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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
dc79a49e-5a3e-4fcc-ad29-9775fd5f9f69 | bts-a-bi-lingual-benchmark-for-text | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Xu_BTS_A_Bi-Lingual_Benchmark_for_Text_Segmentation_in_the_Wild_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Xu_BTS_A_Bi-Lingual_Benchmark_for_Text_Segmentation_in_the_Wild_CVPR_2022_paper.pdf | BTS: A Bi-Lingual Benchmark for Text Segmentation in the Wild | As a prerequisite of many text-related tasks such as text erasing and text style transfer, text segmentation arouses more and more attention recently. Current researches mainly focus on only English characters and digits, while few work studies Chinese characters due to the lack of public large-scale and high-quality Chinese datasets, which limits the practical application scenarios of text segmentation. Different from English which has a limited alphabet of letters, Chinese has much more basic characters with complex structures, making the problem more difficult to deal with. To better analyze this problem, we propose the Bi-lingual Text Segmentation (BTS) dataset, a benchmark that covers various common Chinese scenes including 14,250 diverse and fine-annotated text images. BTS mainly focuses on Chinese characters, and also contains English words and digits. We also introduce Prior Guided Text Segmentation Network (PGTSNet), the first baseline to handle bi-lingual and complex-structured text segmentation. A plug-in text region highlighting module and a text perceptual discriminator are proposed in PGTSNet to supervise the model with text prior, and guide for more stable and finer text segmentation. A variation loss is also employed for suppressing background noise under complex scene. Extensive experiments are conducted not only to demonstrate the necessity and superiority of the proposed dataset BTS, but also to show the effectiveness of the proposed PGTSNet compared with a variety of state-of-the-art text segmentation methods. | ['XiaoHu Qie', 'Ying Shan', 'Honglun Zhang', 'jianqi ma', 'Zhongang Qi', 'Xixi Xu'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['text-style-transfoer'] | ['natural-language-processing'] | [ 3.96768242e-01 -5.94303787e-01 -7.24635422e-02 -3.30067635e-01
-4.31227475e-01 -4.18866038e-01 5.04414022e-01 -2.61307657e-01
-6.33647144e-01 4.17726040e-01 1.74875662e-01 -3.83273065e-01
6.30788505e-01 -5.96760333e-01 -4.17788595e-01 -8.56948674e-01
7.86105216e-01 4.45979387e-01 6.63511455e-01 -5.63260838e-02
4.13031042e-01 1.63592994e-01 -1.05565321e+00 2.73097664e-01
1.50424695e+00 8.24716687e-01 7.50796139e-01 2.93077379e-01
-7.61573493e-01 2.44299352e-01 -8.00803363e-01 -4.12966102e-01
2.52149135e-01 -6.64444208e-01 -5.11519194e-01 4.19457883e-01
2.37674698e-01 -4.03347015e-01 -3.72805089e-01 1.22252417e+00
8.06318045e-01 1.08270876e-01 7.26711512e-01 -7.24554837e-01
-8.38915646e-01 8.58538747e-01 -9.78261650e-01 -3.14439530e-03
-5.67598119e-02 2.14243263e-01 8.78547728e-01 -7.65411377e-01
5.13148248e-01 1.34145904e+00 5.35804689e-01 6.00035667e-01
-7.40863860e-01 -7.25780964e-01 4.30103302e-01 -1.22662832e-03
-1.11499333e+00 -1.85874209e-01 6.47825241e-01 -1.83989197e-01
6.16060436e-01 1.62301660e-01 5.67100406e-01 1.09020746e+00
2.44112551e-01 1.59263432e+00 1.11839902e+00 -4.78944540e-01
-8.82871598e-02 2.01674804e-01 1.33480906e-01 3.35667104e-01
1.66556597e-01 -4.85897213e-01 -1.44377530e-01 6.22401595e-01
