.06367283e-01
4.58209097e-01 -3.11055511e-01 -7.23119855e-01 5.90748847e-01
8.92851174e-01 2.59254444e-02 7.00998724e-01 6.18886590e-01
8.34099472e-01 -5.37699997e-01 -5.09695947e-01 5.15683889e-01
-3.27872843e-01 -1.39100516e+00 -4.91732985e-01 2.93715715e-01
9.92532969e-01 3.08433533e-01 1.76903561e-01 3.14822495e-01
7.72338688e-01 -1.06707382e+00 2.47435242e-01 4.75041598e-01
-3.74963954e-02 -6.75267041e-01 7.93155789e-01 7.55691469e-01
-6.88423097e-01 -8.08644414e-01 -6.25574827e-01 -3.45527142e-01
-2.23519765e-02 -5.02222404e-02 -1.26471686e+00 1.36804178e-01
6.58020496e-01 5.64548850e-01 -2.71984369e-01 1.39558208e+00
1.09546505e-01 1.92016840e-01 -7.05879986e-01 -1.24946117e+00
5.32796621e-01 -2.30646104e-01 1.03577340e+00 5.50564826e-01
4.52811122e-01 -1.19357467e-01 1.83499828e-01 2.69557655e-01
6.50461018e-01 -9.64852236e-03 -5.80360532e-01 1.80052489e-01
4.78064567e-01 1.34698451e+00 -9.93145585e-01 8.92945901e-02
4.70435500e-01 2.58292049e-01 1.82860777e-01 2.36346468e-01
-6.67042136e-01 -6.42323554e-01 8.42845201e-01 -1.06542438e-01
4.01436836e-01 -4.36090380e-01 1.97370481e-02 -5.88964581e-01
-6.94995284e-01 -5.46131372e-01 -1.62275329e-01 -4.15839165e-01
-8.26048374e-01 3.01859081e-01 -1.69799387e-01 -1.27931345e+00
-2.77444959e-01 -8.09656739e-01 -6.65702045e-01 1.98389009e-01
-1.50810039e+00 -7.28115439e-01 -9.23034996e-02 1.10161908e-01
4.07362223e-01 -4.23727840e-01 7.23261535e-01 -2.81285018e-01
-5.81774116e-01 3.08811609e-02 7.98365474e-01 -5.71428418e-01
5.15782595e-01 -1.48759878e+00 -5.25334656e-01 1.52799949e-01
-7.10799277e-01 4.04527724e-01 1.42640138e+00 -4.91581231e-01
-2.22347522e+00 -9.49059546e-01 1.06901471e-02 -4.04600471e-01
8.53470743e-01 -1.85498267e-01 -2.87575990e-01 -1.16792053e-01
2.82512158e-01 -1.01736829e-01 4.41426970e-02 -3.27330083e-01
9.61468637e-01 -2.38918737e-02 -1.13090169e+00 8.68757963e-01
6.01996362e-01 7.20428169e-01 -1.60778947e-02 7.85622239e-01
7.60985851e-01 -7.05208600e-01 -8.76794398e-01 3.15852821e-01
5.69987118e-01 -5.32924116e-01 8.95550668e-01 -5.83462954e-01
7.27054924e-02 -2.19743535e-01 1.65320262e-01 -1.91062224e+00
3.92909162e-02 -1.24000144e+00 1.87531650e-01 4.42811489e-01
4.05723244e-01 -8.99545848e-01 9.75399911e-01 -8.17293152e-02
-1.41087130e-01 -1.04831767e+00 -1.13396382e+00 -9.76294458e-01
5.77859521e-01 3.30160528e-01 2.26135999e-01 6.27028525e-01
4.38712150e-01 1.40106976e-01 -1.46290496e-01 4.08882141e-01
5.46054423e-01 -2.76784599e-01 8.74676585e-01 -1.06328678e+00
-3.80561352e-01 -3.92609894e-01 -1.86170831e-01 -4.53367472e-01
7.29307458e-02 -3.02170187e-01 4.05846566e-01 -1.49406004e+00
-4.96856928e-01 -8.12937498e-01 1.59528360e-01 -1.37481075e-02
1.32384464e-01 -9.82472748e-02 1.33496955e-01 1.54870421e-01
-7.24947929e-01 3.57094884e-01 1.74971163e+00 1.02874547e-01
-7.36380458e-01 3.25429648e-01 -4.44742322e-01 4.85344857e-01
1.13466227e+00 -4.35487747e-01 -2.81547099e-01 -2.92503446e-01
2.46384695e-01 3.43114257e-01 -2.68225968e-02 -1.39008212e+00
5.89160740e-01 -4.53161448e-01 1.16874330e-01 2.42814749e-01
3.13792557e-01 -6.37776673e-01 -3.32471013e-01 1.33801401e+00
-2.93483883e-01 1.05714038e-01 1.23514242e-01 8.69011283e-01
2.44874313e-01 -1.45236343e-01 9.40582454e-01 -3.19863737e-01
-2.84828335e-01 -4.23516966e-02 -9.58193660e-01 -2.52752081e-02
1.56579483e+00 -9.80412681e-03 -4.69787329e-01 -3.84091347e-01
-8.09111357e-01 8.48019958e-01 5.11652887e-01 1.45306900e-01
4.07004714e-01 -5.49957216e-01 -5.53973675e-01 -3.47002856e-02
-4.73307282e-01 2.96266656e-02 -1.50222614e-01 9.37976956e-01
-1.08142900e+00 4.58072126e-01 -4.49735731e-01 -8.03529263e-01
-1.00713420e+00 3.44774544e-01 7.38418341e-01 -1.94908828e-01
-2.77453125e-01 6.69359982e-01 -4.70805109e-01 -5.17855048e-01
1.06019393e-01 -2.38855511e-01 -6.39165461e-01 -3.24558467e-01
3.06618243e-01 5.30572832e-01 -2.62683004e-01 -1.17413320e-01
-1.96303815e-01 8.11773598e-01 2.38739505e-01 -1.57073975e-01
1.65577400e+00 1.70003772e-02 4.76440750e-02 -1.92237809e-01
7.56412625e-01 -2.41780251e-01 -1.61115515e+00 5.64061224e-01
8.30699727e-02 1.27258943e-02 -1.57797083e-01 -5.81907392e-01
-4.27208871e-01 4.92641956e-01 5.69001257e-01 5.15845835e-01
4.86001551e-01 -9.16518196e-02 2.98875570e-01 7.53516495e-01
5.63208163e-01 -1.17343390e+00 6.20675623e-01 7.80280769e-01
6.80318952e-01 -1.19011557e+00 4.31094207e-02 -1.19242258e-01
-5.55572987e-01 1.15031123e+00 5.90273857e-01 -7.91671932e-01
5.79741776e-01 4.95382845e-01 1.32127896e-01 2.53486410e-02
-9.07211244e-01 -2.99589813e-01 -6.66418672e-01 9.81447816e-01
-1.87237114e-02 8.41217414e-02 -3.92303824e-01 -4.52660203e-01
-2.50502080e-01 -2.65681148e-01 9.67158675e-01 1.29355228e+00
-8.44831407e-01 -1.18431890e+00 -4.27844733e-01 1.99003935e-01
-2.79128641e-01 6.70519352e-01 -4.79221195e-02 1.09756470e+00
-6.20182380e-02 1.20243967e+00 -1.70648750e-02 -1.36728287e-01
1.96763724e-01 -7.87203848e-01 4.16446745e-01 -4.91974115e-01
-5.37704647e-01 -2.22135354e-02 -1.97508395e-01 -4.05214816e-01
-4.04245883e-01 -7.76235342e-01 -1.35301518e+00 -2.14151666e-02
-1.92265451e-01 5.79844952e-01 1.06380415e+00 9.63527083e-01
4.08690423e-01 4.93183076e-01 9.67608690e-01 -1.34551024e+00
-9.79333222e-01 -9.02807057e-01 -4.33601350e-01 -1.44939706e-01
5.12088418e-01 -9.20885146e-01 -4.36966091e-01 -5.38461387e-01] | [4.012686252593994, 2.0085997581481934] |
a15ba904-3537-4962-8355-6af1050e83db | provably-personalized-and-robust-federated | 2306.08393 | null | https://arxiv.org/abs/2306.08393v1 | https://arxiv.org/pdf/2306.08393v1.pdf | Provably Personalized and Robust Federated Learning | Clustering clients with similar objectives and learning a model per cluster is an intuitive and interpretable approach to personalization in federated learning. However, doing so with provable and optimal guarantees has remained an open challenge. In this work, we formalize personalized federated learning as a stochastic optimization problem where the stochastic gradients on a client may correspond to one of $K$ distributions. In such a setting, we show that using i) a simple thresholding-based clustering algorithm, and ii) local client gradients obtains optimal convergence guarantees. In fact, our rates asymptotically match those obtained if we knew the true underlying clustering of the clients. Furthermore, our algorithms are provably robust in the Byzantine setting where some fraction of the gradients are corrupted. | ['Martin Jaggi', 'Michael Jordan', 'Sai Praneeth Karimireddy', 'Lie He', 'Mariel Werner'] | 2023-06-14 | null | null | null | null | ['clustering', 'stochastic-optimization', 'personalized-federated-learning'] | ['methodology', 'methodology', 'methodology'] | [-5.20631313e-01 1.16042364e-02 -2.49166936e-01 -5.44737935e-01
-1.06414258e+00 -8.76847327e-01 2.17410356e-01 -5.95642580e-03
-5.50626636e-01 6.99512899e-01 3.12081784e-01 -2.27488890e-01
-3.65964681e-01 -5.84435642e-01 -1.12082744e+00 -1.16480076e+00
-4.34981704e-01 9.29176509e-01 -1.10955246e-01 3.37115407e-01
-6.86763451e-02 4.13607299e-01 -1.16484082e+00 3.97545129e-01
6.23115301e-01 8.40524852e-01 -2.26324067e-01 9.80512619e-01
-3.39357913e-01 8.66236687e-01 -3.82051051e-01 -6.65396512e-01
6.13870382e-01 -3.55460882e-01 -1.10242081e+00 2.24791855e-01
4.19452906e-01 -3.06705713e-01 -3.74766231e-01 1.20326352e+00
3.83484930e-01 5.82082309e-02 4.37046468e-01 -1.30673027e+00
-5.16317427e-01 1.38471019e+00 -3.95224512e-01 4.61740047e-02
3.99323441e-02 1.68581396e-01 1.19119072e+00 -2.14136943e-01
5.17466187e-01 1.04339230e+00 6.47952378e-01 5.51102579e-01
-1.56159925e+00 -3.22746515e-01 3.57747763e-01 2.32281938e-01
-1.22574019e+00 -3.56277913e-01 3.87104690e-01 -1.75385848e-01
3.92515779e-01 4.24141675e-01 1.86420798e-01 6.89424813e-01
-5.77116311e-01 9.48528767e-01 1.02409458e+00 -3.31086129e-01
6.91805303e-01 6.26611650e-01 1.52586848e-01 5.45953155e-01
3.18939567e-01 -2.72642225e-01 -4.14271772e-01 -7.93584466e-01
1.46939784e-01 4.60170776e-01 -3.77762496e-01 -6.39186144e-01
-7.05470502e-01 9.12797034e-01 2.56593913e-01 1.70469031e-01
-4.40124810e-01 7.06172585e-01 2.81369805e-01 4.68068004e-01
3.56442124e-01 -8.61243680e-02 -8.05282176e-01 -1.11608073e-01
-8.22585225e-01 1.68370306e-01 1.37055171e+00 1.04086292e+00
1.12545133e+00 -2.46497154e-01 -1.91341918e-02 4.52158958e-01
9.91650391e-03 5.94564199e-01 2.76656449e-01 -1.58851790e+00
3.84590149e-01 2.08386168e-01 5.30076981e-01 -5.25017321e-01
6.03722967e-02 -2.43458167e-01 -4.43346024e-01 6.15749434e-02
6.76544070e-01 -5.01760244e-01 -3.01767528e-01 2.03985715e+00
5.52203178e-01 8.19775388e-02 -1.02634445e-01 7.51915097e-01
-4.21984941e-01 4.68238890e-01 -1.49543405e-01 -3.54128718e-01
6.80172622e-01 -8.31517220e-01 -4.00923282e-01 3.54803354e-01
5.71812332e-01 -4.02758509e-01 1.07649529e+00 4.29307193e-01
-1.10698390e+00 4.80249256e-01 -4.31097835e-01 4.00653899e-01
-4.63215634e-02 -6.86985731e-01 7.63156712e-01 1.11209655e+00
-1.48621440e+00 8.09697747e-01 -9.65895295e-01 -5.80967665e-01
6.28088295e-01 4.10721868e-01 -1.13302171e-01 -2.91297615e-01
-4.38656271e-01 1.31654367e-01 1.43728212e-01 -7.44887710e-01
-1.04920065e+00 -6.86345756e-01 8.41786503e-04 4.16269630e-01
4.49381948e-01 -8.75664532e-01 1.54683459e+00 -9.26132917e-01
-1.42432261e+00 7.11948097e-01 -2.17775986e-01 -5.32987654e-01
9.02638137e-01 7.28314742e-02 2.87673205e-01 2.65069008e-01
-2.09104657e-01 -3.63547653e-01 6.72409832e-01 -1.63510084e+00
-1.14529455e+00 -8.00198853e-01 2.16273323e-01 7.19728246e-02
-8.59854996e-01 4.01408458e-03 -4.67704028e-01 9.64395143e-03
-1.32352710e-01 -7.44745135e-01 -5.96385539e-01 2.42661074e-01
-1.39877707e-01 -2.84121722e-01 7.93404460e-01 -3.26364636e-01
9.60111082e-01 -2.20429468e+00 -1.55817837e-01 5.27519882e-01
4.28610355e-01 -3.33995849e-01 9.90738422e-02 6.18055761e-01
5.66873431e-01 4.39206660e-01 -9.12532061e-02 -5.78204036e-01
6.29657984e-01 2.76983351e-01 -1.93818510e-01 9.54405427e-01
-9.89705861e-01 4.87954438e-01 -8.72995853e-01 -5.16091526e-01
-1.75720572e-01 1.39645532e-01 -1.04266000e+00 3.85051340e-01
-4.59812582e-01 1.42073497e-01 -5.78974426e-01 4.30428505e-01
6.50471687e-01 -6.08278394e-01 6.83327436e-01 1.96290255e-01
2.72091210e-01 -6.20067120e-02 -1.35733926e+00 1.43507600e+00
-6.60373271e-01 1.20574020e-01 9.67930496e-01 -9.67451572e-01
1.67442679e-01 5.33698142e-01 9.93564010e-01 4.33397070e-02
-5.42779006e-02 2.15862572e-01 -6.76106036e-01 -4.09823358e-01
-2.51428057e-02 -3.17551374e-01 8.05099159e-02 1.30741167e+00
-3.30500258e-03 4.10945654e-01 -2.23326683e-01 6.73111260e-01
1.29523933e+00 -5.62312305e-01 -3.01704615e-01 -3.10210228e-01
2.26814076e-02 -1.11238539e-01 5.81746757e-01 1.26409221e+00
-3.75950187e-01 2.17108116e-01 4.19569969e-01 -3.25488329e-01
-1.15259719e+00 -1.29974282e+00 2.38291457e-01 1.52587712e+00
-6.47544861e-02 -3.24227214e-01 -1.05418694e+00 -8.15545797e-01
3.18679690e-01 5.09940267e-01 -5.25435090e-01 2.31828630e-01
-3.46149266e-01 -7.94593573e-01 3.84425074e-01 1.49900019e-01
2.88089335e-01 -3.70422125e-01 -1.85440764e-01 2.81406820e-01
-1.92791939e-01 -8.66015911e-01 -9.86287892e-01 1.55730367e-01
-8.61007869e-01 -1.20494235e+00 -3.53744298e-01 -5.12730300e-01
6.68701351e-01 5.41579783e-01 9.12787318e-01 -5.00950068e-02
-1.91166252e-01 1.08137608e+00 -2.45864496e-01 1.34212920e-03
-2.56445825e-01 1.10406153e-01 1.78853065e-01 3.35889280e-01
2.86708742e-01 -1.00115538e+00 -8.42035174e-01 -3.19234282e-02
-9.53667819e-01 -6.87547028e-01 1.20989829e-01 5.23679137e-01
4.16532785e-01 1.36681750e-01 2.96984643e-01 -1.35687447e+00
6.91794276e-01 -7.48617113e-01 -5.53279757e-01 4.52188343e-01
-1.06862485e+00 2.03571409e-01 1.17135704e+00 -3.48928034e-01
-1.05810189e+00 3.30311686e-01 2.92531133e-01 -5.99589467e-01
-2.11793274e-01 -1.38889290e-02 -2.91695297e-01 6.68721413e-03
9.10667598e-01 2.65359044e-01 5.26203178e-02 -8.35009873e-01
1.02101636e+00 7.85340548e-01 5.72213531e-01 -1.25856268e+00
8.60835791e-01 1.12863016e+00 -3.85377884e-01 -2.95511067e-01
-6.30113959e-01 -6.07948124e-01 -1.81556746e-01 -9.21559483e-02
2.64109641e-01 -5.61338603e-01 -1.39515471e+00 1.59514770e-01
-7.68818319e-01 -6.98376477e-01 -6.54415488e-01 2.45041624e-01
-8.52580369e-01 7.04637468e-01 -8.96855831e-01 -1.18665111e+00
-5.72670460e-01 -7.46559381e-01 2.69799858e-01 2.63961852e-01
2.62921631e-01 -1.05230486e+00 1.64727703e-01 3.98704886e-01
8.27564955e-01 -2.39424198e-03 7.56063700e-01 -8.80632222e-01
-8.96031737e-01 -3.17506254e-01 3.21388878e-02 1.73408866e-01
1.61572084e-01 -9.79753956e-02 -9.96396482e-01 -7.72809088e-01
1.00240685e-01 -3.98973316e-01 5.49439371e-01 1.62127107e-01
1.58103490e+00 -1.37952745e+00 -2.88573772e-01 9.17425215e-01
1.79441619e+00 -2.35797197e-01 -9.14550647e-02 1.35919258e-01
3.30411792e-01 3.81756514e-01 -8.63341317e-02 1.30767381e+00
5.12753189e-01 2.41923019e-01 4.07151043e-01 4.55344737e-01
3.32396537e-01 -2.88189977e-01 2.63266474e-01 3.87751788e-01
1.34647846e-01 -1.10103391e-01 -6.14377320e-01 9.26956475e-01
-2.09172487e+00 -1.25617564e+00 1.50018603e-01 2.30725956e+00
1.03147876e+00 -4.16654080e-01 5.03891826e-01 -2.32969120e-01
9.18655574e-01 -1.49416596e-01 -9.59338546e-01 -3.37587565e-01
-2.29492217e-01 2.01269135e-01 1.06622577e+00 4.80573177e-01
-5.76204300e-01 7.53694057e-01 6.72337914e+00 6.15116775e-01
-8.27176750e-01 6.73566699e-01 7.29616165e-01 -5.01323700e-01
-5.52733958e-01 2.30085507e-01 -3.95812243e-01 5.76519608e-01
1.08159912e+00 -7.80210197e-01 1.41008067e+00 1.45399022e+00
6.33174405e-02 3.01056325e-01 -1.26441145e+00 9.24993336e-01
-2.92322129e-01 -1.36041605e+00 -2.45708421e-01 2.50254333e-01
1.04462159e+00 2.71163583e-01 3.01071644e-01 6.22882321e-02
1.43340600e+00 -7.18737066e-01 5.39964616e-01 3.15677464e-01
5.20679295e-01 -9.84815359e-01 2.88905740e-01 5.61364591e-01
-7.47469068e-01 -5.68800271e-01 -3.15945148e-01 3.90584052e-01
-3.37882084e-03 5.51061273e-01 -6.79336071e-01 1.50557011e-01
1.22733688e+00 -9.46016833e-02 -1.43192470e-01 1.16499877e+00
1.58145696e-01 1.09439778e+00 -9.36133862e-01 6.25894815e-02
1.24150157e-01 -2.83639073e-01 2.66262025e-01 1.41366518e+00
2.19047014e-02 2.23753899e-01 2.05922589e-01 6.85191095e-01
-7.27333546e-01 2.87162066e-01 -2.79906124e-01 1.39333159e-01
6.89872980e-01 1.26151943e+00 -2.95478076e-01 -4.63459790e-01
-2.82615513e-01 1.19395888e+00 7.81180501e-01 5.89125931e-01
-5.78044951e-01 -1.80474862e-01 1.10184479e+00 1.14017211e-01
5.34669101e-01 1.02315903e-01 -4.19232816e-01 -1.34417582e+00
7.40427850e-03 -7.13976085e-01 1.11974287e+00 7.92977810e-02
-1.79650640e+00 1.50748596e-01 -3.99414003e-01 -3.33321571e-01
-2.23526254e-01 -1.12200834e-01 -7.66884208e-01 5.25121987e-01
-1.08709371e+00 -7.49392569e-01 -4.55279723e-02 1.16910350e+00
-1.20660037e-01 1.34061530e-01 6.66021764e-01 3.99114013e-01
-4.27316844e-01 1.13811159e+00 1.04895258e+00 -2.83255801e-02
7.83968031e-01 -1.44605327e+00 -2.91559935e-01 8.40420961e-01
7.62436315e-02 5.35975099e-01 9.14929330e-01 -6.81335628e-02
-1.80610466e+00 -1.25800419e+00 4.49941665e-01 -2.28131175e-01
7.53966391e-01 -1.86831191e-01 -7.77250588e-01 1.01526189e+00
1.28087014e-01 2.26049945e-01 8.96149695e-01 2.39912197e-01
-6.94489777e-01 -5.43621898e-01 -1.72538567e+00 3.64921123e-01
1.15602505e+00 -5.93362391e-01 1.68727309e-01 9.11913931e-01
8.27170849e-01 1.60777811e-02 -9.13453341e-01 -3.67224813e-01
2.30416179e-01 -9.46836531e-01 5.78197122e-01 -1.12532890e+00
-3.58914763e-01 -4.80752289e-02 -7.04021037e-01 -1.14642429e+00
-3.49622369e-01 -1.13698792e+00 -5.12963414e-01 1.26660764e+00
2.93607414e-01 -8.43244612e-01 1.29978836e+00 1.12953043e+00
3.24570060e-01 -4.43201840e-01 -1.00710738e+00 -9.32210743e-01
3.95269543e-01 -2.90603846e-01 9.09766972e-01 9.53132689e-01
4.62139517e-01 -3.53924632e-01 -2.80543715e-01 2.90411741e-01
1.32734871e+00 3.93137634e-01 8.47177446e-01 -7.71299839e-01
-7.74926364e-01 -5.03005803e-01 7.80413449e-02 -1.02018774e+00
2.57754534e-01 -1.01013768e+00 -3.15191299e-02 -1.26912057e+00
7.72676289e-01 -8.74583960e-01 -3.23842257e-01 3.83463591e-01
-8.86168890e-03 -2.61350453e-01 3.32147151e-01 4.22819823e-01
-1.22938204e+00 3.51798445e-01 3.61842632e-01 -1.05121933e-01
-6.58746958e-02 3.68409514e-01 -1.05224526e+00 2.84444213e-01
9.73730624e-01 -6.35991812e-01 -1.04085185e-01 -4.36337262e-01
4.24388871e-02 1.39616609e-01 3.91711779e-02 -4.72482651e-01
7.89092124e-01 -5.46635211e-01 -2.23585337e-01 -2.69541979e-01
-3.88254263e-02 -9.25911605e-01 3.04317087e-01 5.04994571e-01
-4.69143182e-01 -3.11799586e-01 -5.75885475e-01 9.84307289e-01
4.20933336e-01 -6.07705452e-02 1.04426491e+00 -3.75018001e-01
1.39770627e-01 9.46002960e-01 -4.75217193e-01 4.35310870e-01
1.09044421e+00 1.94080606e-01 -2.00256974e-01 -9.50538158e-01
-7.43092537e-01 4.92566496e-01 1.01116049e+00 -4.52047646e-01
2.14286581e-01 -8.98624063e-01 -6.55180037e-01 -5.97834706e-01
-2.35123008e-01 -9.66259465e-03 2.86906362e-01 6.07552946e-01
-2.41675466e-01 1.96199566e-01 3.17883104e-01 -3.57161015e-01
-9.81038988e-01 1.01338768e+00 6.19838059e-01 -5.46529070e-02
-4.22287136e-01 7.52374053e-01 -2.00408727e-01 -5.38955808e-01
7.82262027e-01 4.21750993e-01 7.55642235e-01 -4.25614268e-01
5.90096354e-01 6.11032248e-01 -1.33123517e-01 -1.63736627e-01
-2.20005408e-01 -1.16354518e-01 -2.17111751e-01 -3.99545550e-01
1.66612482e+00 -4.19578880e-01 -2.58228242e-01 1.66337684e-01
1.49937224e+00 1.72043532e-01 -1.44718313e+00 -7.96908677e-01
1.20782964e-01 -6.90896869e-01 -3.19524556e-01 -5.31559944e-01
-1.38916039e+00 4.09419537e-01 5.49869180e-01 2.95198649e-01
9.10103500e-01 4.70407665e-01 8.03832233e-01 6.47081375e-01
8.01749289e-01 -1.25725543e+00 -1.69144273e-01 -1.94566399e-02
-2.78339867e-04 -9.62842703e-01 -2.88478225e-01 1.48585722e-01
-3.49873781e-01 8.15703928e-01 3.35897952e-01 -3.55438404e-02
8.29619586e-01 1.71997935e-01 -3.07759643e-02 8.07324145e-03
-1.17883277e+00 -7.79036880e-02 -7.88768291e-01 6.33491695e-01
5.40731885e-02 3.35374892e-01 -1.47192568e-01 7.66371071e-01
-5.47740236e-02 -2.79060811e-01 5.07481277e-01 8.88668060e-01
-7.42727637e-01 -1.02604175e+00 -5.76813638e-01 4.86165017e-01
-9.62836444e-01 2.37590760e-01 -2.82262802e-01 4.07655537e-02
-3.74436140e-01 1.24791849e+00 -2.91761607e-01 -1.47085786e-01
-1.00018732e-01 2.09651038e-01 3.09951782e-01 -3.54278088e-01
-6.85358644e-01 -1.34746119e-01 -3.54955316e-01 -6.87393010e-01
-3.09013333e-02 -7.95944035e-01 -1.36710143e+00 -1.09220147e+00
-2.10113481e-01 6.23324275e-01 8.56339812e-01 6.84973300e-01
5.07734895e-01 -3.94393712e-01 1.52144635e+00 -2.29234964e-01
-1.39196920e+00 -4.63057816e-01 -1.09484375e+00 6.75555527e-01
1.47379309e-01 2.07348123e-01 -9.21069801e-01 2.05792353e-01] | [5.88787841796875, 6.1480607986450195] |
d76e73e0-66a2-4d4e-b6b2-3d319fa1b6d7 | aitom-open-source-ai-platform-for-cryo | 1911.03044 | null | https://arxiv.org/abs/1911.03044v2 | https://arxiv.org/pdf/1911.03044v2.pdf | AITom: Open-source AI platform for cryo-electron tomography data analysis | Cryo-electron tomography (cryo-ET) is an emerging technology for the 3D visualization of structural organizations and interactions of subcellular components at near-native state and sub-molecular resolution. Tomograms captured by cryo-ET contain heterogeneous structures representing the complex and dynamic subcellular environment. Since the structures are not purified or fluorescently labeled, the spatial organization and interaction between both the known and unknown structures can be studied in their native environment. The rapid advances of cryo-electron tomography (cryo-ET) have generated abundant 3D cellular imaging data. However, the systematic localization, identification, segmentation, and structural recovery of the subcellular components require efficient and accurate large-scale image analysis methods. We introduce AITom, an open-source artificial intelligence platform for cryo-ET researchers. AITom provides many public as well as in-house algorithms for performing cryo-ET data analysis through both the traditional template-based or template-free approach and the deep learning approach. AITom also supports remote interactive analysis. Comprehensive tutorials for each analysis module are provided to guide the user through. We welcome researchers and developers to join this collaborative open-source software development project. Availability: https://github.com/xulabs/aitom | ['Xiangrui Zeng', 'Min Xu'] | 2019-11-08 | null | null | null | null | ['electron-tomography'] | ['medical'] | [-2.71140754e-01 -6.57832563e-01 3.76365036e-01 -2.53675938e-01
-7.64927268e-01 -6.90613866e-01 4.22625989e-02 1.54697478e-01
-5.57468355e-01 9.79546070e-01 -4.26439226e-01 -2.67080903e-01
3.04777890e-01 -3.12161356e-01 -4.52149868e-01 -1.09003043e+00
-1.79681346e-01 1.04688191e+00 1.26978174e-01 2.30635464e-01
2.96606153e-01 7.94720531e-01 -1.16069901e+00 4.08332407e-01
6.47751451e-01 7.12513685e-01 9.54030216e-01 7.21963942e-01
-1.25190109e-01 1.97239771e-01 -2.47467399e-01 2.68652737e-01
3.29238921e-02 -2.36200854e-01 -8.77754748e-01 -1.19421698e-01
2.91566104e-01 1.00010447e-01 1.54435366e-01 9.24450815e-01
7.11362958e-01 -2.40404934e-01 3.55837017e-01 -6.75847828e-01
-5.42368114e-01 -3.60997230e-01 -3.94146621e-01 7.33727098e-01
3.71556342e-01 1.99848995e-01 6.11814678e-01 -1.23730338e+00
1.23913205e+00 8.35281909e-01 4.28090632e-01 5.29078066e-01
-1.69333112e+00 -6.02996409e-01 -7.61409774e-02 4.92180556e-01
-1.10934150e+00 -5.14326096e-01 5.82994461e-01 -6.13738477e-01
1.12757432e+00 1.61283910e-02 8.56236935e-01 8.44249547e-01
8.06819439e-01 3.28215182e-01 1.53610241e+00 -1.81597918e-01
3.77125710e-01 -2.75844038e-01 1.00785308e-01 7.99303532e-01
-1.02307595e-01 -2.61363685e-01 -1.47110596e-01 -3.92572612e-01
7.35085309e-01 3.26146066e-01 -3.67644101e-01 -5.31857729e-01
-1.32358265e+00 1.31980523e-01 1.10979848e-01 5.66652596e-01
-4.31758702e-01 -2.35775933e-01 5.18781722e-01 2.16397762e-01
3.64548594e-01 4.58866209e-01 -7.73440778e-01 -2.62371115e-02
-8.83532226e-01 1.53876647e-01 2.60271519e-01 4.20162499e-01
7.23157406e-01 -2.70963758e-01 7.78698325e-01 5.32809794e-01
1.98546320e-01 1.87919065e-01 6.28188193e-01 -1.13512504e+00
-2.05282137e-01 6.00309849e-01 2.54596770e-01 -6.30150914e-01
-7.25277722e-01 1.57927617e-01 -7.75412381e-01 4.08656806e-01
6.62389040e-01 -4.46208976e-02 -8.12416613e-01 1.37645948e+00
7.84399271e-01 -9.27271694e-02 -4.55023289e-01 1.20422506e+00
6.62000120e-01 6.00763917e-01 -1.29955277e-01 -4.90948200e-01
1.36338997e+00 -5.77729404e-01 -7.33187973e-01 1.50870919e-01
6.22022927e-01 -6.34063244e-01 9.12630916e-01 3.39018494e-01
-9.66281474e-01 -1.65014043e-01 -9.37900364e-01 -1.82597250e-01
-6.83860123e-01 -5.08790649e-03 4.61057454e-01 3.12005877e-02
-8.75765979e-01 7.57555723e-01 -1.55100775e+00 -5.40392876e-01
5.45703590e-01 4.23227370e-01 -9.88544166e-01 6.37934953e-02
-4.42853093e-01 9.35176194e-01 9.69037786e-02 -9.13735852e-02
-9.54473019e-01 -7.11153209e-01 -3.94761711e-01 3.16766538e-02
-1.33045301e-01 -7.09855378e-01 9.48813617e-01 -3.81520867e-01
-1.21234870e+00 1.28909504e+00 -6.20045602e-01 -3.27420942e-02
-6.65120110e-02 1.54348999e-01 7.13261291e-02 6.04003072e-01
1.95261404e-01 3.69844049e-01 9.62723196e-02 -1.23524463e+00
-2.61265874e-01 -1.08150887e+00 -5.64921498e-01 -2.09565386e-02
3.25138539e-01 4.03943717e-01 -2.74052806e-02 1.24189399e-01
5.79572678e-01 -6.50785267e-01 -1.68891549e-01 3.03604841e-01
-2.26684168e-01 7.22544417e-02 1.37462568e+00 -6.96736753e-01
3.58961701e-01 -1.80858028e+00 3.50853175e-01 -4.25160497e-01
7.59939492e-01 3.51400107e-01 3.07399511e-01 6.88371539e-01
-4.85113233e-01 5.70328087e-02 -7.39001855e-02 -5.07403731e-01
-2.12132365e-01 -3.18547279e-01 2.26720437e-01 9.44036484e-01
-3.94768178e-01 8.04418206e-01 -6.36434734e-01 -6.63132906e-01
4.91796583e-01 6.26021206e-01 1.76001698e-01 1.95905805e-01
-3.87511887e-02 9.48628843e-01 -3.46371591e-01 1.13352787e+00
8.66477251e-01 -6.88078165e-01 8.72178018e-01 -2.74796695e-01
-3.93472761e-01 2.07892865e-01 -6.44604087e-01 1.55637908e+00
5.39170317e-02 4.32407349e-01 9.03545022e-01 -9.65438664e-01
7.62580335e-01 4.26053941e-01 5.87463975e-01 -5.86993217e-01
1.70085346e-03 5.83276115e-02 -1.52222738e-01 -3.49954814e-01
-1.80036232e-01 -5.52766562e-01 3.88081878e-01 7.71042883e-01
1.68607891e-01 1.36887223e-01 1.24935478e-01 2.06032157e-01
9.89349127e-01 3.69020015e-01 4.06298220e-01 -5.70057034e-01
9.05556306e-02 5.91183528e-02 8.16773474e-01 9.03752148e-02
-6.11428022e-01 6.19528651e-01 8.54309872e-02 -9.49287713e-01
-1.47937250e+00 -9.76438284e-01 -5.33491373e-01 8.58539402e-01
2.63348706e-02 -3.74128759e-01 -8.53732884e-01 -3.14374387e-01
-1.60676450e-01 -2.72215039e-01 -5.49759924e-01 4.19571251e-01
-4.91895199e-01 -9.27698374e-01 2.21341923e-01 2.59031415e-01
1.04022592e-01 -1.10189605e+00 -8.62389803e-01 3.41223001e-01
-2.38673329e-01 -1.05181253e+00 -7.81522617e-02 6.82789385e-01
-1.01711452e+00 -1.31028056e+00 -5.41357756e-01 -7.92654097e-01
7.96704769e-01 2.78257936e-01 8.01693320e-01 3.81638139e-01
-6.31148994e-01 2.66095046e-02 -3.69120645e-03 -1.58028454e-02
1.21071868e-01 -1.97083250e-01 4.93459225e-01 -4.08145308e-01
3.16093832e-01 -1.18180501e+00 -6.36141598e-01 3.16296816e-01
-4.92219448e-01 2.45595187e-01 6.39843121e-02 8.28256547e-01
1.31841779e+00 -1.91216946e-01 2.49003559e-01 -8.30098808e-01
3.50261807e-01 -1.51785165e-01 -7.28188097e-01 2.00138599e-01
-4.17436332e-01 -4.32428271e-01 9.47061539e-01 -7.46862665e-02
-7.65438735e-01 1.50088981e-01 -1.98904276e-01 -4.97470737e-01
-7.99596906e-01 4.40559030e-01 -3.79119009e-01 -2.34091282e-01
3.29436928e-01 5.27169526e-01 1.25918180e-01 -6.48120761e-01
-3.53768826e-01 3.25757563e-01 4.29949522e-01 -7.92750001e-01
2.60078579e-01 9.01951194e-01 -4.42567328e-03 -8.30560029e-01
-4.45557117e-01 -4.59352285e-01 -1.16897058e+00 -1.42001241e-01
8.47725511e-01 -5.36639929e-01 -9.74394619e-01 6.23169065e-01
-1.01479733e+00 -5.29782772e-01 2.93899387e-01 3.14121604e-01
-5.10445178e-01 7.39877641e-01 -9.89311814e-01 -4.96299416e-01
-6.35394692e-01 -1.29295444e+00 1.14451671e+00 1.90073416e-01
-1.38722271e-01 -9.69736278e-01 5.04541755e-01 4.22578663e-01
1.69255108e-01 4.14713234e-01 1.03195703e+00 -2.86814690e-01
-7.17351973e-01 -1.57635912e-01 -1.20220654e-01 -3.71253133e-01
2.14564353e-01 3.17190498e-01 -8.23418617e-01 -5.48013628e-01
1.11563519e-01 -5.12230754e-01 5.08192658e-01 6.59678578e-01
1.01721072e+00 1.43131435e-01 -7.73474693e-01 9.99901593e-01
1.35034835e+00 5.40073514e-01 5.68701029e-01 6.80895209e-01
5.71077824e-01 5.20196438e-01 4.33373153e-01 3.02723318e-01
1.15606964e-01 5.61155856e-01 3.81774664e-01 -1.73249558e-01
3.78065914e-01 2.78164148e-01 -1.85016617e-01 7.06775188e-01
-3.59258294e-01 1.36627600e-01 -1.08405101e+00 3.42217565e-01
-1.62545872e+00 -1.11375093e+00 -7.00983778e-02 1.76902258e+00
6.70903921e-01 -2.70172983e-01 1.36764064e-01 -1.06885806e-01
7.72508085e-01 -1.56334430e-01 -8.73348653e-01 -1.02577411e-01
-2.39528283e-01 1.43787175e-01 -4.13124524e-02 3.24826181e-01
-1.01787817e+00 7.90530980e-01 6.88901806e+00 3.94737452e-01
-1.48044288e+00 9.12197158e-02 6.15907371e-01 -1.49556369e-01
2.21213669e-01 7.42042512e-02 -8.24782908e-01 6.62879169e-01
9.59860623e-01 -2.68295586e-01 3.69102806e-01 6.96608186e-01
5.91164529e-01 -3.25263023e-01 -8.71005654e-01 9.23975587e-01
-5.48950315e-01 -2.02349687e+00 -4.48830724e-01 5.67508519e-01
1.81967348e-01 6.22397363e-01 -2.48665720e-01 -4.12543416e-01
5.00787199e-02 -8.15218270e-01 2.96494246e-01 6.99175119e-01
9.05263007e-01 -5.69842339e-01 6.57428384e-01 5.25834620e-01
-1.26843846e+00 1.53378978e-01 -6.90364540e-01 3.18968529e-03
4.52261716e-01 7.35919952e-01 -6.13286436e-01 2.03178123e-01
1.11820889e+00 4.56485212e-01 -2.92200327e-01 9.10043418e-01
3.70970637e-01 2.81396806e-01 -3.82214040e-01 3.88191611e-01
-3.03211957e-01 -7.51435935e-01 3.84570718e-01 1.13049436e+00
-9.00018215e-02 3.91073346e-01 2.37517864e-01 1.12424886e+00
-1.35717532e-02 -2.17920855e-01 -5.08761048e-01 -3.82818520e-01
5.64524114e-01 1.78351498e+00 -1.25558543e+00 -1.15751565e-01
-1.75063148e-01 7.87574887e-01 7.10752487e-01 2.46733710e-01
-4.32409972e-01 -3.72782677e-01 9.80253756e-01 4.79878694e-01
2.40010202e-01 -6.46297395e-01 -1.74807087e-01 -1.18674874e+00
-2.08167851e-01 -7.40695298e-01 2.15206712e-01 -1.09861207e+00
-1.07228577e+00 5.02164662e-01 -3.79365772e-01 -6.35004640e-01
3.52521867e-01 -7.17916369e-01 -8.37299466e-01 7.35961914e-01
-8.57229292e-01 -1.03944719e+00 -3.51454839e-02 2.65490472e-01
2.63370454e-01 1.01837948e-01 1.25775707e+00 2.90284246e-01
-9.02611136e-01 -3.04961428e-02 6.29056573e-01 -8.86345729e-02
6.16291344e-01 -1.37762427e+00 2.10106716e-01 5.36199152e-01
-3.31483632e-01 1.06236625e+00 6.41918719e-01 -7.37040758e-01
-1.55396581e+00 -7.62506127e-01 6.98073328e-01 -6.20604038e-01
4.94832128e-01 -6.43160999e-01 -1.16896653e+00 9.43769693e-01
2.11016893e-01 3.20186913e-01 1.10344648e+00 -6.76548705e-02
6.55880477e-03 2.51321852e-01 -1.54860163e+00 2.67279327e-01
6.37853980e-01 -6.48385704e-01 -3.92404884e-01 4.86868918e-01
2.99907535e-01 -2.70163924e-01 -1.22516704e+00 3.18300016e-02
7.63211668e-01 -1.30423439e+00 7.12618530e-01 -5.68413973e-01
1.94812074e-01 -6.01680338e-01 -2.15524919e-02 -8.91756296e-01
-5.25024772e-01 -2.84638792e-01 4.01429869e-02 6.88898444e-01
2.05127299e-01 -5.93106329e-01 1.09393382e+00 6.93557680e-01
-4.46080923e-01 -1.07485187e+00 -1.10605001e+00 -4.07187849e-01
9.77115855e-02 2.15709940e-01 4.83976156e-01 1.07738078e+00
5.26911497e-01 2.83750713e-01 1.31897524e-01 -7.12794717e-03
7.80579567e-01 6.44378781e-01 5.77498019e-01 -1.29748094e+00
-3.99645455e-02 1.16714381e-01 -4.22948033e-01 -6.73129022e-01
2.26584628e-01 -6.19175494e-01 -2.47967854e-01 -1.48761809e+00
6.74662471e-01 1.42311174e-02 -4.35167272e-03 3.99876863e-01
3.35865110e-01 2.12265700e-01 -1.34004906e-01 6.77707076e-01
-9.16018546e-01 5.31760991e-01 1.60470307e+00 2.28436440e-01
2.06151083e-01 -3.96874726e-01 -3.33775759e-01 6.67850196e-01
9.25473273e-01 -5.21539092e-01 1.27069771e-01 -2.72821724e-01
-2.15544596e-01 3.14658582e-02 3.12505871e-01 -1.02060878e+00
3.72782588e-01 -1.75233036e-01 7.95692503e-01 -1.05136919e+00
6.12176239e-01 -7.60221541e-01 4.82313365e-01 1.70854375e-01
2.75304615e-01 2.33483449e-01 7.75073245e-02 1.88288286e-01
-8.07895325e-03 2.11025834e-01 1.30569327e+00 -9.35280859e-01
-1.61398619e-01 4.04247701e-01 -1.00984299e+00 -4.48176116e-01
8.93075943e-01 -6.26114964e-01 -6.90888464e-01 3.40188742e-02
-1.28388417e+00 1.11949161e-01 1.33976936e+00 -6.75707221e-01
8.21776628e-01 -8.96460891e-01 -1.78610254e-02 2.59045303e-01
-2.62955755e-01 1.18998036e-01 5.00229776e-01 1.12608743e+00
-9.24396992e-01 6.30602479e-01 -8.31986070e-01 -5.74306190e-01
-1.48903167e+00 6.29641473e-01 7.23412395e-01 -2.14917406e-01
-8.58402014e-01 4.97003824e-01 3.05543363e-01 -9.35460865e-01
-2.97025949e-01 7.23996386e-02 -5.27599901e-02 -4.46939051e-01
7.88194180e-01 2.62204647e-01 1.18142858e-01 -7.31508017e-01
-3.40902328e-01 4.39945638e-01 -3.33133936e-01 3.04420918e-01
1.72752297e+00 -6.13626897e-01 -6.57151759e-01 5.71536243e-01
9.66189444e-01 -5.37349403e-01 -1.07433939e+00 1.39367387e-01
-1.55890733e-01 -2.33816385e-01 1.00985639e-01 -6.40501976e-01
-9.23763514e-01 1.12042749e+00 6.42120838e-01 4.84645506e-03
9.11147118e-01 2.15849116e-01 8.50699365e-01 5.41754365e-01
6.94417000e-01 -7.92507529e-01 -4.07215774e-01 5.75825632e-01
4.05212939e-01 -9.22693074e-01 1.74760461e-01 -2.36577198e-01
6.57318532e-02 1.00068474e+00 8.89453828e-01 -2.96188351e-02
5.29452264e-01 7.74425149e-01 4.02319908e-01 -8.02995443e-01
-1.25705612e+00 2.67169356e-01 -5.86314082e-01 9.52457607e-01
7.60420680e-01 -2.16587946e-01 -9.53231752e-03 5.39511204e-01
1.55689269e-01 -5.30954599e-02 4.02281553e-01 1.17624784e+00
-5.49318373e-01 -1.12419152e+00 -4.75241244e-01 3.61647457e-01
-7.14620352e-01 1.79699540e-01 -7.28883684e-01 5.58952689e-01
-2.20283419e-01 2.80189186e-01 1.48503676e-01 -1.39564455e-01
-1.55079573e-01 3.96614492e-01 6.28658116e-01 -5.87853193e-01
-2.40789354e-01 5.21708369e-01 -1.95617408e-01 -6.48868978e-01
-3.76148313e-01 -7.14751959e-01 -1.62961650e+00 -5.28718650e-01
-3.78109813e-01 5.62718630e-01 7.40159333e-01 9.48532641e-01
8.57057989e-01 6.22895621e-02 3.46640617e-01 -1.51088715e+00
3.92483473e-01 -1.01078165e+00 -1.14091659e+00 -9.41247791e-02
1.78086773e-01 -6.39532089e-01 -3.04063737e-01 2.56867409e-01] | [13.494174003601074, -3.0942296981811523] |
e2fbfb0b-4388-488a-87df-27ec8526d19e | drone-path-following-in-gps-denied | 1905.01658 | null | https://arxiv.org/abs/1905.01658v1 | https://arxiv.org/pdf/1905.01658v1.pdf | Drone Path-Following in GPS-Denied Environments using Convolutional Networks | his paper presents a simple approach for drone navigation to follow a predetermined path using visual input only without reliance on a Global Positioning System (GPS). A Convolutional Neural Network (CNN) is used to output the steering command of the drone in an end-to-end approach. We tested our approach in two simulated environments in the Unreal Engine using the AirSim plugin for drone simulation. Results show that the proposed approach, despite its simplicity, has average cross track distance less than 2.9 meters in the simulated environment. We also investigate the significance of data augmentation in path following. Finally, we conclude by suggesting possible enhancements for extending our approach to more difficult paths in real life, in the hope that one day visual navigation will become the norm in GPS-denied zones. | ['M. Shaker', 'M. ElHelw', 'K. Amer', 'M. Samy'] | 2019-05-05 | null | null | null | null | ['drone-navigation'] | ['computer-vision'] | [-2.88796782e-01 1.75883427e-01 3.54409814e-01 -4.25975770e-01
1.90855801e-01 -1.18754065e+00 7.23389149e-01 -3.37092370e-01
-6.74723923e-01 8.32744360e-01 -1.72221914e-01 -1.13553715e+00
-6.33883551e-02 -1.06876957e+00 -5.13816774e-01 -1.45369917e-01
-4.13252056e-01 -1.25560477e-01 3.86670262e-01 -9.28371131e-01
1.10868983e-01 1.07669044e+00 -1.45669043e+00 -4.21123296e-01
4.00388449e-01 8.85392785e-01 -2.22740248e-01 1.27443302e+00
7.90935278e-01 4.19048429e-01 -4.96154398e-01 4.00110297e-02
8.91949236e-01 2.10124366e-02 -2.61283875e-01 -4.23988819e-01
7.14996219e-01 -3.98072690e-01 -7.13169873e-01 5.79788804e-01
6.71226084e-01 6.47540152e-01 2.93872774e-01 -1.34865046e+00
2.40736142e-01 -3.60425949e-01 1.34512827e-01 3.19376856e-01
4.58162218e-01 5.94681144e-01 2.82679856e-01 -5.79331338e-01
6.06110394e-01 7.79571414e-01 1.10253274e+00 4.85214442e-02
-6.72024786e-01 -5.13855875e-01 -3.84821296e-02 -3.82098079e-01
-1.42530596e+00 -4.50420439e-01 1.08068079e-01 -4.23560530e-01
1.28334320e+00 2.04949677e-01 6.72499776e-01 7.63075352e-01
6.46936595e-01 -1.04136750e-01 4.72254634e-01 1.38795674e-01
1.75605357e-01 5.40210027e-03 -4.38418776e-01 1.02886438e+00
4.04677659e-01 9.23104703e-01 -6.97772279e-02 1.14712603e-01
9.04411554e-01 -3.65616679e-01 -5.52853465e-01 -7.57686794e-01
-1.35432947e+00 6.72846973e-01 1.08135223e+00 -3.02882701e-01
-3.66069674e-01 4.98194575e-01 2.81613201e-01 2.96128869e-01
-1.77742988e-01 6.00148618e-01 -4.11593258e-01 -3.62677574e-01
-9.58090246e-01 6.61997437e-01 7.39660859e-01 1.33523142e+00
5.12936771e-01 6.35011256e-01 1.37582749e-01 -1.01914823e-01
3.50558400e-01 5.86803615e-01 1.61649603e-02 -1.11406779e+00
5.81845641e-01 1.21048644e-01 7.43741095e-01 -1.32876313e+00
-9.88896191e-01 -7.95068145e-01 -4.99270797e-01 1.12810695e+00
3.91696692e-01 -1.15996504e+00 -1.02749157e+00 1.22942746e+00
1.74511105e-01 -1.16086178e-01 1.85157269e-01 1.05734158e+00
5.83671689e-01 5.71783185e-01 -3.41236383e-01 8.76608968e-01
7.79524028e-01 -1.00598991e+00 -4.82793301e-01 -5.72228074e-01
8.51666510e-01 -5.37699401e-01 5.40787816e-01 4.93282109e-01
-6.44213974e-01 -8.61707270e-01 -1.75476837e+00 1.28442466e-01
-8.49618495e-01 5.29135764e-01 4.19231117e-01 9.31050420e-01
-1.46160567e+00 5.06744385e-01 -6.66510582e-01 -6.85306728e-01
-1.94280073e-02 6.08864248e-01 -5.27921736e-01 2.82302439e-01
-1.34896684e+00 1.16121173e+00 2.38386929e-01 3.97343069e-01
-8.68940890e-01 -3.26281130e-01 -1.08437550e+00 -2.93301400e-02
1.51758596e-01 -9.90324318e-01 1.41360831e+00 -5.30916989e-01
-1.64488900e+00 2.10763171e-01 1.19781695e-01 -8.65924835e-01
7.52988219e-01 -3.96121293e-01 -7.55061388e-01 -3.03265780e-01
-5.97283095e-02 1.07480419e+00 2.17578217e-01 -1.15086544e+00
-1.20991945e+00 1.00561708e-01 6.22281611e-01 5.53081036e-01
7.13835776e-01 -7.64753163e-01 -4.36958641e-01 -2.47373089e-01
4.23505642e-02 -1.28277862e+00 -7.08845437e-01 3.21167231e-01
-2.84420043e-01 6.29878581e-01 6.13073945e-01 -3.10170591e-01