6.87887907e-01 2.84344349e-02 -2.95120060e-01 7.85661563e-02
-1.37665105e+00 5.99043071e-01 6.18949644e-02 4.89738435e-01
-3.99943367e-02 -2.85989195e-01 7.44426310e-01 -7.51113892e-02
5.33618927e-01 -5.54627292e-02 -5.13095260e-01 -1.90773293e-01
-1.07038999e+00 -4.43074405e-02 5.51411510e-01 1.42087519e+00
5.40217221e-01 2.96412259e-01 -4.82079893e-01 1.23671067e+00
3.52023363e-01 8.82355869e-01 6.82940960e-01 -1.77523196e-01
1.01029861e+00 6.22934818e-01 -2.66581118e-01 -7.62378871e-01
-1.07372016e-01 -2.17969626e-01 -1.12022924e+00 -2.58164197e-01
3.01583141e-01 -4.20482188e-01 -1.32459676e+00 1.02980959e+00
1.39447510e-01 -2.26687506e-01 -3.56729031e-02 9.51905251e-01
8.72405350e-01 1.14927399e+00 1.08844601e-01 -1.63159892e-01
1.46028292e+00 -1.22943568e+00 -9.59100902e-01 -3.78888935e-01
5.14730811e-01 -1.21768868e+00 1.41901195e+00 4.22209144e-01
-8.20348024e-01 -7.02893257e-01 -9.89249110e-01 -3.57718468e-01
-5.51792502e-01 5.81537545e-01 3.81775290e-01 8.76233757e-01
-5.22233129e-01 8.18525776e-02 -7.86395431e-01 -5.52333593e-01
4.70300823e-01 8.97410884e-02 1.95218489e-01 -1.03548035e-01
-1.24793530e+00 5.43405831e-01 8.39377224e-01 3.58888239e-01
-2.99219400e-01 -2.09832370e-01 -7.35295773e-01 -5.23213521e-02
5.83701313e-01 -1.55108988e-01 9.55343246e-01 -1.08295119e+00
-1.68062699e+00 7.81247020e-01 -1.24475144e-01 -1.64285332e-01
9.20608997e-01 -3.60796779e-01 -5.56990087e-01 1.71576366e-01
2.93339670e-01 7.92469442e-01 8.90735924e-01 -9.43221986e-01
-9.09586787e-01 -1.93800583e-01 -6.36686563e-01 6.19950950e-01
-5.55631161e-01 2.15317264e-01 -1.44777763e+00 -1.24646211e+00
1.70925647e-01 -7.33151436e-01 -3.46365944e-02 -1.05942652e-01
-8.91150296e-01 -1.99152887e-01 1.38275826e+00 -9.50346768e-01
1.32764363e+00 -2.14981961e+00 -1.11792445e-01 -1.26102939e-01
-1.91093847e-01 3.98867667e-01 -7.59187043e-02 2.50861257e-01
2.62612313e-01 2.66535789e-01 -4.34108943e-01 -4.29209590e-01
7.54921660e-02 1.86733827e-01 -2.05444917e-01 2.24906608e-01
1.37751997e-02 9.24162865e-01 -4.01397139e-01 -9.91330266e-01
5.18026114e-01 5.21796718e-02 -1.63984865e-01 -6.66348562e-02
-5.20091832e-01 4.41572368e-01 -8.74443591e-01 7.97315896e-01
1.00552547e+00 9.82811861e-03 -1.97562695e-01 -7.71651492e-02
-2.94591635e-01 -5.57951927e-02 -1.22772944e+00 1.64390314e+00
-1.33622095e-01 6.99395359e-01 6.87553212e-02 -8.59597504e-01
9.69534934e-01 -3.93470153e-02 2.80393064e-01 -9.36558783e-01
5.57037055e-01 1.54510900e-01 -7.68260136e-02 -5.17577946e-01
7.97322989e-01 1.87481910e-01 -1.15294136e-01 2.16375723e-01
-3.45502734e-01 -3.56548667e-01 5.93348563e-01 2.68396348e-01
3.40655327e-01 3.59984726e-01 -1.91667080e-01 -3.60611111e-01
6.50398850e-01 9.97062922e-02 6.86051190e-01 5.48696876e-01
-1.11411206e-01 9.03124809e-01 4.28833693e-01 2.00493094e-02
-1.04508710e+00 -6.85978949e-01 -3.49504292e-01 1.11024833e+00
5.48811376e-01 -2.36381590e-01 -1.09235609e+00 -7.96592414e-01
-2.68069863e-01 6.20059073e-01 -4.29889172e-01 3.05890232e-01
-7.19870090e-01 -1.05663431e+00 5.95176220e-01 6.09581947e-01
1.46995890e+00 -1.28512752e+00 -1.30428806e-01 8.40643272e-02
-4.75856096e-01 -1.51086521e+00 -1.07043469e+00 -2.81811245e-02
-7.49313831e-01 -7.10213304e-01 -1.34571457e+00 -1.23875391e+00
6.58959091e-01 2.46780187e-01 6.79556906e-01 -5.78899197e-02
-2.19467387e-01 -6.03241734e-02 -7.00344026e-01 -5.39779246e-01
-4.28572297e-01 3.67749244e-01 -4.44662303e-01 1.23432517e-01
4.07807589e-01 1.23697221e-01 -4.82270867e-01 6.30748868e-01