1.01485085e+00 -1.90075147e+00 -3.27975988e-01 6.24529064e-01
-1.12480950e-02 2.58482695e-01 2.34282225e-01 7.30046332e-01
6.22993223e-02 -1.44980118e-01 6.63125813e-02 -6.49939254e-02
-6.25870153e-02 -1.38891283e-02 -4.70304906e-01 5.95888317e-01
-2.25955606e-01 7.70503104e-01 -8.54667783e-01 4.06374186e-01
4.49709535e-01 5.23657620e-01 -4.69410986e-01 -1.41870216e-01
3.16225708e-01 4.22130942e-01 7.18605667e-02 2.77189195e-01
6.78864896e-01 3.99852067e-01 -1.54064700e-01 1.29815564e-01
-9.19205070e-01 3.41854930e-01 -1.12558019e+00 1.57573450e+00
-5.66460252e-01 1.41851461e+00 1.77114904e-01 -8.86649713e-02
1.03613245e+00 -2.79545458e-03 -3.10955405e-01 -9.34639215e-01
2.94020712e-01 1.19110322e-04 1.15403421e-01 -2.75986567e-02
1.11138988e+00 2.72265643e-01 -1.95332855e-01 2.52984222e-02
-3.88092458e-01 -1.80609196e-01 -7.61611015e-02 5.39378263e-02
1.03179705e+00 2.87185520e-01 3.66354972e-01 -2.85826772e-01
3.32287997e-01 5.48516631e-01 7.87237734e-02 8.49937499e-01
-3.35756302e-01 6.69831991e-01 1.42307729e-01 -6.76638722e-01
-1.03929853e+00 -9.91378665e-01 2.40354791e-01 3.89003485e-01
3.94603819e-01 -3.62657309e-01 -4.10339922e-01 -6.81173384e-01
8.95409286e-02 8.08959663e-01 -5.94481945e-01 1.59168262e-02
-4.11392897e-01 -1.29745007e-01 1.00902510e+00 4.78419453e-01
7.49959409e-01 -5.44618428e-01 -1.28190351e+00 2.06903487e-01
3.20724964e-01 -1.04204834e+00 4.07801196e-02 9.37480778e-02
-5.55374444e-01 -9.19283330e-01 -5.35662651e-01 -3.18530351e-01
6.57415330e-01 6.48996055e-01 6.84769452e-01 6.78834990e-02
3.24467450e-01 4.15831506e-01 -8.94504040e-02 -2.91723430e-01
1.84835225e-01 1.50259003e-01 1.61995426e-01 -5.70114970e-01
-5.13540320e-02 -7.86040053e-02 -1.06662214e+00 6.02674961e-01
-1.97493359e-01 -2.44512856e-02 2.38728032e-01 2.41035536e-01
3.97682115e-02 1.34898007e-01 5.96988201e-02 -4.76527750e-01
6.41197920e-01 -5.99101007e-01 -1.20830929e+00 -4.75262344e-01
-7.05866039e-01 -3.62235606e-01 8.62191260e-01 3.38734329e-01
-7.54460931e-01 3.40098828e-01 -1.58104211e-01 9.86306593e-02
-3.35947245e-01 5.98751307e-01 1.34869725e-01 -5.96673906e-01
8.73281777e-01 -1.46243826e-01 -1.32951155e-01 1.08174525e-01
5.00963151e-01 4.25119460e-01 7.39964724e-01 3.94993722e-01
9.91682231e-01 8.03644061e-01 3.88459921e-01 -8.06037486e-01
9.62004811e-02 -1.41241968e-01 -7.05144048e-01 -3.82657260e-01
5.86942554e-01 -1.19993818e+00 -8.85544479e-01 -2.75451764e-02
-9.12891686e-01 -5.91605246e-01 2.55144566e-01 4.71521050e-01
-5.70110321e-01 6.94041252e-02 1.37262240e-01 -6.28106058e-01
-2.25007348e-02 -9.17409360e-01 5.69840252e-01 5.83470643e-01
-5.45529366e-01 -1.12720346e+00 4.08182681e-01 -3.24125379e-01
7.32856512e-01 3.40782940e-01 5.09527652e-03 -2.33346283e-01
-7.46150255e-01 -6.53573930e-01 -2.10097492e-01 -2.31890067e-01
4.65666037e-03 -2.08988804e-02 -8.83682370e-01 -5.35683513e-01
-7.66086400e-01 3.72138560e-01 5.97100079e-01 3.67487073e-01
4.91468787e-01 7.77333826e-02 -5.75514913e-01 9.86975491e-01
1.51608348e+00 7.96332896e-01 8.41034293e-01 7.68770635e-01
5.84963143e-01 4.34515715e-01 1.11586487e+00 3.08954835e-01
5.20979762e-01 7.15855122e-01 9.16176081e-01 -4.91655737e-01
1.79260001e-01 -4.13214058e-01 2.74976432e-01 -3.32427263e-01
-1.56643063e-01 -8.78397226e-01 -1.06568706e+00 5.89880705e-01
-1.70532072e+00 -5.67864478e-01 -4.47787583e-01 2.46745014e+00
-4.66020167e-01 4.29317504e-01 -1.02368351e-02 -1.38993114e-01
2.00803161e-01 5.70015833e-02 -1.33694842e-01 -8.83020282e-01
2.79498994e-01 -1.20067455e-01 1.44514096e+00 1.01819611e+00
-1.32289612e+00 8.53545487e-01 6.78337526e+00 -1.47967741e-01
-1.34911311e+00 -5.24815977e-01 -1.42322585e-01 -8.48186314e-02
1.46728471e-01 1.93041816e-01 -9.03637767e-01 3.33309472e-01
1.39570844e+00 -2.01837402e-02 2.94000328e-01 8.66209209e-01
5.86423039e-01 -4.88394201e-01 -6.86715782e-01 5.39119780e-01
-2.52277732e-01 -1.48863769e+00 -4.68996823e-01 3.19859505e-01
5.20652652e-01 4.13034081e-01 3.08811367e-01 2.99168319e-01
4.48308736e-01 -1.07807815e+00 7.20635772e-01 5.67608535e-01
8.18749964e-01 -1.32120562e+00 9.31532562e-01 1.84852168e-01
-1.33958459e+00 -8.48252624e-02 -1.92215011e-01 -3.80955905e-01
4.23832834e-01 -1.82721838e-01 -1.53529882e+00 7.96799898e-01
5.73059201e-01 3.94840300e-01 -7.01436341e-01 1.70565677e+00
-3.43580365e-01 2.42330834e-01 -4.12508130e-01 2.07413584e-02
9.52799022e-01 7.52310902e-02 6.36237144e-01 1.14388943e+00
8.99168074e-01 -1.48115978e-01 -1.69881731e-01 3.20260495e-01
6.50335133e-01 -6.10898852e-01 -1.52274656e+00 4.15128022e-01
2.74810880e-01 1.12834835e+00 -7.01028764e-01 -3.94815877e-02
-2.08175778e-01 1.02577436e+00 -2.11096570e-01 7.63936639e-01
-8.54513466e-01 -1.25241256e+00 1.00778913e+00 2.80363560e-01
4.93538588e-01 -8.11978698e-01 -1.82020158e-01 -3.83563787e-01
-2.92775273e-01 -2.94638842e-01 -2.07542926e-01 -1.22360229e+00
-1.98040366e-01 8.55352640e-01 -2.84873903e-01 -1.91861248e+00
-5.24605632e-01 -9.04479623e-01 -5.38434565e-01 1.17431474e+00
-1.48103678e+00 -6.99989080e-01 -6.51581168e-01 2.39598647e-01
2.25383356e-01 -2.40679428e-01 7.57838666e-01 4.08809930e-01
-1.81398600e-01 4.38041836e-01 2.86959093e-02 -5.87211102e-02
5.42832017e-01 -1.25039744e+00 1.29259050e+00 1.19446921e+00
-4.03653592e-01 7.53504932e-01 9.84708011e-01 -7.45932877e-01
-9.52426910e-01 -1.07372117e+00 8.43039632e-01 -5.83530307e-01
3.39026690e-01 -3.45447809e-01 -2.54550874e-01 1.10856795e+00
5.47544777e-01 -1.17898740e-01 3.14713806e-01 -2.80106276e-01
4.54078108e-01 1.36127204e-01 -1.06484854e+00 1.28249598e+00
9.06548142e-01 -1.44577846e-01 3.48567590e-02 -1.81185052e-01
4.83624548e-01 -1.01780248e+00 -2.59449184e-01 3.45038682e-01
7.70568728e-01 -1.24986005e+00 8.83250594e-01 -3.02467495e-01
-8.69971216e-02 -1.06441879e+00 3.63417454e-02 -1.77151859e+00
-3.98438215e-01 -7.42160797e-01 3.27394038e-01 3.72095466e-01
6.48233294e-01 -7.67007709e-01 9.45198119e-01 3.49755436e-01
-3.89749289e-01 -4.21443135e-01 -9.80610371e-01 -5.76460838e-01
-3.87562126e-01 -6.76627517e-01 5.98211825e-01 3.58694077e-01
-3.12703639e-01 5.60335964e-02 -5.76115668e-01 9.67370093e-01
-1.26057252e-01 -5.71203530e-01 1.29501343e+00 -1.01233315e+00
3.56644213e-01 -1.34998292e-01 -7.66394436e-01 -1.46922207e+00
-3.78455073e-01 -5.22007108e-01 1.74101055e-01 -1.72675657e+00
-1.32985246e+00 -3.44576180e-01 1.08731948e-01 2.56269779e-02
5.60435176e-01 2.78833508e-01 6.01631515e-02 -5.35439670e-01
-2.65040547e-01 4.61072236e-01 1.03494430e+00 1.79109856e-01
-2.30301216e-01 6.56994343e-01 -4.07250136e-01 6.31782234e-01
8.78342330e-01 -1.33140475e-01 -7.26464152e-01 -3.57326090e-01
6.97914302e-01 6.80028498e-02 6.25908434e-01 -1.75939655e+00
6.27634346e-01 2.75539607e-01 7.16418684e-01 -1.10689795e+00
5.62585115e-01 -1.22556710e+00 1.07563250e-01 4.73678112e-01
2.14407101e-01 8.17284226e-01 9.41930830e-01 4.34976816e-01
-4.73473705e-02 1.63266376e-01 4.33386743e-01 1.81496143e-01
-1.19181991e+00 1.29762799e-01 -1.16597402e+00 -5.52124918e-01
1.01612270e+00 -7.17502832e-01 -2.95592070e-01 -1.03363359e+00
-6.36113942e-01 5.70680082e-01 7.23038316e-01 4.73719239e-01
7.62923717e-01 -1.33105028e+00 3.55065949e-02 6.58341706e-01
7.79387727e-02 -2.98407167e-01 1.73395798e-01 5.08812785e-01
-1.45518017e+00 1.18290448e+00 -6.00230932e-01 -3.82610738e-01
-9.95088577e-01 1.84363142e-01 8.19184005e-01 3.06989759e-01
-5.11586010e-01 6.02137268e-01 -1.29251257e-01 -9.49269533e-01
1.68054596e-01 -5.91785550e-01 -3.08029473e-01 -2.73524076e-01
3.84178489e-01 4.58742350e-01 2.84444362e-01 -7.14594305e-01
-5.55375695e-01 5.27380764e-01 4.78781372e-01 -6.24410212e-01
6.52991235e-01 -6.24002814e-01 8.39846075e-01 1.00450717e-01
7.43093550e-01 3.75619143e-01 -1.60961974e+00 6.30669057e-01
-2.05432698e-01 -4.20984805e-01 2.60143787e-01 -1.04494739e+00
-7.65487134e-01 8.67352545e-01 1.05212331e+00 -1.09154349e-02
6.07230306e-01 -1.05978799e+00 4.27776605e-01 9.96061862e-01
4.94746357e-01 -6.59936130e-01 -9.45833802e-01 1.07035553e+00
6.96311057e-01 -1.12364018e+00 6.14994504e-02 -6.22701347e-02
-7.40070581e-01 1.34552324e+00 7.63434827e-01 -2.78701246e-01
5.79833508e-01 1.27218470e-01 6.25525832e-01 -3.38158756e-01
-6.26334906e-01 -4.22363609e-01 1.87191054e-01 9.93262708e-01
2.36725628e-01 -9.65296552e-02 2.71723896e-01 3.58874500e-02
-7.70195723e-01 -1.38157487e-01 8.68527949e-01 1.26121402e+00
-5.05511284e-01 -5.53340018e-01 -2.20898062e-01 -2.84963876e-01
3.01575512e-02 -5.67603484e-02 -4.76725161e-01 1.50657141e+00
1.80919200e-01 9.54474211e-01 3.54611665e-01 -7.50908315e-01
8.19738030e-01 -3.59183669e-01 6.59178644e-02 -3.65824074e-01
-7.06543446e-01 -4.25094604e-01 5.85746169e-01 -9.60381091e-01
2.67954439e-01 -1.62772119e-01 -1.36977851e+00 -4.08317894e-01
2.52094388e-01 -1.47031635e-01 8.39828670e-01 3.98991197e-01
7.73945451e-01 7.96925068e-01 4.48475182e-01 -1.20425057e+00
2.89779395e-01 -3.30520391e-01 -2.52514213e-01 -8.10445726e-01
8.16230237e-01 -6.91226482e-01 -1.36303127e-01 -4.73032743e-01] | [7.237626075744629, -1.8911594152450562] |
f82808f7-6c97-45f3-adbb-853b706fb199 | multi-level-matching-and-aggregation-network | 1906.06678 | null | https://arxiv.org/abs/1906.06678v1 | https://arxiv.org/pdf/1906.06678v1.pdf | Multi-Level Matching and Aggregation Network for Few-Shot Relation Classification | This paper presents a multi-level matching and aggregation network (MLMAN) for few-shot relation classification. Previous studies on this topic adopt prototypical networks, which calculate the embedding vector of a query instance and the prototype vector of each support set independently. In contrast, our proposed MLMAN model encodes the query instance and each support set in an interactive way by considering their matching information at both local and instance levels. The final class prototype for each support set is obtained by attentive aggregation over the representations of its support instances, where the weights are calculated using the query instance. Experimental results demonstrate the effectiveness of our proposed methods, which achieve a new state-of-the-art performance on the FewRel dataset. | ['Zhen-Hua Ling', 'Zhi-Xiu Ye'] | 2019-06-16 | multi-level-matching-and-aggregation-network-1 | https://aclanthology.org/P19-1277 | https://aclanthology.org/P19-1277.pdf | acl-2019-7 | ['few-shot-relation-classification', 'few-shot-relation-classification'] | ['methodology', 'natural-language-processing'] | [ 1.51469097e-01 3.28754991e-01 -7.61155009e-01 -4.49744344e-01
-4.44503367e-01 5.73092028e-02 6.65925205e-01 6.78922713e-01
-2.78595865e-01 4.98970360e-01 1.24922611e-01 4.16068882e-01
-5.58988512e-01 -1.21993470e+00 -1.23123221e-01 -4.03133422e-01
-3.89710426e-01 7.50315547e-01 5.86561143e-01 -3.03707212e-01
-1.11802984e-02 4.15037811e-01 -1.89160538e+00 6.14948690e-01
6.38170719e-01 1.36727571e+00 -1.65901005e-01 4.45319027e-01
-3.59328806e-01 9.76954639e-01 -7.42709935e-01 -6.66137278e-01
-1.19245753e-01 -3.74673754e-01 -9.16288197e-01 -6.23924583e-02
5.45689404e-01 1.04250632e-01 -6.94543421e-01 1.13332033e+00
3.23738843e-01 5.58319628e-01 7.94441462e-01 -1.47239840e+00
-8.74606073e-01 9.86361444e-01 -3.87162864e-01 5.93136013e-01
3.97815764e-01 -2.57792443e-01 1.52591586e+00 -1.20991230e+00
8.63760412e-01 1.32887506e+00 2.84302771e-01 1.01305656e-01
-1.23363113e+00 -6.28079355e-01 3.13650519e-01 9.95987058e-01
-1.72461891e+00 -4.19400215e-01 9.01993036e-01 -2.47666210e-01
1.28063798e+00 1.90665811e-01 7.16957510e-01 7.44404495e-01
5.07604256e-02 6.79807365e-01 4.68613416e-01 -3.35461885e-01
3.97536546e-01 1.77483052e-01 1.00777006e+00 7.62935996e-01
2.87153363e-01 -1.07190289e-01 -6.53354287e-01 -4.99632627e-01
3.80569726e-01 2.56308556e-01 -1.76681027e-01 -5.47372282e-01
-9.67421651e-01 8.55382144e-01 9.18792427e-01 7.35520720e-01
-6.56403542e-01 -2.15274125e-01 3.92377049e-01 4.73201811e-01
5.42623937e-01 3.10132384e-01 -1.06719330e-01 3.08775574e-01
-7.69024670e-01 5.19436821e-02 9.89878595e-01 1.15663254e+00
9.19850171e-01 -3.32358420e-01 -7.34408557e-01 9.62534130e-01
7.50194937e-02 -2.92693138e-01 1.75405100e-01 -6.99990988e-01
4.74192798e-01 1.12394023e+00 -2.01944560e-01 -1.37027419e+00
-2.76185691e-01 -6.18232429e-01 -1.03415024e+00 -8.71913061e-02
-1.35736629e-01 4.11302805e-01 -6.16472542e-01 1.60460150e+00
6.38666093e-01 7.03826666e-01 3.92008305e-01 6.91070020e-01
1.35115051e+00 5.88471711e-01 2.86674649e-01 -5.54003656e-01
1.46225035e+00 -1.10683680e+00 -8.81652176e-01 -3.53274308e-02
4.57759231e-01 -3.28748316e-01 6.81115806e-01 -1.93720143e-02
-9.79189217e-01 -7.42816329e-01 -1.36323023e+00 1.34560123e-01
-5.45042932e-01 -3.45438756e-02 5.58847547e-01 1.43590823e-01
-6.14119530e-01 7.61498451e-01 -5.04510939e-01 -4.25441206e-01
6.28446400e-01 2.66160164e-02 -3.36564422e-01 -1.72023892e-01
-1.65269053e+00 1.01275194e+00 8.08394849e-01 -2.23828882e-01
-5.84873259e-01 -7.50710249e-01 -1.05419147e+00 5.33788741e-01
7.22752273e-01 -4.88171935e-01 7.97696471e-01 -4.09083873e-01
-9.32947755e-01 6.56290352e-01 -2.10461885e-01 -5.28944790e-01
-4.46694233e-02 2.06827655e-01 -8.47525835e-01 2.85257906e-01
3.04973517e-02 4.22851622e-01 4.74952489e-01 -1.19120574e+00
-7.24612832e-01 -2.80145884e-01 3.12707156e-01 2.41234943e-01
-5.54184914e-01 -1.26442298e-01 -4.77569699e-01 -4.48862165e-01
2.43944719e-01 -2.69551933e-01 -1.60670504e-01 -3.64710279e-02
-3.79044473e-01 -8.04881394e-01 8.18219185e-01 -1.44164145e-01
1.71180534e+00 -2.05363369e+00 2.72069335e-01 3.42576444e-01
5.61450183e-01 3.23564112e-01 -3.12972367e-01 6.08940840e-01
-1.32168606e-01 -1.88583687e-01 -1.28426522e-01 -4.98453379e-02
-8.42852220e-02 1.68171495e-01 -2.11707368e-01 2.32694492e-01
2.66263872e-01 9.83280361e-01 -1.19852138e+00 -8.81378412e-01
1.84060693e-01 5.60415387e-01 -3.08086365e-01 3.60181361e-01
9.55379084e-02 -5.32847941e-01 -3.40814322e-01 6.40331745e-01
3.42979610e-01 -5.08303106e-01 5.02958417e-01 -7.58588314e-01
5.07766008e-01 2.03923807e-01 -1.35016954e+00 1.43053925e+00
-9.59635973e-02 2.38140687e-01 -4.34691191e-01 -1.16940367e+00
1.28162658e+00 3.00112993e-01 3.56956691e-01 -4.22572404e-01
-5.49728200e-02 -1.50090694e-01 1.27687246e-01 -3.43732923e-01
5.41011930e-01 1.22216254e-01 -2.73975059e-02 3.02762806e-01
4.76417005e-01 3.02790403e-01 4.44736332e-01 6.84570074e-01
1.04690695e+00 -3.01594108e-01 9.15143669e-01 -3.94602753e-02
5.43090224e-01 -1.51102707e-01 5.80172002e-01 9.98267174e-01
-1.59315810e-01 1.97954327e-01 4.99474049e-01 -6.07240975e-01
-5.32043874e-01 -1.07063508e+00 -1.27444118e-01 1.14884341e+00
3.94710660e-01 -8.21871281e-01 -2.49380320e-01 -8.85985911e-01
1.36917666e-01 7.83056915e-01 -8.11246872e-01 -5.63403428e-01
-3.57638389e-01 -5.05901396e-01 2.81710833e-01 4.60777551e-01
6.19172215e-01 -1.32650197e+00 -5.35524547e-01 4.57534999e-01
-1.75156772e-01 -9.20306265e-01 -2.88673252e-01 8.83343592e-02
-5.82741082e-01 -1.23571754e+00 -1.64531365e-01 -9.24760997e-01
4.72049654e-01 2.38178730e-01 1.28740120e+00 4.07360911e-01
-5.35731554e-01 8.89271870e-02 -3.69009852e-01 7.76745826e-02
-1.74367532e-01 1.77926019e-01 3.95732298e-02 3.52906078e-01
6.60400987e-01 -7.45674908e-01 -3.85505557e-01 2.29981795e-01
-7.14495063e-01 -4.51679081e-01 3.33238363e-01 8.49739075e-01
7.89150476e-01 8.17965567e-02 7.96078801e-01 -1.07117426e+00
9.11093712e-01 -7.92091370e-01 -1.78311065e-01 7.57096291e-01
-8.47777665e-01 -2.48649493e-01 3.21776211e-01 -6.35490835e-01
-8.38125944e-01 -3.18553299e-01 4.75946188e-01 -6.71465099e-01
-7.96337277e-02 7.68139899e-01 -1.45477980e-01 1.80408001e-01
7.68221736e-01 -4.81697582e-02 -3.45047683e-01 -2.30186909e-01
5.72511673e-01 6.35961950e-01 7.21534729e-01 -4.30765003e-01
6.57432795e-01 2.08018988e-01 -2.70233899e-02 -7.23714530e-01
-1.33545291e+00 -6.53768539e-01 -8.56897593e-01 -3.29405427e-01
6.42841399e-01 -6.58057451e-01 -6.70204401e-01 4.74347696e-02
-1.19486105e+00 4.68774259e-01 -6.77772105e-01 2.90701419e-01
-1.52229622e-01 3.08622479e-01 -6.08125806e-01 -7.08678901e-01
-3.98033291e-01 -6.99546456e-01 5.91725945e-01 2.91742206e-01
-4.56972361e-01 -8.88036668e-01 1.25953183e-01 7.87396263e-03
1.77510634e-01 1.23692445e-01 1.14913189e+00 -1.18764126e+00
-4.07946765e-01 -4.82903421e-01 -3.13205838e-01 -4.32924852e-02
1.40733629e-01 -1.32510230e-01 -9.16836262e-01 -2.74956346e-01
-2.06127107e-01 -3.65207464e-01 1.07310855e+00 3.31298895e-02
1.02034223e+00 -3.17497134e-01 -7.13635683e-01 2.69610822e-01
1.35684180e+00 1.79938301e-01 4.60840136e-01 -4.23225947e-02
5.41889369e-01 5.99750757e-01 9.83274877e-01 4.53140587e-01
3.78063053e-01 7.73756087e-01 3.20531517e-01 1.25902623e-01
-1.19700655e-01 -1.46537468e-01 -1.15803003e-01 6.73388064e-01
-1.69413358e-01 -3.26896399e-01 -6.28796518e-01 5.89585781e-01
-2.21927714e+00 -1.40657783e+00 2.17130065e-01 2.03667450e+00
6.35291159e-01 3.37577373e-01 3.88071383e-03 1.86902925e-01
9.51345861e-01 5.15681207e-01 -5.67719162e-01 -2.29437739e-01
-1.69355109e-01 3.32602769e-01 -2.30634004e-01 5.28490901e-01
-1.16258717e+00 9.69173253e-01 6.55292273e+00 1.01238978e+00
-2.77333468e-01 5.58456331e-02 4.60557699e-01 -1.63083121e-01
-6.30589351e-02 3.70785315e-03 -8.14077616e-01 4.66245338e-02
8.26588809e-01 -6.07960403e-01 2.53808826e-01 8.77109289e-01
-4.69590276e-01 1.64636876e-02 -1.47226298e+00 8.02017450e-01
3.18768889e-01 -1.83918452e+00 3.39913040e-01 -1.54737264e-01
4.95340019e-01 -3.67459446e-01 -2.35182300e-01 7.37897813e-01
2.45153040e-01 -6.75166190e-01 1.97701886e-01 8.39430273e-01
5.56991756e-01 -1.13411498e+00 8.24605942e-01 2.46421009e-01
-1.75224102e+00 -1.27384171e-01 -6.90689802e-01 -1.33595869e-01
-1.84819959e-02 4.94626701e-01 -7.72141755e-01 6.62089467e-01
5.16635418e-01 9.72106695e-01 -6.53893650e-01 8.83160830e-01
-3.04734349e-01 3.83029968e-01 4.10770439e-02 -2.62760311e-01
1.30218014e-01 -8.87960792e-02 6.13523483e-01 1.35770035e+00
-2.08248906e-02 3.73854965e-01 6.49550557e-01 8.34545612e-01
-2.56973386e-01 1.97350204e-01 -5.73272943e-01 8.75685513e-02
1.13122666e+00 1.50128043e+00 -6.62225306e-01 -9.30495799e-01
-1.83965981e-01 6.48124456e-01 9.23286438e-01 2.59194940e-01
-5.86233437e-01 -9.72881973e-01 8.45431328e-01 -2.14452669e-01
4.83523786e-01 4.94116366e-01 5.95711097e-02 -1.00847697e+00
-1.30120307e-01 -4.29067671e-01 1.16988266e+00 -5.44974506e-01
-1.61280656e+00 7.18689620e-01 2.10826308e-01 -1.30318344e+00
-3.20750296e-01 -8.78560841e-02 -8.00871491e-01 6.62385702e-01
-1.15745759e+00 -9.41735446e-01 -2.46070594e-01 5.96249163e-01
3.49935204e-01 -4.62560117e-01 1.16841650e+00 9.58045572e-02
-5.36630273e-01 7.78216541e-01 -4.32225376e-01 2.78464913e-01
2.11080834e-01 -8.96684945e-01 1.75471425e-01 5.12382448e-01
5.76341450e-01 6.93701446e-01 5.03822565e-01 -6.46900058e-01
-9.48806763e-01 -1.17795515e+00 1.27503896e+00 -2.27725953e-01
7.26908267e-01 -6.80774450e-02 -1.31000423e+00 5.35925984e-01
3.16776872e-01 2.63888836e-01 8.97558868e-01 4.83909607e-01
-6.89289689e-01 -3.67540807e-01 -1.38375354e+00 3.83124202e-01
1.39799643e+00 -6.37705564e-01 -1.08190370e+00 3.26028079e-01
9.85606372e-01 -3.65556851e-02 -1.33584976e+00 5.66074610e-01
3.52994800e-01 -9.02128994e-01 1.18021810e+00 -9.99659002e-01
3.43354672e-01 -3.10442835e-01 -2.81771928e-01 -1.30743718e+00
-8.82362783e-01 -9.06521976e-02 -8.24025095e-01 1.24528575e+00
3.01967889e-01 -4.97933179e-01 4.80907172e-01 2.13034377e-01
7.61410147e-02 -1.10170126e+00 -1.18238747e+00 -6.78988874e-01
-6.86697841e-01 -3.86649013e-01 9.48422074e-01 1.25919724e+00
5.81769347e-01 8.37213993e-01 -2.83241689e-01 3.22091281e-01
8.66661191e-01 6.85042083e-01 2.20918626e-01 -1.57828736e+00
-2.63059527e-01 -3.75186086e-01 -9.59755063e-01 -5.14665902e-01
2.90186584e-01 -1.38364065e+00 -2.28546411e-01 -1.56030715e+00
5.18601537e-01 -2.08848462e-01 -9.75702882e-01 4.49759394e-01
-3.23325723e-01 2.39080399e-01 1.11500271e-01 1.62705898e-01
-9.91282225e-01 4.61605817e-01 9.92885649e-01 -4.89647269e-01
-1.68639466e-01 -2.62930930e-01 -6.24668241e-01 5.67941666e-01
4.34814781e-01 -4.72600132e-01 -5.64469337e-01 6.33149073e-02
-1.32566616e-01 1.95334554e-01 2.96234816e-01 -1.05878842e+00
6.35342062e-01 -2.62710214e-01 4.11952764e-01 -7.82787919e-01
6.86352849e-01 -6.53183222e-01 1.27829492e-01 5.24179041e-01
-7.25808263e-01 -3.94630492e-01 -2.80158162e-01 8.45523059e-01
-3.85981143e-01 -1.38674244e-01 8.55805337e-01 1.43357798e-01
-7.56142616e-01 6.85930371e-01 2.02715829e-01 9.18936282e-02
1.14256680e+00 -1.01917848e-01 -5.15093327e-01 -2.12339342e-01
-1.06537235e+00 3.37796450e-01 -5.95708862e-02 6.16407752e-01
1.17219567e+00 -1.84907925e+00 -7.72844076e-01 2.63179392e-01
7.88220346e-01 -2.87408948e-01 3.08300585e-01 5.15688658e-01
4.28085268e-01 1.02879249e-01 -8.67989287e-02 -3.82238567e-01
-1.32247639e+00 9.21391606e-01 2.27634460e-01 -4.93831307e-01
-6.50639892e-01 9.00723159e-01 -2.60113794e-02 -1.03129700e-01
3.47252309e-01 2.29722522e-02 -8.89089286e-01 6.39426768e-01
9.67260897e-01 5.35227835e-01 -1.29333615e-01 -8.22968006e-01
-5.19495904e-01 3.53629112e-01 -3.29070479e-01 2.39545956e-01
1.39121711e+00 3.39690417e-01 -3.13584268e-01 6.59593165e-01
1.15081239e+00 -5.76972842e-01 -5.39255977e-01 -9.43644404e-01
1.92495793e-01 -5.97189307e-01 -1.10546872e-01 -4.63090032e-01
-1.04915571e+00 3.54185462e-01 4.61812586e-01 2.53861040e-01
8.48937869e-01 5.88626444e-01 3.16420645e-01 7.69862056e-01
4.38869625e-01 -9.51030374e-01 2.33182147e-01 4.31593895e-01
8.40595067e-01 -9.60003972e-01 1.69149395e-02 -5.66471875e-01
-6.71088636e-01 1.00920630e+00 8.94841015e-01 -2.67087609e-01
9.01360095e-01 1.53759932e-02 -3.40517074e-01 -6.63916230e-01
-1.23646164e+00 -2.74205387e-01 6.05289221e-01 5.16360819e-01
3.14030707e-01 2.25412652e-01 -3.58734459e-01 6.75386071e-01
1.83673754e-01 -2.14409873e-01 1.65487841e-01 9.57539201e-01
-8.01535606e-01 -8.03067744e-01 5.66823073e-02 8.23640823e-01
1.04950823e-01 -1.05921380e-01 -5.04294693e-01 5.37441373e-01
4.24032621e-02 1.00565720e+00 3.35896462e-01 -8.04462314e-01
6.62871599e-01 3.72145772e-01 2.93549120e-01 -1.04592347e+00
-5.81292033e-01 -3.83348763e-01 3.29538673e-01 -6.21428907e-01
-4.68573123e-01 -4.50616300e-01 -1.23243296e+00 -1.74152434e-01
-4.08150703e-01 3.09925377e-01 -1.77679583e-01 9.00949776e-01
3.66703302e-01 6.63536310e-01 6.70596302e-01 -5.47970116e-01
-4.72360790e-01 -1.14309311e+00 -7.44642258e-01 5.12788594e-01
5.14336303e-02 -9.75472569e-01 -3.15677166e-01 -3.58237594e-01] | [9.188006401062012, 8.499299049377441] |
044dbbf8-db20-4ccf-85dc-ef01f148cc02 | reproducing-activation-function-for-deep | 2101.04844 | null | https://arxiv.org/abs/2101.04844v2 | https://arxiv.org/pdf/2101.04844v2.pdf | Reproducing Activation Function for Deep Learning | We propose reproducing activation functions (RAFs) to improve deep learning accuracy for various applications ranging from computer vision to scientific computing. The idea is to employ several basic functions and their learnable linear combination to construct neuron-wise data-driven activation functions for each neuron. Armed with RAFs, neural networks (NNs) can reproduce traditional approximation tools and, therefore, approximate target functions with a smaller number of parameters than traditional NNs. In NN training, RAFs can generate neural tangent kernels (NTKs) with a better condition number than traditional activation functions lessening the spectral bias of deep learning. As demonstrated by extensive numerical tests, the proposed RAFs can facilitate the convergence of deep learning optimization for a solution with higher accuracy than existing deep learning solvers for audio/image/video reconstruction, PDEs, and eigenvalue problems. With RAFs, the errors of audio/video reconstruction, PDEs, and eigenvalue problems are decreased by over 14%, 73%, 99%, respectively, compared with baseline, while the performance of image reconstruction increases by 58%. | ['Haizhao Yang', 'Chunmei Wang', 'Liyao Lyu', 'Senwei Liang'] | 2021-01-13 | null | null | null | null | ['video-reconstruction'] | ['computer-vision'] | [-3.72723877e-01 5.57133891e-02 1.46429971e-01 7.98269734e-02
-7.10252047e-01 -8.86258669e-03 5.73042892e-02 -5.35019696e-01
-3.84537935e-01 6.11423254e-01 -2.14613397e-02 -1.97356075e-01
6.19002581e-02 -5.40306628e-01 -1.14475918e+00 -8.09343934e-01
1.71648383e-01 -1.20695010e-02 -2.35376611e-01 -1.60501808e-01
-8.90542343e-02 5.19521058e-01 -9.82870281e-01 2.30762362e-01
1.05225146e+00 1.43489420e+00 -1.38183311e-01 4.30727750e-01
2.24658132e-01 1.00128138e+00 -2.84530580e-01 -1.51735798e-01
4.86059308e-01 -4.85869646e-02 -4.91021097e-01 -3.37108523e-02
2.40468308e-01 -6.53106272e-01 -8.06614876e-01 9.48471308e-01
5.08846104e-01 5.19438505e-01 8.16619158e-01 -1.03135121e+00
-8.30274999e-01 6.08559728e-01 -8.34982932e-01 -1.42798930e-01
-2.10964143e-01 3.96002978e-01 5.99883020e-01 -1.55040252e+00
5.56244180e-02 1.26083124e+00 1.19530749e+00 4.37029541e-01
-1.45311022e+00 -9.82332289e-01 -3.16914879e-02 1.65629685e-01
-1.66948247e+00 -2.24230006e-01 7.47236729e-01 -2.17145056e-01
8.23714554e-01 -8.29192344e-03 9.00741637e-01 1.03136158e+00
5.95587045e-02 9.23943937e-01 5.79380333e-01 -8.35032389e-02
1.07317887e-01 1.20259508e-01 -5.33179268e-02 8.50295365e-01
1.84298446e-03 4.86621726e-03 -4.30162162e-01 -1.25908613e-01
1.50716150e+00 3.88982892e-02 -6.80477858e-01 -1.07115090e-01
-1.10975146e+00 1.04118776e+00 5.96106827e-01 -2.27050468e-01
-5.97727001e-01 4.89283144e-01 5.19136131e-01 7.61048049e-02
6.07759535e-01 5.61935723e-01 -3.59189957e-01 -1.00393623e-01
-6.35470629e-01 3.39404821e-01 6.76059783e-01 8.54901671e-01
8.44815016e-01 1.01521361e+00 -2.49469839e-02 1.09197462e+00
1.37623221e-01 7.17030227e-01 5.45400441e-01 -1.59172571e+00
1.66315675e-01 4.97232616e-01 2.46843714e-02 -1.15319419e+00
-4.27773207e-01 -7.65517414e-01 -1.56678128e+00 6.99021593e-02
1.59526631e-01 -4.76752728e-01 -6.88833117e-01 1.50782561e+00
4.69081819e-01 6.92189813e-01 1.80226892e-01 1.25697422e+00
7.24520683e-01 1.26545095e+00 -4.48737800e-01 -1.86925441e-01
7.67790556e-01 -1.00612140e+00 -4.60960746e-01 1.20267630e-01
6.70838118e-01 -5.63480556e-01 1.28185010e+00 5.19430161e-01
-1.16786468e+00 -5.02537668e-01 -7.26663411e-01 -1.72493339e-01
3.84947360e-01 3.52794200e-01 7.82423317e-01 -4.39944156e-02
-1.20378506e+00 7.85367310e-01 -7.76090324e-01 4.04137433e-01
5.04784107e-01 4.13592100e-01 -7.41300508e-02 1.91962615e-01
-1.00337231e+00 5.15754819e-01 1.56399861e-01 3.64295155e-01
-8.72329473e-01 -1.66481304e+00 -6.39042497e-01 3.30117226e-01
1.50793687e-01 -7.06804454e-01 1.22958446e+00 -1.18365371e+00
-1.93608308e+00 3.15891147e-01 2.01745480e-01 -6.14735425e-01
4.19920146e-01 -4.31722462e-01 -4.36985120e-02 2.54502714e-01
-2.13933364e-01 9.47806358e-01 1.29015827e+00 -9.85937357e-01
-2.25443646e-01 3.16366017e-01 -1.53403461e-01 1.46404386e-01
-6.96543515e-01 -4.00265902e-01 -2.13337675e-01 -8.34860861e-01
-8.71803313e-02 -9.54025209e-01 -4.01862234e-01 4.80286211e-01
-2.74487466e-01 -1.16281144e-01 7.13165998e-01 -8.14016640e-01
7.10032344e-01 -2.43036532e+00 4.40516233e-01 1.84379891e-01
4.99550462e-01 4.64465678e-01 -2.65876174e-01 -5.55224856e-03
-2.58214682e-01 -1.88902333e-01 -1.07969984e-01 -2.20559835e-01
-1.78187355e-01 1.19700164e-01 -6.73750222e-01 4.28148001e-01
1.85959116e-01 9.97295737e-01 -5.85881412e-01 -1.83440670e-01
2.04984188e-01 1.01381791e+00 -9.25021052e-01 1.83652446e-01
-7.88855776e-02 3.16075712e-01 -1.84738442e-01 3.94165099e-01
6.09968245e-01 -3.17869395e-01 -2.40633205e-01 -5.40329576e-01
4.09798287e-02 -1.02029368e-01 -1.29895759e+00 1.43109667e+00
-7.24947929e-01 9.03746009e-01 2.34291926e-01 -1.50595629e+00
8.39914382e-01 3.39560300e-01 6.42901301e-01 -6.10394835e-01
7.77804777e-02 2.63504952e-01 -1.29861414e-01 -3.00369233e-01
1.15851715e-01 9.04599577e-02 4.92323160e-01 1.66146263e-01
-1.25690550e-02 1.40825585e-01 -2.85806388e-01 7.12245107e-02
8.10776532e-01 -9.72618982e-02 -3.53825569e-01 -4.49632972e-01
4.33974355e-01 -1.07143134e-01 7.12306380e-01 4.61349875e-01
3.44209552e-01 5.49782813e-01 4.83517230e-01 -8.05794775e-01
-1.40260661e+00 -8.37267637e-01 -2.07860008e-01 8.81909728e-01
-8.45051184e-02 -1.69445261e-01 -1.01734340e+00 5.02701253e-02
8.07069801e-03 3.80625665e-01 -3.64811659e-01 -5.13243258e-01
-8.33561838e-01 -5.99240184e-01 7.63640583e-01 7.36908257e-01
7.96672761e-01 -9.60724652e-01 -2.55257457e-01 1.88477963e-01
-9.35667157e-02 -1.25276685e+00 -6.61432087e-01 -1.40131131e-01
-1.11670506e+00 -7.02644229e-01 -1.21935976e+00 -8.92524183e-01
7.07170904e-01 1.79190695e-01 6.16415739e-01 -5.26946671e-02
-2.54059881e-01 3.64958078e-01 1.41022325e-01 -4.88714054e-02
-3.41567576e-01 -3.27633731e-02 3.92571300e-01 2.55338550e-01
-4.63575423e-01 -8.80908370e-01 -5.91685474e-01 2.37908006e-01
-8.39290798e-01 4.21258658e-01 6.36329412e-01 1.11513436e+00
6.56964064e-01 -2.91167676e-01 5.12014508e-01 -3.68790805e-01
6.96172655e-01 -4.83362883e-01 -8.15718353e-01 -8.93961340e-02
-5.51433921e-01 2.59972662e-01 1.25879800e+00 -1.20347989e+00
-8.34832013e-01 -5.52747399e-02 -2.22046324e-03 -1.32916093e+00
3.89671743e-01 5.52849829e-01 4.90215778e-01 -5.67420065e-01
8.08243692e-01 4.26280469e-01 2.84189165e-01 -4.76716220e-01
1.82292283e-01 1.25658154e-01 6.46578431e-01 -5.99706292e-01
8.21296215e-01 1.49369240e-01 1.30267248e-01 -1.06494927e+00
-7.52210259e-01 -3.29030991e-01 6.62612170e-02 -4.01606560e-01
3.53258610e-01 -1.22674918e+00 -1.10019588e+00 8.30205023e-01
-1.20410597e+00 -5.98417819e-01 -4.65349019e-01 7.37965882e-01
-3.57808322e-01 2.13339612e-01 -1.21204638e+00 -7.28912294e-01
-6.53326869e-01 -1.09262025e+00 8.65660071e-01 2.92232692e-01
1.39657721e-01 -8.57722640e-01 -2.78153330e-01 1.84158519e-01
5.29798090e-01 2.33884245e-01 8.63502800e-01 1.05694383e-01
-4.34109896e-01 4.93319035e-02 -4.08575505e-01 8.30326378e-01
-2.63897717e-01 6.68498650e-02 -7.18022108e-01 -2.75173813e-01
2.79028744e-01 -4.55044985e-01 7.39401460e-01 6.24442458e-01
1.44931173e+00 -9.14643347e-01 4.21401486e-02 1.32738507e+00
1.25676382e+00 -4.42644022e-02 4.83050108e-01 1.71450675e-02
1.03790259e+00 1.35118619e-01 -2.05255616e-02 6.69238985e-01
-1.78635865e-02 4.85064328e-01 4.10765648e-01 -3.52053463e-01
5.17583191e-02 5.04212789e-02 7.26366401e-01 1.04402506e+00
-3.21357489e-01 4.79017109e-01 -6.64281785e-01 2.88639367e-01
-1.97584808e+00 -5.68916261e-01 -2.27279201e-01 1.93090796e+00
9.41876173e-01 -2.86692977e-01 1.74960285e-01 1.08658209e-01
5.72595298e-01 -3.29127759e-01 -1.12576473e+00 -3.14122826e-01
9.13684964e-02 3.08880895e-01 4.41012681e-01 4.01555479e-01
-6.99602365e-01 7.91493535e-01 6.12073278e+00 9.54680920e-01
-1.59879518e+00 -1.14689678e-01 7.92000711e-01 -4.20691311e-01
-1.67873323e-01 -3.94455463e-01 -4.98037010e-01 2.42977828e-01
9.55268025e-01 1.52823599e-02 1.01290059e+00 1.30597997e+00
3.48906696e-01 4.97760803e-01 -7.67166257e-01 1.44424391e+00
-2.58923799e-01 -1.92402828e+00 1.08079195e-01 9.75660086e-02
9.84536827e-01 8.44972301e-03 2.97386974e-01 4.65527475e-01
8.37024674e-02 -1.21523774e+00 6.92905366e-01 5.89537382e-01
7.42557108e-01 -1.09255004e+00 5.45984268e-01 2.86694527e-01
-9.36869264e-01 -9.51629207e-02 -7.14777052e-01 -1.77449778e-01
-1.11645795e-01 7.13675618e-01 -5.25158763e-01 2.77033329e-01
8.04190218e-01 8.70393574e-01 5.77906482e-02 8.41066658e-01
1.16969742e-01 8.65780711e-01 -4.76917237e-01 -8.99611041e-02
5.34654677e-01 -7.38474071e-01 5.81595480e-01 1.02523518e+00
3.26168299e-01 2.88287729e-01 7.28460923e-02 1.27198756e+00
-3.64982367e-01 1.17569759e-01 -2.20902771e-01 7.34982863e-02
2.28594601e-01 1.35308099e+00 -1.59378231e-01 -1.71125770e-01
-1.78926677e-01 9.61567163e-01 4.47492212e-01 9.02430356e-01
-1.21698022e+00 -3.68411124e-01 9.98223186e-01 2.48970032e-01
2.08702534e-01 -2.87141204e-01 -2.82029718e-01 -1.22675455e+00
1.06975310e-01 -9.73210275e-01 -1.67305470e-01 -8.86828601e-01
-9.63496745e-01 4.68798250e-01 -2.80818820e-01 -1.11485291e+00
-2.00701848e-01 -8.21820736e-01 -6.61921442e-01 7.61659861e-01
-1.18375468e+00 -7.98490286e-01 -3.35065901e-01 6.70149624e-01
3.81098986e-01 -3.07565391e-01 5.81822574e-01 4.64657664e-01
-9.54175174e-01 7.34714925e-01 4.66101915e-01 4.29563880e-01
2.95214981e-01 -7.94029891e-01 1.20460473e-01 5.65517902e-01
-2.89069838e-03 5.33150554e-01 4.69319612e-01 -5.92547469e-02
-1.70816422e+00 -1.13846672e+00 -4.55028526e-02 3.86899292e-01
7.48674572e-01 -1.85261965e-01 -1.27081311e+00 5.87970972e-01
1.62233431e-02 3.42927814e-01 2.56530076e-01 -2.47269318e-01
-5.36033869e-01 -5.27305841e-01 -8.93645227e-01 7.47912824e-01
7.68912435e-01 -3.88434917e-01 3.74613367e-02 4.04468864e-01
9.87582743e-01 -7.69834876e-01 -9.95952368e-01 4.00092721e-01
5.06304264e-01 -7.37934768e-01 1.37282431e+00 -5.94946504e-01
8.16259861e-01 3.21079185e-03 5.31356223e-02 -1.43730748e+00
-4.84380811e-01 -9.47697997e-01 -6.60328150e-01 8.49555910e-01
2.73082227e-01 -7.60383844e-01 5.81822515e-01 5.40309131e-01
-6.44804478e-01 -1.40979004e+00 -9.19375420e-01 -8.50339472e-01
2.84157842e-01 -4.48359996e-01 2.23387495e-01 9.58560467e-01
-3.24429184e-01 2.30407655e-01 -5.15309513e-01 2.06419662e-01
6.10746562e-01 -2.69288182e-01 6.11059904e-01 -8.68077636e-01
-5.08185863e-01 -6.25605822e-01 -1.94905937e-01 -1.27881837e+00
3.59852999e-01 -8.05682778e-01 -1.06985085e-01 -1.24768376e+00
-8.21597502e-02 -4.33443844e-01 -3.52589898e-02 5.12308955e-01
7.19597936e-02 2.90522635e-01 1.10757917e-01 2.20338255e-01
-9.53410268e-02 1.01581693e+00 1.27715981e+00 -1.73323736e-01
-2.19614744e-01 -3.06102872e-01 -5.41063011e-01 9.15937960e-01
5.82907677e-01 -1.69285789e-01 -3.78428042e-01 -6.33913100e-01
2.54787505e-01 5.08202314e-02 5.24521887e-01 -1.16730607e+00
1.49720088e-01 -6.33711740e-02 6.07244730e-01 -4.98750284e-02
4.58465606e-01 -5.09012341e-01 9.44462866e-02 5.75336993e-01
-2.90422231e-01 -1.55561864e-01 4.64429647e-01 3.49686563e-01
-1.61455259e-01 -2.68929526e-02 1.00550258e+00 1.40849143e-01
-3.37051511e-01 5.89745522e-01 -6.02851808e-01 2.09506556e-01
7.14705229e-01 -1.15502678e-01 1.54654428e-01 -6.14759803e-01
-2.96900332e-01 1.03714719e-01 -9.00559500e-03 -4.67097536e-02
8.70134175e-01 -1.71938050e+00 -7.02494681e-01 1.77014798e-01
-5.36382139e-01 6.53113365e-01 2.86933124e-01 9.72901165e-01
-7.51512766e-01 8.49380866e-02 -2.80656014e-02 -7.11586535e-01
-6.08188808e-01 2.31786668e-01 7.22251713e-01 -5.28894812e-02
-1.08953083e+00 1.01785302e+00 4.43107367e-01 -5.48604608e-01
4.63830292e-01 -6.82815075e-01 2.06350207e-01 -3.45736206e-01
5.05734801e-01 7.89217830e-01 -1.16512306e-01 -1.11193158e-01
1.14571480e-02 5.50978363e-01 6.02633445e-05 9.08039734e-02
1.56218433e+00 5.28863549e-01 3.25972922e-02 2.49368742e-01
1.43558049e+00 -3.97236168e-01 -1.85393715e+00 -2.75066406e-01
-6.13592088e-01 7.46536106e-02 4.37706709e-01 -2.11173654e-01
-1.62714243e+00 8.81178200e-01 3.85475129e-01 -1.18295439e-01
1.17006421e+00 -3.38389695e-01 1.27089834e+00 6.78976059e-01
-1.53869614e-01 -1.13999331e+00 4.42407876e-01 6.48207247e-01
1.22416592e+00 -8.51884842e-01 -3.14728200e-01 -2.95293927e-01
-6.02729142e-01 1.40654683e+00 7.98044264e-01 -7.13890731e-01
5.01316071e-01 4.53466445e-01 -2.25693330e-01 2.37432465e-01
-5.77516913e-01 5.56371987e-01 4.26492780e-01 3.36953402e-01
2.29905099e-01 -2.49355793e-01 2.10680991e-01 6.34504795e-01
-9.25200507e-02 -9.53630134e-02 4.73295629e-01 1.56814426e-01
-2.38382205e-01 -4.48801935e-01 -3.06794077e-01 5.46935022e-01
-7.55131692e-02 -3.02600801e-01 6.30024150e-02 7.11880922e-01
-4.15511787e-01 2.71162003e-01 1.52687252e-01 -2.52426445e-01
1.66018695e-01 -1.44058898e-01 3.64454061e-01 -1.51118055e-01
-4.70555514e-01 1.48351550e-01 -2.71237642e-01 -6.30955994e-01
8.61668959e-02 -2.11847812e-01 -1.67314458e+00 -7.71345437e-01
-2.63519555e-01 7.94573575e-02 3.45599651e-01 9.91893709e-01
5.43102324e-01 5.93862116e-01 6.66811347e-01 -1.12424278e+00
-9.17382121e-01 -8.18818152e-01 -4.22588199e-01 1.87767178e-01
3.13966125e-01 -5.56042910e-01 -4.37975407e-01 -1.56629592e-01] | [11.25480842590332, -1.6514348983764648] |
992aab46-f59f-4f0b-9ade-3ffd4bbbd450 | layered-tpot-speeding-up-tree-based-pipeline | 1801.06007 | null | http://arxiv.org/abs/1801.06007v2 | http://arxiv.org/pdf/1801.06007v2.pdf | Layered TPOT: Speeding up Tree-based Pipeline Optimization | With the demand for machine learning increasing, so does the demand for tools
which make it easier to use. Automated machine learning (AutoML) tools have
been developed to address this need, such as the Tree-Based Pipeline
Optimization Tool (TPOT) which uses genetic programming to build optimal
pipelines. We introduce Layered TPOT, a modification to TPOT which aims to
create pipelines equally good as the original, but in significantly less time.