-1.13209093e+00 5.24374008e-01 9.49254930e-01 8.24929237e-01
6.19312167e-01 3.57775390e-01 2.23764479e-01 -1.04738820e+00
5.54520726e-01 4.47477140e-02 -7.18416035e-01 3.78317773e-01
-3.79854858e-01 -3.52936536e-01 9.37921822e-01 -4.63923335e-01
-1.43148303e+00 -8.92657414e-02 -2.53985047e-01 -3.89933735e-02
-3.56280416e-01 4.65791374e-01 -6.74512744e-01 2.14935526e-01
1.06812008e-01 7.54927278e-01 -4.20578539e-01 -5.54323316e-01
1.75808921e-01 1.12378073e+00 4.84052151e-01 -7.04097986e-01
7.10003793e-01 2.31065467e-01 -4.14286435e-01 -1.18114984e+00
-5.68397224e-01 -4.67068285e-01 -8.86376262e-01 8.61364305e-02
1.23899925e+00 -7.74737597e-01 -3.82357597e-01 1.18256140e+00
-1.04618049e+00 -6.64819360e-01 2.85777450e-02 3.98402810e-01
-1.58719361e-01 8.97279501e-01 -9.33863580e-01 -5.47167718e-01
-4.71908718e-01 -1.22878873e+00 1.24392378e+00 5.09599328e-01
4.67767864e-01 -9.10675168e-01 -5.46126544e-01 4.31804895e-01
1.24834433e-01 -9.46252048e-02 8.36758852e-01 -5.94486594e-01
-6.28592134e-01 8.58345069e-03 -5.83720982e-01 6.35977328e-01
2.21989483e-01 1.06798485e-01 -6.40144706e-01 -2.64986485e-01
-2.25593492e-01 -1.82557210e-01 1.08478498e+00 4.37660545e-01
1.39360917e+00 1.46588847e-01 -3.42296839e-01 7.46160746e-01
1.21004355e+00 5.10448039e-01 8.69533896e-01 3.36456358e-01
1.11462331e+00 5.21173775e-01 8.81940365e-01 2.49980494e-01
2.49476746e-01 4.09348577e-01 -1.24700449e-01 -5.12692809e-01
-1.74961910e-01 -1.38807312e-01 1.43113166e-01 1.15261269e+00
2.04064324e-01 -6.71243370e-01 -8.91084313e-01 4.82163519e-01
-1.57381403e+00 -4.85663563e-01 -3.92748058e-01 1.71494389e+00
1.11272430e+00 4.19102579e-01 -5.67859523e-02 7.01254010e-02
1.12539184e+00 2.39611909e-01 -8.14200461e-01 -6.96051568e-02
-5.17653942e-01 2.18734778e-02 6.30170047e-01 -9.32315141e-02
-1.33648384e+00 1.63991690e+00 5.00206566e+00 1.66956484e+00
-1.34469926e+00 -2.34962434e-01 8.98820043e-01 6.64986193e-01
1.91678703e-02 -1.32420450e-01 -1.07718241e+00 8.00449669e-01
2.02172458e-01 1.99789479e-01 1.18965119e-01 6.51588559e-01
3.03223431e-01 -2.45793238e-01 -5.55777133e-01 9.86957371e-01
3.73417959e-02 -1.00481403e+00 2.52408594e-01 -1.48415476e-01
1.07064283e+00 -9.55810249e-02 1.37712359e-01 3.46986413e-01
6.55614361e-02 -7.40154088e-01 8.01012814e-01 -5.04344841e-03
1.11970425e+00 -6.36978090e-01 6.64281130e-01 3.24070871e-01
-1.44353557e+00 3.50110114e-01 -5.16326487e-01 4.08789396e-01
9.89973992e-02 4.30153161e-01 -4.92069691e-01 6.65002882e-01
7.00451732e-01 1.12794113e+00 -7.66259134e-01 1.05295205e+00
-3.12136889e-01 9.03627455e-01 -3.53557616e-01 -3.87265474e-01
6.20801747e-01 -6.16212487e-01 2.60449857e-01 1.76847661e+00
1.26338348e-01 5.86964116e-02 4.65828717e-01 7.07456529e-01
-1.45030141e-01 6.71671152e-01 1.37104988e-02 -2.95602262e-01
3.07743967e-01 1.25054276e+00 -1.53606677e+00 -6.49739265e-01
-6.01088405e-01 1.34312093e+00 -5.06010890e-01 6.56612992e-01
-7.94093251e-01 -9.86466765e-01 -6.29369766e-02 -2.94319272e-01
5.29698610e-01 -2.77088225e-01 -7.34206498e-01 -1.44636488e+00
4.45480049e-02 -1.11953557e+00 2.12538883e-01 -6.90379381e-01
-1.17548442e+00 4.20324028e-01 -1.64393425e-01 -1.25801957e+00
3.22730243e-01 -6.48349047e-01 -6.67882144e-01 1.12884784e+00
-1.53436482e+00 -1.44087064e+00 -4.13574547e-01 7.19259501e-01
1.26744628e+00 -1.28891751e-01 1.21080533e-01 3.10998350e-01
-1.02865124e+00 7.20836639e-01 5.05076408e-01 6.96187019e-01
9.60851729e-01 -1.23826933e+00 8.14225972e-01 1.16778421e+00
-3.07502866e-01 4.54872578e-01 4.51564193e-01 -1.06118810e+00
-1.27070832e+00 -1.22274172e+00 2.58933097e-01 9.94597562e-04
4.99320179e-01 -7.22048104e-01 -1.04284608e+00 6.33635104e-01
4.70064968e-01 -3.74847203e-01 5.23765795e-02 -4.40176845e-01
1.68733597e-01 5.21111228e-02 -8.21042299e-01 9.79840338e-01
7.59852588e-01 -5.36364242e-02 -4.14437741e-01 2.70096958e-01
7.85203755e-01 -6.77940011e-01 -2.75745839e-01 2.75668323e-01
2.05665022e-01 -8.51381660e-01 5.71064234e-01 1.22771181e-01
3.64860862e-01 -3.55560720e-01 1.15253173e-01 -9.20330882e-01
1.08316965e-01 -6.15409434e-01 6.67824686e-01 1.62676787e+00
2.12683216e-01 -6.51618123e-01 9.85492408e-01 2.13712007e-02
-5.16100705e-01 -4.57177758e-01 -5.18387914e-01 -5.63841105e-01
3.70358855e-01 -3.02766204e-01 3.58913898e-01 9.42058146e-01
-6.02016926e-01 4.44920599e-01 -4.89420891e-01 -1.43540621e-01
3.38809252e-01 1.46460280e-01 7.06681073e-01 -9.45823252e-01
1.32445339e-02 -7.24880695e-01 -4.76469435e-02 -1.93178582e+00
6.97191209e-02 -6.00294054e-01 4.16989237e-01 -1.69541228e+00
1.37490779e-01 -5.28558135e-01 2.87178993e-01 2.80498117e-01
-4.60448384e-01 1.91915110e-01 2.27493629e-01 1.40452951e-01
-6.29286706e-01 8.44713330e-01 1.88077295e+00 -3.38378221e-01
-2.26615146e-01 7.21134171e-02 -4.20039147e-01 8.03147256e-01
7.07905591e-01 -6.21512830e-02 -2.59174347e-01 -6.49573982e-01
-2.84958661e-01 -1.73920885e-01 -1.83311567e-01 -8.25228333e-01
3.90074015e-01 -2.61165857e-01 7.29553699e-01 -1.37566400e+00
-4.28350121e-02 -7.38528132e-01 -6.25398219e-01 2.80844927e-01
-1.12468682e-01 -2.23516092e-01 3.57204944e-01 4.27202672e-01
-2.24242404e-01 -3.58825266e-01 9.26040709e-01 -2.31123678e-02
-1.02108836e+00 4.59818214e-01 -5.10876656e-01 3.87988836e-01
8.39692533e-01 -5.80448925e-01 -1.51892096e-01 -4.81163710e-02
-2.51987457e-01 6.29524648e-01 3.63291889e-01 4.82382387e-01
5.55046082e-01 -9.31468964e-01 -7.64888465e-01 2.11267188e-01
-8.03163424e-02 3.73824656e-01 4.46705133e-01 6.87557340e-01
-9.33953047e-01 4.49907899e-01 -1.07153863e-01 -7.60310352e-01
-1.14960027e+00 3.40047240e-01 1.46587655e-01 -1.03349037e-01
-8.21330190e-01 6.18816495e-01 5.84966958e-01 -4.08893883e-01
4.46298808e-01 -6.03262305e-01 -2.72437751e-01 -5.86083978e-02
3.03323120e-01 1.95418596e-01 -8.22117776e-02 -5.63760042e-01
2.51135826e-02 9.89502788e-01 -3.89835298e-01 -1.26431108e-01
7.65015304e-01 -4.64799166e-01 -1.18936323e-01 4.07130539e-01
7.79097080e-01 2.13675633e-01 -1.30346465e+00 -4.15602475e-01
-1.00235976e-01 -4.17357147e-01 -1.01593673e-01 -9.45858240e-01
-1.14091909e+00 1.22813189e+00 4.20321226e-01 1.73016395e-02
1.37902260e+00 -4.44852442e-01 1.25381756e+00 4.11987990e-01
3.16037126e-02 -1.60132694e+00 2.16826364e-01 8.38401437e-01
5.28892040e-01 -1.29960942e+00 -7.70014897e-02 -6.45592451e-01
-9.86984611e-01 1.26478398e+00 9.22234297e-01 1.82704702e-01
2.51994401e-01 2.47528017e-01 3.13847452e-01 4.08210874e-01
-1.13245891e-02 -2.54843891e-01 2.23912969e-01 5.23901045e-01
4.06064570e-01 -2.15605333e-01 -4.57707673e-01 5.88310659e-01
-1.37233987e-01 -3.29622418e-01 4.06973660e-01 7.98754930e-01
-5.50280094e-01 -1.17757261e+00 -5.22925198e-01 5.27541459e-01
-7.16327131e-01 -4.47667927e-01 -2.30488777e-01 9.12953258e-01
1.88035831e-01 8.64435494e-01 4.66210172e-02 -9.47297513e-02
1.67181268e-01 -2.15478688e-01 1.13731712e-01 -6.11278236e-01
-4.25735742e-01 9.00064170e-01 -3.01451445e-01 6.69564828e-02
-3.17313701e-01 -6.92607462e-01 -1.53070617e+00 -2.94326901e-01
-6.54838145e-01 9.35701728e-02 6.11525476e-01 9.78493154e-01
-1.33516088e-01 7.31990635e-01 5.14752567e-01 -6.84381485e-01