This approach evaluates candidate pipelines on increasingly large subsets of
the data according to their fitness, using a modified evolutionary algorithm to
allow for separate competition between pipelines trained on different sample
sizes. Empirical evaluation shows that, on sufficiently large datasets, Layered
TPOT indeed finds better models faster. | ['Randal S. Olson', 'Pieter Gijsbers', 'Joaquin Vanschoren'] | 2018-01-18 | null | null | null | null | ['automated-feature-engineering'] | ['methodology'] | [-1.26285464e-01 3.10798913e-01 1.21595405e-01 -1.89012170e-01
-9.15440381e-01 -8.77009988e-01 1.83707774e-01 1.57669410e-01
-4.44538563e-01 2.64830232e-01 4.67043258e-02 -2.93671638e-01
-1.09254360e-01 -4.89417344e-01 -3.51548016e-01 -4.39031124e-01
1.27576683e-02 9.51633155e-01 8.64900053e-01 1.35687843e-01
5.13702214e-01 4.57935631e-01 -1.45479417e+00 5.25510073e-01
1.01549065e+00 9.80479866e-02 5.91180027e-01 1.11002088e+00
-2.13275284e-01 3.72725546e-01 -5.94658256e-01 -8.14644337e-01
3.71239066e-01 -5.38133323e-01 -1.02982998e+00 -2.82127053e-01
2.31644344e-02 4.70427334e-01 1.39895126e-01 5.74002206e-01
5.15865982e-01 -7.39866272e-02 2.01540321e-01 -9.58221138e-01
6.58653751e-02 5.76226115e-01 -2.38565460e-01 3.86394948e-01
2.77339548e-01 5.38572431e-01 1.15341306e+00 -6.61858737e-01
7.24589348e-01 1.25107086e+00 7.81571567e-01 4.63138819e-01
-1.19978225e+00 -2.38105565e-01 -5.99560618e-01 2.07277834e-01
-1.10575771e+00 -4.12028790e-01 1.73764974e-01 -5.58435977e-01
1.63701546e+00 2.15056553e-01 8.15714002e-01 2.94832081e-01
2.01963335e-01 9.07297969e-01 6.66037440e-01 -7.61265635e-01
3.59679580e-01 7.62093067e-02 3.02113257e-02 7.75404334e-01
4.49764431e-01 -2.97583014e-01 -8.83579254e-01 -2.65533328e-01
1.79366395e-01 -8.01646829e-01 -1.87838692e-02 -6.72190115e-02
-8.16793740e-01 7.95743644e-01 -2.51874477e-01 4.99682993e-01
-3.53352427e-01 -2.67864138e-01 4.38672125e-01 2.52712160e-01
3.10359985e-01 1.48979282e+00 -9.56277251e-01 -5.39354861e-01
-1.25675511e+00 4.84727412e-01 1.14141822e+00 8.08143854e-01
8.91439319e-01 -2.79649585e-01 -2.50623435e-01 5.11958420e-01
5.26586294e-01 -2.76967794e-01 7.38685787e-01 -1.01495719e+00
4.81095284e-01 9.83911932e-01 5.83369918e-02 -6.90095782e-01
-4.80535895e-01 -3.87646146e-02 1.20768152e-01 -1.08742461e-01
6.09896600e-01 -2.73877472e-01 -6.76828682e-01 1.05682635e+00
2.84911156e-01 -7.53650814e-02 3.83167202e-03 3.03167075e-01
4.14609730e-01 7.08511353e-01 1.91195115e-01 9.95907001e-03
1.19626629e+00 -1.01194882e+00 -2.87136525e-01 -4.32459056e-01
1.18013549e+00 -8.37828159e-01 1.24672019e+00 7.61987507e-01
-1.20323956e+00 -3.92344356e-01 -9.19175446e-01 6.46216497e-02
-3.16377223e-01 1.64852634e-01 7.32041836e-01 1.26143479e+00
-1.35781384e+00 1.10318577e+00 -1.12277806e+00 -4.60835576e-01
6.72916114e-01 4.90971118e-01 -1.29557848e-01 -1.40279219e-01
-6.25773072e-01 1.15672541e+00 9.36842203e-01 -1.70947254e-01
-8.05291831e-01 -8.04180562e-01 -6.99370384e-01 1.33633956e-01
1.36750698e-01 -4.32263702e-01 1.34844100e+00 -7.62447298e-01
-1.70993257e+00 8.67918491e-01 -2.04987407e-01 -3.77134770e-01
3.21373045e-01 -1.68138146e-01 -3.94262448e-02 1.97528768e-02
-5.70750348e-02 7.57953644e-01 4.70496356e-01 -5.14305592e-01
-1.04354775e+00 -2.83243597e-01 -2.40437537e-01 2.08942872e-02
-4.18008476e-01 6.40140116e-01 -7.32260644e-01 1.37151971e-01
-2.93270469e-01 -9.23743606e-01 -5.37169337e-01 -4.38152403e-01
-2.93911219e-01 -6.77485824e-01 5.29407322e-01 -9.03289616e-01
1.47545409e+00 -1.77862906e+00 3.66531909e-01 -8.21885839e-02
2.64562741e-02 6.06567800e-01 -3.83742452e-01 4.77560848e-01
2.18178153e-01 4.86015379e-01 -2.78897405e-01 -4.11140949e-01
-1.71698764e-01 3.30465108e-01 5.28366804e-01 -4.21521859e-03
8.02097023e-01 7.21008658e-01 -1.09866440e+00 -8.39014709e-01
-2.79297084e-01 -7.14147976e-03 -7.58779943e-01 3.73746574e-01
-5.33360183e-01 3.67081583e-01 -4.63989168e-01 8.18685234e-01
2.18283251e-01 -4.06675339e-01 3.00125062e-01 2.83843100e-01
-6.88889861e-01 3.35248470e-01 -8.91089797e-01 1.69063568e+00
-2.23994926e-01 9.10607696e-01 -2.28714511e-01 -5.02173603e-01
9.50605869e-01 4.20050234e-01 1.26419649e-01 -1.62328675e-01
-2.83640809e-02 4.39631760e-01 5.39097667e-01 -7.84856856e-01
4.75713611e-01 2.56812066e-01 8.56124982e-02 2.81944931e-01
2.77042061e-01 -3.19560468e-01 8.06772411e-01 5.33611290e-02
1.78467131e+00 6.11248910e-01 2.81611145e-01 -3.32617521e-01
2.70149499e-01 8.17828774e-01 8.37992370e-01 5.43691278e-01
-2.40814060e-01 3.95137489e-01 6.67494655e-01 -5.74693203e-01
-1.42402673e+00 -3.38041455e-01 3.42942595e-01 9.52893913e-01
-4.49012607e-01 -6.62738800e-01 -1.01458383e+00 -6.45749450e-01
-4.18758899e-01 8.13245893e-01 -2.15552300e-01 1.16277784e-01
-7.92493999e-01 -1.19340849e+00 8.14370811e-01 1.99505109e-02
7.73500726e-02 -1.28103161e+00 -1.26693392e+00 3.81875396e-01
1.19453639e-01 -7.62091577e-01 -9.74925682e-02 3.44088346e-01
-1.03452516e+00 -1.14103961e+00 -4.23274010e-01 -5.22551954e-01
4.04716402e-01 -2.09764123e-01 1.52353549e+00 1.99538633e-01
-5.51410913e-01 -1.92348566e-02 -6.62953317e-01 -5.68297803e-01
-8.45759690e-01 4.47134018e-01 -6.01931334e-01 -3.58179212e-01
4.77983177e-01 -1.23134844e-01 -2.26195350e-01 7.33643472e-02
-8.55863512e-01 1.71077609e-01 8.34775984e-01 7.33950138e-01
3.88650656e-01 3.15963656e-01 3.88827473e-01 -8.37743759e-01
7.71914065e-01 -5.37226796e-01 -8.27888727e-01 5.81827402e-01
-7.96065152e-01 2.66229600e-01 5.97343087e-01 -2.30699867e-01
-1.15207303e+00 5.37434876e-01 -1.69921204e-01 -1.72739029e-01
-3.83551582e-03 6.63192570e-01 -4.34090227e-01 -1.36390952e-02
7.28910089e-01 -1.50420025e-01 -1.02264531e-01 -8.00377488e-01
1.15628354e-01 4.43022221e-01 5.78277744e-02 -3.29480320e-01
5.87062597e-01 -9.01660547e-02 -5.91920689e-02 -9.52021599e-01
-2.91866004e-01 -4.96379524e-01 -1.03248036e+00 -1.35970697e-01
8.23464334e-01 -3.71211916e-01 -2.42783427e-01 2.35017657e-01
-1.38238835e+00 -7.89910138e-01 -3.92104350e-02 2.11271644e-01
-1.56669632e-01 2.92139739e-01 -3.04453582e-01 -8.44000757e-01
-5.72495937e-01 -1.27107525e+00 9.20543015e-01 6.77698255e-01
-6.05210662e-01 -8.81206274e-01 5.45076251e-01 2.12006167e-01
1.09766126e-01 8.43085721e-03 1.06779230e+00 -1.06875527e+00
-4.74046260e-01 -3.83523941e-01 2.54666120e-01 2.49627173e-01
-2.43741304e-01 6.74432933e-01 -8.37945700e-01 3.91235575e-04
-4.71860081e-01 -3.15677300e-02 5.48157454e-01 1.26500487e-01
6.72380984e-01 -1.00261666e-01 -5.71992517e-01 3.41238171e-01
1.42523050e+00 3.97862256e-01 8.06576014e-01 7.38510430e-01
5.63241363e-01 1.14310968e+00 8.17925990e-01 3.27353388e-01
9.02314112e-02 4.92846936e-01 5.72854914e-02 2.64305830e-01
2.47156575e-01 -6.69831932e-02 4.35228676e-01 6.07266247e-01
9.87230986e-02 -3.44851077e-01 -1.31961632e+00 7.75824189e-01
-1.77075696e+00 -7.49778330e-01 -5.02330780e-01 2.07030296e+00
9.13059056e-01 1.65087029e-01 2.20002979e-01 6.89656883e-02
3.74521464e-01 -3.64105493e-01 -4.43925411e-01 -7.06449568e-01
-3.87637466e-02 6.95631742e-01 4.48124528e-01 1.50769472e-01
-8.52560520e-01 1.21805644e+00 7.31328154e+00 8.10828984e-01
-6.26871228e-01 2.05363646e-01 7.30657756e-01 -6.25529960e-02
-1.89820051e-01 4.44374293e-01 -9.64162946e-01 5.93122244e-01
1.48421800e+00 -3.02062303e-01 3.57033610e-01 8.53237331e-01
4.80893612e-01 -3.50813568e-01 -9.89502370e-01 3.89414877e-01
-1.85252428e-01 -1.43572855e+00 -4.51164305e-01 -1.70000955e-01
5.96738219e-01 2.24605769e-01 -6.14001334e-01 2.19987080e-01
7.54283905e-01 -8.61083448e-01 6.00337327e-01 4.14302438e-01
-2.35238411e-02 -6.78112030e-01 6.35868430e-01 4.06059027e-01
-8.61246943e-01 -2.26166010e-01 -4.76770312e-01 1.90837570e-02
1.52371749e-01 5.36226988e-01 -1.53557539e+00 4.68090653e-01
1.08858943e+00 2.75752634e-01 -1.22489166e+00 1.59021437e+00
-2.46240631e-01 8.24038088e-01 -2.81998783e-01 -3.79645079e-01
7.49375597e-02 2.18107123e-02 4.96242076e-01 1.37900877e+00
4.34167624e-01 -4.83518541e-01 -1.16578527e-01 6.89264178e-01
3.24729830e-01 1.89826712e-01 -3.20647508e-01 -3.86213481e-01
7.38436520e-01 1.57863462e+00 -1.03182280e+00 -2.38178775e-01
-3.16782027e-01 9.25307393e-01 3.63523901e-01 -2.89663911e-01
-3.70561212e-01 -3.39740098e-01 3.68319958e-01 -2.22653337e-02
2.22819239e-01 -2.39612937e-01 -3.26408386e-01 -5.43820977e-01
-3.57215106e-01 -1.11714268e+00 7.13510871e-01 -7.75995612e-01
-8.96740973e-01 8.20158482e-01 -1.93270333e-02 -7.97076643e-01
-3.00413638e-01 -6.57267749e-01 -7.93849647e-01 9.94498014e-01
-1.15649545e+00 -9.58681047e-01 3.75968292e-02 -1.64593726e-01
8.00722599e-01 -2.14934051e-01 6.93862319e-01 3.75757404e-02
-9.46250439e-01 4.47611868e-01 -1.34509370e-01 -3.55200529e-01
6.48079515e-01 -1.40312147e+00 7.31196582e-01 1.02770936e+00
2.50711963e-02 4.91831154e-01 9.34086502e-01 -8.36859465e-01
-1.68236589e+00 -9.46806848e-01 1.19157147e+00 -8.71359348e-01
5.07518768e-01 4.21587117e-02 -9.14872885e-01 4.26643521e-01
3.10662895e-01 -6.83751643e-01 9.93295670e-01 1.61202885e-02
-2.70996410e-02 3.24710310e-01 -1.21753740e+00 5.89833975e-01
5.70481181e-01 8.17169100e-02 -3.75184894e-01 4.02172983e-01
6.10157669e-01 -1.56505555e-01 -1.01331413e+00 1.64567515e-01
4.92425770e-01 -1.03380823e+00 4.56529737e-01 -5.99547863e-01
4.59218085e-01 -3.73189926e-01 3.57965857e-01 -1.41504788e+00
-3.30490619e-01 -9.90555525e-01 5.47102764e-02 1.16113698e+00
8.11848938e-01 -3.08144659e-01 1.16823983e+00 8.77541661e-01
-5.42419016e-01 -7.84370661e-01 -8.50545824e-01 -8.65081370e-01
-8.34859610e-02 -2.06144050e-01 6.16942048e-01 6.51829600e-01
7.10160881e-02 3.15406501e-01 4.93735634e-02 1.47201791e-01
2.24250451e-01 -2.96162635e-01 7.78268158e-01 -1.48720181e+00
-7.40990281e-01 -7.59927750e-01 -2.29844958e-01 -4.71449137e-01
-2.60653406e-01 -7.01743722e-01 4.14982349e-01 -1.52179015e+00
1.94368660e-01 -4.50864315e-01 2.39823848e-01 6.60373926e-01
-4.40065920e-01 9.99430716e-02 7.69504532e-02 4.52276528e-01
-6.94627404e-01 -2.16852296e-02 9.74185646e-01 1.33441076e-01
-5.88953316e-01 -4.43426296e-02 -7.79397666e-01 6.91710234e-01
9.46031034e-01 -7.83281982e-01 -1.27895549e-01 -5.46557248e-01
6.69899642e-01 -3.27259809e-01 -4.33378100e-01 -1.01868653e+00
2.74986178e-01 -4.85426709e-02 2.29886115e-01 -4.63752806e-01
1.72068272e-02 -2.40656897e-01 5.40606558e-01 6.80427790e-01
-3.20517391e-01 5.93298078e-01 3.67738575e-01 2.02207312e-01
1.13066614e-01 -1.15898347e+00 6.01485968e-01 -4.21518385e-01
-8.72195303e-01 5.52158989e-02 -6.17063582e-01 -2.31851175e-01
9.51757193e-01 -3.07283103e-01 -1.03534229e-01 3.11125398e-01
-2.70099550e-01 3.28373462e-01 7.34246016e-01 2.08010286e-01
1.47695795e-01 -4.17436361e-01 -8.90122592e-01 -1.38221920e-01
4.86971214e-02 1.49275243e-01 -1.67218849e-01 6.81066632e-01
-1.15947914e+00 5.45731544e-01 -6.47949195e-03 -2.88746566e-01
-1.52572799e+00 2.29707867e-01 3.58820349e-01 -8.96044254e-01
-1.83971658e-01 1.23770726e+00 -5.44462383e-01 -4.59395170e-01
-3.38155985e-01 1.30839109e-01 -3.76380533e-01 6.60840869e-02
6.70472741e-01 6.55262530e-01 3.85543555e-01 -2.13602811e-01
-3.06640148e-01 2.75749546e-02 1.35155573e-01 -1.88127264e-01
1.47370338e+00 5.63634753e-01 -3.85238618e-01 5.13001718e-02
6.87763453e-01 -9.19800177e-02 -1.26230371e+00 3.04924041e-01
1.07039547e+00 -5.83855748e-01 -1.53202623e-01 -9.11216140e-01
-5.97015917e-01 7.20504761e-01 3.98764759e-01 3.44392776e-01
1.14760613e+00 7.42249563e-03 3.43565434e-01 3.34990710e-01
3.23455721e-01 -1.10234058e+00 1.67156741e-01 5.53522527e-01
3.54040533e-01 -8.39592993e-01 4.17506211e-02 -2.89107502e-01
-4.70505953e-01 1.12094045e+00 6.79164052e-01 2.21791580e-01
2.72132188e-01 3.81428003e-01 -2.57131666e-01 -2.53750801e-01
-1.00723743e+00 -1.73046812e-01 -1.45274058e-01 8.17379773e-01
5.83919168e-01 -7.87847489e-02 -4.64897513e-01 6.58956051e-01
-1.05916880e-01 -1.47979647e-01 4.35618848e-01 1.04589033e+00
-6.26773298e-01 -1.74440217e+00 -4.85154092e-01 5.89526355e-01
-6.20386302e-01 -1.83838099e-01 -6.82750165e-01 6.41282439e-01
2.59677142e-01 1.18544924e+00 -3.01770940e-02 -4.96001631e-01
3.02602381e-01 3.54848266e-01 5.44540823e-01 -9.17237759e-01
-1.11424410e+00 3.83135229e-02 4.73478347e-01 -3.82398397e-01
-9.49205160e-02 -9.60392118e-01 -1.05581248e+00 2.72583906e-02
-5.23846567e-01 4.12143111e-01 9.08438027e-01 9.42008734e-01
6.73114419e-01 4.81505215e-01 2.74490505e-01 -1.15635264e+00
-2.19971374e-01 -1.04108346e+00 -1.32977605e-01 -1.48889348e-01
-6.15628898e-01 -2.59205043e-01 1.33016864e-02 2.61753142e-01] | [8.466976165771484, 4.423255920410156] |
373f2640-27d0-4d76-a4ef-f8bd3e4394ce | practical-equivariances-via-relational | 2306.10915 | null | https://arxiv.org/abs/2306.10915v1 | https://arxiv.org/pdf/2306.10915v1.pdf | Practical Equivariances via Relational Conditional Neural Processes | Conditional Neural Processes (CNPs) are a class of metalearning models popular for combining the runtime efficiency of amortized inference with reliable uncertainty quantification. Many relevant machine learning tasks, such as spatio-temporal modeling, Bayesian Optimization and continuous control, contain equivariances -- for example to translation -- which the model can exploit for maximal performance. However, prior attempts to include equivariances in CNPs do not scale effectively beyond two input dimensions. In this work, we propose Relational Conditional Neural Processes (RCNPs), an effective approach to incorporate equivariances into any neural process model. Our proposed method extends the applicability and impact of equivariant neural processes to higher dimensions. We empirically demonstrate the competitive performance of RCNPs on a large array of tasks naturally containing equivariances. | ['Luigi Acerbi', 'Samuel Kaski', 'Kevin Sebastian Luck', 'Grégoire Clarté', 'ST John', 'Ulpu Remes', 'Manuel Haussmann', 'Daolang Huang'] | 2023-06-19 | null | null | null | null | ['bayesian-optimization', 'continuous-control'] | ['methodology', 'playing-games'] | [ 4.67286631e-02 2.19958816e-02 -6.79305848e-03 -3.21200341e-01
-9.33772087e-01 -5.34388781e-01 1.17768550e+00 1.82298217e-02
-4.54927474e-01 8.13352406e-01 1.66972548e-01 -5.00442922e-01
-4.29049104e-01 -8.22155476e-01 -9.42516804e-01 -6.66031480e-01
-3.69342327e-01 7.14478672e-01 2.96699345e-01 2.31480092e-01
1.10064253e-01 5.58152854e-01 -1.39352119e+00 3.06941792e-02
6.73312783e-01 7.74957001e-01 -6.64839596e-02 7.13178933e-01
-1.42228737e-01 6.91481709e-01 -7.34069571e-02 -5.58099508e-01
1.92329600e-01 2.44759977e-01 -3.69276732e-01 -4.82829124e-01
2.29474232e-01 -2.49918073e-01 -2.24160805e-01 9.94952679e-01
3.45970094e-01 2.70858943e-01 9.95905697e-01 -1.20021188e+00
-4.89055604e-01 1.00934577e+00 -6.23902202e-01 2.67341763e-01
-2.38792539e-01 1.18291840e-01 1.14910340e+00 -1.27336490e+00
3.91704619e-01 1.92359877e+00 6.86771929e-01 1.91904113e-01
-1.55825782e+00 -7.28990734e-01 4.28253174e-01 -2.82662660e-01
-1.39719522e+00 -3.87042403e-01 3.27022225e-01 -5.33047795e-01
9.08024669e-01 1.09220505e-01 2.37529844e-01 1.39760745e+00
5.93024850e-01 9.19157207e-01 1.16692984e+00 5.23902476e-02
7.30660975e-01 -2.69223362e-01 1.70933753e-01 3.98583829e-01
4.68425870e-01 1.19948395e-01 -7.75079489e-01 -3.90669793e-01
1.10175765e+00 2.39683285e-01 2.07519278e-01 -2.34395370e-01
-1.29824543e+00 7.02727437e-01 -6.32495852e-03 -2.35639393e-01
-3.17092657e-01 8.76254439e-01 2.00156435e-01 1.70970093e-02
6.51154041e-01 4.51523006e-01 -5.07369936e-01 -3.69676113e-01
-7.08590567e-01 3.39407146e-01 8.72552454e-01 1.22648001e+00
6.53191507e-01 6.34211898e-02 -5.83539963e-01 6.29033744e-01
5.28449178e-01 8.01683128e-01 -3.74005549e-02 -1.32243586e+00
5.97222924e-01 1.58448711e-01 2.18364909e-01 -7.39707351e-01
-3.60838830e-01 -3.50578487e-01 -9.35436189e-01 7.99916685e-02
2.95683503e-01 -3.00309479e-01 -9.76501465e-01 1.94797337e+00
3.26657683e-01 3.11642915e-01 8.59031454e-02 4.48603421e-01
1.50664309e-02 7.24856555e-01 3.12487781e-01 -8.88454616e-02
1.13866532e+00 -4.67830062e-01 -6.98030055e-01 -1.13157287e-01
1.34239525e-01 -5.59660375e-01 1.01807213e+00 4.23384964e-01
-1.29720616e+00 -2.49847218e-01 -7.70005524e-01 -9.98417381e-03
-9.77698341e-02 -2.06393152e-01 1.00340390e+00 4.39093649e-01
-1.12464702e+00 9.44725156e-01 -1.50021172e+00 -4.83518578e-02
4.61791039e-01 3.08213890e-01 -1.53824657e-01 3.08568757e-02
-1.01641691e+00 7.46786654e-01 4.72954780e-01 1.90889254e-01
-1.24124372e+00 -7.63205588e-01 -6.26456976e-01 1.94844127e-01
7.64312208e-01 -6.43594086e-01 1.43854117e+00 -3.43799531e-01
-1.52334559e+00 2.58252591e-01 -2.48288572e-01 -5.90068042e-01
8.08525622e-01 -4.69853371e-01 2.24926192e-02 -8.26273263e-02
-1.18010351e-02 6.26891494e-01 9.81595516e-01 -1.11766100e+00
-6.23294711e-01 -5.56838214e-01 1.12828247e-01 2.41660222e-01
5.58255194e-03 1.17096692e-01 -7.45147586e-01 -7.03742385e-01
4.78211582e-01 -1.13533461e+00 -4.97087270e-01 1.11989252e-01
-5.78118622e-01 -4.26264405e-01 5.60844481e-01 -3.58870268e-01
7.27996409e-01 -1.88602269e+00 3.87460262e-01 3.23329508e-01
2.64706016e-01 -4.12283957e-01 4.68508340e-02 2.80301362e-01
4.31250811e-01 3.52649629e-01 -4.31490362e-01 -7.66323090e-01
5.11264265e-01 6.64073467e-01 -5.32737851e-01 4.58042592e-01
6.25107586e-01 9.11913216e-01 -7.90434778e-01 -4.76286411e-01
-4.65888791e-02 3.64273667e-01 -4.15623546e-01 5.18353749e-03
-6.23078883e-01 2.80625105e-01 -3.52148116e-01 7.05933690e-01
5.32227576e-01 -1.89301357e-01 -1.02417655e-01 3.64539355e-01
-9.38056316e-03 9.99389812e-02 -1.42794085e+00 1.44597983e+00
-3.86685163e-01 3.43880266e-01 4.86795232e-02 -2.50931352e-01
5.40136874e-01 1.42131522e-01 3.61174107e-01 1.15604745e-02
-1.72389701e-01 -9.99060422e-02 -6.68908060e-02 6.15111627e-02
6.73757434e-01 -1.90434024e-01 -2.94995099e-01 4.46422547e-01
1.42450720e-01 -2.28593111e-01 1.78856894e-01 2.00132906e-01
1.11454117e+00 5.17236412e-01 6.77343234e-02 -6.05217516e-01
4.37580943e-02 -4.02313232e-01 8.39817584e-01 1.19302177e+00
-1.29885823e-01 4.42234397e-01 7.88019359e-01 -1.06156200e-01
-1.07327867e+00 -1.77394354e+00 -2.66016006e-01 1.36399281e+00
-1.57598019e-01 -3.50420505e-01 -4.74345982e-01 -3.97455812e-01
3.92068829e-03 8.28511596e-01 -5.60228229e-01 1.57307908e-01
-4.00369883e-01 -1.01876712e+00 8.85367990e-01 9.75056887e-01
3.37143570e-01 -6.23881221e-01 -2.43898615e-01 3.05747032e-01
1.27956226e-01 -1.24982083e+00 -9.23320651e-02 3.64969373e-01
-1.14778256e+00 -5.65544724e-01 -4.81581300e-01 -1.82208549e-02
2.30165616e-01 -1.40667632e-01 9.55098212e-01 -8.62730205e-01
8.34779218e-02 4.99940515e-01 3.21199030e-01 -9.29867744e-01
-3.34159851e-01 -6.57500178e-02 4.29972976e-01 -1.57686025e-01
1.31566823e-01 -7.86657870e-01 -2.22849593e-01 1.69587389e-01
-9.09679711e-01 4.74059358e-02 6.91006601e-01 5.37192702e-01
9.99948084e-01 8.52027759e-02 1.87225744e-01 -7.72100985e-01
7.36739695e-01 -5.56250453e-01 -8.85304213e-01 1.07101984e-01
-3.75573546e-01 6.71467364e-01 2.31987715e-01 -5.09604573e-01
-1.44521296e+00 -2.12801442e-01 1.91889077e-01 -6.90932453e-01
-7.13150278e-02 6.68028951e-01 -1.76272705e-01 5.11891603e-01
4.85144019e-01 -9.55417082e-02 -2.58328229e-01 -2.89535373e-01
6.59372211e-01 -1.25943869e-01 5.04503787e-01 -1.20731711e+00
6.44104779e-01 8.64665627e-01 7.03155339e-01 -5.55071235e-01
-5.91511607e-01 -3.49403769e-01 -6.16400123e-01 9.86407772e-02
9.84616935e-01 -1.11624312e+00 -6.11384153e-01 3.54389697e-01
-1.47850156e+00 -3.93727362e-01 -2.51097560e-01 7.29750812e-01
-6.60923541e-01 1.58734336e-01 -8.71638715e-01 -1.30739129e+00
-1.81131780e-01 -1.12452710e+00 1.26778758e+00 3.11586587e-03
5.24982773e-02 -1.11586881e+00 1.81317091e-01 -2.45168254e-01
2.97912478e-01 2.26008475e-01 8.69397104e-01 -8.31406116e-01
-8.77935112e-01 4.54586521e-02 -2.20656589e-01 1.98757663e-01
-3.47104847e-01 2.51558632e-01 -1.26787746e+00 1.21641517e-01
-3.16464971e-03 -8.34068190e-03 1.00276375e+00 5.08406818e-01
1.34441888e+00 -2.92217642e-01 -3.32718909e-01 3.86839718e-01
1.23803890e+00 -8.57758373e-02 4.67305243e-01 -8.67549106e-02
8.94075930e-01 7.84763455e-01 2.81611979e-01 5.08685112e-01
3.46585155e-01 2.45607555e-01 6.07891798e-01 3.80875915e-01
3.84598166e-01 -3.68409336e-01 4.49655563e-01 8.42979074e-01
-5.86248159e-01 -2.48794228e-01 -1.14648318e+00 3.34684819e-01
-2.21834731e+00 -9.77063179e-01 -2.45364785e-01 2.27845263e+00
6.71141922e-01 3.74805033e-01 -3.57114941e-01 -2.60317445e-01
7.55925655e-01 2.96500549e-02 -6.43405437e-01 -2.23880798e-01
-1.61320731e-01 2.05912545e-01 7.27859139e-01 4.25992966e-01
-1.21069443e+00 8.80833387e-01 6.53953552e+00 8.83753777e-01
-5.22409558e-01 3.02991778e-01 4.77444202e-01 -2.27581739e-01
-3.22088212e-01 -5.40500842e-02 -1.20296562e+00 1.63984552e-01
1.25032473e+00 1.79296345e-01 3.06386292e-01 1.06765616e+00
3.32461894e-01 2.66931485e-02 -1.55543804e+00 7.35216737e-01
-3.09232950e-01 -1.06283760e+00 1.26625791e-01 2.48773545e-01
9.94212270e-01 3.39705616e-01 2.68428862e-01 5.48716128e-01
1.14716291e+00 -1.07160413e+00 8.75325441e-01 8.22856665e-01
4.02539283e-01 -8.33584428e-01 6.41265512e-01 2.07622126e-01
-9.75601017e-01 -7.02875182e-02 -5.40231526e-01 2.08788574e-01
3.63774896e-01 7.34224498e-01 -8.48159671e-01 4.51613516e-01
7.43023932e-01 3.35663825e-01 -1.74192533e-01 7.53222227e-01
-2.60435820e-01 7.41928935e-01 -6.36573195e-01 2.63647996e-02
4.61371988e-01 -2.73533851e-01 8.40948939e-01 1.40123260e+00
5.61302125e-01 -1.94696262e-01 -5.90566434e-02 1.16802812e+00
-4.93399091e-02 -1.71001598e-01 -6.49056733e-01 -2.41017833e-01
4.60967511e-01 9.35804307e-01 -6.56151474e-01 -2.91212589e-01
-3.34952742e-01 7.70422995e-01 2.51224965e-01 5.46862602e-01
-8.98300052e-01 1.42141566e-01 9.12639856e-01 2.45319530e-02
3.03076893e-01 -6.86332583e-01 -4.87558126e-01 -1.10656118e+00
-6.34929910e-02 -5.41768789e-01 5.00795424e-01 -8.29191804e-01
-1.58718288e+00 1.61569923e-01 5.72126567e-01 -7.59425163e-01
-3.76726300e-01 -8.46746862e-01 -5.81209660e-01 1.02620602e+00
-1.28243256e+00 -1.36717558e+00 1.86939418e-01 5.18312991e-01
5.67769766e-01 -1.43542914e-02 5.40665030e-01 -2.01052174e-01
-5.57765484e-01 2.11513117e-02 2.94348806e-01 -2.67487198e-01
6.82330847e-01 -1.73584437e+00 5.37182689e-01 1.13102841e+00
8.22003558e-02 9.36908603e-01 8.40638995e-01 -8.71855259e-01
-1.62365150e+00 -1.36021936e+00 3.01596701e-01 -8.45438242e-01
1.10629988e+00 -6.46222949e-01 -8.96583200e-01 9.57593083e-01
-1.70993209e-01 -1.47425935e-01 4.76435930e-01 4.90981579e-01
-5.49332798e-01 3.17412205e-02 -7.69923687e-01 1.02861106e+00
1.06916893e+00 -5.33356369e-01 -5.12601674e-01 2.38494158e-01
1.05303419e+00 -1.70875981e-01 -1.13866866e+00 6.01686060e-01
4.55258965e-01 -5.72010040e-01 1.03737700e+00 -6.80336714e-01
5.30852973e-01 -2.43276179e-01 -5.34881115e-01 -1.08685601e+00
-1.02368563e-01 -7.70201862e-01 -6.52322829e-01 1.08254814e+00
4.45345104e-01 -6.38767719e-01 4.81032282e-01 1.02234340e+00
-1.62033126e-01 -4.57916319e-01 -1.02998269e+00 -8.51411700e-01
4.42890614e-01 -1.15170264e+00 6.61335945e-01 7.11981535e-01
-4.30501997e-01 1.83658779e-01 -2.95210809e-01 4.97753173e-01
9.21497583e-01 -3.56005520e-01 5.55933774e-01 -1.68629706e+00
-7.16171205e-01 -5.96323371e-01 -1.13568962e-01 -8.34341109e-01
1.17275335e-01 -6.65012419e-01 3.08553159e-01 -1.37554932e+00
2.37794578e-01 -5.36805630e-01 -2.80172259e-01 1.44817993e-01
-3.25939149e-01 -2.97395051e-01 7.57345259e-02 4.54989702e-01
-7.81024635e-01 6.83407247e-01 1.04179692e+00 7.60209793e-03
-1.59639463e-01 1.42250031e-01 -4.47991878e-01 1.06054819e+00
8.22171032e-01 -5.72727561e-01 -6.77554250e-01 -4.63535726e-01
5.95846951e-01 9.26681161e-02 3.69224846e-01 -7.82425880e-01
2.85556912e-01 -4.74285334e-01 3.74278724e-01 -6.00973189e-01
7.00774193e-01 -5.45328736e-01 1.70110032e-01 1.58874661e-01
-4.41061407e-01 4.31871504e-01 3.35678786e-01 1.30120170e+00
1.64071798e-01 -1.57043710e-01 3.22623074e-01 -2.45286837e-01
-4.50614274e-01 6.27031147e-01 -5.85352540e-01 -2.81411987e-02
7.75958121e-01 3.06377232e-01 -2.42178395e-01 -3.62118840e-01
-7.97368109e-01 1.22770436e-01 8.97858217e-02 3.12867582e-01
4.05094773e-01 -1.08698678e+00 -6.41511321e-01 -2.63672799e-01
-5.81138432e-02 4.68660384e-01 -9.36273579e-03 8.31854045e-01
-3.18738014e-01 4.10787702e-01 1.15374096e-01 -8.67257893e-01
-8.45964313e-01 5.04082799e-01 2.38896906e-01 -5.41954279e-01
-4.60389733e-01 5.20925164e-01 5.22377729e-01 -6.41947806e-01
2.91974217e-01 -7.11003721e-01 1.89404383e-01 -2.29796767e-01
4.87126887e-01 6.03105128e-01 -8.51551890e-02 -1.36242747e-01
2.16151215e-02 1.12834899e-02 1.50848389e-01 -9.55205321e-01
1.13658381e+00 -1.29787579e-01 -3.32540900e-01 1.02482045e+00
5.88937819e-01 -2.36231953e-01 -1.77319455e+00 -3.58866930e-01
3.03616554e-01 -1.02901399e-01 4.76383120e-02 -3.85563940e-01
-3.91268313e-01 1.24785507e+00 3.46166819e-01 -1.12405056e-02
4.66488570e-01 -1.14290714e-02 2.75650173e-02 9.00672317e-01
4.72048610e-01 -1.18841624e+00 1.27489626e-01 8.12781036e-01
8.56112301e-01 -1.04173326e+00 -2.33078256e-01 -2.31071323e-01
-6.96237922e-01 8.63235950e-01 5.99884391e-01 1.11491727e-02
9.08626199e-01 5.46059728e-01 -6.35522962e-01 -3.16517390e-02
-1.01385343e+00 -2.77311411e-02 1.15754738e-01 4.64988381e-01
-3.36417146e-02 2.75355548e-01 8.53404030e-02 6.21254086e-01
5.81661537e-02 -4.69907165e-01 4.82427180e-01 1.02461851e+00
-3.46403092e-01 -6.40820682e-01 -5.21045446e-01 4.88541275e-01
-7.10721135e-01 -3.65637779e-01 3.56220230e-02 6.14907324e-01
-2.10832387e-01 7.51152098e-01 4.03359652e-01 -6.18150085e-02
3.82000394e-02 1.43702790e-01 3.28427285e-01 -4.82746869e-01
-9.51904580e-02 2.82620400e-01 1.50552914e-01 -5.30484259e-01
-3.11066300e-01 -9.07450438e-01 -1.08685601e+00 -4.74304676e-01
1.94006152e-02 -3.39029402e-01 9.13128197e-01 1.09964502e+00
2.38606989e-01 5.08349419e-01 -7.53951147e-02 -7.49397337e-01
-8.43351364e-01 -1.11674237e+00 -5.72978973e-01 -6.37469664e-02
-4.48152237e-02 -8.11586678e-01 -2.43420288e-01 -4.43596318e-02] | [7.034091949462891, 3.7741708755493164] |
a6944b1c-d4f8-437b-a417-3b9c8a3fa283 | a-robust-and-efficient-method-for-improving | 1407.6705 | null | http://arxiv.org/abs/1407.6705v2 | http://arxiv.org/pdf/1407.6705v2.pdf | A Robust and Efficient Method for Improving Accuracy of License Plate Characters Recognition | License Plate Recognition (LPR) plays an important role on the traffic
monitoring and parking management. A robust and efficient method for enhancing
accuracy of license plate characters recognition based on K Nearest Neighbours
(K-NN) classifier is presented in this paper. The system first prepares a
contour form of the extracted character, then the angle and distance feature
information about the character is extracted and finally K-NN classifier is
used to character recognition. Angle and distance features of a character have
been computed based on distribution of points on the bitmap image of character.