-1.24980494e-01 -1.15577257e+00 -8.34890902e-01 2.76833653e-01
-1.34959027e-01 -1.56364202e-01 5.04421629e-02 3.53870511e-01] | [12.021485328674316, 2.2409257888793945] |
e80802ab-2059-41af-88ae-e7b8a6f5855a | connecting-the-dots-multivariate-time-series | 2005.11650 | null | https://arxiv.org/abs/2005.11650v1 | https://arxiv.org/pdf/2005.11650v1.pdf | Connecting the Dots: Multivariate Time Series Forecasting with Graph Neural Networks | Modeling multivariate time series has long been a subject that has attracted researchers from a diverse range of fields including economics, finance, and traffic. A basic assumption behind multivariate time series forecasting is that its variables depend on one another but, upon looking closely, it is fair to say that existing methods fail to fully exploit latent spatial dependencies between pairs of variables. In recent years, meanwhile, graph neural networks (GNNs) have shown high capability in handling relational dependencies. GNNs require well-defined graph structures for information propagation which means they cannot be applied directly for multivariate time series where the dependencies are not known in advance. In this paper, we propose a general graph neural network framework designed specifically for multivariate time series data. Our approach automatically extracts the uni-directed relations among variables through a graph learning module, into which external knowledge like variable attributes can be easily integrated. A novel mix-hop propagation layer and a dilated inception layer are further proposed to capture the spatial and temporal dependencies within the time series. The graph learning, graph convolution, and temporal convolution modules are jointly learned in an end-to-end framework. Experimental results show that our proposed model outperforms the state-of-the-art baseline methods on 3 of 4 benchmark datasets and achieves on-par performance with other approaches on two traffic datasets which provide extra structural information. | ['Chengqi Zhang', 'Zonghan Wu', 'Xiaojun Chang', 'Jing Jiang', 'Shirui Pan', 'Guodong Long'] | 2020-05-24 | null | null | null | null | ['univariate-time-series-forecasting'] | ['time-series'] | [-1.80095702e-01 -2.57929653e-01 -2.91790336e-01 -3.66524041e-01
1.75861552e-01 -3.81132096e-01 5.70194662e-01 3.25420231e-01
-4.04250920e-02 5.40656984e-01 -1.36783287e-01 -7.84210861e-01
-3.80110055e-01 -1.02624953e+00 -5.52233934e-01 -5.25032222e-01
-8.36434424e-01 3.26399803e-01 2.71544039e-01 -3.53575230e-01
-1.04693025e-01 6.17474318e-01 -1.04650533e+00 -2.52533078e-01
5.80656946e-01 1.29795611e+00 -8.15996493e-04 7.70326734e-01
-3.23656768e-01 1.23662770e+00 -3.35762799e-01 -5.44032812e-01
7.51236007e-02 -2.85471439e-01 -5.00583827e-01 3.64792123e-02
-1.01575255e-01 9.71500501e-02 -7.82699466e-01 7.67133534e-01
1.32330373e-01 3.49827200e-01 4.58074838e-01 -1.52386320e+00
-9.81694460e-01 7.14705408e-01 -7.71052003e-01 6.83218062e-01
-2.38889948e-01 1.06541030e-01 1.22613454e+00 -3.92608970e-01
2.98191667e-01 1.24757564e+00 6.44077897e-01 -4.97717559e-02
-1.08003056e+00 -6.14890575e-01 8.06697726e-01 4.98335540e-01
-1.05395687e+00 1.67405203e-01 1.29795837e+00 -4.58457172e-01
1.26588571e+00 1.17979579e-01 4.83290046e-01 8.93043399e-01
4.77247834e-01 3.39663178e-01 7.36413419e-01 -7.36830607e-02
-2.53195371e-02 -1.92393094e-01 2.27265924e-01 6.06008589e-01
-1.85196087e-01 7.69486278e-02 -4.15218055e-01 9.84216109e-02