In K-NN method, the Euclidean distance between testing point and reference
points is calculated in order to find the k-nearest neighbours. We evaluated
our method on the available dataset that contain 1200 sample. Using 70% samples
for training, we tested our method on whole samples and obtained 99% correct
recognition rate.Further, we achieved average 99.41% accuracy using
three/strategy validation technique on 1200 dataset. | ['Reza Azad', 'Hamed Amiri', 'Hamid Reza Shayegh'] | 2014-07-24 | null | null | null | null | ['license-plate-recognition'] | ['computer-vision'] | [ 9.32447240e-02 -7.74812520e-01 -2.09259585e-01 -4.17239010e-01
-5.77392340e-01 -5.80539465e-01 4.32977378e-01 -2.75474757e-01
-6.39614463e-01 6.36589944e-01 -1.95630863e-01 -3.36097360e-01
-6.81650788e-02 -9.74263072e-01 -2.46059909e-01 -6.00760996e-01
2.50914931e-01 4.35402036e-01 7.16471136e-01 -1.40589714e-01
1.20497000e+00 1.24890268e+00 -1.39417958e+00 1.26865268e-01
6.49575949e-01 9.17804062e-01 3.63470465e-02 9.36830640e-01
-1.50953069e-01 4.86811906e-01 -5.62822700e-01 -3.66822124e-01
6.35766447e-01 4.84709907e-03 -4.42538649e-01 3.30785513e-01
8.43824372e-02 -1.81617782e-01 -5.00383317e-01 9.76599336e-01
2.51157105e-01 2.91420221e-01 1.27877355e+00 -1.06198716e+00
-6.32239401e-01 -9.76088718e-02 -5.40144861e-01 1.58361942e-01
7.85936564e-02 -1.22898981e-01 3.73479486e-01 -1.23803020e+00
3.95357788e-01 3.48449916e-01 9.41408634e-01 3.82206976e-01
-6.28114402e-01 -6.51348591e-01 -9.67335820e-01 6.65985167e-01
-1.95899308e+00 -3.72151583e-01 8.63178968e-01 -4.34485465e-01
7.22950101e-01 4.85284507e-01 6.27023816e-01 -1.87590662e-02
2.45859489e-01 6.56417310e-01 1.19384050e+00 -8.09023976e-01
-2.39157099e-02 5.99029720e-01 5.69165230e-01 3.43768001e-01
1.31940141e-01 -6.68307319e-02 1.59000233e-01 1.70529976e-01
8.39296639e-01 1.38272762e-01 2.47246832e-01 1.97590768e-01
-7.76825726e-01 6.81323111e-01 -5.94996214e-02 3.28137457e-01
-2.76399374e-01 -3.88649106e-01 1.08480565e-01 -1.58382490e-01
-4.33963567e-01 -1.90549884e-02 6.99891429e-03 -4.89976466e-01
-1.14608848e+00 -1.17706954e-01 5.36445320e-01 1.06384742e+00
8.06870401e-01 4.62594554e-02 -3.97191523e-03 1.10539818e+00
3.38703066e-01 7.95291901e-01 6.11390591e-01 -4.24443573e-01
5.46562552e-01 6.28307283e-01 1.22301310e-01 -1.30596268e+00
3.57212499e-02 1.49183989e-01 -5.68953335e-01 4.93748218e-01
2.55969495e-01 2.70727538e-02 -1.00517619e+00 2.45920688e-01
-2.07423761e-01 1.75774708e-01 3.08813274e-01 6.59658968e-01
4.75423694e-01 1.08551860e+00 -3.57831240e-01 3.33532579e-02
1.32310545e+00 -7.88432360e-01 -3.83365899e-01 1.11515932e-01
2.27823362e-01 -1.19514084e+00 6.16960287e-01 3.51193517e-01
-6.25069797e-01 -6.06592596e-01 -1.14609289e+00 3.61220509e-01
-5.62783360e-01 8.92452955e-01 2.09295861e-02 8.36929560e-01
-6.69173479e-01 4.39241916e-01 -2.07902312e-01 -1.95885569e-01
1.73363745e-01 5.76986611e-01 -6.23867989e-01 -9.21896324e-02
-8.48356962e-01 1.03257966e+00 2.38755524e-01 2.70554215e-01
1.24046147e-01 8.62130076e-02 -4.54013258e-01 -3.25614214e-01
-2.97501802e-01 5.47474325e-01 5.62185466e-01 -6.33953512e-01
-1.57321703e+00 7.35914767e-01 -6.70432076e-02 -1.63765699e-01
3.14658999e-01 5.39662063e-01 -1.18102968e+00 1.21873766e-01
-1.05705537e-01 1.70049354e-01 7.13813484e-01 -6.91183269e-01
-8.15240264e-01 -3.02751482e-01 -9.08601820e-01 2.73708791e-01
1.67765766e-02 2.19402924e-01 -6.34524465e-01 -2.79629081e-01
4.12306398e-01 -8.24260473e-01 1.02446005e-01 -3.17641556e-01
-2.68224835e-01 -2.72370249e-01 1.03697920e+00 -9.40867543e-01
1.12156594e+00 -2.39129281e+00 -1.06669259e+00 9.67393935e-01
-3.12035531e-01 9.47413862e-01 4.32628661e-01 3.57127100e-01
-5.81906289e-02 -2.30813965e-01 -1.20647743e-01 4.18496996e-01
-1.01806253e-01 -1.16972879e-01 -8.60058814e-02 6.73786223e-01
2.19011769e-01 5.45291245e-01 8.78103673e-02 -7.50974059e-01
5.37299931e-01 4.16674554e-01 3.39920148e-02 -1.55296087e-01
6.56627417e-01 -4.51659441e-01 -4.75632042e-01 8.40215027e-01
1.26202297e+00 4.23732370e-01 -3.91050458e-01 -2.52178222e-01
-5.17541349e-01 -4.69807982e-01 -1.54631782e+00 2.96019644e-01
-1.52975721e-02 1.11525059e+00 -5.36662936e-01 -8.06361377e-01
1.93591988e+00 9.24758017e-02 2.31154710e-02 -9.67621624e-01
4.12548780e-02 4.75376815e-01 -1.90154403e-01 -6.50520980e-01
9.24784780e-01 2.43851751e-01 8.20222683e-03 -7.96290785e-02
-5.74959576e-01 5.42421602e-02 3.33178043e-02 -2.57238805e-01
6.32441163e-01 -2.67353743e-01 2.68934727e-01 -1.02389716e-01
1.15773368e+00 4.84897196e-01 2.03688204e-01 3.92723918e-01
-4.71089214e-01 4.51591134e-01 3.53307985e-02 -5.42837262e-01
-1.42436004e+00 -7.04320908e-01 -6.40205979e-01 2.19749168e-01
2.24845827e-01 3.41293722e-01 -4.75265294e-01 -1.54237121e-01
3.28698568e-02 6.35170043e-01 -1.44329324e-01 3.20454895e-01
-7.93595135e-01 -4.75732386e-01 8.48561645e-01 4.68726128e-01
8.22201133e-01 -9.87482071e-01 -1.65693238e-01 1.31198838e-01
4.40402180e-01 -9.87843812e-01 -3.96864891e-01 -4.12740290e-01
-7.69810438e-01 -1.02888882e+00 -8.09348643e-01 -1.27899694e+00
8.74119282e-01 2.21654862e-01 2.86090046e-01 -1.11754656e-01
-3.95484358e-01 1.63844991e-02 -1.46110401e-01 -8.13497752e-02
-3.92164409e-01 -4.48589414e-01 -2.10263938e-01 2.54765898e-01
9.48404253e-01 -1.72211707e-01 -3.91796440e-01 8.43222499e-01
-3.99115950e-01 -4.15889442e-01 1.07951939e+00 1.06135666e-01
5.10729671e-01 4.64931279e-01 2.89418966e-01 -4.06293839e-01
7.29273260e-01 -2.68467963e-01 -1.00118530e+00 2.07322791e-01
-5.06307602e-01 -1.87371507e-01 7.06293702e-01 -1.33281156e-01
-5.52017510e-01 2.60737300e-01 -2.92125762e-01 -1.43647462e-01
-4.01232153e-01 5.70370406e-02 -2.74647512e-02 -5.08178234e-01
4.85634297e-01 1.15786958e+00 2.52121121e-01 -1.85880795e-01
-1.87142506e-01 1.66709530e+00 8.30643833e-01 -2.34220400e-01
8.31341207e-01 1.73551366e-01 1.83830425e-01 -1.36879706e+00
6.48417354e-01 -9.89222229e-01 -6.47131860e-01 -6.27768338e-01
7.42344379e-01 -4.48377103e-01 -9.91134703e-01 9.76198196e-01
-8.12346935e-01 4.21166927e-01 2.69845486e-01 8.16507936e-01
-2.66411036e-01 6.49591386e-01 -2.35736057e-01 -1.22094977e+00
-2.14555755e-01 -1.04701507e+00 3.51962626e-01 5.42134523e-01
2.86958218e-01 -7.34944880e-01 1.25471637e-01 4.54480112e-01
5.80982566e-01 -1.12152137e-01 5.30575514e-01 -1.12260020e+00
-5.80093801e-01 -1.22038913e+00 -4.89762485e-01 4.85934049e-01
-3.20150703e-03 5.01443207e-01 -5.68194687e-01 4.30034190e-01
-3.40722233e-01 1.63875625e-01 5.72555363e-01 1.67157799e-01
8.49285841e-01 -2.55963236e-01 -2.69518554e-01 3.22868228e-01
1.70048881e+00 5.87042630e-01 1.46610272e+00 6.72368765e-01
4.58289623e-01 1.35757178e-01 7.71935523e-01 2.16550440e-01
2.49807984e-01 6.35874748e-01 -1.94818467e-01 2.98870862e-01
1.34184808e-01 -1.10612012e-01 3.12080234e-01 7.86300242e-01
-3.58449817e-01 1.14644028e-01 -1.10569298e+00 2.41074756e-01
-1.16837907e+00 -1.23557889e+00 -6.83266521e-01 2.38046670e+00
5.58437824e-01 2.25459278e-01 3.67203563e-01 7.06644773e-01
1.32277846e+00 -3.02127481e-01 -1.58602923e-01 -7.81039596e-01
-8.14045295e-02 -5.36771230e-02 1.10950315e+00 6.01726234e-01
-9.50008750e-01 7.80231953e-01 5.96969986e+00 1.23084426e+00
-1.48640871e+00 -7.24836409e-01 4.39550042e-01 7.57498205e-01
2.15401292e-01 -2.09237486e-01 -1.37707460e+00 7.81987309e-01
8.72014821e-01 -8.92529041e-02 3.17770034e-01 1.01591086e+00
9.16786566e-02 -3.75549525e-01 -3.18696767e-01 1.30821526e+00
2.76741296e-01 -1.49666810e+00 -1.08123064e-01 3.90748441e-01
4.45705235e-01 -1.09511271e-01 -2.40091365e-02 2.21558392e-01
-2.59308457e-01 -9.51433718e-01 3.56372267e-01 9.77069020e-01
1.03147066e+00 -1.11805320e+00 9.26259339e-01 4.54474777e-01
-1.13330209e+00 1.55696362e-01 -7.48723507e-01 1.74680054e-01
-2.29070202e-01 1.95479438e-01 -1.39592767e+00 1.09898485e-01
5.22852540e-02 4.07733738e-01 -6.30501449e-01 1.40438592e+00
4.43933606e-01 5.69184899e-01 -4.34538186e-01 -5.94645262e-01
3.78452927e-01 -6.79901898e-01 3.17386597e-01 1.48248231e+00
4.67076212e-01 1.85956731e-01 -5.11691749e-01 4.86074865e-01
2.68096894e-01 4.91794556e-01 -4.31735843e-01 1.10750884e-01
7.95630038e-01 1.17736721e+00 -8.59279335e-01 -2.84857273e-01
-3.27832133e-01 7.14255154e-01 -1.30982220e-01 -5.33190072e-02
-7.53145516e-01 -1.07505965e+00 -1.10615991e-01 2.41557971e-01
6.08355045e-01 -3.99180353e-01 -2.59115517e-01 -4.74339485e-01
1.53616607e-01 -5.33914924e-01 1.55714184e-01 -7.58978248e-01
-9.00847435e-01 4.66599882e-01 -3.33932787e-01 -1.70095277e+00
-7.52040073e-02 -1.11434639e+00 -1.00644875e+00 1.20594239e+00
-1.30407739e+00 -8.39464903e-01 -9.47650075e-02 7.92941213e-01
3.74578893e-01 -7.78262615e-01 6.48301661e-01 2.45139360e-01
-5.82566917e-01 6.89597785e-01 7.04498470e-01 8.41398895e-01
3.24419707e-01 -7.85493433e-01 1.86510429e-01 8.81447375e-01
-2.24316955e-01 4.00428742e-01 2.18881637e-01 -7.93879390e-01
-1.27982068e+00 -8.75652075e-01 1.00718534e+00 -1.94139689e-01
3.86936575e-01 7.45505020e-02 -6.44119918e-01 4.56305206e-01
-3.42620760e-01 2.17599228e-01 7.51896620e-01 -7.51729965e-01
2.99725700e-02 -2.00969920e-01 -1.57357824e+00 3.38727564e-01
-4.13781777e-02 -4.41687852e-01 -5.26522756e-01 -9.65682194e-02
-3.55208725e-01 -1.42633110e-01 -8.65808427e-01 -6.44587576e-02
8.52700055e-01 -9.39415991e-01 9.12380099e-01 -5.08984132e-03
6.14412427e-02 -8.16039860e-01 -4.76657927e-01 -4.18833941e-01
-3.87460768e-01 -5.72959408e-02 6.94814682e-01 1.25709844e+00
7.65938878e-01 -7.06188142e-01 9.84278560e-01 9.03740287e-01
-7.10002407e-02 -6.13358021e-01 -8.39827120e-01 -6.52681470e-01
-1.10114932e-01 -4.74779427e-01 4.66861874e-01 7.21224427e-01
-1.12449564e-01 -2.20829293e-01 -4.03771818e-01 4.14095461e-01
6.23001695e-01 -2.14897111e-01 7.23571718e-01 -9.55918372e-01
2.09380239e-01 -4.07907724e-01 -1.36139071e+00 -7.27543592e-01
-2.02062652e-01 -6.88410759e-01 -1.33928567e-01 -1.30501246e+00
5.70684187e-02 -6.46625400e-01 7.11161271e-02 2.40614060e-02
3.44767511e-01 7.21792758e-01 6.70100525e-02 7.24415541e-01
-3.16325128e-01 -2.74476614e-02 8.92499864e-01 7.53723234e-02
-1.44640774e-01 5.94431996e-01 -4.43056338e-02 4.60666478e-01
9.91266727e-01 -1.95733696e-01 1.14347287e-01 2.01591134e-01
-2.27601632e-01 7.41078481e-02 1.80041656e-01 -1.30512846e+00
6.74706995e-01 -1.75337046e-01 9.41987514e-01 -1.30226564e+00
3.69711369e-01 -1.01817942e+00 1.70662120e-01 3.29771787e-01
1.41227618e-01 3.96187976e-02 9.12580043e-02 1.76667571e-01
-1.98093712e-01 -8.52428436e-01 8.91596973e-01 1.36174321e-01
-1.34127414e+00 -5.37090711e-02 -5.89434683e-01 -4.74889249e-01
1.33217955e+00 -1.35552287e+00 -2.38630101e-01 3.58985877e-03
-3.37601662e-01 -4.10829484e-01 5.45240760e-01 -6.08701967e-02
9.16045904e-01 -1.58664751e+00 -6.94511294e-01 5.89976728e-01
2.67908752e-01 -6.40292406e-01 6.32591322e-02 5.77243805e-01
-1.44913936e+00 7.50388682e-01 -6.35513723e-01 -4.81053025e-01
-1.42174566e+00 1.39236569e-01 3.78209114e-01 3.96332026e-01
-4.17370409e-01 3.13365966e-01 -1.15106106e+00 -3.55090141e-01
-2.79893816e-01 1.24176599e-01 -7.68394828e-01 -2.49494210e-01
6.60465777e-01 8.78470123e-01 2.51275301e-01 -1.19608521e+00
-6.73139632e-01 1.36780226e+00 -1.68284342e-01 -3.20262998e-01
1.13667417e+00 2.33379230e-01 -9.33595300e-02 1.39762372e-01
1.35021877e+00 5.83767414e-01 -1.02422762e+00 -3.25592561e-03
-1.13887563e-02 -7.81190395e-01 -2.32562676e-01 -5.40744901e-01
-7.09589839e-01 5.92495024e-01 6.19349599e-01 1.19108193e-01
7.04716027e-01 -5.85615158e-01 6.27585828e-01 4.99522597e-01
1.59080908e-01 -1.49819744e+00 -5.75459778e-01 5.23339212e-01
4.72370118e-01 -1.09275663e+00 -2.22467110e-02 -1.39772862e-01
-9.95269299e-01 1.80904663e+00 2.50460327e-01 -6.22900486e-01
7.75382400e-01 2.96662182e-01 7.74378255e-02 1.31546587e-01
1.29176176e-03 2.27249742e-01 3.75430405e-01 6.96034968e-01
9.17842388e-02 2.95515060e-01 -3.16591501e-01 3.78025949e-01
-3.51671845e-01 1.51945263e-01 7.40699172e-01 9.72843170e-01
-9.73888338e-01 -7.94728398e-01 -8.93307090e-01 5.16344965e-01
-1.47695109e-01 -4.60312888e-02 -1.90701425e-01 9.60425079e-01
-8.22218731e-02 6.81343198e-01 4.67926770e-01 -6.01548254e-01
2.97982424e-01 2.03784764e-01 2.32268572e-01 2.45438248e-01
1.15686983e-01 -2.97232270e-01 -3.01156323e-02 1.59663260e-01
-9.59019084e-03 -8.39626729e-01 -1.41890168e+00 -8.11391950e-01
-4.00792092e-01 5.24101138e-01 1.27733362e+00 7.91834950e-01
1.16821751e-01 -2.37234145e-01 1.16800714e+00 -4.76153910e-01
-3.80993545e-01 -6.47239447e-01 -8.33324194e-01 1.35807931e-01
-1.39797509e-01 -1.15267985e-01 -4.52110976e-01 7.17427060e-02] | [9.791455268859863, -5.002038955688477] |
3acee780-3888-495c-bed1-0e05c8b2f544 | few-shot-text-generation-with-pattern | 2012.11926 | null | https://arxiv.org/abs/2012.11926v2 | https://arxiv.org/pdf/2012.11926v2.pdf | Few-Shot Text Generation with Pattern-Exploiting Training | Providing pretrained language models with simple task descriptions in natural language enables them to solve some tasks in a fully unsupervised fashion. Moreover, when combined with regular learning from examples, this idea yields impressive few-shot results for a wide range of text classification tasks. It is also a promising direction to improve data efficiency in generative settings, but there are several challenges to using a combination of task descriptions and example-based learning for text generation. In particular, it is crucial to find task descriptions that are easy to understand for the pretrained model and to ensure that it actually makes good use of them; furthermore, effective measures against overfitting have to be implemented. In this paper, we show how these challenges can be tackled: We introduce GenPET, a method for text generation that is based on pattern-exploiting training, a recent approach for combining textual instructions with supervised learning that only works for classification tasks. On several summarization and headline generation datasets, GenPET gives consistent improvements over strong baselines in few-shot settings. | ['Hinrich Schütze', 'Timo Schick'] | 2020-12-22 | null | null | null | null | ['headline-generation'] | ['natural-language-processing'] | [ 5.91680169e-01 3.40595335e-01 -2.97982395e-01 -3.06285590e-01
-1.24327636e+00 -4.14998949e-01 9.70169485e-01 1.27825677e-01
-3.24289858e-01 1.01437235e+00 5.60239792e-01 -1.53910905e-01
1.94059163e-01 -7.46581614e-01 -7.50856340e-01 -5.16663373e-01
3.34809691e-01 8.68361533e-01 8.51988420e-02 -5.55239975e-01
4.14627373e-01 9.40214191e-03 -1.55193222e+00 6.29133761e-01
1.00702012e+00 5.08274198e-01 3.88806581e-01 8.39293540e-01
-3.64192456e-01 1.02502513e+00 -7.98835218e-01 -5.81010163e-01
-1.24242514e-01 -7.24442840e-01 -1.00683820e+00 3.16156209e-01
4.81549591e-01 -3.21291089e-01 1.89276874e-01 5.29921293e-01
7.24315405e-01 3.99951279e-01 9.87032413e-01 -9.46656466e-01
-6.10495389e-01 1.06181026e+00 -1.88991636e-01 1.60875037e-01
4.53711957e-01 3.13980371e-01 1.23759842e+00 -7.93919981e-01
8.19719911e-01 1.15750372e+00 4.67074037e-01 8.81258011e-01
-1.52663255e+00 -3.38331580e-01 2.57895980e-02 -1.70765206e-01
-7.08117187e-01 -5.84952533e-01 5.31920552e-01 -3.50177348e-01
1.31697881e+00 1.63688749e-01 4.05891895e-01 1.55807316e+00
4.65306975e-02 1.12172604e+00 8.35273623e-01 -7.89178967e-01
2.35140219e-01 3.14741582e-01 3.23987901e-01 5.12574792e-01
2.73314208e-01 -1.42952174e-01 -6.89759314e-01 -7.51319751e-02
2.56464481e-01 -2.55919725e-01 -3.90284598e-01 -2.88203955e-01
-1.19494998e+00 1.10180771e+00 1.53451473e-01 4.69887555e-01
-2.28993997e-01 2.60222197e-01 6.29872918e-01 2.95578152e-01
6.78194642e-01 1.20406866e+00 -4.01122719e-01 -3.57756466e-01
-1.04560196e+00 5.19819379e-01 1.09553480e+00 1.24456620e+00
6.34609222e-01 1.62298024e-01 -6.45449340e-01 9.34882939e-01
-2.70662218e-01 2.32947052e-01 8.12204123e-01 -6.81843758e-01
6.65045857e-01 3.38157177e-01 7.65573904e-02 -4.06787157e-01
-2.46556595e-01 -2.99262434e-01 -6.15889549e-01 -1.14468597e-01
4.96215314e-01 -3.95699471e-01 -8.06528330e-01 1.59665608e+00
-8.01500082e-02 -8.31608631e-05 3.79800111e-01 3.25022072e-01
7.61243761e-01 8.67303371e-01 2.67041158e-02 -3.40436101e-01
1.17827797e+00 -1.20227277e+00 -7.02870071e-01 -6.31871223e-01
1.05500460e+00 -7.21827567e-01 1.40554786e+00 2.87179589e-01
-1.12922657e+00 -5.17901719e-01 -8.79043221e-01 -2.11837322e-01
-3.60116333e-01 1.39323473e-01 6.32117987e-01 5.30291617e-01
-9.50593352e-01 7.76165664e-01 -6.14193916e-01 -6.22647345e-01
4.21711296e-01 1.05242312e-01 -1.06489688e-01 -2.07398444e-01
-9.99810755e-01 9.85212862e-01 7.55885482e-01 -4.79393929e-01
-8.84913445e-01 -6.41474307e-01 -1.08913767e+00 2.13397175e-01
6.20975614e-01 -8.18269134e-01 1.68997180e+00 -7.78201759e-01
-1.40639555e+00 5.81859052e-01 -1.92420408e-01 -6.53211534e-01
3.24954778e-01 -3.06246102e-01 1.70456961e-01 1.88989460e-03
1.65740788e-01 7.74328351e-01 8.46655667e-01 -1.12112415e+00
-4.87017125e-01 -6.73948005e-02 -7.51511939e-03 2.07014695e-01
-4.88206416e-01 -8.40390772e-02 -2.22718045e-01 -6.93097413e-01
-5.01636624e-01 -8.29156399e-01 -3.94000351e-01 -6.27299726e-01
-6.53245807e-01 -5.86640775e-01 3.86663020e-01 -5.70479035e-01
1.02399576e+00 -1.72757292e+00 1.32667914e-01 -3.44757140e-01
-1.69672340e-01 4.32314217e-01 -3.47839057e-01 7.32601166e-01
1.48722365e-01 2.69599289e-01 -2.03575760e-01 -5.57965338e-01
9.45183933e-02 1.68392092e-01 -6.14021063e-01 -3.28542292e-01
5.92065990e-01 1.18108761e+00 -1.18764675e+00 -4.40592706e-01
8.92287877e-04 1.39715046e-01 -6.01516366e-01 4.06081378e-01
-7.34790623e-01 2.19062626e-01 -3.60146999e-01 1.89186722e-01
2.98391487e-02 -3.25993091e-01 -2.63438653e-03 1.29549682e-01
1.98926181e-01 6.86405063e-01 -7.99899459e-01 1.86388850e+00
-6.77227676e-01 5.88448942e-01 -6.59143865e-01 -1.15065074e+00
7.53129721e-01 5.50284505e-01 -3.17792632e-02 -4.23415482e-01
1.24555416e-01 2.63980210e-01 -2.31019676e-01 -6.26831293e-01
8.05383503e-01 -3.48130375e-01 -2.59158492e-01 1.02742684e+00
6.23499334e-01 -5.94340563e-01 8.71284425e-01 4.33762044e-01
1.10659099e+00 4.00968313e-01 5.06776750e-01 6.60558837e-03
1.56997785e-01 3.62851143e-01 1.03755988e-01 1.12729263e+00
3.68736744e-01 7.63845086e-01 4.12248701e-01 -2.81556129e-01
-1.08451569e+00 -5.17134964e-01 1.46637186e-01 1.35407937e+00
-5.37964106e-01 -6.45872831e-01 -8.66352022e-01 -8.27240646e-01
-4.40707505e-01 1.27193975e+00 -5.49863160e-01 -3.28694910e-01
-5.61791182e-01 -7.49447763e-01 3.75755876e-01 6.69851840e-01
2.30104551e-01 -1.35295677e+00 -5.25527060e-01 4.61531699e-01
-2.67331630e-01 -1.21738911e+00 -4.36142474e-01 4.33841795e-01
-1.03593981e+00 -8.25238407e-01 -9.17112470e-01 -8.90184462e-01
6.28158033e-01 2.29242548e-01 1.57140613e+00 1.62255138e-01
-2.43805707e-01 2.16003567e-01 -5.98505974e-01 -5.93971014e-01
-9.45627689e-01 6.03182018e-01 -4.22735155e-01 -3.76051515e-01
3.23585927e-01 -4.34752226e-01 -6.07717223e-02 -1.25328690e-01
-8.79958153e-01 2.88955808e-01 7.39698112e-01 1.17199683e+00
2.40926340e-01 -2.23659977e-01 6.83245957e-01 -1.35291266e+00
1.00578988e+00 -2.39620194e-01 -1.85972109e-01 4.75332975e-01
-5.14897525e-01 5.32706320e-01 8.80927324e-01 -3.92169088e-01
-1.15286648e+00 9.01649240e-03 -1.30224749e-01 -3.19044515e-02
-3.93618703e-01 4.10294622e-01 1.43762097e-01 5.66687167e-01
1.13846874e+00 4.65540409e-01 -7.94984475e-02 -4.06881511e-01
6.64647043e-01 6.42431259e-01 2.65911937e-01 -6.18402719e-01
6.30972147e-01 8.42005983e-02 -3.54582459e-01 -8.85219216e-01
-1.37325978e+00 -3.94008458e-01 -5.62463343e-01 1.90553933e-01
8.09588730e-01 -7.83804536e-01 8.12046751e-02 1.24391705e-01
-1.36505759e+00 -7.08054602e-01 -5.57114661e-01 2.06963703e-01
-8.69968295e-01 1.32421404e-01 -5.80097556e-01 -8.22681069e-01
-5.59458017e-01 -9.77235734e-01 1.36035395e+00 1.26049578e-01
-6.64853871e-01 -1.22988832e+00 1.38690934e-01 3.22803229e-01
4.09054190e-01 -1.96618610e-03 1.03321254e+00 -1.16007423e+00
-4.50355858e-01 -1.70338020e-01 1.72310337e-01 5.13194621e-01
1.95161089e-01 -1.97030753e-01 -9.63023663e-01 -1.56594053e-01
2.45453976e-02 -1.00118196e+00 1.00857341e+00 2.19118491e-01
9.96984303e-01 -5.53000987e-01 -2.56291449e-01 2.27125704e-01
1.13436842e+00 -8.53357539e-02 5.39936244e-01 9.85985100e-02
6.10974014e-01 7.93965518e-01 5.76635599e-01 2.11452767e-01
2.05118269e-01 6.84827864e-01 -1.74343199e-01 4.35575545e-02
-1.68590084e-01 -3.70918840e-01 3.96358073e-01 6.83339894e-01
2.04317290e-02 -4.30175781e-01 -7.18464553e-01 5.57826161e-01
-1.80890262e+00 -1.22288048e+00 -3.28746364e-02 2.18290186e+00
1.27304912e+00 2.91559517e-01 2.35224113e-01 9.69393924e-02
4.80853677e-01 8.19370225e-02 -2.12822139e-01 -5.66292167e-01
4.16826904e-02 4.99347717e-01 -1.00766681e-02 3.21897805e-01
-1.00741220e+00 1.12057126e+00 6.54742479e+00 8.88541520e-01
-9.04991210e-01 1.33032456e-01 6.26195192e-01 -8.54720324e-02
-3.11675817e-01 7.18065426e-02 -1.02610230e+00 3.61297101e-01
1.18681753e+00 -4.71767068e-01 2.59863585e-01 9.73892152e-01
9.37275514e-02 -1.45080969e-01 -1.44652843e+00 6.86914146e-01
4.76019353e-01 -1.42789626e+00 2.38689423e-01 -2.32050598e-01
1.03956282e+00 -1.59802675e-01 -1.75791070e-01 7.50010967e-01
6.50703371e-01 -1.09744394e+00 3.54502350e-01 1.24708086e-01
5.03746390e-01 -6.25316501e-01 6.37878299e-01 7.05405176e-01
-6.57796741e-01 1.21286660e-01 -5.86085856e-01 -1.03812389e-01
3.28677803e-01 5.37627459e-01 -1.41505027e+00 5.15053689e-01
8.81138667e-02 7.36864924e-01 -6.01630569e-01 9.21324193e-01
-6.12769544e-01 6.54027700e-01 1.11923302e-02 -4.37636435e-01
2.87563235e-01 2.04120532e-01 3.35061401e-01 1.41571212e+00
3.46548706e-01 -2.28193566e-01 3.04257780e-01 8.70301902e-01
-1.76652417e-01 3.24545145e-01 -1.01500762e+00 -3.68481845e-01
1.46925941e-01 1.30421495e+00 -6.09045863e-01 -6.25404418e-01
-2.25200281e-01 9.81748879e-01 5.71888447e-01 2.15388700e-01
-4.71104205e-01 -4.19412881e-01 7.64504746e-02 8.53315592e-02
2.49279484e-01 -8.66563618e-02 -1.86688930e-01 -1.24547946e+00
-2.12535307e-01 -1.09448409e+00 3.17439914e-01 -8.08439672e-01
-1.25469184e+00 5.75229347e-01 1.11193649e-01 -1.10247612e+00
-1.03822422e+00 -4.89355743e-01 -8.66415560e-01 5.93721628e-01
-1.47265363e+00 -1.00497293e+00 -1.57213032e-01 3.00541788e-01
1.21277642e+00 -1.50727600e-01 1.02308834e+00 -2.79798806e-01
-4.10071135e-01 5.17913640e-01 -5.81016354e-02 2.48012822e-02
8.72475326e-01 -1.56573188e+00 6.61462188e-01 8.06061327e-01
5.64476788e-01 5.20520806e-01 9.15949225e-01 -4.82633442e-01
-1.17407084e+00 -1.23883438e+00 1.12890899e+00 -7.58070648e-01
4.76937413e-01 -5.55915177e-01 -8.98562670e-01 7.78923571e-01
4.83470380e-01 -4.83129382e-01 7.36987174e-01 3.31668198e-01
-1.27372876e-01 2.66619891e-01 -6.31052732e-01 7.35049307e-01
9.00838971e-01 -2.40005210e-01 -9.65205550e-01 9.97521877e-01
8.46138835e-01 -3.63741010e-01 -4.40101355e-01 4.61890968e-03
1.26208499e-01 -7.55169392e-01 5.89622855e-01 -8.76909018e-01
8.75459373e-01 2.58617133e-01 3.08411419e-01 -1.74718237e+00
8.89206827e-02 -8.71514142e-01 -1.84745997e-01 1.46422112e+00
8.17632675e-01 -3.25428396e-01 5.42623818e-01 4.53622758e-01
-3.20618153e-01 -6.55109227e-01 -3.40749949e-01 -9.54548001e-01
2.71636605e-01 -3.58180404e-01 3.22564632e-01 6.37446582e-01
2.55414635e-01 1.17119384e+00 -5.69142520e-01 -6.58205032e-01
3.75601918e-01 6.70561451e-04 1.16978812e+00 -1.07186902e+00
-4.80703175e-01 -3.40324521e-01 8.54505673e-02 -1.02687097e+00
3.20890278e-01 -1.12306392e+00 4.67787504e-01 -1.86436319e+00
3.85459960e-01 -2.53895372e-01 2.03052461e-01 5.66299081e-01
-4.87092048e-01 8.08888003e-02 9.52170342e-02 1.49323285e-01
-7.26824045e-01 4.49443102e-01 1.25759935e+00 -1.08878620e-01
-2.57162035e-01 1.06971264e-01 -1.00663674e+00 3.96696180e-01
8.22121799e-01 -4.53400612e-01 -5.45171857e-01 -4.14295524e-01
1.90689504e-01 -8.80400985e-02 9.08534857e-04 -9.53480840e-01
1.76656112e-01 -1.28202792e-02 1.01779670e-01 -3.89808923e-01
2.49801308e-01 -1.71403468e-01 -3.49615961e-01 3.41350317e-01
-8.42784762e-01 -1.42789736e-01 4.42394800e-02 5.12954414e-01
-9.21354368e-02 -7.39255905e-01 6.81368470e-01 -4.78318065e-01
-6.07058227e-01 -5.08718425e-04 -3.49072963e-01 5.47936022e-01
8.40940773e-01 1.32464573e-01 -4.44300622e-01 -6.25099242e-01
-5.14035344e-01 2.07531616e-01 3.58520150e-01 3.88862431e-01
3.76627862e-01 -1.12727499e+00 -9.68663931e-01 -2.11032294e-02
3.54482383e-01 8.91590565e-02 -6.43057600e-02 7.27083683e-01
-1.11843161e-01 6.99431658e-01 7.68454298e-02 -5.45413613e-01
-1.01440191e+00 5.80731213e-01 2.20140745e-03 -8.13899100e-01
-6.24222755e-01 8.29948366e-01 7.76525289e-02 -3.60957056e-01
2.25029320e-01 -2.46821240e-01 -3.40259641e-01 1.80343941e-01
9.17418718e-01 -1.00613877e-01 2.33878687e-01 -1.54861271e-01
2.54442692e-01 9.65764895e-02 -4.67598230e-01 -2.92662382e-01
1.40171945e+00 2.22985044e-01 2.33216673e-01 5.96991658e-01
8.82872581e-01 -2.40789086e-01 -1.21071780e+00 -3.07465255e-01
3.27441961e-01 -1.96698084e-01 -2.42696092e-01 -9.13818121e-01
-5.45730710e-01 1.08518183e+00 -1.84695750e-01 4.33323503e-01
7.65143931e-01 1.61233336e-01 8.67067218e-01 7.94486821e-01
2.77854860e-01 -1.19038415e+00 7.22491860e-01 7.50656366e-01
9.33361411e-01 -1.36957645e+00 -1.30262360e-01 -2.59535164e-01
-9.42198515e-01 1.14276183e+00 5.74317276e-01 -1.48201898e-01
2.24546958e-02 2.87560701e-01 -2.16631502e-01 -8.36853907e-02
-1.19164145e+00 -4.48716789e-01 3.93820375e-01 7.09621549e-01
8.10114920e-01 -2.60266095e-01 -3.25648725e-01 4.54944551e-01
-5.18678308e-01 2.02523228e-02 7.77860343e-01 1.00695527e+00
-7.51217782e-01 -1.37836993e+00 -7.57196099e-02 7.08775163e-01
-4.37363088e-01 -3.52751404e-01 -5.04444003e-01 7.19150245e-01
-4.06523257e-01 1.02858758e+00 -7.03187957e-02 -1.69465333e-01
1.79496512e-01 6.78935885e-01 5.28498650e-01 -1.48663211e+00
-6.01735055e-01 -6.58140099e-03 5.26077926e-01 -2.19513744e-01
-2.02078193e-01 -4.86622453e-01 -9.36322331e-01 7.84811974e-02
-5.47451675e-01 2.58387566e-01 5.69298267e-01 1.30122304e+00
1.80573240e-01 6.32982194e-01 4.72702473e-01 -9.66694474e-01
-1.01939785e+00 -1.23105049e+00 -2.59557575e-01 5.48460722e-01
-1.24181351e-02 -3.27301323e-01 -3.96250188e-01 2.24037781e-01] | [11.668352127075195, 8.864684104919434] |
27a1893e-995c-43a4-b811-7e35042f3b68 | combining-metric-learning-and-attention-heads | 2209.06585 | null | https://arxiv.org/abs/2209.06585v2 | https://arxiv.org/pdf/2209.06585v2.pdf | Combining Metric Learning and Attention Heads For Accurate and Efficient Multilabel Image Classification | Multi-label image classification allows predicting a set of labels from a given image. Unlike multiclass classification, where only one label per image is assigned, such a setup is applicable for a broader range of applications. In this work we revisit two popular approaches to multilabel classification: transformer-based heads and labels relations information graph processing branches. Although transformer-based heads are considered to achieve better results than graph-based branches, we argue that with the proper training strategy, graph-based methods can demonstrate just a small accuracy drop, while spending less computational resources on inference. In our training strategy, instead of Asymmetric Loss (ASL), which is the de-facto standard for multilabel classification, we introduce its metric learning modification. In each binary classification sub-problem it operates with $L_2$ normalized feature vectors coming from a backbone and enforces angles between the normalized representations of positive and negative samples to be as large as possible. This results in providing a better discrimination ability, than binary cross entropy loss does on unnormalized features. With the proposed loss and training strategy, we obtain SOTA results among single modality methods on widespread multilabel classification benchmarks such as MS-COCO, PASCAL-VOC, NUS-Wide and Visual Genome 500. Source code of our method is available as a part of the OpenVINO Training Extensions https://github.com/openvinotoolkit/deep-object-reid/tree/multilabel | ['Vladislav Sovrasov', 'Kirill Prokofiev'] | 2022-09-14 | null | null | null | null | ['multi-label-image-classification'] | ['computer-vision'] | [ 3.9410645e-01 2.1827719e-01 -6.6705465e-01 -7.9869729e-01
-1.0690231e+00 -7.5656700e-01 4.4660670e-01 6.7020881e-01
-6.1381567e-01 1.0210927e+00 -2.5332949e-01 -3.8771400e-01
-1.2126815e-01 -7.1423131e-01 -6.4633447e-01 -8.7874377e-01
1.1386984e-01 5.9752470e-01 1.4228781e-02 6.6624790e-02
-4.0619593e-02 3.8711160e-01 -1.4449512e+00 2.1159397e-01
6.2522835e-01 1.2824361e+00 -2.0916207e-01 5.7743394e-01
2.7567541e-02 8.3686423e-01 -2.4354380e-01 -7.5416404e-01
3.2104316e-01 -3.3791414e-01 -9.7054869e-01 -1.7351270e-01
1.0869733e+00 5.5771895e-02 -1.4751987e-01 1.2252389e+00
5.4124528e-01 1.6464317e-02 1.0152528e+00 -1.4887483e+00
-4.9698672e-01 5.2916116e-01 -6.3700658e-01 -1.7404762e-01
3.3727103e-01 -8.9406386e-02 1.4671277e+00 -5.3818935e-01
7.7996916e-01 1.1905991e+00 7.6380622e-01 4.4766459e-01
-1.5070640e+00 -7.5129825e-01 8.0900803e-02 1.1430608e-01
-1.3646215e+00 -1.4929707e-01 2.9579806e-01 -3.7028781e-01
6.5059865e-01 4.9287930e-01 2.1065664e-01 8.7171000e-01
2.2253168e-01 6.6951615e-01 1.5253590e+00 -6.0540748e-01
-1.4884718e-01 2.7217153e-01 4.9450833e-01 1.1196369e+00
2.1046939e-01 -6.7223720e-02 -2.3581177e-01 -2.5534892e-01
1.6850640e-01 -1.4740588e-01 -1.7987037e-01 -6.9695359e-01
-1.2292125e+00 1.0749091e+00 5.3211951e-01 1.3411480e-01
1.4727788e-01 3.4745502e-01 7.2361451e-01 4.9250498e-01
4.7253662e-01 8.0865085e-02 -4.3613896e-01 3.1222242e-01
-7.0201379e-01 5.5001959e-02 6.2693924e-01 8.0040359e-01
1.0196817e+00 -4.8443168e-01 -2.6607558e-01 1.0572650e+00
1.9174033e-01 3.8298136e-01 3.2722744e-01 -8.6255258e-01
2.0045395e-01 6.1894679e-01 -4.4330013e-01 -7.0186156e-01
-6.8084967e-01 -4.5887393e-01 -7.5515401e-01 3.2792184e-01
6.8041778e-01 1.5578093e-01 -1.0252682e+00 1.9273914e+00
3.6042157e-01 -3.7184691e-01 -2.9080281e-01 5.9698874e-01
7.6753390e-01 2.2372222e-01 4.0004605e-01 -2.5396882e-02
1.4653202e+00 -9.2137283e-01 -5.5562013e-01 -9.1480240e-02
1.2205439e+00 -6.2185872e-01 1.0249686e+00 3.3779734e-01
-5.6920242e-01 -1.3530502e-01 -9.9874723e-01 -2.6627246e-01
-8.0317163e-01 3.9157938e-02 8.4042627e-01 7.3573142e-01
-1.1023327e+00 4.2909104e-01 -4.3275744e-01 -6.3330305e-01
5.2999508e-01 4.5625964e-01 -8.4858084e-01 -3.4548447e-01
-1.1800475e+00 1.1005535e+00 5.6124246e-01 -3.7745646e-01
-7.6656055e-01 -5.1064342e-01 -1.0354100e+00 -2.0334285e-01
3.1466249e-01 -4.5137009e-01 9.4400322e-01 -9.0067434e-01
-9.0690118e-01 1.4751328e+00 2.9577427e-02 -4.0471110e-01
5.0843084e-01 3.9607942e-01 -1.2412179e-01 5.2863162e-02
3.0160514e-01 1.1777686e+00 4.8268366e-01 -1.2342966e+00
-6.9586945e-01 -4.3224216e-01 3.5012200e-01 1.2035094e-01
-3.0038375e-01 8.0646193e-03 -6.1439477e-02 -4.8071182e-01
1.8359246e-02 -1.1660236e+00 3.0004608e-02 1.4093308e-01
-6.3953966e-01 -3.6691916e-01 5.2289844e-01 -4.0843180e-01
9.7928876e-01 -1.9556983e+00 3.2769952e-02 2.4663644e-01
2.3091605e-01 -7.4922442e-05 -3.1094763e-01 1.7290662e-01
-3.6302006e-01 3.0355126e-01 -3.1700468e-01 -4.7600931e-01
9.9577993e-02 3.0401662e-01 9.7937472e-02 8.1120294e-01
-4.8143864e-02 7.5953031e-01 -8.4619462e-01 -8.2017505e-01
2.4266252e-01 2.8615180e-01 -3.9335999e-01 -1.9518650e-01
-5.9005704e-02 2.5534585e-01 -1.0306929e-01 9.3150020e-01
6.4090055e-01 -4.5826700e-01 3.6396939e-01 -2.9609883e-01
3.3245823e-01 -1.1963048e-01 -1.0677437e+00 1.6871101e+00
-5.5556977e-01 4.5804375e-01 -3.0167190e-02 -1.3122085e+00
4.6294981e-01 1.9969364e-01 4.1783440e-01 -7.1399772e-01
3.0103877e-01 2.4444623e-01 -3.1104234e-01 2.3729874e-02
1.4814554e-01 -2.1640658e-01 -2.2909120e-01 2.0012407e-01
5.1588339e-01 4.8590582e-04 5.7745188e-01 3.8679519e-01
1.0091424e+00 -1.2047537e-02 6.7157000e-01 -3.3104196e-01
5.8398491e-01 -9.5457941e-02 5.7933098e-01 7.6769233e-01
-3.4633741e-01 4.9792424e-01 8.0022103e-01 -1.3137236e-01
-6.8660063e-01 -1.0432928e+00 -6.9715130e-01 1.4651059e+00
-4.7128224e-03 -3.4966999e-01 -5.0217295e-01 -1.2667030e+00
1.6359919e-01 6.0204631e-01 -6.4167994e-01 1.5027303e-02
-2.3801930e-01 -1.0406770e+00 8.0673617e-01 1.5387024e-01
1.3840291e-01 -7.0371085e-01 -1.1027002e-02 -2.0494248e-01
-2.3508091e-01 -1.0375612e+00 -5.6110835e-01 7.1995783e-01
-5.3982872e-01 -1.2061235e+00 -6.0568148e-01 -8.5191578e-01
8.5201222e-01 -8.0091447e-02 1.0583717e+00 -4.4071801e-02
-4.8522815e-01 4.1284686e-01 -3.4017751e-01 -2.0413721e-01
-4.0376809e-01 1.4352483e-01 -2.2017737e-01 -1.3780374e-02
1.4104620e-01 -3.1482989e-01 -4.6469525e-01 3.6707896e-01
-9.2037708e-01 -7.4272163e-02 4.6441987e-01 1.0748725e+00
8.8555336e-01 -3.5850960e-01 4.4577959e-01 -1.1820195e+00
3.8065571e-01 -5.0434166e-01 -6.1863905e-01 5.3157240e-01
-1.0059888e+00 6.3789852e-02 5.9505850e-01 -2.6353720e-01
-5.2036440e-01 2.6430649e-01 5.5602859e-03 -1.1970232e-01
-2.6061711e-01 4.7493476e-01 -1.1809508e-01 -4.7631407e-01
7.8990293e-01 -1.0751993e-01 -5.1956526e-03 -2.0256047e-01
6.0939610e-01 5.5502951e-01 2.3065661e-01 -6.3193578e-01
4.3863159e-01 4.5218542e-01 4.0401328e-01 -4.0928093e-01
-1.0424913e+00 -7.1510464e-01 -6.7311788e-01 -2.0298816e-01
1.0827390e+00 -8.2229608e-01 -7.4163592e-01 3.1504351e-01
-8.3935142e-01 -2.4975428e-01 -2.0800442e-01 4.1138986e-01
-5.3506726e-01 4.6381739e-01 -7.4405384e-01 -4.6590880e-01
-2.1196496e-01 -1.3946632e+00 1.1327825e+00 2.6927680e-02
-8.8742539e-02 -1.2699511e+00 1.0467227e-01 6.1518329e-01
9.3184590e-02 4.6151233e-01 1.1858928e+00 -9.7012949e-01
-2.7344334e-01 -5.0484264e-01 -4.8497158e-01 6.6615921e-01
-4.7941003e-02 -3.0554888e-01 -1.0690953e+00 -5.7748145e-01
-6.2120992e-01 -9.7014076e-01 1.0027282e+00 2.5056821e-01
1.2304599e+00 3.2777403e-02 -4.6099955e-01 6.1995196e-01
1.6115745e+00 -1.2804468e-01 3.4307915e-01 2.7855793e-01
8.3479428e-01 6.1016238e-01 7.5289446e-01 9.0995342e-02
6.2948948e-01 6.6689467e-01 7.4988979e-01 -1.9803329e-01
-2.1787240e-01 5.6031261e-02 1.8758851e-01 5.2391768e-01
1.8863274e-01 -5.3385013e-01 -9.0259176e-01 3.1594750e-01
-1.7157015e+00 -6.9962168e-01 -1.6456926e-01 2.3922970e+00
9.3095189e-01 -1.5499715e-02 7.5615220e-02 -1.3325898e-02
7.5502849e-01 1.9557889e-01 -4.4066754e-01 -4.1461933e-01
-4.1811740e-01 2.6043543e-01 9.8575169e-01 6.8554270e-01
-1.3718270e+00 7.2783124e-01 5.6255541e+00 1.2805009e+00
-1.1022826e+00 3.9017224e-01 9.2682141e-01 -3.8806640e-02
-1.7603248e-01 1.5771212e-01 -1.0011864e+00 2.1631017e-01
1.0435249e+00 1.5744273e-01 3.3051893e-01 6.8852395e-01
-4.1633764e-01 -1.6806166e-01 -1.2428672e+00 8.8799047e-01
7.1857914e-02 -9.8436195e-01 5.7457615e-02 2.1847999e-01
4.9368358e-01 2.4214177e-01 4.7492448e-02 2.8436813e-01
3.0044621e-01 -1.1136897e+00 7.8698391e-01 1.3772637e-01
1.2278069e+00 -5.4914552e-01 8.4059238e-01 1.4438878e-01
-1.1408496e+00 5.1475838e-02 -2.4973258e-01 3.4004253e-01
-7.7031739e-02 8.6642343e-01 -6.6278166e-01 7.3182422e-01
3.3526337e-01 5.8976126e-01 -9.3581921e-01 8.9667505e-01
-1.6797921e-01 4.8568210e-01 -2.3831095e-01 1.7253564e-01
1.9471098e-01 -1.1515220e-01 2.3778771e-01 1.4123480e+00
1.5232711e-02 -4.6195734e-01 4.6663764e-01 2.2008535e-01
-3.6313176e-01 4.6860895e-01 -5.8730066e-01 4.2374704e-02
2.5097567e-01 1.6922671e+00 -8.7005615e-01 -3.1123209e-01
-6.5746874e-01 8.6200708e-01 6.7756015e-01 1.6396776e-01
-1.0446118e+00 -4.5454341e-01 2.5821060e-01 4.3497145e-02
-4.9268991e-02 1.5360236e-01 -1.4407194e-01 -1.1150525e+00
-2.0783558e-01 -8.2757825e-01 1.0006001e+00 -4.5850649e-01
-1.3213471e+00 3.4583700e-01 5.8981981e-02 -1.1352026e+00
-2.4700232e-02 -9.5738012e-01 -7.9727612e-02 6.9757771e-01
-1.7059556e+00 -1.5555416e+00 -1.4595552e-01 4.6333584e-01
3.2972980e-02 8.6377680e-02 1.1073973e+00 7.0869297e-01
-4.7645399e-01 9.9416369e-01 1.5824851e-01 1.3542312e-01
1.2487414e+00 -1.5976247e+00 -2.7568084e-01 4.7757909e-01
1.7403795e-01 2.9747704e-01 4.5362696e-01 -2.8095290e-01
-8.9922941e-01 -1.1856514e+00 9.1623265e-01 -4.2640725e-01
7.7158135e-01 -3.7490818e-01 -6.0228890e-01 9.7029763e-01
1.6104929e-01 4.3377757e-01 9.4186473e-01 1.2353702e-01
-7.4086952e-01 -1.9216256e-01 -1.3002028e+00 1.9908923e-01
1.0278255e+00 -5.1389152e-01 2.4336999e-02 8.9004111e-01
4.1322267e-01 -2.8162503e-01 -1.2351714e+00 5.5719298e-01
6.2390655e-01 -8.3358729e-01 9.4796479e-01 -4.8408732e-01
2.2045898e-01 -2.0393802e-01 -4.9239331e-01 -1.2542529e+00
-2.5417969e-01 5.2219667e-03 3.8573682e-01 1.2171516e+00
7.4425912e-01 -1.0292125e+00 6.0058075e-01 2.1202287e-01
-8.1451148e-02 -9.0075028e-01 -1.0592322e+00 -7.6432711e-01
3.8819292e-01 -2.7811971e-01 1.4863855e-01 1.2420927e+00
-1.2647915e-01 2.3950966e-01 -1.6636129e-01 -3.6668766e-02
9.6649140e-01 -7.5768523e-02 4.8738036e-01 -1.2657328e+00
-1.7174301e-01 -4.2950496e-01 -7.2180480e-01 -4.7496018e-01
6.5211254e-01 -1.6618444e+00 -2.2055617e-01 -1.3199013e+00
5.0168997e-01 -7.0181102e-01 -6.9929999e-01 1.0216997e+00
-8.1571620e-03 6.9944447e-01 1.7578386e-01 -4.6050970e-02
-8.7165880e-01 -2.2182504e-03 1.1046480e+00 -4.6811694e-01
3.9314529e-01 -3.0825278e-01 -6.2471128e-01 5.5112469e-01
7.5527078e-01 -7.6437747e-01 -2.8553522e-01 -1.1733993e-01
2.9932928e-01 7.1843811e-03 3.8053554e-01 -7.5923294e-01
-3.0215089e-03 1.7058905e-02 1.6306277e-01 -1.6906819e-01
2.2957987e-01 -6.5086275e-01 1.5029101e-01 5.2190161e-01
-5.6058955e-01 -1.9638868e-01 -2.2111824e-02 4.8846221e-01
-2.7549043e-01 -3.6674342e-01 1.1053398e+00 -1.3743407e-01
-6.1937201e-01 4.6908242e-01 -3.6887735e-02 1.5237275e-01
1.1236329e+00 -4.2912055e-02 -6.8588930e-01 -1.2079896e-01
-8.6950713e-01 3.9942953e-01 5.1257825e-01 1.7115891e-01
6.1867040e-02 -1.3039515e+00 -6.5035367e-01 -2.2185692e-01
6.2167072e-01 -3.9735049e-01 1.1974909e-01 1.1265122e+00
-5.4805928e-01 5.7414281e-01 -1.3733864e-01 -6.3320678e-01
-1.5499127e+00 4.8853326e-01 4.3513197e-01 -5.7159138e-01
-1.3995109e-01 9.0674055e-01 4.0088928e-01 -1.0356634e+00
2.5720987e-01 -8.5091390e-02 -1.1518538e-01 4.4151479e-01
7.6133423e-02 3.4188375e-01 3.1274021e-01 -8.2586884e-01
-4.7269022e-01 6.3491392e-01 -1.9436488e-01 1.8387184e-01
8.4281325e-01 -1.8822482e-02 -4.3831041e-01 4.5237666e-01
1.6435813e+00 -1.4396545e-01 -6.6307563e-01 -4.6632800e-02
-6.9056287e-02 -1.9166224e-01 1.0839677e-01 -9.6394223e-01
-9.4140840e-01 7.8660816e-01 1.0012399e+00 1.5350278e-01
9.6695709e-01 1.9259696e-01 3.6888802e-01 2.8841245e-01
4.0518922e-01 -9.1908270e-01 -2.2440672e-01 2.4267976e-01
5.2628493e-01 -1.6898253e+00 6.8422824e-02 -5.4861963e-01
-4.2324939e-01 8.7387455e-01 4.0379193e-01 2.2744690e-01
5.7560182e-01 6.8918079e-02 1.4155711e-01 -1.2575899e-01
-4.9779078e-01 -2.8790894e-01 3.4162468e-01 4.2262125e-01
6.9525909e-01 3.8452336e-01 -5.9607697e-01 1.2798409e-01
5.6523811e-02 -3.2576689e-01 3.8510340e-01 8.7576813e-01
-2.0104289e-01 -1.3938897e+00 -2.2852522e-01 8.7813526e-01
-7.2605181e-01 -2.0485121e-01 -1.8391311e-01 8.6566710e-01
2.1582185e-01 8.4282839e-01 -2.2656082e-01 -1.8284743e-01
1.3529466e-01 2.7573079e-01 6.2361270e-01 -5.8846116e-01
-3.8850287e-01 -2.9478091e-01 2.9485705e-01 -5.7642114e-01
-5.1108778e-01 -5.8103806e-01 -1.2809299e+00 -2.5828832e-01
-4.8072937e-01 5.1769245e-02 6.9414276e-01 5.9072918e-01
7.3817745e-02 3.7707520e-01 3.4283969e-01 -5.1120758e-01
-6.0342348e-01 -8.2627988e-01 -7.3704833e-01 6.2445915e-01
9.9525027e-02 -7.9877508e-01 -6.4966208e-01 -6.7337036e-02] | [9.58692741394043, 4.110672950744629] |
bd491c7a-c482-41dc-8dc4-3f56014f35fa | run-off-election-improved-provable-defense | 2302.02300 | null | https://arxiv.org/abs/2302.02300v3 | https://arxiv.org/pdf/2302.02300v3.pdf | Run-Off Election: Improved Provable Defense against Data Poisoning Attacks | In data poisoning attacks, an adversary tries to change a model's prediction by adding, modifying, or removing samples in the training data. Recently, ensemble-based approaches for obtaining provable defenses against data poisoning have been proposed where predictions are done by taking a majority vote across multiple base models. In this work, we show that merely considering the majority vote in ensemble defenses is wasteful as it does not effectively utilize available information in the logits layers of the base models. Instead, we propose Run-Off Election (ROE), a novel aggregation method based on a two-round election across the base models: In the first round, models vote for their preferred class and then a second, Run-Off election is held between the top two classes in the first round. Based on this approach, we propose DPA+ROE and FA+ROE defense methods based on Deep Partition Aggregation (DPA) and Finite Aggregation (FA) approaches from prior work. We evaluate our methods on MNIST, CIFAR-10, and GTSRB and obtain improvements in certified accuracy by up to 3%-4%. Also, by applying ROE on a boosted version of DPA, we gain improvements around 12%-27% comparing to the current state-of-the-art, establishing a new state-of-the-art in (pointwise) certified robustness against data poisoning. In many cases, our approach outperforms the state-of-the-art, even when using 32 times less computational power. | ['Soheil Feizi', 'Atoosa Chegini', 'Kiarash Banihashem', 'Keivan Rezaei'] | 2023-02-05 | null | null | null | null | ['data-poisoning'] | ['adversarial'] | [ 1.68892384e-01 -1.10787414e-01 1.86281335e-02 -8.91612396e-02
-1.19422245e+00 -8.86291325e-01 5.45917869e-01 4.35925275e-01
-5.72274089e-01 8.53185713e-01 -1.40046567e-01 -6.08482242e-01
-9.67843831e-03 -9.68907773e-01 -9.72996533e-01 -8.87991250e-01
-1.03928052e-01 5.51242113e-01 6.47983611e-01 -2.49427631e-01
2.73338765e-01 5.34200549e-01 -9.82852995e-01 7.34474540e-01
6.52614534e-01 1.03849447e+00 -9.44884479e-01 5.23883581e-01
4.85577613e-01 8.79184544e-01 -7.82372057e-01 -8.07882845e-01
5.98001182e-01 -2.47127131e-01 -8.19350362e-01 -3.86400402e-01
3.17899019e-01 -5.45704365e-01 -4.33251619e-01 9.22877908e-01
5.32749355e-01 -3.24274093e-01 4.37622666e-01 -1.48007107e+00
-2.87022501e-01 9.88694072e-01 -6.93586111e-01 -5.81857115e-02
5.12161627e-02 2.96192318e-01 8.16218734e-01 -3.71409774e-01
2.57506877e-01 1.08909190e+00 7.20008135e-01 7.88987160e-01
-1.51103580e+00 -1.19642067e+00 2.35640645e-01 2.23459601e-01
-1.38707900e+00 -2.42051005e-01 5.83954811e-01 -6.40586391e-02
8.35937977e-01 6.03849113e-01 1.24715015e-01 1.16821122e+00
1.92552224e-01 5.55584311e-01 1.51077485e+00 -1.68992355e-01
6.08809590e-01 -4.39761430e-02 5.62704265e-01 3.49370003e-01
6.78365111e-01 3.09438735e-01 -3.94706249e-01 -1.12892330e+00
-1.04778878e-01 -1.27956793e-01 -2.83835799e-01 -1.51053086e-01
-6.50441229e-01 8.91677201e-01 4.82851177e-01 -9.54228267e-02
-3.75498474e-01 2.75381178e-01 4.81841266e-01 2.72146225e-01
4.06632870e-01 3.32609624e-01 -5.29113650e-01 3.44176352e-01
-1.00024724e+00 5.29430628e-01 7.97461450e-01 1.95065603e-01
3.63230944e-01 -1.89677417e-01 -2.88029820e-01 3.85819711e-02
2.11367354e-01 6.16886795e-01 1.33397400e-01 -6.90417349e-01
4.94539112e-01 3.69335204e-01 2.01471210e-01 -9.07297254e-01
-2.74872422e-01 -5.41065395e-01 -8.34701002e-01 3.34796399e-01
4.59190011e-01 -2.39134967e-01 -8.25876653e-01 2.05346942e+00
6.79980814e-01 4.47098106e-01 2.14538723e-01 3.87780815e-01
1.35636017e-01 3.03795636e-01 2.81325012e-01 -1.87059775e-01
1.34163439e+00 -4.89395082e-01 -4.63544838e-02 2.35744193e-01
7.26497710e-01 -1.16649665e-01 5.87843716e-01 8.69183540e-01
-7.40050316e-01 -1.24214709e-01 -1.14584696e+00 4.41188782e-01
-3.52770478e-01 -4.20493364e-01 2.75374800e-01 1.23297346e+00
-8.82505298e-01 5.06412625e-01 -8.86904895e-01 1.48293346e-01
7.90426135e-01 7.43829608e-01 -5.06571591e-01 2.64127087e-02
-1.35451329e+00 6.66687906e-01 1.41417608e-01 -3.64284635e-01
-1.29770374e+00 -6.74914598e-01 -3.32621694e-01 1.07749186e-01
3.93530935e-01 -7.14945912e-01 8.59523594e-01 -1.46895587e-01
-1.23348033e+00 4.33727145e-01 -4.00650725e-02 -1.36178923e+00
7.07613587e-01 -2.38149598e-01 -2.82575697e-01 9.14672241e-02
-3.66060734e-01 3.99745375e-01 5.54181039e-01 -1.53917480e+00
-6.74713910e-01 -5.80556631e-01 4.72302526e-01 -3.82770300e-01
-3.09583843e-01 1.06734678e-01 2.91662574e-01 -3.86875391e-01
-1.51048914e-01 -1.15112388e+00 -4.60355729e-01 -3.18930566e-01
-7.93181241e-01 -8.82426724e-02 1.13412964e+00 -4.91175264e-01
1.45993233e+00 -1.95846021e+00 -3.48699659e-01 5.36289275e-01
4.13975239e-01 6.10311270e-01 -3.00776381e-02 3.22867483e-01
1.16969638e-01 5.30358136e-01 -5.34660697e-01 -4.34930682e-01
7.30582103e-02 -7.53450319e-02 -9.51425850e-01 5.89762747e-01
-1.74498141e-01 4.23949599e-01 -5.99752128e-01 4.43564095e-02
1.23776458e-02 2.72034407e-01 -8.34640265e-01 -2.73311287e-01
-2.12809354e-01 5.73238015e-01 -3.44849467e-01 3.27275723e-01
1.03547382e+00 -4.39008884e-02 4.60781753e-01 -1.08560681e-01
2.66288042e-01 5.77685893e-01 -1.07860625e+00 9.55936790e-01
-1.27528444e-01 -1.31983710e-02 -1.32691011e-01 -7.61608243e-01
6.61547005e-01 3.94516766e-01 3.36421400e-01 -3.82960886e-01
2.50753522e-01 2.23405898e-01 1.26351684e-01 2.44770825e-01
3.07637483e-01 5.29681444e-02 -4.50593352e-01 8.23338985e-01
-5.22749662e-01 2.46209338e-01 -1.47264540e-01 3.11055630e-01
1.51756835e+00 -3.59798700e-01 2.03375801e-01 -2.45615542e-01
6.55318677e-01 -1.39636695e-01 6.75015390e-01 1.18696690e+00
-2.45328709e-01 2.79992402e-01 5.74240088e-01 -5.66959441e-01
-7.75455832e-01 -9.57187057e-01 4.41403314e-02 7.57511199e-01
7.05016479e-02 -4.83491361e-01 -1.10578489e+00 -1.35183167e+00
4.56714211e-03 9.15530086e-01 -6.41304016e-01 -2.92486966e-01
-5.94493151e-01 -1.18182719e+00 1.35982800e+00 3.00015479e-01
9.36561763e-01 -5.71092606e-01 -6.41324282e-01 9.62225348e-02
-2.08482936e-01 -9.89036679e-01 -1.02319665e-01 2.14857042e-01
-6.45959258e-01 -1.25222278e+00 1.93029881e-01 1.02526143e-01
6.16681159e-01 9.20915082e-02 7.03072131e-01 2.15649396e-01
2.88047493e-01 -8.99523050e-02 -2.41373956e-01 -5.28492391e-01
-5.64613104e-01 2.30332762e-01 3.97832841e-01 2.02635765e-01
2.55573153e-01 -7.06009269e-01 -5.61212420e-01 3.18296522e-01
-1.04586506e+00 -3.37154299e-01 2.02719346e-01 5.80218196e-01
4.79370326e-01 1.98993355e-01 7.60373652e-01 -1.19688535e+00
3.81596804e-01 -5.87878048e-01 -7.21415043e-01 3.74146163e-01
-5.86583316e-01 1.75104678e-01 9.01214182e-01 -5.98515749e-01
-4.67262000e-01 -5.93703501e-02 -1.76291838e-01 -4.19708848e-01
-5.72845228e-02 8.07342585e-03 -3.97194505e-01 -2.13523254e-01
9.42534029e-01 1.30229011e-01 -3.27263832e-01 -2.77368814e-01
3.43013167e-01 6.84184909e-01 5.24235368e-01 -7.77315259e-01
1.04385781e+00 7.92299867e-01 1.51541978e-01 -7.97915831e-02
-6.05190575e-01 1.26426235e-01 -2.71031141e-01 3.62316757e-01
4.95790124e-01 -7.55061507e-01 -1.03755617e+00 8.19260776e-01
-1.03318572e+00 -3.07787240e-01 -1.66950822e-01 9.31638554e-02
-2.54261792e-02 6.17717266e-01 -6.53081357e-01 -8.80485952e-01
-7.75867403e-01 -1.26654005e+00 6.83804393e-01 -5.24418317e-02
-9.32831168e-02 -5.43064415e-01 -7.73024419e-03 4.44107205e-01
3.33914101e-01 5.75014830e-01 1.00427878e+00 -1.41909897e+00
-3.82261425e-01 -3.42786998e-01 8.41263980e-02 3.51454645e-01
-2.00574815e-01 -2.21906990e-01 -1.19107664e+00 -6.66916430e-01
1.52574718e-01 -3.78424078e-01 1.02157593e+00 4.99593653e-03
1.47405970e+00 -7.49840796e-01 -3.75570804e-01 3.22439075e-01
1.21928215e+00 2.46339664e-01 8.39413464e-01 3.68551493e-01
3.42647016e-01 1.16427340e-01 3.04996222e-01 5.50373018e-01
5.97447395e-01 6.05890512e-01 6.70794487e-01 1.34702012e-01
2.10984007e-01 -1.87016055e-01 5.11047423e-01 -2.60019396e-02
1.91246912e-01 -4.56341922e-01 -7.63203025e-01 2.81543851e-01
-1.67818224e+00 -1.08170819e+00 -1.01308562e-01 2.71500969e+00
9.28021193e-01 4.21550512e-01 2.36762762e-01 6.43070400e-01
6.18210554e-01 7.94447660e-02 -6.26563728e-01 -3.98424566e-01
-6.70602173e-02 5.25948882e-01 8.57917964e-01 4.44339633e-01
-1.34672821e+00 7.51192987e-01 5.89270544e+00 1.16993821e+00
-1.23270106e+00 4.41645920e-01 9.80647087e-01 -2.29870006e-01
-1.96571052e-01 3.15826803e-01 -9.05521572e-01 5.80253303e-01
1.17036200e+00 1.36215433e-01 4.44669008e-01 6.22880816e-01
-2.16208830e-01 -5.26840761e-02 -1.05334282e+00 4.13056701e-01
-6.65668324e-02 -1.22770321e+00 1.07621945e-01 4.67355788e-01
6.30994499e-01 7.96757787e-02 2.80149281e-01 3.40093732e-01
8.77220929e-01 -8.53629887e-01 7.54578590e-01 1.13726169e-01
5.29240012e-01 -1.04781497e+00 6.96699619e-01 6.40006840e-01
-8.12482893e-01 -4.48704630e-01 -2.47959614e-01 7.81988204e-02
-1.66513219e-01 6.70985341e-01 -5.89054465e-01 9.30918634e-01
6.99075103e-01 -2.43369743e-01 -5.76841772e-01 6.71557724e-01
-3.45821410e-01 1.16899943e+00 -7.21003234e-01 3.20569068e-01
1.83112100e-01 5.24317801e-01 6.32688403e-01 8.87206256e-01
5.83174601e-02 3.14854473e-01 3.83986562e-01 4.49635565e-01
-5.36361158e-01 -4.04707700e-01 -2.01713935e-01 5.96905053e-01
9.36278343e-01 9.84924912e-01 -4.36855048e-01 -4.88668323e-01
2.27069601e-01 4.89506394e-01 1.07380427e-01 -5.91294058e-02
-1.15916693e+00 -4.84701246e-02 5.19717693e-01 1.69543445e-01
3.21959049e-01 8.76760185e-02 -4.41134810e-01 -9.58779991e-01
-2.14156881e-01 -1.39875364e+00 7.22249866e-01 -1.03748225e-01
-1.35118806e+00 8.11789453e-01 7.97730833e-02 -1.09296596e+00
-4.80598174e-02 -2.12016776e-01 -6.52415216e-01 5.51100135e-01
-1.05706966e+00 -1.27949095e+00 1.90401167e-01 5.93964517e-01
-8.74362886e-02 -4.76448908e-02 9.74029243e-01 1.18274041e-01
-5.94371736e-01 1.21249306e+00 1.26516029e-01 1.09098174e-01
5.41971028e-01 -8.87811363e-01 5.80430984e-01 1.13348472e+00
2.67769367e-01 7.08068669e-01 7.30991662e-01 -6.52378738e-01
-8.39378357e-01 -1.22250235e+00 6.37820959e-01 -6.26341283e-01
4.14150923e-01 -4.60289180e-01 -9.58487511e-01 5.93116164e-01
-5.33596575e-02 2.92168349e-01 8.66692662e-01 -8.71623009e-02
-1.07839310e+00 -3.21719080e-01 -1.82211041e+00 3.48491758e-01
6.91127896e-01 -3.70398372e-01 -2.39631087e-01 -5.64533435e-02
5.51461816e-01 -3.29983056e-01 -5.51183760e-01 7.44064450e-01
6.56434953e-01 -1.02444971e+00 8.86959672e-01 -7.50881851e-01
-3.58047374e-02 -6.06889188e-01 -5.28740883e-01 -1.02213156e+00
-1.48663953e-01 -6.64291024e-01 -3.72273356e-01 1.20231736e+00
4.04426903e-01 -1.02044487e+00 6.34295940e-01 6.44175351e-01
3.35371912e-01 -9.45984304e-01 -1.20274806e+00 -7.09935069e-01
5.47458172e-01 -4.06328857e-01 1.09541690e+00 8.96947801e-01
-3.05135697e-01 -1.43573865e-01 -5.14258862e-01 7.69325733e-01
8.96234393e-01 -3.87830958e-02 1.06516922e+00 -1.21206093e+00
-5.22931337e-01 -1.53327242e-01 -2.57645041e-01 -5.47069907e-01
8.96456987e-02 -7.16305375e-01 -4.25687701e-01 -9.83297706e-01
4.36057687e-01 -5.39268136e-01 -6.42597437e-01 1.06427193e+00
-1.35349184e-01 6.16151035e-01 5.52127898e-01 3.43994558e-01
-6.14148915e-01 1.11737281e-01 2.73048222e-01 -2.84361601e-01
2.06400864e-02 2.81651989e-02 -1.06179368e+00 6.16960704e-01
9.43219662e-01 -1.03012383e+00 -2.09646448e-01 -5.51032051e-02
-6.00786656e-02 -2.54192263e-01 6.70476079e-01 -1.28857112e+00
3.99380535e-01 6.62394911e-02 3.37979048e-01 -6.07639372e-01
2.71225154e-01 -6.84211552e-01 4.04148310e-01 1.03026986e+00
-4.30113435e-01 9.90778022e-03 1.89087749e-01 6.50762320e-01
2.75182664e-01 1.83273658e-01 9.64541018e-01 1.62680432e-01
1.71095833e-01 3.02515835e-01 -2.40901858e-01 -1.60091102e-01
1.15004468e+00 1.32603511e-01 -9.60713029e-01 -1.76313430e-01
-5.28971732e-01 2.05269665e-01 5.21673143e-01 -6.97924569e-02
2.94888437e-01 -9.02933419e-01 -9.60251272e-01 4.88719344e-02
-2.23818064e-01 -1.20036274e-01 3.11146915e-01 8.81124794e-01
-1.97192460e-01 1.37870416e-01 7.02201650e-02 -2.67138720e-01
-1.58880711e+00 5.82506239e-01 4.08823609e-01 -1.10526586e+00
-1.49037033e-01 6.45123363e-01 1.52195394e-01 -4.46016341e-01
1.25051871e-01 -1.77170541e-02 1.26174703e-01 -3.06443632e-01
6.04768217e-01 4.72811759e-01 2.00538263e-01 -4.33192670e-01
-6.19718373e-01 2.97595896e-02 -3.57465595e-01 -1.90404370e-01
1.13055539e+00 2.10330442e-01 -1.76051751e-01 -1.61989018e-01
7.07871199e-01 4.23953474e-01 -9.56005752e-01 -1.77850857e-01
-7.43404552e-02 -2.60021210e-01 -5.87701872e-02 -1.17701352e+00
-9.87456203e-01 6.70258939e-01 5.54397166e-01 1.65444300e-01
1.30353856e+00 -2.69789129e-01 1.00490427e+00 4.89369541e-01
8.29193771e-01 -4.53506172e-01 -3.12875867e-01 2.40683392e-01
4.24191117e-01 -7.26107061e-01 1.99694961e-01 -1.83597550e-01
-5.95819175e-01 5.56489646e-01 4.68847245e-01 -2.76754498e-01
6.37725592e-01 3.42809021e-01 -2.95513421e-01 8.35694298e-02
-1.04583776e+00 2.85437942e-01 -1.42054737e-01 4.59531844e-01
-3.43537301e-01 2.02767566e-01 -4.05486524e-01 1.09327292e+00
-9.89128426e-02 -2.34983161e-01 4.92624640e-01 8.23243678e-01
-3.07928324e-01 -1.56036747e+00 -6.03784144e-01 3.44752073e-01
-1.07691586e+00 -1.43451303e-01 -5.80189884e-01 6.02543294e-01
3.22541952e-01 1.31364989e+00 -2.94220477e-01 -9.20421839e-01
9.12974030e-02 3.38602513e-02 2.84950584e-01 -2.18665615e-01
-1.22674394e+00 -3.74024898e-01 3.57244499e-02 -4.70113456e-01
-9.55200568e-02 -5.02323508e-01 -1.20948827e+00 -7.11792946e-01
-5.82687736e-01 4.57228482e-01 4.72653896e-01 9.39402223e-01
6.68062210e-01 8.91877264e-02 1.10149193e+00 -4.49842811e-01
-1.10819709e+00 -8.22745144e-01 -3.18326741e-01 1.75156191e-01
1.53960988e-01 -4.69969422e-01 -6.00246727e-01 -4.30979580e-01] | [5.827052593231201, 7.4989705085754395] |
1d816422-1c81-48b0-902c-fc3b57ea0c22 | hypergraph-artificial-benchmark-for-community | 2210.15009 | null | https://arxiv.org/abs/2210.15009v3 | https://arxiv.org/pdf/2210.15009v3.pdf | Hypergraph Artificial Benchmark for Community Detection (h-ABCD) | The Artificial Benchmark for Community Detection (ABCD) graph is a recently introduced random graph model with community structure and power-law distribution for both degrees and community sizes. The model generates graphs with similar properties as the well-known LFR one, and its main parameter can be tuned to mimic its counterpart in the LFR model, the mixing parameter. In this paper, we introduce hypergraph counterpart of the ABCD model, h-ABCD, which produces random hypergraph with distributions of ground-truth community sizes and degrees following power-law. As in the original ABCD, the new model h-ABCD can produce hypergraphs with various levels of noise. More importantly, the model is flexible and can mimic any desired level of homogeneity of hyperedges that fall into one community. As a result, it can be used as a suitable, synthetic playground for analyzing and tuning hypergraph community detection algorithms. | ['François Théberge', 'Paweł Prałat', 'Bogumił Kamiński'] | 2022-10-26 | null | null | null | null | ['community-detection'] | ['graphs'] | [ 3.34642194e-02 3.20153236e-01 -3.27903442e-02 4.44875330e-01
-1.01244457e-01 -9.80822980e-01 6.49394810e-01 4.40285951e-01
1.18013673e-01 7.04325497e-01 -1.22994334e-01 -4.24432188e-01
-2.26726115e-01 -1.36651528e+00 -4.39984053e-01 -7.07095742e-01
-6.05883718e-01 8.97486508e-01 7.85313785e-01 -2.31063530e-01
9.96779576e-02 4.68428910e-01 -1.23978674e+00 -4.58510295e-02
8.06570888e-01 8.19917396e-02 8.83637741e-02 7.18676269e-01
1.76164031e-01 6.72942281e-01 -7.05358684e-01 -4.82923597e-01
3.90086353e-01 -5.21834195e-01 -5.66328585e-01 3.19545716e-01
-8.31059963e-02 4.11584467e-01 -7.82313704e-01 1.23680210e+00
4.75788236e-01 -4.12029177e-01 9.32920575e-01 -1.72711301e+00
-5.00665545e-01 1.07324815e+00 -1.30155909e+00 3.78517397e-02
6.22812510e-01 1.15854308e-01 1.04306245e+00 -3.53962719e-01
1.00180924e+00 1.51734054e+00 7.36945391e-01 2.73063809e-01
-1.82127976e+00 -7.18525887e-01 -1.10915847e-01 5.27519472e-02
-1.80913353e+00 4.61486995e-01 5.64051688e-01 -7.73749411e-01
1.95096135e-01 2.50444233e-01 1.15430558e+00 1.12911999e+00
1.57121241e-01 5.31659007e-01 1.16895342e+00 -5.31046212e-01
3.90199095e-01 7.00867251e-02 1.28731117e-01 7.21397638e-01
1.29909694e+00 -2.51778793e-02 -2.41675377e-02 -6.07057214e-01
7.87945271e-01 -2.64587969e-01 -5.06984949e-01 -9.25782740e-01
-1.30677366e+00 9.81273115e-01 4.34077710e-01 4.44715589e-01
-2.43553910e-02 1.18191727e-01 2.31097192e-01 3.91459227e-01
1.09179586e-01 4.38507766e-01 3.50557059e-01 5.39405465e-01
-6.96579993e-01 3.25889498e-01 1.16907513e+00 1.27705979e+00
6.61210835e-01 -1.95677072e-01 -3.30410212e-01 4.47228700e-01
3.03875655e-01 6.45327270e-01 8.46160278e-02 -7.18316495e-01
1.30082279e-01 1.03453469e+00 -3.19480225e-02 -1.39804828e+00
-4.60847288e-01 -6.24792457e-01 -1.46307015e+00 1.05457462e-01
5.18857121e-01 2.37471998e-01 -7.04324543e-01 1.46281707e+00
3.48075092e-01 1.84589073e-01 -6.83003590e-02 7.93012083e-01
8.38899076e-01 3.76216531e-01 -4.00636584e-01 -2.20479026e-01
1.30122375e+00 -6.29165888e-01 -5.37378848e-01 7.07921237e-02
2.45031863e-01 -5.11795342e-01 9.68129039e-01 1.73848763e-01
-8.49473774e-01 -6.51016925e-03 -1.11065316e+00 7.21173048e-01
-9.77609679e-02 -3.69909853e-01 5.03189027e-01 7.35704958e-01
-1.20111775e+00 2.86680251e-01 -2.89067000e-01 -5.22117972e-01
-9.28072333e-02 3.40414830e-02 -4.61462200e-01 -3.92500490e-01
-1.16000497e+00 3.68022054e-01 4.91288394e-01 -1.37674287e-01
-1.12207830e+00 -3.04173470e-01 -6.57130420e-01 6.28128573e-02
5.35866439e-01 -8.05243731e-01 6.88688219e-01 -5.87569952e-01
-7.53524601e-01 1.11489832e+00 3.62266123e-01 -3.75840783e-01
7.26951957e-01 8.62902105e-01 -2.02385977e-01 2.25270018e-01
1.15826659e-01 1.76473632e-01 6.21637106e-01 -1.56381786e+00
2.69627552e-02 -4.54918295e-02 1.65520519e-01 -2.83377379e-01
5.96766099e-02 -2.78082132e-01 -2.98009723e-01 -6.20434403e-01
1.62940487e-01 -1.06694782e+00 -3.43966037e-01 -4.10291046e-01
-8.71560693e-01 -6.72443304e-03 3.86908531e-01 1.14657082e-01
1.49988091e+00 -1.80723655e+00 2.14418724e-01 8.77442539e-01
1.03716946e+00 2.47321185e-02 -4.48698878e-01 1.00143731e+00
-6.08681254e-02 6.26111329e-01 -3.08445424e-01 2.92545527e-01
-1.35106556e-02 1.57521646e-02 1.70474067e-01 6.09293461e-01
7.92533681e-02 5.06200969e-01 -1.36648130e+00 -5.76623678e-01
-2.62281775e-01 1.89729020e-01 -4.89098430e-01 -7.05969855e-02
1.99995395e-02 1.69856414e-01 -4.15366828e-01 4.07029867e-01
9.25224543e-01 -7.43735015e-01 7.71432519e-01 1.59295440e-01
1.87368125e-01 -4.31843668e-01 -1.56554127e+00 7.29009509e-01
2.02736482e-01 6.55483067e-01 2.96206713e-01 -6.39354587e-01
1.26185513e+00 1.35340944e-01 3.20015460e-01 -1.19036190e-01
1.41786367e-01 8.49711224e-02 5.33070982e-01 -7.81602785e-02
1.64793566e-01 2.08388209e-01 -1.71514116e-02 7.39516735e-01
-1.53659672e-01 -3.13949347e-01 8.47600281e-01 8.66387784e-01
1.87660813e+00 -7.76729524e-01 4.18025345e-01 -7.24597812e-01
6.16416097e-01 -3.00774723e-02 3.54193628e-01 1.05382216e+00
-6.34446293e-02 7.95726478e-01 1.12294579e+00 -1.16343081e-01
-1.32170546e+00 -1.18824840e+00 8.11641887e-02 3.44220042e-01
5.52482724e-01 -6.40887082e-01 -8.03746045e-01 -2.74095803e-01
1.34452537e-01 3.40206251e-02 -8.13302219e-01 -2.23832980e-01
-3.39208484e-01 -9.35108304e-01 4.97385323e-01 -1.27069980e-01
2.85160184e-01 -1.07882738e+00 -4.08176193e-03 -1.25834476e-02
-3.68119478e-01 -1.00026166e+00 -6.15791619e-01 -1.25585869e-01
-5.05886912e-01 -1.91201401e+00 -7.73274720e-01 -8.95678937e-01
8.23385835e-01 5.02022207e-01 1.68460488e+00 6.44780815e-01
-3.36647689e-01 3.86373699e-01 -6.47460520e-01 -3.08471750e-02
-9.92175281e-01 2.77229995e-01 -1.68570593e-01 2.25464895e-01
3.82192656e-02 -7.51585782e-01 -5.48138797e-01 6.54377222e-01
-1.07827961e+00 -3.70659195e-02 3.10706913e-01 7.58041739e-01
4.54138547e-01 3.46458763e-01 4.59916711e-01 -1.18025589e+00
1.17055678e+00 -7.16537416e-01 -7.88352132e-01 3.60706538e-01
-6.89017653e-01 8.61776993e-03 4.08395857e-01 -5.55026293e-01
-3.25386167e-01 -3.19164604e-01 5.10822356e-01 -3.76173854e-01
2.87743390e-01 5.09463012e-01 -1.05096385e-01 -6.66235164e-02
7.86067843e-01 7.34915510e-02 1.67838428e-02 -1.66428387e-01
2.28622019e-01 4.78839785e-01 3.04151952e-01 -3.64698350e-01
1.31683993e+00 3.55928779e-01 3.32177043e-01 -6.97209418e-01
-1.83146939e-01 -5.06296575e-01 -5.53597271e-01 -3.32975626e-01
1.99052498e-01 -8.88258398e-01 -6.79922938e-01 6.69894397e-01
-9.74388063e-01 -3.81266713e-01 -2.40476832e-01 1.75415620e-01
-3.05680990e-01 5.28165698e-01 -7.68025219e-01 -8.48870397e-01
-6.51232228e-02 -8.41394186e-01 8.70889127e-01 3.11316568e-02
-1.20846145e-02 -1.02465725e+00 4.28435862e-01 -2.72720039e-01
2.05697045e-01 7.84985304e-01 9.64832604e-01 -5.26521802e-01
-6.97433650e-01 -4.16391760e-01 -3.07815075e-01 -2.25234732e-01
-9.27496254e-02 6.46232128e-01 -3.41854781e-01 -7.49306560e-01
-6.08009517e-01 -2.16671117e-02 6.86411858e-01 3.93259883e-01
6.95379376e-01 -2.67454386e-01 -6.68754101e-01 2.52134234e-01
1.52604425e+00 -1.05924606e-01 9.42593098e-01 3.39075297e-01
3.34246278e-01 4.47953999e-01 1.67795241e-01 3.21276128e-01
2.51318544e-01 4.88121390e-01 5.18947244e-01 -2.75197029e-01
3.52043174e-02 -3.90693158e-01 7.06754299e-03 8.67880702e-01
-1.82188213e-01 -6.57076836e-01 -1.13233423e+00 3.38441461e-01
-1.81739724e+00 -1.20050573e+00 -8.31438780e-01 2.33249831e+00
6.59568608e-01 2.89708525e-01 5.80264568e-01 4.68008280e-01
1.56815624e+00 -8.56250431e-03 -2.70465970e-01 -1.07006781e-01
-5.69956899e-01 -1.24732643e-01 4.27041501e-01 4.38769311e-01
-6.95531368e-01 4.35859114e-01 7.07511806e+00 8.37235153e-01
-4.67248648e-01 -1.54759869e-01 4.76311386e-01 3.43561381e-01
-6.40985072e-01 1.25951171e-01 -4.50225741e-01 3.59754503e-01
4.82345015e-01 -8.60194862e-01 5.70053875e-01 6.26508236e-01
7.16268942e-02 1.61485989e-02 -7.89815187e-01 7.76358902e-01
-2.07516253e-01 -1.12071645e+00 -3.66335101e-02 4.67611790e-01
1.00329208e+00 -2.42373481e-01 -5.28063357e-01 1.42959639e-01
9.08041418e-01 -9.34359312e-01 5.36008179e-01 2.49899954e-01
9.34902132e-01 -6.34227097e-01 8.88360322e-01 5.15970230e-01
-1.58305311e+00 -1.90950677e-01 -4.44426864e-01 1.93674833e-01
1.23419790e-02 8.12817395e-01 -7.21990943e-01 5.48548460e-01
5.72778940e-01 3.53919327e-01 -9.35803413e-01 1.51326728e+00
-1.70480087e-01 6.83109581e-01 -4.30413693e-01 -3.10330421e-01
-8.51097032e-02 -6.27802968e-01 9.29332852e-01 1.08383512e+00
1.93457812e-01 -2.91776299e-01 2.86401570e-01 9.10622537e-01
-1.90554276e-01 1.05579570e-03 -9.51416194e-01 -1.33393198e-01
9.28333580e-01 1.42636979e+00 -1.26195455e+00 -1.69842720e-01
5.44665158e-02 3.16699445e-01 2.95303375e-01 3.76912683e-01
-8.17709506e-01 -3.89761895e-01 1.18817173e-01 9.66610730e-01
1.78235725e-01 1.57031398e-02 2.77607858e-01 -1.11843491e+00
-3.59808385e-01 -1.16774523e+00 4.18336928e-01 -7.10874319e-01
-1.68337154e+00 9.88571167e-01 2.00435549e-01 -1.29065239e+00
1.14991978e-01 -3.95252913e-01 -8.76983404e-01 3.86762470e-01
-8.52445185e-01 -9.29198802e-01 -7.05748498e-01 5.58621943e-01
-3.87305379e-01 -2.10824326e-01 5.55166781e-01 1.31360009e-01
-5.41141093e-01 4.33910936e-01 2.14790061e-01 1.50718972e-01
3.38377535e-01 -1.55830729e+00 4.85652566e-01 8.38834703e-01
-1.60186276e-01 5.96042633e-01 9.41252112e-01 -7.52136588e-01
-1.05468500e+00 -1.26936328e+00 3.71298999e-01 -1.87708601e-01
9.48373437e-01 -7.33172238e-01 -7.46060252e-01 2.98194110e-01
1.79760620e-01 -4.63415682e-02 3.21909428e-01 -8.64940695e-03
-3.99598569e-01 5.83553612e-02 -1.27894187e+00 7.28453934e-01
1.42811704e+00 -1.49787456e-01 -2.66893748e-02 6.04860127e-01
7.26262152e-01 1.00544490e-01 -8.69343102e-01 4.46595460e-01
3.06239337e-01 -1.15275550e+00 8.91317964e-01 -2.79411614e-01
9.19563919e-02 -5.24420559e-01 2.50197917e-01 -1.26428628e+00
-8.41119230e-01 -9.71917570e-01 -4.72176820e-02 1.19578171e+00
3.98282617e-01 -8.25518727e-01 6.55236125e-01 -4.13793713e-01
4.85901326e-01 -4.80870306e-01 -7.05503047e-01 -9.41244245e-01
-2.84636207e-02 3.97519797e-01 8.09414744e-01 1.05154240e+00
-6.77177310e-02 6.87573254e-01 -7.43963420e-02 -1.36421807e-02
1.02731514e+00 2.24528790e-01 1.03005815e+00 -1.73325181e+00
-4.69655961e-01 -6.94719970e-01 -5.41200399e-01 -3.59229743e-01
-1.18048124e-01 -1.05320656e+00 -2.51843303e-01 -1.46176851e+00
7.97160983e-01 -5.25265694e-01 2.70035297e-01 2.58475728e-02
-1.56966463e-01 4.31319118e-01 8.57843757e-02 1.93392560e-01
-4.86589938e-01 1.31838024e-01 1.56714201e+00 -3.79297316e-01
-2.69762784e-01 8.34240690e-02 -5.44760823e-01 4.45273012e-01
7.65792847e-01 -6.80917859e-01 -4.64907944e-01 5.00598550e-01
7.04343796e-01 1.57851130e-01 3.53935868e-01 -8.07223082e-01
9.93124768e-02 -2.81329881e-02 -2.09618106e-01 -5.16140580e-01
-3.43637735e-01 -5.55261374e-01 9.98456538e-01 6.57254457e-01
-5.17186001e-02 4.30808276e-01 -1.92405492e-01 9.20030177e-01
-2.15613946e-01 -3.30749393e-01 8.86034250e-01 -3.22829098e-01
1.12213092e-02 3.45096499e-01 -7.22055197e-01 3.94827485e-01
1.22833431e+00 -5.38457155e-01 -7.23070681e-01 -6.58489048e-01
-6.55907512e-01 4.37981665e-01 9.48213160e-01 1.97181150e-01
3.76902759e-01 -1.29534662e+00 -1.08570063e+00 3.29277106e-02
3.54939938e-01 -1.01349592e-01 -1.90072209e-01 7.75490880e-01
-9.97058749e-01 -2.00855419e-01 -1.80558115e-02 -8.21534336e-01
-1.28088760e+00 8.86172295e-01 4.81433004e-01 -8.17859113e-01
-5.50361812e-01 2.24713683e-01 3.66159171e-01 -6.38204634e-01
7.45303780e-02 1.96979418e-01 -1.83322206e-01 -3.31437774e-02
4.07531440e-01 4.14644510e-01 -2.47116178e-01 -3.90727788e-01
-3.45189154e-01 3.59635055e-01 4.15095419e-01 1.86167479e-01
1.23089850e+00 -4.50223535e-02 -4.56593007e-01 3.88830662e-01
6.13963366e-01 3.56836945e-01 -5.83702981e-01 6.57427609e-02
3.96629982e-02 -4.22351301e-01 -4.81415957e-01 -4.50078070e-01
-1.14674520e+00 4.33203548e-01 2.86458969e-01 1.12839711e+00
8.86104107e-01 2.40782827e-01 -9.59916562e-02 8.79238099e-02
6.96486056e-01 -4.12101448e-01 2.54536808e-01 3.10294539e-01
1.01731384e+00 -6.53620601e-01 8.98828506e-02 -8.11374068e-01
-3.85462523e-01 8.26193511e-01 5.59535801e-01 -4.00154322e-01
7.96301961e-01 5.10307550e-01 -3.65834445e-01 -4.88948703e-01
-7.42104948e-01 -2.65599042e-01 6.08788766e-02 1.02380669e+00
1.46549344e-01 2.73372352e-01 -5.13866365e-01 2.68561006e-01
-1.90304995e-01 -3.11067313e-01 1.11791289e+00 3.68313849e-01
-4.14718598e-01 -1.08967471e+00 -5.42341411e-01 4.82784152e-01
5.82060665e-02 1.16966456e-01 -8.81272614e-01 1.16910660e+00
-1.13026567e-01 1.04428554e+00 -4.87712137e-02 -5.12975931e-01
4.17625755e-01 -5.36499858e-01 5.49154460e-01 -6.28562272e-01
-5.61130822e-01 -7.20917955e-02 5.78434132e-02 -1.63774461e-01
-2.58956283e-01 -2.58865654e-01 -6.76843703e-01 -1.02007568e+00
-8.93594980e-01 5.27781308e-01 -1.58131257e-01 1.59896433e-01
1.36257365e-01 3.53010774e-01 8.25422049e-01 -5.91401219e-01
-2.66767234e-01 -1.14806497e+00 -1.24122691e+00 5.70668697e-01
-1.03178978e-01 -7.94989884e-01 -7.81809509e-01 -3.11508834e-01] | [6.958712577819824, 5.252319812774658] |
197bf601-8385-4b0c-817a-3cfb66644a79 | topic-detection-in-continuous-sign-language | 2209.02402 | null | https://arxiv.org/abs/2209.02402v1 | https://arxiv.org/pdf/2209.02402v1.pdf | Topic Detection in Continuous Sign Language Videos | Significant progress has been made recently on challenging tasks in automatic sign language understanding, such as sign language recognition, translation and production. However, these works have focused on datasets with relatively few samples, short recordings and limited vocabulary and signing space. In this work, we introduce the novel task of sign language topic detection. We base our experiments on How2Sign, a large-scale video dataset spanning multiple semantic domains. We provide strong baselines for the task of topic detection and present a comparison between different visual features commonly used in the domain of sign language. | ['Xavier Giro-i-Nieto', 'Jordi Torres', 'Francesc Moreno-Noguer', 'Gerard I. Gallego', 'Laia Tarres', 'Alvaro Budria'] | 2022-09-01 | null | null | null | null | ['sign-language-recognition'] | ['computer-vision'] | [ 2.55130261e-01 -3.59375596e-01 -5.27705312e-01 -3.67907822e-01
-8.45871031e-01 -5.34065127e-01 8.98169756e-01 -6.35002851e-01
-3.93751889e-01 4.66700464e-01 9.64049459e-01 -9.84750502e-03
3.35444421e-01 -7.31595606e-02 -4.56578881e-01 -4.71648425e-01
1.00870669e-01 1.84800848e-01 5.82519710e-01 8.40165317e-02
2.74364054e-01 9.54036936e-02 -1.89992249e+00 4.04916108e-01
5.80210865e-01 6.39972270e-01 -1.96969971e-01 6.26425982e-01
-2.07701117e-01 8.90237212e-01 -7.13902831e-01 -1.04282416e-01
1.24445908e-01 -8.05283129e-01 -5.26366830e-01 1.53349206e-01
1.26868999e+00 -4.60361183e-01 -7.23054171e-01 8.98877442e-01
8.24365199e-01 -1.51077509e-01 8.15466106e-01 -1.45065320e+00
-5.01351714e-01 4.66853589e-01 -2.62091011e-01 -4.81002629e-02
4.41369951e-01 3.91978920e-01 9.97634590e-01 -9.26403463e-01
1.31500483e+00 1.24214172e+00 4.46354270e-01 1.06826961e+00
-5.12268603e-01 -9.50753689e-01 4.79437947e-01 3.78896058e-01
-1.08828402e+00 -4.85375106e-01 7.97453225e-01 -7.51369298e-01
7.29613423e-01 -1.25800267e-01 9.00704026e-01 1.53809643e+00
-6.06013477e-01 1.56306851e+00 1.42627490e+00 -4.95515227e-01
6.02034442e-02 -2.95136720e-01 4.42971200e-01 6.46566451e-01
4.04013038e-01 -9.21845585e-02 -1.12766385e+00 -7.89211318e-03
7.43859410e-01 -3.12206805e-01 -4.66076136e-01 -4.33178395e-01
-1.24777639e+00 6.99688554e-01 2.21608579e-02 2.83570439e-01
-1.80385724e-01 5.10252297e-01 6.41298532e-01 3.51477027e-01
1.39238611e-01 1.20004028e-01 -2.78654516e-01 -6.18353903e-01
-8.35307300e-01 4.23926979e-01 9.58459437e-01 1.02585447e+00
-2.18761310e-01 -6.47257045e-02 -3.68866235e-01 6.94404542e-01
5.66658616e-01 8.89529228e-01 6.46852553e-01 -6.02018893e-01
5.39547801e-01 3.89773130e-01 -8.16431046e-02 -2.99117029e-01
-7.47874305e-02 3.31911534e-01 -1.09350137e-01 3.74633998e-01
6.50154948e-01 1.78770255e-03 -1.67951596e+00 1.51349366e+00
1.12971194e-01 3.35546046e-01 -1.98171418e-02 1.14674711e+00