6.79309547e-01 4.87778515e-01 6.41184375e-02 -1.12325273e-01
-1.06177771e+00 8.46191764e-01 8.22963357e-01 4.03215617e-01
-6.77514851e-01 3.33027005e-01 4.31296915e-01 5.34782171e-01
9.60422575e-01 -1.24666043e-01 -5.16396940e-01 -3.62745821e-02
-4.83308077e-01 -1.16594687e-01 8.01555097e-01 1.00902700e+00
7.92359948e-01 2.37783059e-01 -9.88942105e-03 5.05262911e-01
1.77038193e-01 4.47889566e-01 1.01155579e-01 -1.69785574e-01
8.35083783e-01 8.64261448e-01 -3.69645089e-01 -1.76604593e+00
-6.28440857e-01 -6.84112430e-01 -1.23392105e+00 -2.41618365e-01
1.66415542e-01 -2.80415773e-01 -8.28020036e-01 1.63425910e+00
2.07422405e-01 8.92492473e-01 -2.53269494e-01 7.36925483e-01
6.67527258e-01 7.64302254e-01 6.62217662e-02 -1.85022131e-01
8.46972883e-01 -1.00744522e+00 -8.88170242e-01 -2.22314909e-01
4.99824762e-01 -3.96804810e-01 5.73972285e-01 -2.31911358e-03
-5.89697897e-01 -6.32849038e-01 -8.29650104e-01 1.31466776e-01
-7.53044128e-01 -4.26345259e-01 1.16565537e+00 2.32179433e-01
-1.13789845e+00 7.23392785e-01 -1.02928913e+00 -3.14564884e-01
4.31183428e-01 4.11226034e-01 -2.10473835e-01 1.41114444e-01
-1.39795959e+00 6.38933420e-01 1.33659884e-01 5.30275345e-01
-6.21868908e-01 -4.60512012e-01 -8.61052871e-01 1.08772516e-01
4.34014738e-01 -6.98341012e-01 7.48994172e-01 -8.68990898e-01
-1.16082358e+00 3.23772609e-01 -2.41917625e-01 -6.82492495e-01
2.08545789e-01 2.30299458e-02 -1.11273623e+00 3.58000398e-03
-5.43569587e-02 -4.54786159e-02 9.56854522e-01 -8.32602262e-01
-6.36944890e-01 -4.75750476e-01 2.68914580e-01 -1.85931966e-01
-5.46535790e-01 6.50678501e-02 -6.77690625e-01 -7.61470675e-01
-4.56323884e-02 -8.57809365e-01 -5.65350294e-01 -3.33821714e-01
-3.77781093e-01 -3.96351814e-01 1.12532926e+00 -6.52486801e-01
1.72866261e+00 -2.03627849e+00 1.96552739e-01 5.30085862e-01
5.99053383e-01 -4.69324552e-02 -1.60086200e-01 7.16579258e-01
-4.40170079e-01 6.56634197e-02 -2.54249036e-01 -4.16272342e-01
-8.62473343e-03 3.62422615e-01 -5.12714207e-01 5.87706983e-01
4.09538478e-01 1.14360452e+00 -1.05638993e+00 -2.35736072e-01
2.05273405e-01 7.81867564e-01 -1.89929798e-01 2.02187151e-01
-2.43068233e-01 5.50768733e-01 -5.99310279e-01 5.06137431e-01
5.81261575e-01 -8.40815783e-01 1.23196840e-01 5.26894964e-02
9.17389020e-02 1.84884936e-01 -7.90038705e-01 1.48786259e+00
-3.29927653e-01 7.44675040e-01 -4.37050790e-01 -1.35090792e+00
1.09630299e+00 2.67574131e-01 7.72489190e-01 -8.51811111e-01
1.61531612e-01 -1.67889372e-01 8.88864100e-02 -6.15607977e-01
2.76744515e-01 3.09044998e-02 -1.43004218e-02 1.62879065e-01
-1.76934734e-01 3.24344277e-01 4.06624228e-01 3.24774772e-01
1.21704578e+00 -1.14140749e-01 -5.43998666e-02 2.53240988e-02
6.90552533e-01 -2.03855261e-01 5.98147273e-01 2.96665221e-01
-3.70153561e-02 2.24092409e-01 9.02351379e-01 -6.61646426e-01
-5.49892128e-01 -9.46021795e-01 3.81277055e-01 1.04244089e+00
1.92747433e-02 -5.61963677e-01 -3.27645838e-02 -7.98036575e-01
1.55594885e-01 5.13272703e-01 -8.16093504e-01 3.38080749e-02
-6.57064855e-01 -5.73365211e-01 1.72452435e-01 7.55700171e-01
3.53047997e-01 -1.02334273e+00 3.85138541e-02 4.16899770e-01
2.86284834e-03 -1.44494247e+00 -3.69129479e-01 1.58910900e-01
-9.36013818e-01 -9.31914151e-01 -3.59130472e-01 -6.53049648e-01
4.90562677e-01 4.09749955e-01 1.33215737e+00 2.15267897e-01
-3.16236578e-02 2.82508105e-01 -3.30343962e-01 -4.31288093e-01
6.78738728e-02 1.65765300e-01 -2.70395458e-01 4.98622596e-01
5.62320054e-01 -1.18507743e+00 -6.13812327e-01 1.44311398e-01
-8.37817013e-01 -2.00121582e-01 4.03385103e-01 5.66202819e-01
5.49942315e-01 3.96989167e-01 5.84292412e-01 -1.03619742e+00
6.15008771e-01 -9.50998664e-01 -6.69179797e-01 1.77672744e-01
-6.16684914e-01 -7.25468472e-02 9.80704665e-01 -4.75507468e-01
-6.10503912e-01 -3.23767871e-01 2.97958821e-01 -7.46874928e-01
1.12220133e-02 1.22423947e+00 6.60393164e-02 1.76070109e-01
2.09081829e-01 5.89406155e-02 -5.87566197e-02 -3.53365034e-01
3.73020917e-01 1.40087858e-01 4.53993946e-01 -3.31511140e-01
1.02847517e+00 6.58553064e-01 4.64011222e-01 -7.20815897e-01
-6.31383300e-01 -5.66253662e-01 -7.74294138e-01 -3.33606094e-01
8.60813975e-01 -8.61503482e-01 -7.17690170e-01 5.66731215e-01
-1.09449887e+00 -3.42665493e-01 1.16990395e-01 5.25264442e-01
-1.94917679e-01 3.98397595e-01 -6.34725571e-01 -6.86820090e-01
-7.73597434e-02 -7.00863063e-01 7.70290494e-01 1.37956217e-01
-1.79973803e-02 -1.88306212e+00 -1.40931919e-01 -7.05029592e-02
6.97743773e-01 6.82566941e-01 1.04016685e+00 -4.84576225e-01
-7.67626464e-01 -3.82695258e-01 -3.16286504e-01 2.01332241e-01
3.24487835e-01 3.13752666e-02 -6.78608954e-01 -2.15094745e-01
-2.65913069e-01 2.92164952e-01 9.68944371e-01 3.89486998e-01
1.34361827e+00 -2.47931644e-01 -4.79902297e-01 7.72265196e-01
1.30806065e+00 7.85028860e-02 4.60389644e-01 7.56279752e-02
1.23778403e+00 7.00120509e-01 1.53073862e-01 5.53510070e-01
9.08507705e-01 3.16541404e-01 6.77724242e-01 -3.44896168e-01
1.13217451e-01 -8.10705423e-02 2.51224190e-01 1.19006503e+00
-3.17067772e-01 -5.93291581e-01 -9.48114455e-01 6.09218776e-01
-2.13780046e+00 -1.13830590e+00 -5.99275291e-01 1.75078189e+00
2.53335029e-01 3.72647434e-01 2.54062980e-01 3.55737656e-02
6.78136647e-01 7.44452059e-01 -6.75945163e-01 -3.06370974e-01
-1.89025417e-01 2.31180027e-01 7.72726595e-01 3.58659953e-01
-1.27727950e+00 7.86774099e-01 5.33567476e+00 3.27868968e-01
-1.44441676e+00 -2.41560582e-02 5.52937925e-01 1.45657569e-01
-4.02595878e-01 1.64709017e-01 -4.71693873e-01 4.37016010e-01
1.16514874e+00 -3.79976183e-01 5.77764213e-01 4.92978007e-01
3.49624008e-01 5.35664678e-01 -9.91941094e-01 9.00677383e-01
-1.23065665e-01 -1.23071527e+00 -1.42960817e-01 -1.43825440e-02
6.62354171e-01 3.03493768e-01 1.87039703e-01 3.91707122e-01
4.09789383e-01 -1.14827216e+00 2.89303631e-01 7.99063981e-01
3.68912697e-01 -8.44836652e-01 5.71594000e-01 1.16144374e-01
-1.85863519e+00 -8.97936672e-02 9.78765544e-03 -5.21122694e-01
4.06349272e-01 7.53892004e-01 -6.25489891e-01 9.89041030e-01
6.67129874e-01 1.45557237e+00 -7.23415434e-01 8.33320200e-01
-3.23008001e-01 8.20331931e-01 -1.31632879e-01 -2.22822018e-02
5.04401922e-01 -3.47110063e-01 4.38167572e-01 1.20229399e+00
1.43004909e-01 -4.58898470e-02 2.07492188e-01 7.97342658e-01
8.83227214e-02 -1.20394804e-01 -7.02047527e-01 -4.98306721e-01
2.18060672e-01 1.25811303e+00 -8.88324261e-01 -2.23913074e-01
-9.01071668e-01 7.22152352e-01 4.07049656e-01 9.33024168e-01
-1.09467709e+00 -4.27567124e-01 6.43203199e-01 3.13552432e-02
5.93713701e-01 -6.32739782e-01 -2.39728734e-01 -1.13732147e+00
2.33292878e-01 -4.56362396e-01 7.67055333e-01 -4.97711509e-01
-1.74293983e+00 7.92120337e-01 -1.37845963e-01 -1.18508589e+00
-4.99196351e-01 -5.32495320e-01 -7.45121658e-01 1.23603499e+00
-1.93901217e+00 -1.32527030e+00 -3.88851255e-01 1.00735998e+00
2.07552880e-01 -1.07484162e-01 5.79000771e-01 4.40250516e-01
-7.79406846e-01 2.80560434e-01 -2.19290033e-01 4.58168417e-01
2.90608197e-01 -1.18246484e+00 7.95676470e-01 1.16209829e+00
1.14231840e-01 6.11620665e-01 5.30327976e-01 -7.18179643e-01
-1.64029777e+00 -1.45667553e+00 1.17670071e+00 -2.98346519e-01