1.28946781e+00 5.90755418e-02 3.62767100e-01 3.39834899e-01
1.54743648e+00 -9.13831890e-01 -8.23290408e-01 -1.44908696e-01
4.47770715e-01 -8.85761321e-01 1.21921253e+00 4.92832035e-01
-5.97271204e-01 -2.58076075e-03 -7.63005614e-01 -3.38773608e-01
-4.94134218e-01 3.74600232e-01 6.36103928e-01 5.55314004e-01
-7.44712412e-01 -6.83773383e-02 -8.45222831e-01 -1.09746993e+00
8.52554321e-01 -1.41623408e-01 -2.90412635e-01 -3.52484375e-01
-7.80842662e-01 8.80222976e-01 1.87434978e-03 -1.66575238e-01
-8.07164073e-01 -4.18173254e-01 -7.75871813e-01 -7.33287036e-01
2.86193579e-01 -6.54867589e-01 1.45990205e+00 -7.67839432e-01
-1.62526429e+00 1.47005367e+00 -3.85460198e-01 -3.24681640e-01
8.86817336e-01 -6.66631162e-01 -4.68233228e-01 8.70284513e-02
-1.25915900e-01 6.32367015e-01 1.11072671e+00 -1.06238437e+00
-6.92081273e-01 -2.72610158e-01 -2.15496838e-01 4.93200906e-02
-7.73387253e-02 9.46547806e-01 -5.35800338e-01 -9.82136846e-01
2.05538258e-01 -9.57114697e-01 1.40097141e-01 4.36544538e-01
-2.48697743e-01 -5.27562797e-01 1.14773393e+00 -8.30940962e-01
1.04327333e+00 -2.29055166e+00 4.43278998e-02 9.75251850e-03
1.84374481e-01 4.30481642e-01 -1.60667449e-01 3.37452233e-01
3.47724557e-01 -2.04266999e-02 -2.05379888e-01 -2.67785847e-01
2.44607151e-01 3.05225670e-01 -7.39672422e-01 4.79005188e-01
6.74682707e-02 1.04322386e+00 -8.55983138e-01 -6.64446414e-01
1.88156500e-01 4.97317612e-01 -3.85135561e-01 -2.62825601e-02
-4.06023115e-01 3.24326545e-01 -5.58785737e-01 1.20634925e+00
2.41521820e-01 -1.25952125e-01 -8.31316411e-03 -5.60655929e-02
-1.38117746e-01 4.58457619e-01 -9.13109422e-01 1.77573550e+00
-2.74062514e-01 1.43821406e+00 8.64119157e-02 -5.85451961e-01
5.10108054e-01 4.62182254e-01 4.03699845e-01 -2.34579816e-01
2.42867902e-01 5.31753004e-01 -6.27937391e-02 -8.17742825e-01
2.75912136e-01 -7.07320822e-03 3.95346060e-02 4.69002128e-01
3.80168669e-02 -3.05388361e-01 5.12409508e-01 5.14894500e-02
1.09565425e+00 3.25162083e-01 2.77773082e-01 1.57986477e-01
2.54395097e-01 1.96045429e-01 7.12767541e-02 6.20324731e-01
-7.15843499e-01 8.61013353e-01 1.82294056e-01 -1.10897295e-01
-6.61039770e-01 -1.19117320e+00 -1.51132718e-01 1.01388466e+00
1.38660580e-01 -7.23620653e-01 -2.95239449e-01 -1.03282022e+00
2.66587794e-01 2.87521511e-01 -3.86946142e-01 4.00786817e-01
-1.10831606e+00 -2.41471007e-01 8.91498804e-01 8.33229542e-01
5.94442606e-01 -1.30622661e+00 -7.50772417e-01 -2.67784804e-01
-4.73865569e-02 -1.37322080e+00 -6.47541404e-01 -4.47625995e-01
-7.73321867e-01 -1.38379633e+00 -1.13972199e+00 -1.19979167e+00
5.16101122e-01 8.57561752e-02 9.01121855e-01 -2.33820245e-01
-7.14558721e-01 7.73665488e-01 -4.51765597e-01 -9.33230639e-01
-2.35015392e-01 -2.64597505e-01 -2.07676411e-01 -1.68623254e-01
7.85450459e-01 -2.53727823e-01 -4.34540659e-01 3.39818984e-01
-6.42882228e-01 -1.45373553e-01 5.73292077e-01 8.67666483e-01
2.23752141e-01 -9.58389461e-01 1.51209339e-01 -2.94565260e-01
5.75226247e-01 1.72913019e-02 -4.81590748e-01 3.15231532e-01
-1.19128250e-01 5.23731932e-02 -1.38884276e-01 -6.47314906e-01
-8.68352413e-01 -1.08825468e-01 8.77984092e-02 -2.58610934e-01
-5.00426888e-01 1.99021235e-01 -5.29111847e-02 -1.85239926e-01
3.94028783e-01 3.95411074e-01 1.61666036e-01 -6.92710817e-01
5.91982126e-01 8.95574510e-01 3.69568288e-01 -2.72909015e-01
6.48072362e-01 9.50052559e-01 -2.97668040e-01 -1.21128404e+00
-5.55580795e-01 -8.36832821e-01 -6.49775684e-01 -4.50047851e-01
7.46321023e-01 -1.02225053e+00 -3.55532229e-01 9.79457915e-01
-1.03775883e+00 -3.25635940e-01 -5.76809764e-01 7.69093394e-01
-6.30033493e-01 5.00734448e-01 -2.44770557e-01 -7.32175112e-01
-7.64905736e-02 -7.82577813e-01 1.52018070e+00 3.29161286e-02
-3.95966321e-01 -5.39035261e-01 3.28255683e-01 3.57430995e-01
3.74404192e-01 2.23594517e-01 4.74817246e-01 -4.57511127e-01
-9.23069656e-01 -2.54116029e-01 -3.27515513e-01 4.23456728e-01
1.38448551e-01 6.55900761e-02 -1.04891300e+00 -8.58636275e-02
-7.32158899e-01 -7.75014758e-01 1.51071143e+00 5.34882784e-01
7.63579249e-01 6.46335557e-02 -5.01540124e-01 6.15271926e-01
9.85887766e-01 9.11836475e-02 3.25003356e-01 2.38627881e-01
6.55521333e-01 5.55013895e-01 5.47567368e-01 2.46953577e-01
5.07450223e-01 7.46536255e-01 -2.40900934e-01 -6.14303984e-02
-1.02050710e+00 -5.27742624e-01 6.17206693e-01 7.95602739e-01
-4.36700940e-01 -2.86907017e-01 -9.66728866e-01 9.31448221e-01
-2.04347920e+00 -9.58052933e-01 1.42473817e-01 1.71832550e+00
7.57354140e-01 -3.09785903e-01 3.96729261e-01 -4.48295362e-02
3.69153857e-01 5.37980974e-01 -4.66915727e-01 7.25257993e-02
-5.46236098e-01 1.95001617e-01 4.83932763e-01 1.67622447e-01
-1.63866103e+00 1.29634166e+00 7.29412889e+00 3.78690243e-01
-1.34910715e+00 6.48110285e-02 -4.87957269e-01 -3.31547230e-01
8.09331536e-02 -1.24955982e-01 -8.61089408e-01 2.96888798e-01
2.29988635e-01 -1.00266635e-01 -1.54195324e-01 8.94190311e-01
1.20854415e-01 -2.39646882e-02 -1.12666857e+00 1.29755902e+00
7.47813106e-01 -9.43371177e-01 1.60272643e-01 1.95370615e-02
8.45574498e-01 5.18093050e-01 -5.30496463e-02 7.83891305e-02
6.25454545e-01 -9.58138824e-01 8.44006002e-01 4.28132474e-01
9.18672562e-01 1.64097458e-01 4.42795336e-01 -1.93725333e-01
-1.16580129e+00 8.51818770e-02 3.77353638e-01 1.17753781e-01
7.09419549e-01 -8.12535882e-02 -5.92258275e-01 -5.16441986e-02
7.41702437e-01 1.26906443e+00 -3.23194593e-01 1.44792140e+00
-7.58432090e-01 1.04237258e+00 -7.35752821e-01 -3.91778260e-01
1.01839341e-01 1.09439135e-01 8.51738334e-01 1.32249379e+00
1.67403176e-01 -5.63316531e-02 4.71465290e-01 3.36447328e-01
-5.38473688e-02 3.62105578e-01 -7.42290914e-01 -3.84995759e-01
-2.31360625e-02 3.15540373e-01 -5.07440269e-01 -6.66849077e-01
-7.39618480e-01 9.01087046e-01 -2.70859510e-01 5.17583728e-01
-6.18578255e-01 -3.23440343e-01 1.23491859e+00 5.87045662e-02
4.59559858e-01 -5.53517580e-01 -2.59466588e-01 -1.61005735e+00
4.24792260e-01 -8.47009063e-01 5.16264200e-01 -4.49021697e-01
-1.40254843e+00 -9.70400311e-03 6.92565143e-02 -1.42066264e+00
-3.17466676e-01 -1.12756360e+00 -5.61095476e-01 4.06759828e-01
-1.81011426e+00 -1.55289268e+00 -5.32682180e-01 7.33946085e-01
8.57420981e-01 -3.53619605e-01 5.42646587e-01 3.26233715e-01
-6.07432015e-02 4.52487975e-01 6.97633019e-03 4.98301655e-01
1.16651928e+00 -9.76519644e-01 4.19928640e-01 8.38243484e-01
5.27122021e-01 4.08172518e-01 5.72567165e-01 -7.73213387e-01
-1.18100214e+00 -6.51702404e-01 1.19692910e+00 -7.80487418e-01
8.69302213e-01 -4.01661783e-01 -2.81343162e-01 6.66128933e-01
8.94521922e-03 -1.36453658e-02 4.16628450e-01 -3.49545181e-02
-7.02460349e-01 4.57477540e-01 -7.26362944e-01 7.24531651e-01
1.74427056e+00 -7.70530522e-01 -8.87354076e-01 5.58027267e-01
1.02682285e-01 -2.93361992e-01 -3.65587413e-01 3.93624395e-01
1.22897661e+00 -2.23296583e-01 9.39104259e-01 -1.06184506e+00
3.78828138e-01 -4.28112119e-01 -8.94290507e-02 -8.19733083e-01
3.85120243e-01 -4.27741081e-01 -3.00871313e-01 9.16155398e-01
6.12751693e-02 -4.86373216e-01 8.97735536e-01 3.79000098e-01
6.53345138e-02 -1.74738362e-01 -9.26379323e-01 -1.17138541e+00
-1.19586870e-01 -5.84949970e-01 1.04742572e-01 6.95903182e-01
5.25989011e-02 -1.98545381e-01 -4.95146424e-01 -3.73913258e-01
8.04646790e-01 4.92884129e-01 1.22309077e+00 -1.38241875e+00
1.53777236e-02 -6.09022200e-01 -9.09182787e-01 -1.37197638e+00
1.60211235e-01 -7.36851811e-01 3.86366069e-01 -1.71389043e+00
9.36266854e-02 2.90656835e-01 -1.43326789e-01 6.13852024e-01
4.74411994e-02 3.42864007e-01 4.54961717e-01 4.18790072e-01
-7.29396462e-01 4.88560706e-01 1.25788844e+00 -3.45997423e-01
-1.03953019e-01 -1.79259479e-01 -2.33953834e-01 1.13252294e+00
6.47382677e-01 -1.41198128e-01 1.75034776e-02 -4.75255728e-01
-4.55472499e-01 -8.05030644e-01 5.84856153e-01 -8.00253510e-01
2.15456098e-01 -2.44777258e-02 -2.98539549e-01 -7.70433664e-01
3.78521889e-01 -4.05532241e-01 -7.33024776e-01 4.67511415e-01
-3.77852798e-01 -3.89733523e-01 3.07601150e-02 5.98706543e-01
-7.44556785e-01 2.29667053e-01 5.68458021e-01 -1.16502509e-01
-1.48000240e+00 7.97365457e-02 -4.70046580e-01 5.52840054e-01
9.44378197e-01 -3.19600016e-01 -3.47671688e-01 -4.38212812e-01
-7.46357739e-01 2.40305230e-01 3.81254494e-01 9.69669938e-01
7.99855351e-01 -1.22555327e+00 -8.52317750e-01 1.99057728e-01
7.31630862e-01 -5.24267137e-01 -1.32635921e-01 7.70432711e-01
-5.27051628e-01 6.27238154e-01 -1.50647745e-01 -6.26468897e-01
-2.08486938e+00 -1.91084281e-01 1.42132742e-02 2.17707083e-01
-1.06411624e+00 9.70080435e-01 1.07771359e-01 -8.94793347e-02
7.50176787e-01 -9.32491362e-01 -2.18357086e-01 1.00990787e-01
4.68745738e-01 2.43241012e-01 -5.61838269e-01 -7.31492817e-01
-6.54042184e-01 1.10631430e+00 1.67680562e-01 -3.67066115e-01
1.00590682e+00 2.27908954e-01 2.07381457e-01 5.38233280e-01
9.54845190e-01 -3.91414538e-02 -1.03783643e+00 -4.23292726e-01
3.55033726e-01 -6.44876420e-01 -1.71552286e-01 -9.27340746e-01
-6.59533858e-01 8.16457093e-01 7.56502688e-01 -6.38615310e-01
8.41585457e-01 5.65831542e-01 9.02458727e-01 5.49205124e-01
4.65171486e-01 -1.45457089e+00 -2.23618764e-02 7.53203213e-01
1.20015264e+00 -1.60572541e+00 -1.27733752e-01 -4.69864041e-01
-6.40400171e-01 9.70667422e-01 3.75434428e-01 -1.52293578e-01
1.01275146e+00 1.64303869e-01 7.83302128e-01 -5.01002133e-01
-3.90918970e-01 -8.55650425e-01 7.25081503e-01 7.05161035e-01
5.49821615e-01 1.56168953e-01 -6.76725268e-01 2.54002035e-01
-2.50824183e-01 5.20698905e-01 2.49046952e-01 1.32940733e+00
-3.86551201e-01 -1.05422175e+00 -2.68999308e-01 3.30666542e-01
-2.02069044e-01 -4.86005060e-02 -1.01649094e+00 1.03734684e+00
-4.00491524e-03 5.54073155e-01 -1.29878968e-01 1.17144018e-01
5.09258628e-01 4.87983972e-01 8.48475933e-01 -4.78822678e-01
-1.05801202e-01 -8.98765996e-02 4.14467454e-01 -6.85700059e-01
-8.79107833e-01 -1.30107045e+00 -1.09435070e+00 4.06160682e-01
-9.47397649e-02 -4.81936097e-01 8.82381201e-01 1.00943065e+00
1.72048748e-01 1.69413209e-01 -2.90877819e-01 -4.72407520e-01
-4.86103207e-01 -1.10369098e+00 -6.69893265e-01 7.26769984e-01
4.17805165e-01 -8.64452302e-01 -3.47995520e-01 4.53608692e-01] | [9.154878616333008, -6.467281818389893] |
70c903e4-3e48-4d5e-b6f8-fd9d0af13920 | dont-just-scratch-the-surface-enhancing-word | 1908.09282 | null | https://arxiv.org/abs/1908.09282v3 | https://arxiv.org/pdf/1908.09282v3.pdf | Don't Just Scratch the Surface: Enhancing Word Representations for Korean with Hanja | We propose a simple yet effective approach for improving Korean word representations using additional linguistic annotation (i.e. Hanja). We employ cross-lingual transfer learning in training word representations by leveraging the fact that Hanja is closely related to Chinese. We evaluate the intrinsic quality of representations learned through our approach using the word analogy and similarity tests. In addition, we demonstrate their effectiveness on several downstream tasks, including a novel Korean news headline generation task. | ['Sang-goo Lee', 'Kang Min Yoo', 'Taeuk Kim'] | 2019-08-25 | dont-just-scratch-the-surface-enhancing-word-1 | https://aclanthology.org/D19-1358 | https://aclanthology.org/D19-1358.pdf | ijcnlp-2019-11 | ['headline-generation'] | ['natural-language-processing'] | [-1.13991179e-01 -7.25862607e-02 -5.62578559e-01 -3.84317040e-01
-1.48083401e+00 -6.29606307e-01 4.98441368e-01 7.19767958e-02
-9.16675150e-01 1.08208489e+00 1.07982516e+00 -4.18665469e-01
4.02788401e-01 -8.22012722e-01 -6.82424843e-01 -1.54105335e-01
1.98106840e-01 1.33959249e-01 -2.50858545e-01 -7.64558017e-01
2.34338805e-01 -4.46517430e-02 -4.21409011e-01 4.61531371e-01
1.25802350e+00 3.73679668e-01 2.74403363e-01 4.42726791e-01
9.01631936e-02 3.18297416e-01 -7.94685364e-01 -5.30235708e-01
-7.21043488e-03 -5.94561458e-01 -8.28244150e-01 -3.71385425e-01
2.58104622e-01 -3.35585177e-01 -1.96622103e-01 8.68733525e-01
5.99007785e-01 5.89225709e-01 6.04394734e-01 -3.28205198e-01
-1.65032923e+00 1.32039881e+00 -2.65790194e-01 3.97836357e-01
4.14840847e-01 -2.10627601e-01 1.51518774e+00 -1.18893206e+00
7.04880416e-01 1.12539792e+00 8.27648878e-01 8.66103470e-01
-1.09465873e+00 -7.73056030e-01 5.68543434e-01 2.19236106e-01
-1.24413574e+00 -3.58925074e-01 7.50311673e-01 -3.16281505e-02
1.29178917e+00 -2.83855110e-01 6.90986738e-02 1.41813803e+00
-1.17681526e-01 7.48688400e-01 8.44964206e-01 -7.78241396e-01
-7.66971335e-02 6.53065145e-02 3.63770902e-01 6.01250172e-01
3.03530037e-01 -1.41925246e-01 -5.45721114e-01 -4.02677841e-02
4.76494044e-01 -4.06325519e-01 -5.33256829e-01 5.21911144e-01
-1.26026821e+00 1.30957353e+00 6.21591210e-01 5.84709346e-01
-2.12583303e-01 2.75847226e-01 6.71828806e-01 1.06790036e-01
7.18507886e-01 8.39622915e-01 -9.02597487e-01 -2.51952391e-02
-4.42516983e-01 -2.07492098e-01 3.78596097e-01 1.00110042e+00
5.72702527e-01 3.11374575e-01 -2.02872902e-01 1.12289035e+00
2.34790593e-01 4.24050272e-01 1.28299868e+00 -4.47048038e-01
6.90312028e-01 -3.51183414e-02 1.57286432e-02 -4.58322823e-01
-5.58941066e-02 -2.44522467e-01 -4.75620478e-01 -5.17766058e-01
9.56186354e-02 -5.42371809e-01 -9.17648971e-01 2.13462138e+00
-2.94256747e-01 -6.85247034e-02 6.64437175e-01 6.40100956e-01
7.22909629e-01 1.07257581e+00 4.58891779e-01 -2.80528784e-01
1.13997483e+00 -1.32560551e+00 -1.02396810e+00 -2.31008902e-01
1.11532235e+00 -7.18607545e-01 1.47356820e+00 9.00076106e-02
-8.11966896e-01 -7.04128206e-01 -1.00046313e+00 -3.59200448e-01
-6.22586012e-01 4.78305489e-01 7.61926234e-01 5.66788197e-01
-7.68948913e-01 5.11466205e-01 -6.25610173e-01 -4.01243985e-01
3.91301438e-02 -4.13675942e-02 -4.36706096e-01 -2.37345323e-01
-1.69942713e+00 1.06966007e+00 7.77475417e-01 -6.65874109e-02
-5.09266734e-01 -7.95476079e-01 -1.22666442e+00 -8.25945064e-02
4.90888543e-02 -5.70673704e-01 1.33055961e+00 -9.08262849e-01
-1.64832306e+00 5.91527760e-01 -2.52893031e-01 -4.85237330e-01
-7.48576596e-02 -8.90484095e-01 -7.33529031e-01 -7.58775175e-02
1.05203144e-01 6.19700968e-01 3.15410912e-01 -1.08050597e+00
-2.25645840e-01 -6.82068244e-02 4.62294184e-02 2.51271397e-01
-7.63994277e-01 1.25169754e-01 -2.12969452e-01 -1.41161394e+00
-4.82022494e-01 -7.46936083e-01 -3.61545086e-01 -5.59600770e-01
-4.17967409e-01 -5.99462926e-01 2.50599027e-01 -1.18535340e+00
1.50480127e+00 -2.06581593e+00 7.11242110e-02 6.04358362e-03
-4.19612139e-01 3.88449579e-01 -6.32626414e-01 6.76678002e-01
2.55817454e-02 6.40579283e-01 -1.08773842e-01 -2.07094073e-01
-1.29361778e-01 3.29709649e-01 -6.90033793e-01 -8.49281773e-02
4.10705209e-01 1.31842458e+00 -1.08417928e+00 -1.50644571e-01
-4.49914515e-01 4.25922722e-01 -8.02966654e-01 1.49094507e-01
-2.18749478e-01 4.47016321e-02 -3.87031078e-01 3.79038453e-01
2.34878108e-01 1.47015639e-02 4.75312263e-01 -2.86578387e-01
7.63244852e-02 1.04129839e+00 -2.53133118e-01 2.01911378e+00
-9.71796572e-01 4.05120850e-01 -6.14254653e-01 -7.19278932e-01
1.06644404e+00 3.57619673e-01 -2.18667328e-01 -7.19985127e-01
-1.18675396e-01 2.60171890e-01 9.39562991e-02 -1.05077915e-01
9.80291188e-01 -4.78322089e-01 -4.09978747e-01 6.55674517e-01
3.56503129e-01 8.25554952e-02 8.04173201e-02 3.49121571e-01
6.39865696e-01 3.91721040e-01 7.80444503e-01 -3.93267930e-01
2.63767362e-01 -2.34616026e-02 4.83079940e-01 4.19708014e-01
1.32821605e-01 5.03350019e-01 9.83567536e-02 -1.13918856e-01
-7.28601813e-01 -1.10977805e+00 1.72221586e-01 1.39373410e+00
-2.02536151e-01 -8.73895049e-01 -6.19628191e-01 -1.14712286e+00
-1.84806973e-01 1.31579101e+00 -7.82687545e-01 -5.04323840e-01
-9.02556419e-01 -5.52641869e-01 7.62956500e-01 1.19564867e+00
4.42069434e-02 -1.12587965e+00 1.63837999e-01 3.36727560e-01
-2.48490945e-01 -1.14660895e+00 -1.03259230e+00 1.14315420e-01
-8.63194227e-01 -5.32652259e-01 -8.26504171e-01 -1.05109882e+00
5.88215649e-01 4.54328179e-01 1.12921596e+00 4.35777418e-02
1.29275948e-01 3.74155909e-01 -7.78899968e-01 -2.17368677e-01
-3.28000367e-01 4.92461592e-01 1.22045942e-01 -5.90683162e-01
4.27924961e-01 -9.57428142e-02 -1.30960032e-01 -2.02508435e-01
-7.73796380e-01 -1.83145747e-01 6.11743093e-01 1.05687249e+00
4.65865940e-01 -5.61305940e-01 1.12059116e+00 -1.17442501e+00
1.19935954e+00 -6.16360724e-01 -1.93189695e-01 4.21024173e-01
-6.18181527e-01 2.69207984e-01 7.83597171e-01 -6.23030543e-01
-1.32896423e+00 -3.38774264e-01 -4.44029629e-01 1.56843424e-01
1.00021563e-01 1.14304876e+00 -1.55663222e-01 5.00354290e-01
6.21114135e-01 9.89110097e-02 -3.51557195e-01 -7.38375306e-01
1.04157019e+00 3.57753158e-01 3.22777659e-01 -6.87406659e-01
7.80464351e-01 -2.68452391e-02 -8.53836775e-01 -6.93204701e-01
-1.24576151e+00 -1.61841124e-01 -6.14617348e-01 3.41875881e-01
1.07640576e+00 -1.15829539e+00 -4.90135141e-02 -4.06776351e-04
-1.59051645e+00 -3.99877876e-01 -2.67945796e-01 7.53355384e-01
-1.80245459e-01 1.85490593e-01 -1.04104710e+00 -3.09353203e-01
-5.59175551e-01 -5.28585255e-01 8.77842784e-01 1.05688917e-02
-5.41799366e-01 -1.41656435e+00 4.10691917e-01 2.11889490e-01
4.78151530e-01 -2.00150639e-01 1.34695375e+00 -1.02010632e+00
-4.58897948e-02 -1.23350851e-01 -1.44293711e-01 6.02818906e-01
3.65426898e-01 -1.81418806e-01 -7.62126207e-01 -3.22080702e-01
-5.36446571e-01 -6.67487562e-01 1.11416340e+00 -3.45308892e-02
9.63277876e-01 -5.16118050e-01 -1.31833360e-01 3.73100847e-01
1.50823843e+00 1.11384176e-01 4.60069418e-01 1.51736215e-01
7.09774017e-01 4.68279809e-01 6.39469802e-01 1.60591975e-01
4.73481357e-01 6.00309670e-01 -3.30561817e-01 -6.92521334e-02
-3.40350091e-01 -8.24266016e-01 8.08389604e-01 1.77210677e+00
-1.75411791e-01 -4.12098080e-01 -6.56926036e-01 6.80719554e-01
-1.59993017e+00 -5.94857752e-01 2.42128551e-01 1.85391009e+00
1.19258368e+00 -4.70995903e-04 -2.39878431e-01 -4.77028131e-01
4.74765867e-01 1.10273257e-01 -8.72949660e-02 -5.95638692e-01
-2.46474653e-01 6.68336689e-01 3.73607278e-01 7.66719639e-01
-8.34013760e-01 1.69657314e+00 7.10383892e+00 7.70734429e-01
-9.72195745e-01 4.16177392e-01 5.13300240e-01 3.58233392e-01
-8.10381234e-01 -1.75829276e-01 -1.05047119e+00 8.67488831e-02
9.57712710e-01 -5.60392976e-01 1.67355433e-01 8.68768632e-01
1.51032945e-02 2.85896689e-01 -1.00350332e+00 4.70430017e-01
5.76082051e-01 -1.46047258e+00 6.64086044e-01 -2.42889449e-01
1.00528169e+00 -2.46497188e-02 -1.44088855e-02 7.36434937e-01
5.31513631e-01 -9.71913934e-01 3.38575453e-01 3.47077161e-01
8.06500196e-01 -9.06341076e-01 8.00322533e-01 -6.02518655e-02
-1.00249541e+00 2.86669761e-01 -4.97633547e-01 7.91663751e-02
4.58162010e-01 1.44475952e-01 -7.97333598e-01 6.79551303e-01
2.19108507e-01 8.73387575e-01 -5.83586693e-01 3.72280449e-01
-9.05131459e-01 9.94301260e-01 2.67299682e-01 -1.83110729e-01
3.26714873e-01 -1.13647789e-01 1.63021520e-01 1.70681214e+00
5.99858880e-01 1.87790647e-01 1.46152839e-01 8.18507075e-01
-5.47384202e-01 5.46364486e-01 -8.01055431e-01 -3.15071762e-01
4.04257029e-01 1.05108285e+00 -2.36615241e-01 -5.55927396e-01
-4.36250895e-01 1.21367383e+00 7.45248497e-01 5.71781874e-01
-7.04031050e-01 -5.55844188e-01 8.51790071e-01 -3.68576467e-01
5.62653661e-01 -7.29640543e-01 -6.08411953e-02 -1.25660503e+00
-1.94930911e-01 -7.96580136e-01 4.43682581e-01 -8.71405482e-01
-1.84543145e+00 9.16966856e-01 -9.62551609e-02 -1.06620347e+00
-5.75234950e-01 -8.53114665e-01 -8.02010894e-01 9.57462370e-01
-1.85038662e+00 -1.29648292e+00 1.63324475e-01 5.84108949e-01
8.17487359e-01 -3.13967913e-01 1.32507515e+00 2.22883910e-01
-4.97186780e-01 1.00520873e+00 -1.59239367e-01 4.13635224e-01
1.08021939e+00 -1.08144629e+00 6.05902433e-01 7.94197083e-01
4.75944430e-01 1.00297534e+00 3.49478461e-02 -7.02621043e-01
-1.01518512e+00 -1.16317785e+00 1.32431197e+00 -3.08144778e-01
9.64698493e-01 -4.20537412e-01 -1.05264688e+00 8.69050682e-01
8.22724998e-01 -1.91510811e-01 1.30092382e+00 4.26809341e-01
-8.54486763e-01 1.56509221e-01 -6.07297778e-01 7.80295372e-01
9.55796599e-01 -7.93266952e-01 -1.16090989e+00 2.20649272e-01
1.56051552e+00 2.09130511e-01 -6.46913469e-01 1.95667386e-01
2.39891425e-01 -1.77207708e-01 7.35703230e-01 -1.41339779e+00
5.48863232e-01 6.45730272e-02 -4.30606514e-01 -2.01235437e+00
-5.50719798e-01 -5.80490351e-01 1.00730345e-01 1.40538192e+00
9.36364412e-01 -5.58451653e-01 3.36193204e-01 9.00443494e-02
-4.34525192e-01 -4.39711273e-01 -6.58964217e-01 -1.06202543e+00
6.55920863e-01 -3.01519841e-01 2.59585857e-01 1.43111539e+00
2.66705722e-01 8.48727584e-01 -3.03969353e-01 -4.88074571e-02
-5.07317819e-02 -1.44668236e-01 2.22769514e-01 -6.93073332e-01
-3.73813540e-01 -2.50749677e-01 1.58542961e-01 -1.18658721e+00
7.67872572e-01 -1.39221179e+00 2.57601261e-01 -1.38171029e+00
2.93796286e-02 -2.76595592e-01 -7.02734768e-01 5.72150946e-01
-5.62717259e-01 3.63889009e-01 2.96711206e-01 -1.18532382e-01
-4.37731206e-01 7.95600414e-01 9.58270431e-01 1.08452057e-02
-3.72515060e-02 -4.47994709e-01 -9.99128163e-01 5.90370774e-01
1.16878557e+00 -3.94154817e-01 -1.25301093e-01 -9.32521760e-01
2.40150899e-01 -4.24604297e-01 -2.97136784e-01 -4.49472725e-01
-1.76783830e-01 -4.55411933e-02 2.85796762e-01 -2.09897265e-01
2.81782508e-01 -2.82058746e-01 -7.26044595e-01 3.09453428e-01
-7.81327188e-01 5.81163466e-01 5.22699952e-01 5.13300002e-01
-4.22193944e-01 -3.71113837e-01 5.11100829e-01 -1.78500473e-01
-1.01094091e+00 -5.29059842e-02 -4.97997284e-01 3.64906788e-01
7.07006574e-01 2.51422554e-01 -4.58715230e-01 -5.77173114e-01
-4.33480859e-01 -8.30301493e-02 6.58628419e-02 8.71865153e-01
8.65802348e-01 -1.66810954e+00 -8.77492368e-01 1.23358563e-01
4.28593338e-01 -7.99022198e-01 -2.40961581e-01 3.03108662e-01
-9.66863409e-02 4.14903432e-01 -1.43956497e-01 3.09329599e-01
-9.06011403e-01 6.54752672e-01 -1.06853954e-01 -4.11634415e-01
-2.15209454e-01 9.12397921e-01 3.50962989e-02 -5.76783597e-01
-5.58150820e-02 -4.43841696e-01 -3.56761813e-01 1.23451129e-01
5.92897296e-01 2.27863982e-01 -1.31612331e-01 -4.47730362e-01
-2.62954354e-01 3.98477197e-01 -4.05656785e-01 -3.82414877e-01
1.29767406e+00 4.22394462e-02 -6.41792342e-02 5.58264494e-01
1.37898350e+00 6.74533248e-01 -4.97717351e-01 -2.96970338e-01
5.00382781e-01 -1.16911925e-01 -1.63022086e-01 -9.04011071e-01
-7.56935179e-01 8.51970434e-01 1.55111089e-01 -3.16986173e-01
8.06285381e-01 1.37829319e-01 1.20608616e+00 6.39636993e-01
2.18129590e-01 -1.29588628e+00 2.64635384e-01 9.72200215e-01
1.09234250e+00 -1.25008798e+00 -3.95442471e-02 -4.08117652e-01
-9.83124614e-01 1.15551424e+00 7.00787902e-01 -3.66047889e-01
5.34247935e-01 1.30532039e-02 4.02302533e-01 3.18458110e-01
-8.90092015e-01 -1.79651782e-01 4.97792184e-01 7.30687976e-01
1.11802483e+00 2.22677618e-01 -8.25897217e-01 1.11530447e+00
-2.21009880e-01 -3.09864849e-01 4.88437682e-01 5.64354241e-01
-2.77578712e-01 -1.38649738e+00 1.19664550e-01 1.66477770e-01
-5.35419762e-01 -8.29924166e-01 -5.39939046e-01 8.34087074e-01
-2.03296423e-01 6.89013481e-01 -6.07188698e-03 -3.54776591e-01
2.65824616e-01 2.52339512e-01 1.88460723e-01 -1.16765928e+00
-5.91673255e-01 8.64276588e-02 4.99769241e-01 -3.48880351e-01
-2.89790064e-01 -1.70546964e-01 -1.37985528e+00 2.87139267e-01
-3.31419349e-01 4.27588284e-01 5.43903172e-01 7.67939150e-01
4.08667207e-01 4.65513587e-01 4.84321892e-01 -6.18867278e-01
-7.69372880e-01 -1.31928849e+00 -3.50261241e-01 4.86197203e-01
-5.27020963e-03 -1.70466602e-01 -1.60340577e-01 1.46432430e-01] | [10.951261520385742, 9.206174850463867] |
3b5f5d43-3887-43e5-a551-2c2f2fa5191f | real-time-and-robust-3d-object-detection | 2204.00132 | null | https://arxiv.org/abs/2204.00132v2 | https://arxiv.org/pdf/2204.00132v2.pdf | Real-Time and Robust 3D Object Detection Within Road-Side LiDARs Using Domain Adaptation | This work aims to address the challenges in domain adaptation of 3D object detection using infrastructure LiDARs. We design a model DASE-ProPillars that can detect vehicles in infrastructure-based LiDARs in real-time. Our model uses PointPillars as the baseline model with additional modules to improve the 3D detection performance. To prove the effectiveness of our proposed modules in DASE-ProPillars, we train and evaluate the model on two datasets, the open source A9-Dataset and a semi-synthetic infrastructure dataset created within the Regensburg Next project. We do several sets of experiments for each module in the DASE-ProPillars detector that show that our model outperforms the SE-ProPillars baseline on the real A9 test set and a semi-synthetic A9 test set, while maintaining an inference speed of 45 Hz (22 ms). We apply domain adaptation from the semi-synthetic A9-Dataset to the semi-synthetic dataset from the Regensburg Next project by applying transfer learning and achieve a 3D mAP@0.25 of 93.49% on the Car class of the target test set using 40 recall positions. | ['Alois Knoll', 'Marcus Grabler', 'Walter Zimmer'] | 2022-03-31 | null | null | null | null | ['robust-3d-object-detection'] | ['computer-vision'] | [-2.53884882e-01 -8.08655545e-02 7.94446766e-02 -5.39011121e-01
-1.06827986e+00 -5.66345453e-01 7.13236809e-01 -1.48159057e-01
-8.31997693e-01 4.99855161e-01 -8.10068429e-01 -6.51542127e-01
2.51366258e-01 -1.09217286e+00 -1.20687532e+00 -2.21737668e-01
-1.62579566e-01 1.08209074e+00 1.15142560e+00 1.86410751e-02
-1.08814746e-01 1.03365958e+00 -1.73214900e+00 8.40171203e-02
5.94347298e-01 9.78870928e-01 5.09754241e-01 9.23567176e-01
9.11415592e-02 1.11419842e-01 -6.57048047e-01 -9.57849696e-02
7.22518563e-01 6.70457065e-01 -2.24398345e-01 -3.92219812e-01
1.05507803e+00 -5.41426182e-01 -2.75823474e-01 6.98972166e-01
8.01310182e-01 -2.07445443e-01 8.97419393e-01 -1.76124227e+00
1.59916878e-02 -1.57501519e-01 -7.28977978e-01 4.65977043e-01
9.20286551e-02 3.41946691e-01 3.77585083e-01 -1.36452401e+00
4.10282940e-01 1.61298680e+00 9.84225094e-01 2.96329618e-01
-1.26628053e+00 -1.42021239e+00 -2.47577488e-01 3.73673350e-01
-1.80711019e+00 -2.52718061e-01 1.65691599e-01 -5.88003695e-01
1.06425345e+00 -1.49158642e-01 3.51553440e-01 1.03031874e+00
1.79955944e-01 8.60318184e-01 9.47317839e-01 -1.59671918e-01
3.59730899e-01 6.09575212e-01 1.39461439e-02 7.11278081e-01
4.55344141e-01 6.86561882e-01 -4.36097205e-01 1.18221371e-02
3.24822277e-01 -3.53871763e-01 5.06830990e-01 -5.90488017e-01
-1.05613601e+00 9.30894971e-01 6.96458220e-01 -4.14079517e-01
-1.55044779e-01 1.14229277e-01 3.45775843e-01 9.84162539e-02
4.46607769e-01 -2.72771329e-01 -5.22767663e-01 2.75888324e-01
-1.02135932e+00 6.02585196e-01 7.06393898e-01 1.59145117e+00
9.53050017e-01 -1.38395369e-01 -1.06962748e-01 3.77885640e-01
5.18462360e-01 1.43873310e+00 -3.49095643e-01 -8.87893558e-01
6.65175855e-01 4.11923915e-01 3.68040770e-01 -2.30619013e-01
-2.59504646e-01 -4.91123423e-02 -1.60661668e-01 7.60307610e-01
2.52728611e-01 -3.43533128e-01 -1.25319743e+00 1.27137101e+00
5.84143877e-01 3.50414574e-01 -1.78543985e-01 4.94612992e-01
8.67158592e-01 6.28113866e-01 1.89574942e-01 4.29912537e-01
1.21640313e+00 -4.98384595e-01 4.68595885e-02 -4.55242217e-01
5.63647866e-01 -6.79304421e-01 7.90937304e-01 6.16415367e-02
-6.76591277e-01 -1.02360463e+00 -1.18701589e+00 1.83521196e-01
-6.85762346e-01 2.49959454e-01 6.10983707e-02 9.19766963e-01
-1.01162422e+00 -5.93695156e-02 -7.10871637e-01 -7.67709196e-01
6.61013186e-01 4.67004299e-01 -1.86984032e-01 -1.63693056e-01
-9.48144734e-01 1.23778987e+00 3.38171780e-01 -5.97117066e-01
-1.23936117e+00 -9.54634845e-01 -6.17728174e-01 -3.22515279e-01
3.32795531e-01 -5.17988980e-01 1.39629948e+00 3.88514489e-01
-8.98945987e-01 1.17885280e+00 9.27279890e-03 -9.03541386e-01
7.19767988e-01 -1.84034944e-01 -4.73850340e-01 -6.01267535e-03
6.62367642e-01 1.34082592e+00 5.83637297e-01 -1.26773655e+00
-1.36153710e+00 -4.36992794e-01 -2.25442812e-01 -2.32554987e-01
3.47748518e-01 -2.87410766e-01 -4.01477903e-01 4.02603038e-02
-2.73515940e-01 -1.23234534e+00 8.52326229e-02 2.44814619e-01
-2.68535554e-01 -4.49986726e-01 1.41271138e+00 7.77321681e-02
3.22630525e-01 -2.15803242e+00 -8.48249078e-01 9.44503620e-02
-1.67361408e-01 5.89173794e-01 -1.57086462e-01 2.35639736e-01
2.28570268e-01 -2.92198844e-02 -4.37723175e-02 -3.72529805e-01
2.01291274e-02 4.56678867e-01 -4.45558220e-01 3.95093530e-01
3.87351662e-01 7.34422803e-01 -9.74303424e-01 -8.07132006e-01
7.78881192e-01 4.56285566e-01 -3.97957981e-01 1.91720918e-01
-2.88034379e-02 1.59239024e-01 -4.55138981e-01 8.26474607e-01
1.33276856e+00 4.19448823e-01 -6.42854989e-01 -7.25933537e-02
-4.20293272e-01 2.64185637e-01 -1.06079042e+00 1.32028246e+00
-5.10052502e-01 7.89948285e-01 2.15848356e-01 -5.81888020e-01
1.09810579e+00 -2.01777637e-01 2.66368359e-01 -8.51168990e-01
-1.78043619e-01 2.32313842e-01 -3.47854555e-01 -2.89338976e-01
5.58615208e-01 -5.04217446e-02 -1.43805817e-01 8.98503810e-02
1.13953941e-01 -7.59050131e-01 3.48894268e-01 3.88426423e-01
1.30314827e+00 1.26360357e-01 -6.54613003e-02 -1.78267464e-01
4.51106668e-01 2.90501654e-01 1.83705211e-01 1.18117797e+00
-4.84658331e-01 3.56070250e-01 -1.95625633e-01 -3.85010570e-01
-9.95323479e-01 -1.79438281e+00 -7.65979469e-01 8.93791378e-01
1.14535309e-01 -1.63447946e-01 -5.91761470e-01 -1.09424412e+00
8.17816794e-01 1.11469376e+00 -3.51813853e-01 6.84413463e-02
-4.26889658e-01 -4.66296673e-01 1.03793216e+00 6.24961734e-01
7.99601138e-01 -6.12318933e-01 -9.23360407e-01 1.27055747e-02
2.33700186e-01 -1.60593867e+00 1.46080047e-01 2.89759427e-01
-4.59945589e-01 -1.14726901e+00 -2.59506464e-01 -6.30255818e-01
1.06502466e-01 7.72620440e-01 1.21698713e+00 -4.70116407e-01
-2.59534061e-01 5.11643589e-01 1.77720431e-02 -1.27779770e+00
-5.48338830e-01 1.32688180e-01 2.36069769e-01 -6.42826140e-01
1.09749138e+00 -4.55463529e-01 -4.44947273e-01 7.13559330e-01
-2.13172629e-01 -1.99807644e-01 7.61045754e-01 2.40937471e-01
6.45656168e-01 -3.70788932e-01 5.31866491e-01 -5.16822934e-01
3.58095914e-02 -6.98273897e-01 -1.18723845e+00 -2.72008508e-01
-6.87448978e-01 -2.51412243e-01 -1.27145015e-02 -1.84789836e-01
-7.76748002e-01 5.19120812e-01 -1.46852285e-01 -6.59844279e-01
-4.39401269e-01 -3.19101602e-01 -1.70862526e-01 -1.96481243e-01
1.08860004e+00 7.63255507e-02 -1.28381670e-01 -4.80473429e-01
4.87450927e-01 1.05584300e+00 7.83267438e-01 -6.84677720e-01
1.37222767e+00 7.44573116e-01 2.78941303e-01 -1.08685648e+00
-5.93429446e-01 -6.48395181e-01 -7.17668533e-01 -1.82776675e-01
6.05926812e-01 -1.50388861e+00 -7.95713305e-01 1.46302387e-01
-1.29920745e+00 -3.99717987e-01 -4.23427224e-01 3.77989501e-01
-5.79349041e-01 -3.21549538e-04 5.33860996e-02 -1.00477898e+00
-1.28374249e-01 -1.04706120e+00 1.61253142e+00 2.34699864e-02
1.71597078e-01 -2.58610427e-01 2.43219420e-01 1.59947306e-01
1.40004352e-01 2.63499111e-01 4.56696063e-01 -5.69900393e-01
-9.01093423e-01 -5.81293285e-01 -7.27862179e-01 4.54240024e-01
-3.58586103e-01 -1.55922353e-01 -1.19417667e+00 -2.18832076e-01
-4.49324250e-01 -1.34062856e-01 9.46995974e-01 1.21987864e-01
8.53320956e-01 5.41705370e-01 -8.92859399e-01 2.67713606e-01
1.13071811e+00 9.70933288e-02 4.29867268e-01 4.11399782e-01
2.37596884e-01 1.92364320e-01 1.17460144e+00 1.30244434e-01
8.54459047e-01 7.50243485e-01 7.75642693e-01 -6.33921940e-04
-2.53448188e-01 -4.17202532e-01 3.50527018e-01 -1.08394369e-01
2.08483621e-01 -1.20163716e-01 -1.25819612e+00 8.44090044e-01
-1.53680933e+00 -8.02363098e-01 -4.24721003e-01 2.30970311e+00
3.30633432e-01 6.90777779e-01 5.16928315e-01 -1.94952488e-02
7.17738152e-01 -2.88476586e-01 -4.38112408e-01 -1.91861704e-01
3.64744693e-01 3.65165174e-01 1.19790697e+00 3.46403003e-01
-1.21996164e+00 1.08568001e+00 6.06453228e+00 7.85885334e-01
-8.60906780e-01 3.88271183e-01 -2.24450126e-01 -2.98173964e-01
4.69275862e-01 -2.49319300e-01 -1.55075133e+00 6.53590620e-01
1.55638158e+00 -1.97366521e-01 -2.22912524e-02 1.21742749e+00
2.79210597e-01 -3.79293233e-01 -1.32222307e+00 6.88270807e-01
-2.93559045e-01 -1.11147273e+00 -2.48479024e-01 2.89876252e-01
3.50735009e-01 1.15768409e+00 -1.15245571e-02 1.10904813e+00
6.03724957e-01 -7.08333433e-01 9.53047276e-01 5.61565049e-02
1.13429821e+00 -8.50614667e-01 6.11256003e-01 9.05396402e-01
-1.29541802e+00 1.06496394e-01 -6.64707720e-01 2.90993899e-01
1.09154083e-01 4.49356437e-01 -1.86492479e+00 2.30084151e-01
1.00544941e+00 2.28830129e-01 -8.42066526e-01 1.17885828e+00
-3.64870459e-01 6.33732438e-01 -9.53142583e-01 1.00321425e-02
3.07558447e-01 2.59768069e-01 7.96063185e-01 1.71197867e+00
4.11896944e-01 -4.78496701e-01 2.73271382e-01 9.01333630e-01
-1.00100622e-01 -6.36314571e-01 -1.08859301e+00 6.98766172e-01
1.02123916e+00 1.16871524e+00 -1.91792712e-01 -3.32397163e-01
-3.30118328e-01 3.27519983e-01 -3.13627021e-03 1.01336263e-01
-1.28459561e+00 -3.75681698e-01 7.26804793e-01 7.61913121e-01
6.31882668e-01 -3.01246762e-01 7.66824037e-02 -3.51814866e-01
-1.44729838e-01 -3.80212843e-01 1.77624226e-01 -8.71539593e-01
-1.10858262e+00 3.41759622e-01 7.44758427e-01 -1.29786742e+00
-4.00049865e-01 -8.95193577e-01 -5.13502121e-01 8.44305098e-01
-1.83852148e+00 -1.56541789e+00 -5.54072559e-01 5.65207243e-01
4.69223678e-01 -3.07689518e-01 5.95159173e-01 6.26340210e-01
-1.28053352e-01 4.12146866e-01 -1.82651654e-01 -9.78762656e-02
7.96016932e-01 -1.39754927e+00 1.17820442e+00 6.03339255e-01
7.61428326e-02 1.86757743e-01 6.64759576e-01 -6.73027396e-01
-1.07534385e+00 -1.73344839e+00 5.36743164e-01 -1.06746256e+00
4.01747048e-01 -7.68698633e-01 -5.86201787e-01 8.92047524e-01
-1.71959817e-01 4.14593607e-01 1.48602232e-01 -1.11247636e-01
-4.06589627e-01 -4.00033146e-01 -1.62795293e+00 1.25253722e-01
1.25736117e+00 -4.97631609e-01 -7.57776082e-01 4.49347973e-01
6.85061038e-01 -5.74014485e-01 -4.62931424e-01 7.03990817e-01
5.81794858e-01 -1.04660082e+00 1.36633658e+00 -1.28057525e-01
-3.04682255e-01 -6.95240438e-01 -4.55196977e-01 -1.10444427e+00
4.33757231e-02 -1.68966472e-01 -2.85894185e-01 1.02082503e+00
2.06132621e-01 -6.20760024e-01 9.85002995e-01 -1.25226364e-01
-1.63385987e-01 -1.72851697e-01 -1.49919832e+00 -1.29486382e+00
1.98233664e-01 -9.20394778e-01 5.90290487e-01 2.93387920e-02
-9.24180627e-01 5.80069602e-01 3.00396234e-01 5.12568533e-01
1.05735004e+00 -1.87662631e-01 1.58410454e+00 -1.51288462e+00
3.29845846e-01 2.32837737e-01 -6.62600994e-01 -1.12165487e+00
8.30673277e-02 -8.58012140e-01 8.41384679e-02 -1.42893887e+00
-1.27733648e-01 -6.99642122e-01 -1.00347854e-01 3.13384056e-01
3.55442852e-01 4.38648820e-01 1.34333223e-01 9.10118148e-02
-5.25403500e-01 2.86714941e-01 5.60783207e-01 -2.71298975e-01
1.16547480e-01 4.95650381e-01 -1.48359463e-01 8.41127813e-01
5.39332330e-01 -8.76241624e-01 -2.16908842e-01 -3.71485025e-01
-2.41607383e-01 -3.80894959e-01 8.07445705e-01 -1.61218715e+00
3.34492117e-01 1.73348457e-01 5.63362300e-01 -1.79656947e+00
5.79780877e-01 -9.47858930e-01 -2.75161058e-01 5.63732743e-01
5.76139688e-01 -3.79864685e-02 7.58322299e-01 6.83187902e-01
3.54028076e-01 1.20643251e-01 9.72900748e-01 8.70381147e-02
-1.00961530e+00 2.13652521e-01 -3.87710452e-01 -7.32689723e-02
1.36828566e+00 -4.82944727e-01 -3.35812479e-01 6.37670308e-02
-2.49395847e-01 6.31600499e-01 3.67238790e-01 5.59329689e-01
4.93007839e-01 -1.13622463e+00 -7.90071011e-01 5.48868716e-01
5.19093633e-01 3.65577757e-01 -4.35525417e-01 3.54407609e-01
-3.77545059e-01 6.96795702e-01 -1.91981092e-01 -1.31709099e+00
-1.17088437e+00 3.89180034e-01 3.22315454e-01 5.34746572e-02
-4.17879730e-01 6.24226391e-01 -4.82674539e-02 -9.74855602e-01
1.94053948e-01 -4.21642393e-01 3.28213006e-01 -7.56708011e-02
3.40647459e-01 6.86934888e-01 3.42876822e-01 -6.97358608e-01
-9.58395779e-01 3.71763259e-01 1.37476325e-01 -1.81148261e-01
1.16116238e+00 4.52710167e-02 7.04102159e-01 1.02601364e-01
7.45841801e-01 2.95173153e-02 -1.32547510e+00 -1.94953620e-01
1.00964725e-01 -4.38548297e-01 1.98111266e-01 -7.93429375e-01
-4.38460588e-01 1.08596134e+00 1.31408203e+00 3.49281393e-02
3.73434395e-01 1.59739181e-01 6.28267050e-01 8.10306966e-01
8.74849021e-01 -8.56128454e-01 -1.99389637e-01 6.69497430e-01
7.45369613e-01 -1.51561916e+00 3.60718705e-02 -2.45660618e-01
-2.11442515e-01 6.10030651e-01 8.32082033e-01 -2.73094743e-01
6.26407683e-01 6.20664537e-01 -3.27120945e-02 -2.51100570e-01
-4.67349350e-01 -7.11971760e-01 -5.60482815e-02 1.25850976e+00
-3.54196429e-01 1.33909270e-01 3.94129038e-01 -3.35850902e-02
-2.39170849e-01 1.61734641e-01 1.41196772e-01 9.41313028e-01
-8.07185590e-01 -9.41425860e-01 -5.43052197e-01 2.72016555e-01
1.67369038e-01 2.90177226e-01 -4.24771041e-01 1.56461012e+00
6.41223192e-01 1.04085720e+00 2.86202908e-01 -4.26726580e-01
9.77001607e-01 5.38014360e-02 5.62935174e-01 -9.16626036e-01
-1.49255514e-01 -2.92340904e-01 6.34598210e-02 -5.08665025e-01
-1.47912294e-01 -7.69112229e-01 -1.32204568e+00 -4.65393186e-01
-4.02741194e-01 -6.30076304e-02 1.19844043e+00 6.08362734e-01
6.21916950e-01 2.59636253e-01 5.10927796e-01 -1.21527565e+00
-7.52623200e-01 -1.08239555e+00 -2.72240907e-01 -2.02647373e-01
3.05000335e-01 -9.87107754e-01 -1.75683305e-01 -3.93955976e-01] | [7.771960735321045, -2.4629149436950684] |
a094abf8-47ac-415d-8418-e4f98cdcd9de | discriminative-sentence-modeling-for-story | 1912.09008 | null | https://arxiv.org/abs/1912.09008v1 | https://arxiv.org/pdf/1912.09008v1.pdf | Discriminative Sentence Modeling for Story Ending Prediction | Story Ending Prediction is a task that needs to select an appropriate ending for the given story, which requires the machine to understand the story and sometimes needs commonsense knowledge. To tackle this task, we propose a new neural network called Diff-Net for better modeling the differences of each ending in this task. The proposed model could discriminate two endings in three semantic levels: contextual representation, story-aware representation, and discriminative representation. Experimental results on the Story Cloze Test dataset show that the proposed model siginificantly outperforms various systems by a large margin, and detailed ablation studies are given for better understanding our model. We also carefully examine the traditional and BERT-based models on both SCT v1.0 and v1.5 with interesting findings that may potentially help future studies. | ['Wei-Nan Zhang', 'Yiming Cui', 'Shijin Wang', 'Ting Liu', 'Wanxiang Che', 'Guoping Hu'] | 2019-12-19 | null | null | null | null | ['cloze-test'] | ['natural-language-processing'] | [ 3.39939028e-01 -1.81181803e-01 -5.92764020e-01 -5.88332951e-01
-6.68889999e-01 -7.49382317e-01 6.36011720e-01 4.00607893e-03
-2.34288812e-01 6.61019206e-01 8.30590069e-01 -4.90868054e-02
5.14551103e-02 -6.31710947e-01 -2.79922664e-01 -2.77465194e-01
2.93870747e-01 4.07490104e-01 9.84502211e-02 -4.86095428e-01
6.81246281e-01 -6.53748438e-02 -1.47262216e+00 8.22860897e-01
9.25014377e-01 9.71773624e-01 2.89177984e-01 3.63713354e-01
-2.24333122e-01 1.33596754e+00 -7.18542039e-01 -5.93769193e-01
-1.15852617e-01 -8.14669430e-01 -8.84486973e-01 -2.72831857e-01
1.89780489e-01 -3.30213457e-01 -3.84381860e-01 8.11713994e-01
3.17908347e-01 2.42677420e-01 9.20580685e-01 -1.13410282e+00
-9.99760807e-01 1.31482649e+00 -3.62906456e-01 2.04003140e-01
7.16665626e-01 3.07095684e-02 1.34239221e+00 -8.34737778e-01
7.93985963e-01 1.12786996e+00 8.08681309e-01 5.72584152e-01
-7.58868694e-01 -7.36509979e-01 2.71244079e-01 1.00268400e+00
-1.36093819e+00 -2.09259123e-01 1.31465054e+00 -3.74642551e-01
7.72986352e-01 4.74549949e-01 6.09120250e-01 1.71895182e+00
4.57257368e-02 1.29303849e+00 1.18304420e+00 -1.55219048e-01
5.59561662e-02 -1.08432390e-01 5.35991549e-01 1.33137077e-01
-1.08364567e-01 -1.86004296e-01 -7.24153280e-01 1.15561269e-01
4.59410489e-01 -3.04261029e-01 -3.65610898e-01 2.17958361e-01
-1.04913306e+00 1.09262586e+00 3.27779591e-01 4.68735546e-01
-3.47958878e-02 2.12944195e-01 7.79935122e-01 2.02015013e-01
3.01763386e-01 6.60508215e-01 -1.37271613e-01 -4.95761961e-01
-8.91393304e-01 4.81479794e-01 5.02257824e-01 8.51558387e-01
-2.63533741e-02 -8.89118668e-03 -5.95390260e-01 1.05749309e+00
1.53645545e-01 -6.74906373e-02 7.38029480e-01 -7.36565113e-01
4.42991734e-01 4.41017151e-01 -9.92268175e-02 -1.07323766e+00
-3.69658947e-01 -5.39502800e-01 -6.92084789e-01 -2.02449411e-01
3.14104021e-01 9.78050977e-02 -6.00437045e-01 1.96061254e+00
-8.02867144e-05 3.85708511e-01 4.54620048e-02 9.20803428e-01
1.41818643e+00 6.08715177e-01 4.87372279e-02 -1.62015200e-01
1.42806053e+00 -9.72937703e-01 -1.01260686e+00 -5.32485127e-01
7.60539234e-01 -6.85301900e-01 1.56710362e+00 3.36008012e-01
-8.75950992e-01 -4.41397011e-01 -1.47611201e+00 -4.57220018e-01
-3.46518308e-01 3.03057134e-01 6.38614953e-01 4.83126819e-01
-5.74885070e-01 6.64030910e-01 -2.40560204e-01 -4.13542986e-01
3.58531445e-01 -2.44303316e-01 2.11701915e-01 -1.53337047e-01
-1.73517108e+00 1.10620093e+00 8.42408657e-01 -5.48598953e-02
-8.08822453e-01 -3.69382530e-01 -1.00052834e+00 -1.00400932e-02
3.36556047e-01 -5.53967595e-01 1.15497565e+00 -8.68590951e-01
-1.37736964e+00 9.35058951e-01 -2.51232207e-01 -5.43617368e-01
6.18491352e-01 -3.25237185e-01 -6.38483703e-01 -4.57357466e-02
2.91735113e-01 4.46947038e-01 6.04662240e-01 -1.08775651e+00
-4.30371672e-01 -4.68393862e-02 2.20199823e-01 3.96368980e-01
-1.26332670e-01 2.88424015e-01 -1.42692924e-01 -1.16060638e+00
2.39239573e-01 -6.13362610e-01 2.53419399e-01 -5.54301143e-01
-4.96576548e-01 -6.67481542e-01 7.92331398e-01 -7.76554525e-01
1.48580778e+00 -2.11989927e+00 -1.01577424e-01 -2.88298607e-01
6.04292639e-02 -1.95622906e-01 -1.19604826e-01 4.42132294e-01
-1.53037012e-02 1.40489072e-01 -2.34432191e-01 -2.32052326e-01
3.20730321e-02 -2.49988269e-02 -5.17786086e-01 1.51849121e-01
-3.37931812e-02 9.55324769e-01 -9.02648568e-01 -5.35191536e-01
-1.17998719e-01 1.53721899e-01 -3.77924442e-01 3.91950682e-02
-4.24853623e-01 3.01416099e-01 -4.21564579e-01 6.18365288e-01
3.87181222e-01 -8.01131278e-02 5.10338359e-02 5.99619932e-02
2.30907589e-01 7.34830737e-01 -9.69302475e-01 1.76219058e+00
-2.49412388e-01 1.06158423e+00 -6.37497127e-01 -6.93233013e-01
9.63971317e-01 1.50867939e-01 -1.39149666e-01 -5.61610103e-01
5.89693546e-01 4.00908738e-02 -7.05250278e-02 -7.20421255e-01
8.32044601e-01 -3.85823578e-01 -5.72388887e-01 4.75957781e-01
-2.26321608e-01 -1.13597274e-01 2.72062004e-01 1.02603160e-01
9.27824974e-01 3.29750478e-01 3.42157692e-01 6.34608883e-03
2.79357582e-01 1.81660816e-01 7.79756188e-01 8.79507244e-01
-3.37712467e-01 8.46605122e-01 6.32823944e-01 -3.25375229e-01
-6.08201146e-01 -7.62491167e-01 -1.89322069e-01 1.11947060e+00
5.18208563e-01 -6.10676348e-01 -6.10612988e-01 -8.74033332e-01
-4.28192466e-01 1.78918529e+00 -8.75978351e-01 -2.29004174e-01
-6.16348505e-01 -4.83829916e-01 7.17658699e-01 7.91238666e-01
5.85601985e-01 -1.19546223e+00 -4.90814060e-01 1.19882062e-01
-7.94681072e-01 -1.05968761e+00 -3.58690321e-01 2.22590402e-01
-5.30792713e-01 -1.01809764e+00 -2.45884269e-01 -8.80408943e-01
5.86386211e-02 2.72708952e-01 1.00386095e+00 -1.75963230e-02
4.21832740e-01 -2.15362117e-01 -8.26003730e-01 -3.24464172e-01
-2.94040680e-01 2.13637762e-02 -2.12626934e-01 -2.63156116e-01
7.54769504e-01 -6.06092989e-01 -2.87871927e-01 1.97538808e-01
-7.36362576e-01 5.39815664e-01 9.53174233e-02 8.12520027e-01
2.84401983e-01 2.11124212e-01 8.11034203e-01 -8.83384109e-01
1.06023979e+00 -7.32573807e-01 1.42901391e-01 2.14548379e-01
-2.92051792e-01 -1.20405927e-01 8.58365715e-01 -5.37294090e-01
-1.12381959e+00 -3.39128017e-01 -3.63416374e-01 3.94322798e-02
-7.38759339e-02 7.87607014e-01 -4.79053974e-01 6.03557467e-01
6.60126746e-01 3.38916838e-01 -6.73936725e-01 -3.62006545e-01
4.34741586e-01 7.14871883e-01 7.91893005e-01 -4.67531681e-01
3.77713382e-01 2.58314669e-01 -5.46707988e-01 -4.42582428e-01
-1.26756895e+00 -1.88169956e-01 -4.14836586e-01 -4.67425346e-01
7.41105199e-01 -8.46827149e-01 -4.10851479e-01 3.62424821e-01
-1.50383306e+00 -1.28614023e-01 -1.42247796e-01 4.30741161e-01
-7.11189866e-01 1.15752243e-01 -7.55870938e-01 -4.30124760e-01
-6.10875264e-02 -9.06892002e-01 5.93676209e-01 2.94822842e-01
-9.78779614e-01 -9.17821884e-01 1.24544814e-01 7.50172853e-01
6.33166656e-02 4.26922321e-01 1.24241138e+00 -1.06236267e+00
-1.28545344e-01 -2.45777965e-01 -1.78961486e-01 6.23934492e-02
-1.59735262e-01 -2.15836212e-01 -1.01242042e+00 2.02644154e-01
3.99001062e-01 -7.23357797e-01 1.20841312e+00 3.46543163e-01
1.19181979e+00 -1.11447170e-01 -7.92442113e-02 5.39156616e-01
1.25581193e+00 3.97213340e-01 8.32113624e-01 5.68798304e-01
7.02830851e-01 5.18124461e-01 8.50888431e-01 4.01016712e-01
6.05522811e-01 4.88559395e-01 3.35506767e-01 5.08449078e-01
-3.66288543e-01 -5.92271924e-01 4.21251774e-01 7.02940166e-01
1.56065315e-01 -4.65827078e-01 -7.71151841e-01 6.39036477e-01
-2.02551913e+00 -1.25771761e+00 -3.45877260e-01 1.54714143e+00
9.35048223e-01 4.79140580e-01 -8.54895040e-02 2.67682433e-01
8.67714405e-01 6.55037761e-01 -7.34143615e-01 -5.72097123e-01
-5.10906816e-01 -1.81471735e-01 -1.95240945e-01 2.17615724e-01
-1.06250548e+00 1.25365543e+00 7.13634586e+00 1.31139016e+00
-9.59258735e-01 1.29931793e-01 7.35017300e-01 -2.19107822e-01
-4.76529062e-01 1.15089558e-01 -5.56112170e-01 5.17733753e-01
5.72075486e-01 -2.53424704e-01 3.02907765e-01 1.11041069e+00
2.82246679e-01 -9.54866633e-02 -1.25215697e+00 1.13117337e+00
4.93102252e-01 -1.44404280e+00 8.91003385e-02 -6.23632312e-01
6.59508049e-01 -1.50301725e-01 -9.37286243e-02 7.10862160e-01
2.74663687e-01 -1.11552119e+00 1.09356654e+00 2.59698868e-01
6.65975928e-01 -8.93849015e-01 7.84309030e-01 4.68798250e-01
-1.00876641e+00 5.22555411e-02 -1.55701607e-01 -5.90234399e-01
2.04302073e-01 2.11846828e-01 -6.66969299e-01 2.93814123e-01
4.09016997e-01 1.03920674e+00 -6.52352571e-01 9.07904983e-01
-7.88019896e-01 9.31160748e-01 1.06126525e-01 -3.71153325e-01
3.59979868e-01 1.13213755e-01 6.62158847e-01 1.19873595e+00
2.85089523e-01 2.48770759e-01 1.04426362e-01 1.02276778e+00
-1.73673719e-01 1.78583473e-01 -4.71845150e-01 -2.13596295e-03
5.13317406e-01 8.49640548e-01 -1.00409091e+00 -3.42269242e-01
-7.07302317e-02 1.09308505e+00 3.71895254e-01 2.37872168e-01
-1.20354056e+00 -3.81454468e-01 4.93857294e-01 -4.29843292e-02
7.35835880e-02 2.85535748e-03 -1.00099635e+00 -1.16646123e+00
-3.83743979e-02 -6.55280828e-01 5.53315520e-01 -1.09577763e+00
-1.38526344e+00 6.57646298e-01 6.92316741e-02 -1.29556715e+00
-3.50356638e-01 -2.40531296e-01 -9.96130347e-01 3.23227912e-01
-1.18994713e+00 -1.18800402e+00 -1.80438489e-01 3.42173487e-01
1.04982853e+00 5.90772927e-02 6.17204309e-01 -2.08933309e-01
-5.62178493e-01 6.80612922e-01 1.02525018e-01 3.15524161e-01
6.73632920e-01 -1.06019664e+00 3.94241333e-01 8.49798560e-01
1.60263509e-01 2.95449913e-01 1.01018524e+00 -7.89486289e-01
-6.25640631e-01 -9.11011696e-01 1.26980352e+00 -3.27246815e-01
6.77527308e-01 -2.77088791e-01 -8.30140710e-01 6.42075598e-01
5.01355052e-01 -7.99441159e-01 1.06219518e+00 5.31472981e-01
-6.73643947e-01 4.18518454e-01 -1.09067059e+00 1.05708623e+00
1.30002499e+00 -4.71767575e-01 -1.28939164e+00 1.83342457e-01
9.03028488e-01 -2.75810093e-01 -3.24520737e-01 2.50523269e-01
5.95464051e-01 -1.02055240e+00 6.19872272e-01 -5.73721051e-01
1.07621121e+00 -1.59263685e-01 -2.92363822e-01 -1.37176132e+00
-4.35885489e-01 -3.68002117e-01 -2.14978214e-02 1.38819349e+00
3.18287313e-01 -1.63396448e-01 6.20308220e-01 5.05686939e-01
-2.72627831e-01 -8.29871356e-01 -8.36119652e-01 -9.30998504e-01
2.05586582e-01 -1.14625335e+00 7.51910150e-01 1.38177907e+00
3.11260402e-01 8.48060489e-01 -5.40197015e-01 -2.61026531e-01
8.51396769e-02 2.62335211e-01 2.66792774e-01 -9.12579060e-01
-3.06309313e-01 -8.10232759e-01 -2.36533195e-01 -1.25802827e+00
4.01631445e-01 -1.10270905e+00 -7.55034834e-02 -1.75301576e+00
5.19878685e-01 -1.06671974e-01 -2.39755064e-01 3.29470634e-01
-3.65261316e-01 2.74287999e-01 2.06465319e-01 5.57153821e-02
-7.83133388e-01 8.11826885e-01 1.11363316e+00 -3.79242837e-01
-2.09409148e-01 -1.91649586e-01 -1.02603030e+00 1.15135002e+00
9.97970283e-01 -5.94335973e-01 -6.59239769e-01 -3.75806391e-01
4.03929621e-01 6.74352422e-02 2.89700836e-01 -9.58560169e-01
-8.49680081e-02 -5.09519458e-01 3.76438707e-01 -1.04180205e+00
3.50221723e-01 -4.83391672e-01 2.42819693e-02 -3.22194882e-02
-8.26715827e-01 -1.90027863e-01 9.97768342e-02 3.33506674e-01
-2.69049257e-01 -6.64867222e-01 5.37597179e-01 6.19970784e-02
-9.24252689e-01 -2.92821586e-01 -4.59742725e-01 3.94871891e-01
1.06267130e+00 -5.53689837e-01 -4.19561565e-01 -8.40880036e-01
-6.65814221e-01 3.09644222e-01 3.96925032e-01 7.33497739e-01
8.92681479e-01 -1.84503329e+00 -1.06207085e+00 -3.35089594e-01
3.99780840e-01 -4.71030563e-01 4.46300685e-01 3.82960409e-01
-3.48049402e-01 2.01430410e-01 -1.23169318e-01 -2.24037930e-01
-1.23961854e+00 4.72871214e-01 -6.24210611e-02 -4.39762533e-01
-5.18862903e-01 9.67568576e-01 1.53170317e-01 -6.66769892e-02
1.94919243e-01 -2.27775723e-01 -6.90352917e-01 2.87836611e-01
8.17645371e-01 2.80382782e-01 -3.37204367e-01 -8.43573928e-01
-3.00563514e-01 2.94307828e-01 -2.80766845e-01 -7.26775751e-02
1.23671365e+00 -1.20136634e-01 1.18608892e-01 8.69191527e-01
1.08863413e+00 -1.45199522e-01 -1.09328842e+00 -1.49203390e-01
2.27995828e-01 -4.21613276e-01 -2.90078130e-02 -1.11183953e+00
-5.72416186e-01 6.73935711e-01 8.11184421e-02 2.30955839e-01
1.19039273e+00 1.35842085e-01 1.14411938e+00 2.25114942e-01
2.41380498e-01 -1.36663139e+00 1.74864590e-01 9.06973839e-01
1.33626330e+00 -1.13734210e+00 -1.11770608e-01 -3.66565615e-01
-1.13826525e+00 1.07441545e+00 6.69698358e-01 -2.59053856e-01
2.31454089e-01 -3.82953696e-02 1.45372897e-02 -2.12976262e-01
-8.54122519e-01 -1.51816815e-01 2.11679548e-01 4.49250221e-01
6.55952156e-01 2.23711520e-01 -7.85668254e-01 1.33446276e+00
-6.14017844e-01 -3.02619487e-01 6.11497581e-01 7.07648873e-01
-4.27068770e-01 -8.70177388e-01 -4.11209106e-01 4.77481186e-01
-2.45143205e-01 -3.60836565e-01 -9.04155552e-01 7.02028573e-01
3.15721840e-01 1.20790744e+00 -1.06306769e-01 -8.25679004e-01
1.29252717e-01 2.53640026e-01 3.54966789e-01 -4.32465494e-01
-7.25700855e-01 -6.87511340e-02 5.39208710e-01 -3.21884036e-01
-2.47120619e-01 -8.10254693e-01 -1.42035592e+00 -4.22392756e-01
-3.07072014e-01 -1.08369932e-01 3.12633812e-01 1.18623006e+00
9.90868658e-02 5.54197431e-01 4.74853665e-01 -4.12731111e-01
-1.31487593e-01 -1.16923761e+00 -5.42605281e-01 4.35786963e-01
-2.05302954e-01 -8.01011086e-01 -4.16541994e-01 -5.83832860e-02] | [11.27477741241455, 8.841788291931152] |
8a62fb45-2c97-4a03-a1e0-026680bb3677 | mris-a-multi-modal-retrieval-approach-for | 2303.10249 | null | https://arxiv.org/abs/2303.10249v1 | https://arxiv.org/pdf/2303.10249v1.pdf | MRIS: A Multi-modal Retrieval Approach for Image Synthesis on Diverse Modalities | Multiple imaging modalities are often used for disease diagnosis, prediction, or population-based analyses. However, not all modalities might be available due to cost, different study designs, or changes in imaging technology. If the differences between the types of imaging are small, data harmonization approaches can be used; for larger changes, direct image synthesis approaches have been explored. In this paper, we develop an approach based on multi-modal metric learning to synthesize images of diverse modalities. We use metric learning via multi-modal image retrieval, resulting in embeddings that can relate images of different modalities. Given a large image database, the learned image embeddings allow us to use k-nearest neighbor (k-NN) regression for image synthesis. Our driving medical problem is knee osteoarthritis (KOA), but our developed method is general after proper image alignment. We test our approach by synthesizing cartilage thickness maps obtained from 3D magnetic resonance (MR) images using 2D radiographs. Our experiments show that the proposed method outperforms direct image synthesis and that the synthesized thickness maps retain information relevant to downstream tasks such as progression prediction and Kellgren-Lawrence grading (KLG). Our results suggest that retrieval approaches can be used to obtain high-quality and meaningful image synthesis results given large image databases. | ['Marc Niethammer', 'Boqi Chen'] | 2023-03-17 | null | null | null | null | ['metric-learning', 'metric-learning'] | ['computer-vision', 'methodology'] | [ 2.96484232e-01 -3.06683797e-02 -3.69489878e-01 -3.86092573e-01
-1.51859248e+00 -1.60562605e-01 3.79897118e-01 1.77855372e-01
-5.51710963e-01 6.24943793e-01 5.21821558e-01 1.04443379e-01
-5.20931363e-01 -7.20508933e-01 -4.39872354e-01 -6.40928388e-01
-4.10506278e-02 6.24454141e-01 2.68319845e-01 -8.03692415e-02
3.62854958e-01 4.54096347e-01 -1.57586920e+00 5.21584868e-01
5.61496854e-01 9.66029823e-01 3.97517025e-01 8.02584648e-01
-9.07006264e-02 5.26143789e-01 -3.69123459e-01 -1.40911132e-01
3.07426423e-01 -5.21770537e-01 -8.72253120e-01 1.68509424e-01
5.92506051e-01 -3.53426397e-01 -3.10016036e-01 7.83503652e-01
9.46657538e-01 -6.33835942e-02 1.06336737e+00 -9.43210483e-01
-7.24400342e-01 2.13383615e-01 -4.30259168e-01 6.95291013e-02
1.45510018e-01 1.52339973e-02 8.23215127e-01 -8.66587639e-01
9.13906634e-01 1.08830237e+00 4.40973759e-01 6.32368803e-01
-1.27607179e+00 -2.50243843e-01 -4.94025350e-01 4.63412642e-01
-1.29337776e+00 -1.36431277e-01 8.16196442e-01 -7.01923549e-01
4.22109425e-01 4.69002545e-01 6.43867910e-01 8.49891067e-01
7.22077012e-01 5.20053506e-01 1.45355892e+00 -5.76020300e-01
1.58360630e-01 -7.90330917e-02 -3.21066231e-01 7.06543028e-01
1.06473695e-02 -6.98567135e-03 -4.90983963e-01 -1.59791261e-01
7.91524887e-01 1.02640584e-01 -3.25470805e-01 -5.20015895e-01
-1.65256095e+00 9.28477824e-01 5.58520436e-01 4.81108546e-01
-4.50146228e-01 1.15238525e-01 4.86901164e-01 4.29845303e-01
3.12296152e-01 6.11657262e-01 -2.20002308e-02 3.07424188e-01
-1.02779245e+00 3.58091891e-01 3.24429214e-01 3.44426811e-01
3.96273553e-01 -3.62343788e-01 -9.78595912e-02 1.33287024e+00
2.31047988e-01 4.44723248e-01 1.28377521e+00 -1.21791565e+00
1.38385952e-01 4.99278784e-01 -2.87193805e-01 -1.04286373e+00
-4.32056308e-01 1.55308336e-01 -9.02459681e-01 4.19513822e-01
3.98267031e-01 3.01073879e-01 -9.92050171e-01 1.41069722e+00
1.87382206e-01 -1.07247837e-01 -2.13051140e-02 1.25426650e+00
6.32816017e-01 2.28061587e-01 -1.36360407e-01 -1.69035569e-01
1.60978258e+00 -8.72515917e-01 -4.93118763e-01 7.71143585e-02
9.23227072e-01 -9.09000218e-01 1.16663098e+00 3.54440123e-01
-9.87417817e-01 -4.55969065e-01 -1.06811094e+00 -1.34951577e-01
-4.30318713e-01 6.20703921e-02 2.68638074e-01 3.52088273e-01
-9.74968255e-01 5.28583527e-01 -7.48727322e-01 -2.89753020e-01
1.97856188e-01 2.87144125e-01 -7.87618220e-01 -4.85675275e-01
-1.11897492e+00 1.10008121e+00 1.60697877e-01 -4.48988341e-02
-6.27053559e-01 -8.69205713e-01 -7.92409122e-01 -5.28160691e-01
6.11357987e-02 -9.84748960e-01 9.30026889e-01 -5.81949770e-01
-1.26424778e+00 1.25894070e+00 1.95620686e-01 -2.02520609e-01
5.29719353e-01 -4.03569825e-03 -4.07765120e-01 5.84092855e-01
1.17841028e-01 8.86423469e-01 7.15686083e-01 -1.12634873e+00
-3.88671607e-01 -6.04005456e-01 -2.41083786e-01 3.55618477e-01
-3.25094104e-01 -2.94977482e-02 -2.20047265e-01 -7.71536827e-01
3.28621507e-01 -1.11113286e+00 -4.48387206e-01 4.83140737e-01
-1.59061193e-01 1.41139045e-01 5.36857307e-01 -8.08096707e-01
1.07632101e+00 -2.00687122e+00 5.19347072e-01 2.61152685e-01
2.83912450e-01 -2.71165967e-01 -1.26036674e-01 1.00347534e-01
-1.79879889e-01 2.58470178e-01 -3.30139011e-01 -8.83173048e-02
-2.28240877e-01 1.74640507e-01 4.07377720e-01 4.09318388e-01
9.92047042e-02 7.81221509e-01 -6.03340387e-01 -9.78871346e-01
3.90721440e-01 2.93206692e-01 -5.72307348e-01 2.39438519e-01
2.02173352e-01 4.92979050e-01 -3.52265358e-01 7.75690675e-01
1.32001221e-01 -3.63204837e-01 7.47773796e-02 -6.45050764e-01
3.15951526e-01 -2.70359516e-01 -9.27113533e-01 2.12758088e+00
-7.71534562e-01 4.08088744e-01 -2.90036917e-01 -1.08125615e+00
6.91281974e-01 3.71907234e-01 9.49308813e-01 -8.98061037e-01
-2.32966617e-03 6.01653814e-01 2.05601037e-01 -7.14289844e-01
1.05470806e-01 -3.17840874e-01 -5.44181243e-02 4.06404972e-01
-9.57668573e-02 -5.92536092e-01 1.32638901e-01 -7.84334689e-02
1.04241216e+00 -3.27678740e-01 2.10363865e-01 -6.78543374e-02
4.52373266e-01 4.63849455e-02 2.41775289e-01 4.68000174e-01
-2.05298871e-01 1.25367332e+00 2.97050536e-01 -3.53004575e-01
-1.51614976e+00 -1.33934140e+00 -4.16293740e-01 4.55461711e-01
-1.06789410e-01 -1.57823443e-01 -5.14964521e-01 -3.25835288e-01
-2.58848500e-02 -5.63327447e-02 -7.62423575e-01 -2.77933955e-01
-7.01828122e-01 -7.73439288e-01 3.35290700e-01 3.06538820e-01
9.81350094e-02 -8.24173272e-01 -4.80488360e-01 1.87874228e-01
-1.06625505e-01 -6.69914603e-01 -6.14478469e-01 -1.84099883e-01
-1.28190422e+00 -1.26781559e+00 -1.56266010e+00 -9.57373023e-01
6.47652149e-01 -2.93362886e-04 6.71115875e-01 -1.29718617e-01
-8.55991125e-01 6.49275541e-01 -2.46542513e-01 1.44336611e-01
-6.26253009e-01 -5.01608029e-02 2.15720788e-01 -3.04566417e-02
-1.92866087e-01 -5.00967681e-01 -1.16308987e+00 2.96047330e-01
-1.25137269e+00 8.23074430e-02 1.00836992e+00 1.10403895e+00
1.05488753e+00 -2.23724037e-01 4.95773643e-01 -8.05852473e-01
9.01775837e-01 -3.31651032e-01 -5.16428426e-02 6.31659329e-01
-9.47073221e-01 3.26014459e-01 1.61000773e-01 -5.16975403e-01
-6.66389942e-01 -1.95670038e-01 6.19804822e-02 -6.25328541e-01
-1.26890829e-02 8.11815143e-01 3.03140253e-01 -2.00442389e-01
8.51105988e-01 1.10358916e-01 6.08270168e-01 -4.21105742e-01
3.50674808e-01 8.59154463e-01 4.43742096e-01 -5.39187074e-01
2.05905274e-01 4.32331413e-01 1.77224576e-01 -6.51189029e-01
-3.46861541e-01 -2.64902711e-01 -5.59918582e-01 -3.22338104e-01
8.76626194e-01 -7.59937346e-01 -1.29637450e-01 3.42046976e-01
-6.96960330e-01 3.07841506e-03 -1.50311083e-01 9.49932814e-01
-8.54307652e-01 4.14850473e-01 -7.27786779e-01 -8.54503214e-02
-4.45588142e-01 -1.74164939e+00 1.23310161e+00 -5.64051159e-02
-2.80307502e-01 -9.01535749e-01 3.50946426e-01 5.93467236e-01
4.48054343e-01 2.94165462e-01 1.36777949e+00 -4.73060369e-01
-4.06797349e-01 -2.54573554e-01 -2.07531840e-01 5.59121609e-01
6.39071941e-01 -1.24879122e-01 -4.51693177e-01 -1.17848873e-01
-1.13344692e-01 -4.55455810e-01 6.02229178e-01 6.51580691e-01
1.35741282e+00 -3.60379554e-02 -7.37562776e-02 3.55787873e-01
1.44148397e+00 1.71182118e-02 7.02566147e-01 5.46210885e-01
6.50881886e-01 8.55366707e-01 7.20858514e-01 1.40121281e-01
2.64571398e-01 7.85803854e-01 3.15061510e-02 -8.33622459e-03
-5.92984557e-01 1.36482865e-01 3.57067920e-02 1.26464450e+00
2.59045474e-02 2.33442277e-01 -1.10496926e+00 6.59550369e-01
-1.60312510e+00 -5.73875189e-01 1.00666106e-01 2.17525291e+00
1.02800846e+00 -8.16611759e-03 -6.30952716e-02 -2.36625634e-02
6.37337267e-01 2.42430158e-03 -4.04432118e-01 -2.58568555e-01
-9.75360870e-02 2.99993157e-01 4.78086829e-01 2.74066150e-01
-7.93528020e-01 1.92936108e-01 6.83542156e+00 8.68348300e-01
-1.22322655e+00 2.59066910e-01 7.04183817e-01 -3.24119121e-01
-3.65666509e-01 -3.18553507e-01 -1.43263161e-01 3.36242944e-01
1.06816351e+00 -3.27283233e-01 7.75973126e-02 4.93704021e-01
2.93775320e-01 -2.26617530e-01 -1.10777473e+00 1.25542963e+00
2.02210054e-01 -1.57242489e+00 2.72153437e-01 1.71552636e-02
6.04757905e-01 -2.75180824e-02 2.95916140e-01 -1.40688181e-01
-2.60650456e-01 -8.81498516e-01 1.30952716e-01 1.07783437e+00
9.68861759e-01 -4.89240021e-01 7.09125042e-01 -2.83733189e-01
-8.68496418e-01 2.13757500e-01 -2.41601229e-01 5.87307930e-01
6.55062050e-02 5.61748743e-01 -6.63246393e-01 5.98835111e-01
6.30442858e-01 5.26946545e-01 -7.35065341e-01 1.16818690e+00
3.77305746e-01 6.99092075e-02 -1.72156155e-01 3.09086740e-01
2.50918488e-03 -1.67333871e-01 2.96245158e-01 6.44887209e-01
6.70358658e-01 -6.03437200e-02 6.93399757e-02 5.47846258e-01
5.59122711e-02 5.60737431e-01 -7.53655314e-01 4.35036570e-02
1.10534593e-01 1.08738136e+00 -5.86472809e-01 -2.20741332e-01
-5.81649601e-01 8.89062107e-01 -1.66028515e-01 6.52522966e-02
-4.78885323e-01 -2.45927840e-01 3.76452148e-01 2.60452986e-01
-1.87339693e-01 5.10843471e-02 -1.20567784e-01 -1.00213599e+00
1.32624000e-01 -1.09758365e+00 6.07661843e-01 -9.60071683e-01
-1.51611590e+00 4.81357694e-01 1.92153037e-01 -1.53946829e+00
-4.17455375e-01 -7.08117723e-01 -6.04011491e-02 6.81359053e-01
-1.23891926e+00 -1.17638981e+00 -5.63713424e-02 4.80253041e-01
4.21275884e-01 -3.60678583e-01 8.90954077e-01 4.73803997e-01
-1.28093019e-01 3.92666221e-01 3.73273760e-01 9.29702222e-02
1.16659081e+00 -1.26318371e+00 -5.01856089e-01 1.36601418e-01
-1.12573922e-01 3.30979705e-01 4.52129036e-01 -5.75096309e-01
-1.26670086e+00 -8.52359772e-01 5.83261847e-01 -1.80130929e-01
7.88326859e-01 2.60400563e-01 -7.69701004e-01 1.82280481e-01
-7.81766791e-03 3.04940939e-01 1.08702767e+00 -3.18461180e-01
-2.61413634e-01 -1.76143304e-01 -1.15752816e+00 8.37813795e-01
5.74006319e-01 -5.81878662e-01 -7.19837308e-01 3.75781178e-01
5.28336585e-01 -2.85558581e-01 -1.76128531e+00 4.89369184e-01
8.34434628e-01 -5.62268257e-01 1.04912996e+00 -6.67310596e-01
8.27192605e-01 -7.45680630e-02 -5.73118687e-01 -1.25439537e+00
9.16335434e-02 2.38627732e-01 2.60953069e-01 7.32553184e-01
3.67444396e-01 -3.45022708e-01 6.60655499e-01 7.54143417e-01
-1.95545435e-01 -8.39203298e-01 -1.05785441e+00 -6.37002707e-01
4.05960798e-01 -1.42290041e-01 1.74874872e-01 7.97711372e-01
-1.21575728e-01 -1.81977928e-01 -3.37105751e-01 -3.64729110e-03
5.57975888e-01 2.92148173e-01 3.85087103e-01 -1.00703943e+00
-3.86661708e-01 -5.31501532e-01 -7.76684523e-01 -2.34310180e-01
-1.24208763e-01 -1.13182282e+00 -1.66266233e-01 -1.57747471e+00
3.83368552e-01 -4.50102866e-01 -5.88785291e-01 1.41561151e-01
1.77804112e-01 6.15491152e-01 8.19195211e-02 4.29207355e-01
-2.13709354e-01 4.51249719e-01 1.71323812e+00 -5.37475944e-01
-2.72113178e-02 -2.75953531e-01 -3.42910290e-01 4.80148911e-01
6.18391573e-01 -6.19942844e-01 -4.82822627e-01 -2.97367692e-01
1.81153454e-02 3.47885489e-01 4.54031587e-01 -1.05410552e+00
1.53658926e-01 -6.13966919e-02 3.63567501e-01 -2.68067956e-01
2.81728923e-01 -7.43292212e-01 3.17752898e-01 6.28018260e-01
-5.19076169e-01 1.20095834e-01 -1.38458416e-01 3.84292245e-01
-7.11588860e-01 -3.26927006e-01 8.22061062e-01 -3.34443390e-01
-5.53231359e-01 3.97494048e-01 -2.39908755e-01 -4.91547212e-02
9.01890278e-01 -3.40484917e-01 -8.92868415e-02 -1.25258267e-01
-1.17942071e+00 -4.06366475e-02 4.26159024e-01 6.91476047e-01
1.09625101e+00 -1.78566623e+00 -8.19089651e-01 -1.61670417e-01
5.31986237e-01 -2.35951439e-01 4.35238242e-01 1.19872642e+00
-7.83723891e-01 3.32828581e-01 -4.90439594e-01 -9.02937114e-01
-1.25236368e+00 2.56585896e-01 4.26468045e-01 -4.04718786e-01
-5.97293019e-01 2.74497390e-01 1.42685220e-01 -4.64161277e-01
-9.97777134e-02 -2.50519574e-01 1.93909667e-02 2.99072057e-01
4.44800377e-01 1.47759601e-01 3.29522789e-01 -5.22283316e-01
-1.80757537e-01 9.17082429e-01 -4.21921998e-01 -3.29603255e-01
1.28636861e+00 -1.78849220e-01 -4.83407676e-02 7.28011787e-01
1.64084053e+00 -4.05032992e-01 -6.24407291e-01 -2.32453644e-01
-6.17279857e-02 -5.75942039e-01 3.28960627e-01 -5.13937294e-01
-1.33998299e+00 7.33880341e-01 1.41745830e+00 -2.31120259e-01
1.22097445e+00 1.88552752e-01 7.58854985e-01 2.19022393e-01
4.98370349e-01 -1.07799828e+00 2.81750619e-01 -2.67796934e-01
1.08148885e+00 -1.45582795e+00 6.10683486e-02 8.35685432e-03
-5.65985918e-01 1.21508026e+00 3.48075867e-01 -1.09939054e-01
6.46539390e-01 -1.19210705e-02 3.11957717e-01 -2.98786283e-01
-4.97522593e-01 -2.08902415e-02 5.49949348e-01 4.81815517e-01
4.94191229e-01 8.79991427e-02 -7.72685766e-01 2.41212785e-01
4.20713723e-02 4.88890596e-02 3.49239081e-01 9.10484076e-01
-2.67739296e-01 -1.60422969e+00 -5.88553905e-01 9.97115374e-01
-3.95477176e-01 9.15046036e-02 -1.39170974e-01 9.27832067e-01
-1.05474398e-01 3.41721147e-01 -5.34477904e-02 -2.75507390e-01
3.53741169e-01 5.51141426e-02 8.03843021e-01 -5.18770993e-01
-1.11496404e-01 1.69408932e-01 -1.63223058e-01 -6.45468354e-01
-6.39296532e-01 -6.90781355e-01 -1.04841924e+00 -5.07422313e-02
-1.47201329e-01 -1.66277856e-01 7.06034541e-01 7.06683874e-01
2.28691667e-01 5.63487172e-01 6.73838139e-01 -6.37435675e-01
-3.33415449e-01 -7.84914553e-01 -7.36793756e-01 7.05761194e-01
1.46825701e-01 -8.75742257e-01 -2.35837698e-01 2.48180151e-01] | [14.356610298156738, -1.6409391164779663] |
d35a09da-26f7-47d0-9428-840c66b7f04d | neural-machine-translation-for-low-resource-3 | 2304.07869 | null | https://arxiv.org/abs/2304.07869v2 | https://arxiv.org/pdf/2304.07869v2.pdf | Neural Machine Translation For Low Resource Languages | Neural Machine translation is a challenging task due to the inherent complex nature and the fluidity that natural languages bring. Nonetheless, in recent years, it has achieved state-of-the-art performance in several language pairs. Although, a lot of traction can be seen in the areas of multilingual neural machine translation (MNMT) in the recent years, there are no comprehensive survey done to identify what approaches work well. The goal of this paper is to investigate the realm of low resource languages and build a Neural Machine Translation model to achieve state-of-the-art results. The paper looks to build upon the mBART language model and explore strategies to augment it with various NLP and Deep Learning techniques like back translation and transfer learning. This implementation tries to unpack the architecture of the NMT application and determine the different components which offers us opportunities to amend the said application within the purview of the low resource languages problem space. | ['Utsa Chattopadhyay', 'Kannan Girija Ravikumar', 'Parvathy Krishnaswamy', 'Kartikay Goyle', 'Vakul Goyle'] | 2023-04-16 | null | null | null | null | ['nmt'] | ['computer-code'] | [ 2.21417710e-01 -2.41222866e-02 -5.88741422e-01 -2.29945928e-01
-1.03275001e+00 -4.44118649e-01 1.00756514e+00 -4.68267977e-01
-5.56676388e-01 1.09405708e+00 1.71688661e-01 -9.42266405e-01
1.19173117e-01 -6.04502916e-01 -8.33653629e-01 -4.15892422e-01
3.35895330e-01 9.99361157e-01 -2.96125948e-01 -7.63859034e-01
1.03816073e-02 4.29183960e-01 -1.04028726e+00 5.69656253e-01
8.83607566e-01 4.60909039e-01 1.38742030e-01 3.21660489e-01
-4.59370136e-01 5.35035968e-01 -3.27434540e-01 -7.65019000e-01
4.09456134e-01 -7.02244341e-01 -1.12145829e+00 -4.99254107e-01
4.40963984e-01 -8.40719715e-02 -8.80172551e-02 8.57072473e-01
7.42584884e-01 -1.88004509e-01 4.67281371e-01 -9.80469048e-01
-1.07883656e+00 7.60555267e-01 -3.24167460e-01 2.81415254e-01
3.68154377e-01 -2.89751999e-02 8.04332912e-01 -1.18529058e+00
7.55696833e-01 1.23977828e+00 7.03806102e-01 4.39438105e-01
-1.04599667e+00 -7.18865454e-01 -1.91285193e-01 2.53434211e-01
-1.02827990e+00 -4.95948941e-01 4.70200837e-01 -2.08031029e-01
1.49268293e+00 6.70619980e-02 3.93226892e-01 1.38212550e+00
5.70887506e-01 7.60960340e-01 1.55539775e+00 -1.01609588e+00
-3.06892753e-01 3.77437621e-01 -1.47196546e-01 5.24627507e-01
-1.53436750e-01 2.89476335e-01 -6.91788614e-01 5.17598428e-02
6.71725631e-01 -4.33011323e-01 8.00132379e-02 1.22258421e-02
-1.51348734e+00 7.86151052e-01 2.63450265e-01 9.89440024e-01
-2.00226247e-01 -1.28856108e-01 4.72608298e-01 8.40812266e-01
7.58315980e-01 6.06530666e-01 -6.97750688e-01 -3.86039853e-01
-1.11711705e+00 1.30272269e-01 9.49123025e-01 7.74974942e-01
7.60229528e-01 1.12694763e-01 -2.95148827e-02 9.59371924e-01
3.69916577e-03 4.39130604e-01 5.80975175e-01 -4.76858109e-01
9.10055637e-01 4.21839595e-01 -2.15843543e-01 -6.54049397e-01
-1.75735936e-01 -5.11644244e-01 -8.62984478e-01 9.25779417e-02
5.12667179e-01 -2.85891742e-01 -6.63984120e-01 1.63906729e+00
-5.96723370e-02 -3.98981005e-01 1.26866192e-01 7.97369957e-01
5.87951660e-01 9.23393786e-01 -1.76142454e-01 -7.58417100e-02
1.01657319e+00 -1.10448277e+00 -7.02387094e-01 -3.56360555e-01
7.23079622e-01 -1.23138690e+00 1.08298922e+00 9.69058871e-02
-1.23409235e+00 -5.82170129e-01 -9.20966685e-01 -3.53541374e-01
-6.72570527e-01 1.16918996e-01 7.04485834e-01 6.62818968e-01
-1.30676234e+00 5.94142020e-01 -7.99102128e-01 -9.05622065e-01
-2.35071834e-02 5.89489460e-01 -5.44418693e-01 -1.21040545e-01
-1.50845301e+00 1.74028659e+00 3.41821164e-01 2.46808693e-01
-4.17274714e-01 -4.08001035e-01 -5.05156875e-01 -3.45639586e-01
1.33339763e-01 -9.21169400e-01 1.28396010e+00 -1.35356879e+00
-1.72929263e+00 1.20569360e+00 -1.87586755e-01 -4.17437464e-01
7.11988330e-01 -1.96273685e-01 -3.44132364e-01 -4.20110554e-01
1.01272546e-01 8.10088933e-01 2.45319128e-01 -7.70368874e-01
-5.30439854e-01 -3.12302232e-01 -1.15948103e-01 2.94740736e-01
-3.20894271e-01 7.74448872e-01 -2.49213368e-01 -6.64931118e-01
-1.78077474e-01 -1.18543577e+00 -4.12123203e-02 -5.48078239e-01
-8.14721957e-02 -2.31242463e-01 3.50006461e-01 -8.42419505e-01
9.94725883e-01 -1.67820656e+00 5.61961174e-01 -2.69472361e-01
-2.39626691e-01 3.67817789e-01 -3.14545453e-01 9.78093922e-01
-9.14433971e-02 5.73031157e-02 -2.98400730e-01 -3.32330316e-01
9.28545278e-03 3.11333507e-01 -2.95083195e-01 3.39460701e-01
3.37780416e-01 1.35574925e+00 -6.83363676e-01 -3.22810531e-01
-1.56097347e-02 5.02641141e-01 -1.77791551e-01 9.05146226e-02
-1.01114906e-01 5.66109002e-01 -1.70833647e-01 7.07212746e-01
3.98049384e-01 2.78427035e-01 1.93414271e-01 1.77111819e-01
-5.28851867e-01 6.86575413e-01 -4.90700543e-01 1.88360333e+00
-5.15151918e-01 1.00799632e+00 3.99786383e-02 -1.14557981e+00
9.40454721e-01 5.68673313e-01 2.39547804e-01 -9.43027258e-01
1.76903576e-01 9.48853493e-01 3.87419939e-01 -4.67922032e-01
6.15498543e-01 -4.27283674e-01 6.70767855e-03 5.47666907e-01
2.10824713e-01 2.63380744e-02 2.00057961e-02 -3.78876776e-01
5.79195142e-01 6.00896835e-01 3.61734152e-01 -3.75706762e-01
4.86461043e-01 2.92237818e-01 2.27071494e-01 4.05107945e-01
-4.75445623e-03 3.15802634e-01 1.96959637e-02 -7.78777897e-01
-1.40809214e+00 -7.95656562e-01 -3.78779881e-02 1.19066560e+00
-5.09056151e-01 6.32884800e-02 -7.35799789e-01 -4.37333286e-01
-4.10842240e-01 6.03627145e-01 -5.14863074e-01 1.49423167e-01
-1.06404531e+00 -1.07570028e+00 8.84266853e-01 3.03400159e-01
5.59420407e-01 -1.26934445e+00 -2.13862300e-01 3.30295622e-01
-4.32607174e-01 -1.01269591e+00 -1.65555224e-01 2.53976375e-01
-9.82530236e-01 -4.58318830e-01 -8.97574067e-01 -1.19151974e+00
1.67082503e-01 1.95994657e-02 1.40166974e+00 -3.23963523e-01
2.36043483e-01 -1.10832378e-01 -3.07154566e-01 -4.78671491e-01
-8.56097102e-01 8.07759464e-01 1.83249533e-01 -3.60798091e-01
8.47089112e-01 -7.06629992e-01 9.38742310e-02 6.70378283e-02
-6.68924630e-01 2.41688609e-01 9.84259546e-01 9.13061380e-01
3.55735421e-01 -5.97671449e-01 6.41108453e-01 -7.57586420e-01
9.46579516e-01 -4.28088665e-01 -3.18660766e-01 3.74509126e-01
-6.72257364e-01 -6.07395731e-02 5.25815666e-01 -4.16242003e-01
-7.42220104e-01 -1.55306667e-01 -2.54721105e-01 -1.37732672e-02
-1.14347063e-01 7.03672707e-01 -7.02955993e-03 -1.20849803e-01
7.07100213e-01 4.05745804e-01 1.07470013e-01 -5.31309426e-01
4.71484542e-01 8.86308134e-01 3.39715332e-01 -7.57489383e-01
7.18722403e-01 -7.26267472e-02 -1.00109473e-01 -5.58092117e-01
-4.45192814e-01 -8.74884501e-02 -9.97957349e-01 -2.06777956e-02
7.80835509e-01 -7.43622005e-01 -1.02435332e-02 3.08255821e-01
-1.46500814e+00 -3.52547854e-01 4.80794720e-02 5.46554685e-01
-7.18406856e-01 9.11257342e-02 -1.01232731e+00 -4.95750695e-01
-7.16257393e-01 -1.46478140e+00 8.36706877e-01 -2.40668893e-01
-4.49863076e-01 -1.16898263e+00 2.51286626e-01 3.80782992e-01
7.80309141e-01 -3.88354808e-02 1.15880334e+00 -7.17231333e-01
-5.69750547e-01 -4.38195691e-02 -2.01411217e-01 3.50990713e-01
9.97672305e-02 -2.34589115e-01 -7.80958414e-01 -2.53302306e-01
3.29934694e-02 -2.38431975e-01 4.90357488e-01 2.59301901e-01
2.15711579e-01 -2.59377182e-01 -1.82300150e-01 4.97424304e-01
1.35478640e+00 1.39233530e-01 6.45701826e-01 8.15338612e-01
4.72888052e-01 8.40143621e-01 3.19088072e-01 -4.74192351e-01
4.16056395e-01 8.19360971e-01 9.26927850e-02 -3.96797150e-01
-2.32397258e-01 -2.01764971e-01 6.15753174e-01 1.35199296e+00
-2.94767141e-01 1.78116595e-03 -1.24166703e+00 4.87544239e-01
-1.85382247e+00 -7.95768321e-01 -2.05427185e-01 1.86896348e+00
9.53168511e-01 -2.20034588e-02 8.21030065e-02 -1.23973168e-01
5.39574623e-01 -9.16510373e-02 -1.27378898e-02 -1.08056331e+00
-4.65387493e-01 2.85158783e-01 3.13904822e-01 5.71042180e-01
-7.12211311e-01 1.45748389e+00 7.12131834e+00 6.59821510e-01
-1.33002222e+00 3.30497473e-01 5.93116939e-01 5.79468161e-02
-1.40195340e-01 -1.13961950e-01 -8.47378194e-01 1.42225921e-01
1.22833848e+00 -6.06737807e-02 8.70360434e-01 4.50037241e-01
1.92857891e-01 2.88466841e-01 -1.31262350e+00 6.28919303e-01
3.29681128e-01 -1.32757652e+00 2.15184748e-01 2.47671828e-01
8.59264731e-01 7.03778803e-01 2.81657696e-01 6.80664480e-01
2.37848878e-01 -1.17734218e+00 5.09561181e-01 1.85395598e-01
8.25707793e-01 -6.69168353e-01 8.80375504e-01 6.85042262e-01
-5.17125309e-01 1.02146432e-01 -3.83945107e-01 -4.49104100e-01
2.51234397e-02 2.59880781e-01 -8.71812999e-01 7.96690106e-01
5.98438978e-01 4.32778567e-01 -2.47679874e-01 7.09271491e-01
-1.35254040e-01 5.00324070e-01 -1.68419808e-01 -1.49914309e-01
6.93439782e-01 -5.45256972e-01 4.16526109e-01 1.51029718e+00
5.55407107e-01 -3.72646272e-01 9.72557440e-02 7.32978284e-01
-2.55393565e-01 7.17220426e-01 -9.22647536e-01 -2.35496506e-01
-1.53540308e-03 9.49824035e-01 -3.44410121e-01 -2.53938407e-01
-6.93648636e-01 1.11816680e+00 6.01703167e-01 3.65224004e-01
-5.32459140e-01 -3.70667665e-03 5.11953115e-01 5.11848517e-02
-2.22242177e-01 -3.93265098e-01 -4.90542978e-01 -1.19543135e+00
8.62597227e-02 -1.39326727e+00 -2.79000644e-02 -7.42779136e-01
-1.29864013e+00 9.96086419e-01 -6.85282797e-02 -1.05193663e+00
-7.16113389e-01 -7.35416949e-01 -3.22309881e-01 1.40449965e+00
-1.51365077e+00 -1.54189384e+00 4.33111757e-01 2.37546816e-01
7.31042027e-01 -6.66709900e-01 1.21737909e+00 5.82516491e-01
-2.86725312e-01 6.94410205e-01 3.78500074e-01 2.18324170e-01
9.38010335e-01 -8.86902988e-01 6.91888869e-01 7.88972378e-01
3.42794031e-01 7.95705140e-01 6.76735401e-01 -5.11347234e-01
-1.47287941e+00 -7.86876082e-01 1.78671134e+00 -8.03057611e-01
8.56843352e-01 -3.68694574e-01 -6.44987941e-01 8.84386957e-01
8.86750579e-01 -5.32270670e-01 6.89074218e-01 2.24083588e-01
-3.51292700e-01 8.17565396e-02 -7.55290866e-01 8.21731031e-01
7.97943056e-01 -7.58005857e-01 -8.87041926e-01 3.76620591e-01
6.29461467e-01 -4.33745205e-01 -7.82776356e-01 4.78884995e-01
6.49047613e-01 -7.01373696e-01 7.53781438e-01 -8.31716120e-01
7.50107884e-01 -2.01064665e-02 -2.49682397e-01 -1.57745266e+00
-3.43571991e-01 -7.47974396e-01 2.96705246e-01 1.24519098e+00
9.05649066e-01 -8.28497350e-01 5.75708926e-01 2.45406643e-01
-1.72145545e-01 -9.59154904e-01 -1.23686683e+00 -7.07440436e-01
1.01840818e+00 -3.86970371e-01 5.72002590e-01 1.29516792e+00
1.23760730e-01 7.63481438e-01 -5.67504168e-01 -3.44474822e-01
1.33643776e-01 1.60359904e-01 8.04202199e-01 -1.00750613e+00
-3.89802009e-01 -6.76546276e-01 -3.95000353e-02 -7.05593705e-01
3.52429092e-01 -1.39990091e+00 -2.60072708e-01 -1.76169038e+00
2.22334802e-01 -1.91690028e-01 -1.74731821e-01 5.17757893e-01
2.29236912e-02 4.87324059e-01 3.40569556e-01 4.31269079e-01
-7.58507103e-02 2.49103248e-01 1.15402544e+00 -2.22429231e-01
-8.42705593e-02 -1.82398167e-02 -6.39241934e-01 2.94958860e-01
9.87472773e-01 -4.27352369e-01 -1.07938321e-02 -1.16904783e+00
4.63147849e-01 -3.06799859e-02 -1.24427378e-01 -5.97842038e-01
9.17666182e-02 3.34934443e-02 3.33658010e-01 -4.59574550e-01
3.07465762e-01 -7.78491497e-01 2.03884244e-01 3.42377484e-01
-3.93714637e-01 7.89989889e-01 3.56181234e-01 -1.75869510e-01
-4.15129572e-01 1.33523578e-02 5.61036587e-01 -4.92151499e-01
-3.47713679e-01 1.05428360e-02 -5.54553092e-01 -2.02566326e-01
5.45369506e-01 -2.17913792e-01 -7.82500952e-02 -2.03490183e-01
-3.18110526e-01 -2.48359852e-02 5.10341167e-01 8.27583909e-01
6.49290681e-02 -1.31258941e+00 -1.09257042e+00 8.68296251e-02
1.14270531e-01 -5.67135870e-01 -3.79326403e-01 9.00370300e-01
-5.21158695e-01 9.07582939e-01 -5.66230059e-01 -3.38435590e-01
-9.21820879e-01 5.22016585e-01 5.92606723e-01 -5.17552912e-01
-4.22126800e-01 5.75133502e-01 -2.45667532e-01 -1.03504550e+00
-4.16131504e-02 -1.64694935e-02 -3.63623910e-02 -3.70974429e-02
2.76767582e-01 2.23498285e-01 3.23800921e-01 -7.98577189e-01
-9.52747613e-02 4.88908231e-01 -1.00286603e-01 -4.24548388e-01
1.32272673e+00 -2.43114568e-02 -6.79201066e-01 6.14514053e-01
1.11745441e+00 -1.60690323e-01 -3.95548016e-01 -3.64025146e-01
2.60322422e-01 -8.74398351e-02 -1.90377027e-01 -1.16772354e+00
-5.95205963e-01 1.28079641e+00 5.59378862e-01 -2.37591371e-01
9.35743868e-01 -3.02234024e-01 8.74065757e-01 4.54591423e-01
4.99519080e-01 -1.08391285e+00 -5.28564513e-01 9.55567241e-01
8.57240558e-01 -1.17148316e+00 -3.63105983e-01 -1.10836200e-01
-2.50820935e-01 1.23168254e+00 3.92874986e-01 -5.80858323e-04
1.67567343e-01 4.38562781e-01 4.25642461e-01 3.07708174e-01
-6.51468635e-01 -1.76977869e-02 1.58705398e-01 5.36181271e-01
1.07336235e+00 1.75790206e-01 -7.81833053e-01 1.26562357e-01
-5.72217762e-01 1.42791837e-01 1.46041885e-01 6.52137816e-01
-1.70603156e-01 -1.83423209e+00 -4.63137537e-01 1.23964593e-01
-9.70555723e-01 -5.05668581e-01 -8.43626618e-01 9.98169780e-01
1.01231508e-01 9.13538694e-01 -2.72915274e-01 -3.81913215e-01
2.19779164e-01 5.11335015e-01 6.82885289e-01 -5.17659128e-01
-1.16018260e+00 -1.98211614e-02 3.51342738e-01 -9.23700109e-02
-4.54567671e-01 -6.20915473e-01 -7.43993223e-01 -5.03411055e-01
-1.33136988e-01 2.07519040e-01 1.03919721e+00 1.17841113e+00
1.90233365e-01 2.03107506e-01 1.04949772e-01 -9.74158406e-01
-5.87298989e-01 -1.24451017e+00 1.33843794e-01 2.01536361e-02
8.98516402e-02 -7.86881298e-02 1.43800244e-01 -1.50973484e-01] | [11.541364669799805, 10.300251007080078] |
a1344d5f-ec22-469b-8952-9c5771c7b80d | the-best-path-algorithm-automatic-variables | 2211.07267 | null | https://arxiv.org/abs/2211.07267v2 | https://arxiv.org/pdf/2211.07267v2.pdf | The Best Path Algorithm automatic variables selection via High Dimensional Graphical Models | This paper proposes a new algorithm for an automatic variable selection procedure in High Dimensional Graphical Models. The algorithm selects the relevant variables for the node of interest on the basis of mutual information. Several contributions in literature have investigated the use of mutual information in selecting the appropriate number of relevant features in a large data-set, but most of them have focused on binary outcomes or required high computational effort. The algorithm here proposed overcomes these drawbacks as it is an extension of Chow and Liu's algorithm. Once, the probabilistic structure of a High Dimensional Graphical Model is determined via the said algorithm, the best path-step, including variables with the most explanatory/predictive power for a variable of interest, is determined via the computation of the entropy coefficient of determination. The latter, being based on the notion of (symmetric) Kullback-Leibler divergence, turns out to be closely connected to the mutual information of the involved variables. The application of the algorithm to a wide range of real-word and publicly data-sets has highlighted its potential and greater effectiveness compared to alternative extant methods. | ['Consuelo R. Nava', 'Maria G. Zoia', 'Luigi Riso'] | 2022-11-14 | null | null | null | null | ['variable-selection'] | ['methodology'] | [ 2.74102747e-01 1.56217009e-01 -3.57907474e-01 -4.53667581e-01
-5.52273214e-01 -3.35766673e-01 7.00366974e-01 4.84682530e-01
-3.94640237e-01 9.55652475e-01 -8.70461911e-02 -4.20544207e-01
-1.02101159e+00 -9.13602412e-01 5.72805703e-02 -8.30067098e-01
-5.01836121e-01 5.85704505e-01 -3.19763087e-02 1.33745730e-01
5.17074049e-01 5.50179541e-01 -1.81121123e+00 -4.76766229e-01
7.54181802e-01 8.57716441e-01 3.32739949e-01 4.98537809e-01
-3.08644086e-01 3.48190129e-01 -2.77936727e-01 -5.98315656e-01
-2.30768956e-02 -7.01031864e-01 -8.97696912e-01 -7.50167072e-02
-2.63285488e-01 3.07609916e-01 1.65506616e-01 8.96522462e-01
4.86562937e-01 4.90886234e-02 1.14571667e+00 -1.22082996e+00
-1.44354537e-01 6.03543103e-01 -3.73456627e-01 1.49932802e-01
5.02202392e-01 -4.15472567e-01 1.51312816e+00 -9.24723685e-01
7.51627088e-01 1.18131745e+00 4.96119946e-01 -1.17218494e-01
-1.51858974e+00 -5.06020844e-01 -4.32470262e-01 1.97141394e-01
-1.59520030e+00 -3.18694152e-02 1.00207591e+00 -6.98118567e-01
6.66153491e-01 5.07676780e-01 5.24498284e-01 3.81614625e-01
1.76225588e-01 3.34215879e-01 1.14242220e+00 -7.56495297e-01
3.61172229e-01 3.83980930e-01 2.42811084e-01 6.89325571e-01
4.24090505e-01 2.54410505e-01 -5.04110456e-01 -5.93009472e-01
4.39951330e-01 -3.47917676e-01 -3.87468897e-02 -8.79642725e-01
-8.03723454e-01 1.15312374e+00 1.24458462e-01 5.91909528e-01
-3.12678516e-01 -3.64768177e-01 1.55941516e-01 3.56084824e-01
3.95000070e-01 4.18106973e-01 -4.79046464e-01 1.32631501e-02
-9.00892913e-01 1.60212830e-01 9.90555644e-01 5.27090728e-01
8.44732046e-01 -5.16928911e-01 6.65735379e-02 5.51806867e-01
8.24824154e-01 1.88884556e-01 2.24446714e-01 -2.28014737e-01
2.67752528e-01 9.59223866e-01 -1.45984501e-01 -1.52109969e+00
-5.34095466e-01 -3.73735636e-01 -7.33651757e-01 2.76379853e-01
2.99516469e-01 -2.54476726e-01 -4.43638921e-01 1.66408849e+00
5.38422465e-01 -2.93685019e-01 -1.20938681e-02 5.93396902e-01
7.35383749e-01 1.76602900e-01 2.05569774e-01 -6.43544912e-01
1.03386307e+00 -3.52690667e-02 -7.86641657e-01 2.66596824e-01
4.35726225e-01 -6.57457590e-01 3.30320626e-01 1.71854123e-01
-6.55147970e-01 -2.35855207e-01 -1.03765965e+00 4.85978842e-01
-4.89111125e-01 -7.93766454e-02 8.48797143e-01 7.19360352e-01
-1.08512259e+00 6.90875769e-01 -2.81229079e-01 -3.61978024e-01
-6.49189800e-02 7.27249146e-01 -5.32600403e-01 3.79129976e-01
-1.43695605e+00 1.07073975e+00 4.15493697e-01 -7.94451535e-02
1.04237795e-02 -6.38582036e-02 -8.38351190e-01 -7.41705224e-02
1.30167410e-01 -7.59027004e-01 5.23170590e-01 -9.06921685e-01
-1.34840333e+00 8.04542899e-01 -2.31280252e-01 -3.40088665e-01
5.52340984e-01 1.55307695e-01 -3.84551018e-01 7.65684322e-02
1.22614853e-01 1.50904268e-01 5.99474788e-01 -1.13810027e+00
-5.26136637e-01 -6.41670108e-01 -4.68517572e-01 1.97979197e-01
-2.37908870e-01 6.15588389e-02 -3.98369610e-01 -4.70098704e-01
5.94383657e-01 -7.01834798e-01 -3.67311418e-01 -2.49472618e-01
-3.22237283e-01 -3.43303829e-01 4.94403601e-01 -3.88772249e-01
1.71701074e+00 -1.83492148e+00 4.63837951e-01 1.02697992e+00
3.18247825e-01 -1.01049021e-02 3.43764901e-01 8.13115954e-01
-2.93381423e-01 2.63042420e-01 -5.40084422e-01 1.02673970e-01
-1.80081978e-01 -8.12155306e-02 2.81051487e-01 6.37962222e-01
2.58697718e-01 3.20791125e-01 -7.54873335e-01 -9.50113714e-01
5.62919080e-01 4.97968227e-01 -1.16549209e-01 8.76437724e-02
1.61933303e-01 1.49286017e-01 -8.04911852e-01 2.22390771e-01
5.50332546e-01 -2.24536955e-01 4.37943429e-01 1.36141479e-01
-2.18437135e-01 3.30376066e-02 -1.50853896e+00 1.10244191e+00
-2.35908199e-02 5.49036443e-01 -4.04838443e-01 -7.68503904e-01
1.34875631e+00 3.52105111e-01 7.09801018e-01 -4.84119564e-01
3.78511161e-01 2.79944718e-01 -5.19548021e-02 -4.97777164e-01
3.06287766e-01 -2.52650291e-01 3.39543149e-02 2.64307559e-01
1.01917498e-01 -3.22136134e-02 3.15770388e-01 1.23401396e-01
7.61837721e-01 6.65977597e-02 1.13797057e+00 -4.34147239e-01
8.05763900e-01 -2.12426871e-01 2.97206283e-01 6.29882932e-01
5.02867848e-02 2.38373399e-01 7.51739383e-01 -1.12303905e-01
-6.40397966e-01 -7.61562049e-01 -5.58312833e-01 4.99013364e-01
1.29961688e-02 -4.35040891e-01 -4.40043330e-01 -5.01833975e-01
1.36566624e-01 6.62077367e-01 -8.22767556e-01 1.26654997e-01
1.19091712e-01 -7.40094781e-01 1.56160325e-01 -2.44710565e-01
2.81283353e-02 -7.98440516e-01 -6.20721877e-01 3.09335351e-01
1.45252600e-01 -2.33794421e-01 4.49043840e-01 4.52106833e-01
-1.08319306e+00 -1.30731201e+00 -4.29040551e-01 -4.73970383e-01
5.45185864e-01 1.44273811e-03 1.02383637e+00 -1.02939449e-01
-2.50381351e-01 1.35926828e-02 -4.66415405e-01 -2.27400035e-01
-2.55408615e-01 -8.87612626e-03 -2.12384716e-01 2.96112746e-02
5.31050146e-01 -5.11747062e-01 -1.48820013e-01 3.34922999e-01
-6.23258948e-01 -2.57045180e-01 7.30801761e-01 8.98038089e-01
6.42069101e-01 3.46638083e-01 5.31719267e-01 -8.80142510e-01
8.10089052e-01 -7.71153629e-01 -7.12428510e-01 1.16655394e-01
-1.05902839e+00 4.55479234e-01 1.22771233e-01 7.55394474e-02
-7.55099237e-01 2.05822587e-01 1.02326550e-01 2.02043980e-01
-9.49251279e-02 9.42272305e-01 -2.14522064e-01 -6.28196076e-02
3.78950328e-01 -3.39287668e-02 6.20953254e-02 -5.57322562e-01
2.17710242e-01 5.42819440e-01 3.53001803e-02 -1.16929926e-01
5.69026470e-01 2.93293875e-02 6.23359382e-01 -8.22005033e-01
-2.45732486e-01 -5.67392826e-01 -9.75173056e-01 -2.81081080e-01
7.00559616e-01 -2.70266861e-01 -7.20257640e-01 2.03943431e-01
-7.44445682e-01 4.96365219e-01 9.51320007e-02 7.46799469e-01
-4.81260151e-01 3.03645909e-01 1.49573997e-01 -1.21657431e+00
-1.91561192e-01 -1.03930008e+00 5.73385894e-01 4.48328489e-03
-6.43034756e-01 -1.24064970e+00 5.22877932e-01 -1.58254907e-01
9.45087597e-02 4.71245736e-01 1.29050052e+00 -8.83675754e-01
-2.00252403e-02 -5.69261849e-01 -5.10807671e-02 -1.47532299e-01
2.60782927e-01 3.79405856e-01 -6.99977040e-01 -6.10906165e-03
-8.83145481e-02 1.78836852e-01 6.27434075e-01 5.27261078e-01
6.07886434e-01 -1.32864732e-02 -5.07784903e-01 2.23795801e-01
1.62589157e+00 1.88037798e-01 3.90798450e-01 4.68221098e-01
2.04098120e-01 7.22273767e-01 8.29745412e-01 7.53382981e-01
3.21845591e-01 7.63046801e-01 4.47958976e-01 -2.02256883e-03
3.77110988e-01 -1.26744345e-01 -3.91927399e-02 5.79190314e-01
5.81787080e-02 -1.62085339e-01 -1.02055991e+00 3.79313916e-01
-1.67472100e+00 -9.83907700e-01 -5.10212481e-01 2.41923308e+00
5.65049350e-01 1.33669630e-01 2.69289136e-01 5.36569357e-01
7.84975946e-01 -2.12100074e-02 -1.74094230e-01 -3.99599761e-01
-2.25677192e-01 4.58704010e-02 4.05241638e-01 5.93596637e-01
-1.13258445e+00 4.41062152e-01 6.74484444e+00 5.99872470e-01
-7.15364516e-01 -4.80020642e-01 5.19119322e-01 2.58953631e-01
-3.13313097e-01 1.66513026e-01 -6.22584581e-01 4.06312436e-01
9.71259356e-01 -3.82541656e-01 -1.27644658e-01 7.07974017e-01
3.58208120e-01 -5.69000244e-01 -9.04954851e-01 6.53387070e-01
6.48374930e-02 -7.80392706e-01 -7.80547261e-02 3.37843001e-01
4.84303236e-01 -3.71550560e-01 2.32184379e-04 -3.60276550e-01
2.03668907e-01 -9.50972259e-01 4.48509276e-01 7.14342952e-01
4.75707978e-01 -1.06307650e+00 9.58701372e-01 6.98596314e-02
-1.08593261e+00 1.19874738e-02 -1.34323854e-02 -1.74910039e-01
1.91870537e-02 9.67101932e-01 -1.10752618e+00 6.96816742e-01
2.63524562e-01 4.88378942e-01 -6.26542807e-01 1.29348326e+00
-1.93151966e-01 4.29832667e-01 -3.01308423e-01 -4.33088362e-01
1.37844488e-01 -4.09078985e-01 7.06478536e-01 1.14713442e+00
4.00330424e-01 -1.07729726e-01 -2.76652306e-01 5.49471736e-01
4.84252363e-01 6.71317041e-01 -8.24419916e-01 2.33034566e-02
6.12556994e-01 9.86472666e-01 -7.45976031e-01 -8.43630210e-02
-3.46853226e-01 5.28382063e-01 1.37519643e-01 -1.15292124e-01
-3.20516080e-01 -4.70696837e-01 3.38355303e-01 -1.76203907e-01
3.82115483e-01 -5.45427874e-02 -6.03339851e-01 -4.07795727e-01
-3.92596163e-02 -5.94413877e-01 7.17126489e-01 -2.91406512e-01
-1.06833160e+00 5.78869522e-01 9.34340656e-02 -1.25631332e+00
-6.44839048e-01 -3.93567592e-01 -2.68159926e-01 1.25491583e+00
-1.09057033e+00 -6.26299679e-01 1.41004905e-01 5.16450524e-01
8.38458613e-02 -6.80566058e-02 1.11508179e+00 1.39977679e-01
-4.20866907e-01 2.07861394e-01 4.03145969e-01 -2.45309517e-01
3.77576411e-01 -1.39829481e+00 -7.55736232e-02 5.82482100e-01
-8.31028074e-02 6.35773897e-01 1.07201636e+00 -8.21233809e-01
-1.25263703e+00 -4.75030631e-01 1.48147285e+00 -2.23216593e-01
8.49788964e-01 4.07515056e-02 -5.91013014e-01 1.70964792e-01
-1.16847947e-01 -5.32407701e-01 1.02405632e+00 4.21253711e-01
9.31612104e-02 2.17841849e-01 -1.20488870e+00 3.78856033e-01
5.99924088e-01 -3.01164269e-01 -4.81467664e-01 2.27645859e-02
9.88479704e-02 2.28613570e-01 -1.09327030e+00 2.30528295e-01
7.47892141e-01 -1.03401613e+00 6.96956575e-01 -4.88681644e-01
1.96885571e-01 -2.84625381e-01 -1.56315774e-01 -1.04114091e+00
-5.34783125e-01 -5.61760604e-01 1.26898706e-01 1.36840224e+00
7.72721887e-01 -4.18984294e-01 4.72252637e-01 6.95148349e-01
6.83278918e-01 -7.64329851e-01 -9.83472705e-01 -3.39154363e-01
-3.60232115e-01 -3.86617810e-01 5.53127229e-01 8.80049407e-01
3.23117375e-01 4.74998593e-01 -4.45289671e-01 -6.81132376e-02
7.50014603e-01 2.71287680e-01 6.32797778e-01 -1.87282729e+00
-7.65644014e-02 -5.70063114e-01 -8.90202105e-01 -1.75274819e-01
-6.42856723e-03 -9.06075418e-01 -2.39692867e-01 -1.45477235e+00
3.22455645e-01 -4.90510792e-01 -3.83695662e-01 1.35793397e-02
-1.78368762e-01 -2.12631509e-01 3.66384052e-02 1.32861346e-01
-2.28969783e-01 2.42432013e-01 9.33483005e-01 3.50727648e-01
-4.99674916e-01 4.28349048e-01 -6.63256228e-01 6.20628774e-01
6.03799701e-01 -6.35948896e-01 -3.04522276e-01 2.74910569e-01
4.25638705e-01 3.57678384e-01 7.06944689e-02 -6.04726791e-01
1.37696207e-01 -2.94461578e-01 2.90566772e-01 -7.03426778e-01
3.84426899e-02 -1.09744847e+00 6.57784343e-01 4.42495227e-01
-5.03638804e-01 2.93628514e-01 -2.57369906e-01 6.51366413e-01
-1.97418153e-01 -5.27116060e-01 5.00295520e-01 1.80334777e-01
-7.13872015e-01 -3.16440016e-02 -3.35362285e-01 -4.51772839e-01
1.14735258e+00 -3.57882947e-01 3.51186901e-01 -3.77939463e-01
-8.72137547e-01 -1.27506983e-02 2.92023242e-01 2.76728690e-01
5.25888741e-01 -1.17326105e+00 -6.71658814e-01 1.91560969e-01
1.49703875e-01 -5.55658221e-01 -1.41921043e-01 8.37841868e-01
-1.56694025e-01 6.16804183e-01 -2.01923192e-01 -4.01150376e-01
-1.56828177e+00 4.32228714e-01 7.68102193e-03 -4.63659495e-01
-3.12406898e-01 6.39155984e-01 -3.70966852e-01 -3.29926461e-02
7.34829828e-02 3.08713406e-01 -7.55048990e-01 6.48654461e-01
3.04092258e-01 5.94491661e-01 7.01849833e-02 -1.04174364e+00
-6.35314822e-01 4.92404222e-01 3.37662727e-01 -3.57451409e-01
1.48058558e+00 -3.07381064e-01 -3.85938853e-01 6.20609760e-01
1.43382168e+00 -1.44111276e-01 -7.62879431e-01 -3.21751833e-01
4.10092503e-01 -5.51001787e-01 2.69991368e-01 -6.55938447e-01
-7.60857344e-01 4.91093665e-01 7.13466763e-01 5.09256661e-01
1.08445728e+00 -5.78802302e-02 -3.01403314e-01 1.00306936e-01
3.55782479e-01 -9.06816900e-01 -6.05560362e-01 3.09621930e-01
6.78183377e-01 -1.26006854e+00 3.21948946e-01 -4.85554963e-01
-5.69913924e-01 1.25999308e+00 -7.05750808e-02 1.81820586e-01
9.59373355e-01 6.75678253e-03 -2.07251966e-01 -4.27940518e-01
-5.95933378e-01 -3.85869145e-01 4.61672723e-01 4.78857785e-01
7.58391500e-01 3.04440677e-01 -1.22144973e+00 1.94863886e-01
-3.28068525e-01 -6.30876720e-02 1.52134141e-02 7.24142492e-01
-4.19706672e-01 -1.21385586e+00 -3.43419671e-01 7.51282990e-01
-3.02964807e-01 -4.85220999e-02 -7.06851542e-01 9.40107465e-01
-4.74720076e-02 1.13625550e+00 -2.33366475e-01 -3.68137717e-01
-1.13365497e-03 1.31140798e-01 2.10074246e-01 -2.47199804e-01
-4.43983585e-01 -1.21485302e-02 1.77094206e-01 -2.25064203e-01
-6.10951900e-01 -1.05564880e+00 -9.28761363e-01 -1.19308770e-01
-7.70053983e-01 4.67273682e-01 9.49751794e-01 1.02344060e+00
1.67147264e-01 2.33437344e-01 8.67956519e-01 -6.60720348e-01
-3.79495680e-01 -7.95108676e-01 -8.84931445e-01 2.82368302e-01
1.08480275e-01 -1.03015113e+00 -4.07856047e-01 -2.50375539e-01] | [7.8263726234436035, 4.701794624328613] |
959bf045-2827-485c-8e16-cfa17c8c7b4b | flow-edge-guided-video-completion | 2009.01835 | null | https://arxiv.org/abs/2009.01835v1 | https://arxiv.org/pdf/2009.01835v1.pdf | Flow-edge Guided Video Completion | We present a new flow-based video completion algorithm. Previous flow completion methods are often unable to retain the sharpness of motion boundaries. Our method first extracts and completes motion edges, and then uses them to guide piecewise-smooth flow completion with sharp edges. Existing methods propagate colors among local flow connections between adjacent frames. However, not all missing regions in a video can be reached in this way because the motion boundaries form impenetrable barriers. Our method alleviates this problem by introducing non-local flow connections to temporally distant frames, enabling propagating video content over motion boundaries. We validate our approach on the DAVIS dataset. Both visual and quantitative results show that our method compares favorably against the state-of-the-art algorithms. | ['Jia-Bin Huang', 'Johannes Kopf', 'Ayush Saraf', 'Chen Gao'] | 2020-09-03 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/1715_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123570698.pdf | eccv-2020-8 | ['video-inpainting'] | ['computer-vision'] | [-4.62598540e-02 -5.30119538e-01 -2.16406003e-01 1.00110807e-01
-2.05345482e-01 -8.80353868e-01 3.83122534e-01 -2.27446839e-01
-3.68890405e-01 9.64590788e-01 3.57960820e-01 -1.42983347e-01
1.98544860e-01 -5.99822462e-01 -4.58244681e-01 -1.85525060e-01
-6.23273313e-01 -2.11683959e-01 9.81100619e-01 -1.86317302e-02
4.73403305e-01 5.74269950e-01 -9.49401140e-01 5.09600222e-01
8.56071949e-01 4.91520196e-01 -2.43652835e-02 1.10064554e+00
-3.13389868e-01 9.71663535e-01 -2.80184537e-01 -8.15239102e-02
4.85288382e-01 -6.20541155e-01 -1.03401077e+00 2.72420347e-01
7.95995533e-01 -9.40075338e-01 -7.63228059e-01 7.29581177e-01
-2.04442471e-01 3.68329644e-01 3.71114194e-01 -1.23941076e+00
-5.56097031e-01 1.86754111e-02 -1.04467463e+00 5.10590613e-01
7.56887317e-01 1.54425323e-01 7.07232296e-01 -9.81246352e-01
1.17441177e+00 1.13705695e+00 7.46757329e-01 6.60902321e-01
-1.33358216e+00 -5.01937568e-01 3.17285210e-01 1.62012145e-01
-1.19309926e+00 -4.91842806e-01 7.84551501e-01 -5.07742226e-01
7.95664430e-01 2.60735065e-01 8.58264923e-01 5.10861158e-01
3.91730309e-01 6.04833186e-01 6.10763133e-01 -2.77764380e-01
1.10085942e-01 -5.80958366e-01 -1.27128273e-01 8.28281403e-01
7.43796453e-02 1.62174821e-01 -6.28422737e-01 -8.92997682e-02
1.45003402e+00 2.42186375e-02 -7.36534297e-01 -3.63306791e-01
-1.39402521e+00 4.33528990e-01 3.56018931e-01 1.79527283e-01
-2.75969565e-01 4.89249825e-01 1.71071321e-01 1.57729968e-01
5.35681009e-01 -1.13046840e-01 -1.14223035e-02 -2.74202943e-01
-1.53000700e+00 2.75273114e-01 7.13014126e-01 1.01150107e+00
1.04827452e+00 1.12668738e-01 -2.16189653e-01 3.93859059e-01
1.72389701e-01 3.55688259e-02 -3.95830691e-01 -1.90546334e+00
3.68571907e-01 6.49319366e-02 5.36878049e-01 -1.22117913e+00
-1.01355828e-01 4.09307599e-01 -7.14698732e-01 7.57316232e-01
7.93656826e-01 -3.14256966e-01 -8.02329063e-01 1.32087529e+00
3.18982899e-01 7.44505405e-01 -2.68132418e-01 1.02967453e+00
5.50653279e-01 9.63152766e-01 8.96626189e-02 -3.21480781e-01
6.97960258e-01 -1.20125937e+00 -8.27260673e-01 -8.64354223e-02
2.72164941e-01 -1.03046775e+00 6.97256386e-01 3.25646222e-01
-1.58678889e+00 -5.80908775e-01 -8.27840626e-01 -2.00449616e-01
3.65268171e-01 -3.83813322e-01 6.81702793e-01 3.46192151e-01
-1.51021791e+00 8.12605500e-01 -9.62046862e-01 -8.25156644e-02
4.44049746e-01 1.85477257e-01 -5.91603816e-01 -4.19394612e-01
-9.30865884e-01 3.45846146e-01 1.27859980e-01 8.96868929e-02
-8.18442523e-01 -9.89542127e-01 -8.98180664e-01 -3.49150710e-02
1.49580702e-01 -8.83316934e-01 9.72974539e-01 -1.09641612e+00
-1.27337325e+00 4.85972166e-01 -6.44570231e-01 -1.42930359e-01
1.00551450e+00 -3.28053147e-01 -2.08217934e-01 8.16947997e-01
-6.30249158e-02 1.20517504e+00 8.43385816e-01 -1.41371226e+00
-1.08963549e+00 3.70318115e-01 2.57964909e-01 4.76081558e-02
-2.29555219e-01 1.39426544e-01 -9.59433317e-01 -8.63354385e-01
1.49987908e-02 -5.85845530e-01 -2.44132996e-01 8.25043738e-01
-6.97229654e-02 7.74244294e-02 1.15252340e+00 -8.15511703e-01
1.53919256e+00 -2.12851214e+00 1.07855909e-01 8.20529833e-02
5.20519495e-01 8.88716653e-02 -2.29744926e-01 3.20326060e-01
1.16124697e-01 2.17472509e-01 -4.16254252e-01 -3.12993377e-01
-6.44281685e-01 1.49373397e-01 -1.73561215e-01 5.12273192e-01
1.79026634e-01 6.16050720e-01 -1.26390803e+00 -6.30492985e-01
3.68224889e-01 8.07153165e-01 -9.28218782e-01 9.02806371e-02
5.90095762e-03 5.64460516e-01 -2.39133894e-01 4.28974122e-01
1.01792002e+00 -2.86565572e-02 -4.19410923e-03 -6.69769049e-02
-4.53183591e-01 1.79783609e-02 -1.38267267e+00 2.02155876e+00
-5.96296564e-02 1.06336069e+00 3.37508023e-01 -2.96383649e-01
6.06543422e-01 3.69228601e-01 9.01726961e-01 -1.95790306e-01
-4.09835070e-01 2.98426542e-02 -3.48095566e-01 -4.77807432e-01
8.29482794e-01 1.51594402e-02 6.51490867e-01 2.71652281e-01
-3.43727112e-01 6.39931113e-02 6.40941322e-01 5.90369165e-01
1.12081587e+00 5.05966842e-01 -2.48333007e-01 -2.96923071e-01
5.40509582e-01 1.52863696e-01 9.57859516e-01 4.79505390e-01
-6.69522882e-01 1.22390068e+00 5.30273974e-01 -5.71408749e-01
-1.12966514e+00 -1.41790199e+00 1.33598179e-01 6.31371617e-01
5.80884099e-01 -6.48233593e-01 -7.31722713e-01 -6.69881701e-01
-1.82995275e-01 1.37028590e-01 -4.92009968e-01 3.88009489e-01
-9.72599626e-01 9.57586020e-02 1.87649995e-01 6.95855796e-01
5.24013758e-01 -9.02673721e-01 -5.29118478e-01 5.63652456e-01
-4.45012152e-01 -1.30327427e+00 -1.00985193e+00 -7.29980290e-01
-1.27780020e+00 -1.12230027e+00 -1.16461194e+00 -9.14888322e-01
7.87232161e-01 7.65875578e-01 1.14212000e+00 6.09488964e-01
-3.03767174e-01 2.74218291e-01 -1.86144918e-01 6.89550221e-01
-3.99680465e-01 -3.19591910e-01 -4.28568423e-01 -5.74609786e-02
-5.40558668e-03 -4.82642502e-01 -1.15816152e+00 3.62683147e-01
-1.04666829e+00 2.24656180e-01 -2.53959090e-01 6.24912977e-01
3.43315333e-01 -1.03935756e-01 2.53212601e-01 -5.09950161e-01
4.07350391e-01 -1.97787777e-01 -6.33842111e-01 1.02976330e-01
3.72631662e-02 -2.25357056e-01 4.67141926e-01 -3.22832048e-01
-1.33882475e+00 2.48275146e-01 1.57987297e-01 -7.54782975e-01
-1.20854229e-01 -2.37075314e-02 3.91989648e-01 -1.81642160e-01
4.96220380e-01 -2.20910802e-01 -2.06703722e-01 -2.04032362e-01
5.82915843e-01 -1.02856802e-02 9.22095597e-01 -4.90006924e-01
7.62108862e-01 1.05695558e+00 -1.92520674e-02 -8.52192283e-01
-5.85071146e-02 -5.97634554e-01 -8.13673675e-01 -6.46806717e-01
9.43444908e-01 -6.83094025e-01 -5.48195422e-01 3.84970218e-01
-1.48659515e+00 -5.20412862e-01 -1.53288409e-01 5.24546385e-01
-4.70783204e-01 9.76295412e-01 -1.37836349e+00 -6.22881055e-01
5.47017716e-02 -9.82486010e-01 6.02386773e-01 3.30675662e-01
-3.40429604e-01 -1.34176755e+00 2.59416729e-01 -1.80782437e-01
4.10615921e-01 3.67005885e-01 4.28020567e-01 6.78625405e-01
-9.55963850e-01 1.74670935e-01 -3.84485781e-01 -5.35410903e-02
3.76511216e-01 7.98339605e-01 -6.84638083e-01 -2.80747890e-01
-4.47161436e-01 1.86923698e-01 1.18223739e+00 6.50633454e-01
8.42769980e-01 -7.78762475e-02 -3.45976502e-01 7.93863297e-01
1.47106767e+00 1.36666447e-01 1.09971583e+00 2.66392767e-01
7.85969973e-01 6.51797593e-01 3.74005795e-01 4.10086602e-01
1.39876097e-01 3.37669343e-01 2.09889099e-01 -5.46804428e-01
-5.36924541e-01 -2.24449724e-01 4.56922531e-01 5.42168319e-01
-5.26719689e-01 -2.28545189e-01 -6.91422462e-01 7.04316378e-01
-1.88718879e+00 -1.21869636e+00 -6.54720306e-01 2.09271646e+00
7.42358446e-01 1.58755660e-01 1.75650522e-01 1.19279504e-01
7.93498278e-01 2.51825094e-01 -7.20038861e-02 -3.52507979e-01
5.46438359e-02 3.57612558e-02 3.61320794e-01 1.20985734e+00
-1.08351886e+00 1.12853336e+00 7.54751253e+00 4.57468271e-01
-7.90533721e-01 -6.10716008e-02 5.68289101e-01 -7.48768374e-02
-5.85160434e-01 3.30756992e-01 -4.25977528e-01 1.90521851e-01
1.69029862e-01 -2.73488904e-03 3.94238859e-01 2.71476299e-01
7.21567810e-01 -5.12177944e-01 -1.05558646e+00 8.57975781e-01
-2.19575182e-01 -1.77198350e+00 9.79534723e-03 -2.46831700e-01
1.15653586e+00 -3.28563243e-01 -2.44847029e-01 -3.95770609e-01
3.32469344e-01 -8.03170204e-01 7.24187851e-01 4.21617687e-01
8.99846315e-01 -6.86619639e-01 8.58332962e-02 -1.95998847e-01
-1.63757646e+00 3.19963694e-01 -2.52733946e-01 -2.81434447e-01
7.87957609e-01 6.36463523e-01 -1.82828873e-01 5.30267060e-01
7.23571241e-01 1.08697116e+00 -2.06390738e-01 1.46653712e+00
-2.87549496e-01 4.60151315e-01 -2.09928393e-01 7.08685935e-01
4.01611745e-01 -4.78428155e-01 6.00356817e-01 1.55823934e+00
2.10124120e-01 1.14038266e-01 3.70072752e-01 1.03272831e+00
7.90302828e-03 -1.43334746e-01 -6.33649290e-01 3.97235960e-01
3.25814843e-01 1.14033258e+00 -1.23972845e+00 -5.62575579e-01
-7.98030138e-01 1.39745867e+00 3.66701297e-02 1.00459683e+00
-7.31546581e-01 -3.58691901e-01 8.61379206e-01 2.66045988e-01
2.17989162e-01 -7.54982591e-01 -2.51850247e-01 -1.32377696e+00
-3.56803252e-03 -2.90463954e-01 3.89279306e-01 -7.28758216e-01
-1.09672534e+00 3.69800985e-01 -1.04121178e-01 -1.33717155e+00
-1.81953073e-01 -2.34398559e-01 -8.47192764e-01 7.92352736e-01
-1.71633112e+00 -7.13955462e-01 -4.18110698e-01 9.15982544e-01
7.51057923e-01 6.31747067e-01 3.21976960e-01 4.92383510e-01
-1.94088235e-01 1.17954180e-01 -1.32402465e-01 2.92301059e-01
9.15046453e-01 -9.30887222e-01 5.65374374e-01 1.49697208e+00
-1.00477472e-01 3.43376160e-01 4.56449568e-01 -9.13613319e-01
-9.89302039e-01 -8.00888121e-01 7.91193366e-01 -9.19135585e-02
4.93075281e-01 -1.07455835e-01 -1.19655371e+00 6.86232626e-01
5.18711209e-01 4.02960539e-01 4.92081046e-02 -3.95271212e-01
-2.52089202e-01 1.17256857e-01 -9.40393567e-01 7.27461934e-01
1.16016877e+00 -2.99517155e-01 -3.60170096e-01 -1.08369343e-01
5.06859362e-01 -3.27058256e-01 -4.90137190e-01 1.18626945e-01
6.07741058e-01 -1.26800501e+00 1.14278865e+00 -4.10807550e-01
7.71594167e-01 -6.71581984e-01 3.31536710e-01 -1.03425467e+00
-5.46224654e-01 -1.23487556e+00 -1.16831660e-01 1.16945446e+00
7.27828145e-02 -1.02123089e-01 9.60266232e-01 8.24583828e-01
-1.15973376e-01 -2.21948147e-01 -6.86254203e-01 -7.16956735e-01
7.61769712e-02 -3.30269158e-01 -7.74421617e-02 1.20048964e+00
2.30303645e-01 -2.25283250e-01 -3.48512381e-01 -4.93081696e-02
7.26566076e-01 -1.11460306e-01 5.75421929e-01 -8.09823513e-01
-6.34878501e-02 -6.35432303e-01 -2.68335909e-01 -1.45114398e+00
-4.39989083e-02 -4.19734776e-01 3.82594354e-02 -1.84747112e+00
7.60976002e-02 -4.71972257e-01 -3.83967273e-02 3.28206509e-01
-3.76167029e-01 5.79187155e-01 4.79641199e-01 1.57939017e-01
-4.39682454e-01 1.36326447e-01 1.64465296e+00 2.30295192e-02
-5.61279714e-01 -5.70778072e-01 4.67593037e-02 9.10283029e-01
6.25084221e-01 -3.75070721e-01 -5.36469817e-01 -6.70762479e-01
-6.21875525e-02 3.46725315e-01 2.87056029e-01 -9.55588937e-01
2.96328038e-01 -4.20743555e-01 5.96340418e-01 -4.93265152e-01
4.46151420e-02 -6.57501221e-01 2.37867773e-01 5.59863389e-01
-1.10017478e-01 2.91513592e-01 2.87768841e-01 4.92412329e-01
-2.53504306e-01 -1.08662933e-01 7.46765971e-01 -9.60031897e-02
-1.06890082e+00 4.22232240e-01 -7.66670883e-01 2.70498961e-01
1.07103133e+00 -5.88052094e-01 -2.01727852e-01 -4.80492204e-01
-8.04258406e-01 3.07720482e-01 8.65553558e-01 3.92187029e-01
9.77304578e-01 -1.25032008e+00 -8.11392546e-01 2.82863319e-01
-4.84762818e-01 -3.71481702e-02 3.12311113e-01 7.58791804e-01
-1.39423192e+00 -6.92841411e-02 -4.24451411e-01 -7.41936982e-01
-1.25103533e+00 6.83482349e-01 3.32316190e-01 1.18417293e-01
-1.14658761e+00 6.24513566e-01 5.31942546e-01 6.32538140e-01
2.13521197e-01 -4.53695506e-01 1.10867433e-01 -2.27605984e-01
8.43271732e-01 7.25900173e-01 -3.97782743e-01 -3.87824446e-01
-3.65799844e-01 8.79854560e-01 5.07916138e-02 -5.67341447e-01
9.29643452e-01 -4.93305534e-01 7.08106160e-03 2.18037087e-02
1.14511716e+00 2.78584152e-01 -2.12026596e+00 3.26961905e-01
-2.29476258e-01 -1.19508564e+00 -1.06881402e-01 -2.02178180e-01
-1.49702203e+00 9.71842706e-01 1.52262330e-01 1.26456842e-01
1.18849063e+00 -4.78417993e-01 1.02241385e+00 -4.11713153e-01
2.67419785e-01 -8.06642473e-01 1.63159341e-01 4.62507010e-01
5.54031193e-01 -8.49682987e-01 -3.44273336e-02 -9.94505405e-01
-2.50834525e-01 1.53087449e+00 5.08702278e-01 -3.75340700e-01
5.44056535e-01 5.74466825e-01 1.20366551e-01 2.40544692e-01
-6.46138608e-01 1.01609617e-01 1.90959170e-01 8.33694458e-01
6.88953459e-01 -3.89706224e-01 -5.41111708e-01 -5.08626342e-01
5.38531184e-01 4.61069286e-01 9.17161644e-01 1.05679524e+00
-4.90326643e-01 -1.09359944e+00 -6.03178561e-01 -2.99089938e-01
-4.56543058e-01 -1.48036361e-01 -1.53057560e-01 8.10434818e-01
-2.29511291e-01 1.18698299e+00 3.96921486e-01 -1.15987994e-01
8.84448290e-02 -2.50392377e-01 5.95758557e-01 -7.99473226e-02
-3.99519622e-01 5.12901306e-01 5.36427163e-02 -9.73112047e-01
-7.38639772e-01 -3.82701695e-01 -1.73023117e+00 -8.26487720e-01
8.91162381e-02 -6.89579695e-02 6.06040843e-02 4.05452013e-01
2.89767504e-01 4.72567767e-01 6.07656062e-01 -1.16943395e+00
4.58918780e-01 -2.84690112e-01 -3.07647288e-01 8.45845222e-01
6.55338347e-01 -3.72090280e-01 -3.87991071e-01 7.66586304e-01] | [10.643553733825684, -1.4595510959625244] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.