1.29389298e+00 -3.25626582e-01 -1.09342253e+00 9.11671638e-01
3.11463803e-01 4.75177944e-01 4.12142932e-01 4.09890145e-01
-4.84045118e-01 -4.33438092e-01 -3.45253944e-01 5.00839710e-01
1.12187779e+00 -7.88094282e-01 -4.72985320e-02 4.90937352e-01
9.32037055e-01 -1.79711968e-01 -9.10665929e-01 3.51746470e-01
9.05077234e-02 -8.12182128e-01 8.66623938e-01 -8.32424641e-01
2.54799962e-01 -3.45685482e-01 -4.73771729e-02 -1.25375164e+00
-5.24841070e-01 -8.98938596e-01 -4.89737451e-01 1.17955077e+00
4.74287301e-01 -1.03145373e+00 4.28326190e-01 6.01256847e-01
-8.69729668e-02 -7.29670167e-01 -6.05665445e-01 -9.55873609e-01
-1.52586877e-01 -8.28167915e-01 8.27829480e-01 1.24022782e+00
-1.72006205e-01 6.82466745e-01 -3.77748966e-01 5.24142563e-01
5.32629907e-01 3.62476408e-01 8.11977446e-01 -1.32574558e+00
-3.58652085e-01 -7.55968809e-01 -7.60254264e-01 -1.06655359e+00
3.90097946e-01 -8.94420624e-01 -5.40942848e-01 -1.68473494e+00
-2.76566327e-01 -5.16321182e-01 -8.30599368e-01 3.12527090e-01
-2.46102169e-01 -2.17865139e-01 -6.93235546e-02 2.46765278e-02
-6.11064196e-01 6.40086710e-01 1.30392766e+00 -2.39952311e-01
-1.98667318e-01 7.19847456e-02 -4.72080916e-01 4.31892842e-01
7.55978584e-01 -3.98482770e-01 -7.64851868e-01 -5.64567804e-01
3.23334992e-01 3.69337708e-01 4.67478752e-01 -7.29275048e-01
5.16863406e-01 -2.95317799e-01 9.21892747e-02 -7.99295664e-01
1.51969001e-01 -1.10244882e+00 2.51017332e-01 -5.26038632e-02
7.79433735e-03 7.91064024e-01 1.81551725e-01 9.60411549e-01
-4.35690105e-01 6.96874917e-01 8.65720809e-02 1.34883091e-01
-1.04900098e+00 1.05931640e+00 9.64686275e-02 -6.21726550e-02
9.89781618e-01 5.41800261e-02 -2.69477040e-01 -6.26534581e-01
-6.47414386e-01 5.51364243e-01 -1.10118873e-01 9.22087193e-01
6.24290586e-01 -1.35138619e+00 -6.32487953e-01 2.98765153e-01
8.46399590e-02 -1.62247136e-01 3.26131910e-01 1.18415558e+00
-2.51552314e-01 3.98658812e-01 1.19367018e-01 -7.40144551e-01
-9.08735454e-01 1.03432810e+00 1.41841739e-01 -5.11472821e-01
-7.68256485e-01 7.36891031e-01 1.19934052e-01 -1.58820361e-01
2.94653505e-01 -5.86952031e-01 -3.33903491e-01 1.13240015e-02
3.16034436e-01 1.97301865e-01 -7.15339407e-02 -7.67688036e-01
-4.24693614e-01 4.20648038e-01 2.04350278e-01 2.02730849e-01
1.73122787e+00 -3.34211022e-01 -3.46696943e-01 7.97527909e-01
1.35494661e+00 -2.36674756e-01 -1.12105644e+00 -5.87055862e-01
1.64517477e-01 -3.10303181e-01 -5.99390082e-03 -3.70663822e-01
-1.64676774e+00 1.08493578e+00 7.20903724e-02 9.00581777e-01
1.52453971e+00 -5.54233193e-02 8.80475998e-01 1.06752321e-01
1.88232183e-01 -7.67575383e-01 -3.81172560e-02 7.84373701e-01
7.31094003e-01 -1.14799821e+00 -1.54424205e-01 -4.82144147e-01
-4.64995980e-01 1.33188069e+00 4.20584053e-01 -2.90159196e-01
1.24930859e+00 2.06250951e-01 -9.85661447e-02 -5.65502822e-01
-1.07350504e+00 -4.80065525e-01 5.44810712e-01 4.87619936e-01
4.74041730e-01 2.51073509e-01 7.05315098e-02 4.12164211e-01
3.65088554e-03 -1.65561199e-01 2.24629901e-02 5.94647706e-01
1.46706030e-01 -9.36071932e-01 3.49381901e-02 3.78453106e-01
-4.11952585e-01 -7.10419491e-02 -2.48841956e-01 9.46970224e-01
-1.81775749e-01 1.19289708e+00 2.49780282e-01 -6.65583014e-01
3.87344539e-01 -2.77384341e-01 1.20662078e-01 -2.79693156e-01
-5.88393569e-01 6.58468008e-02 1.11750931e-01 -6.86905146e-01
-6.52340889e-01 -4.38752770e-01 -1.28158402e+00 -6.00721121e-01
-1.40762508e-01 -1.93841271e-02 3.94755125e-01 9.25494671e-01
5.26772559e-01 1.05057120e+00 8.42175901e-01 -4.94331926e-01
-1.16167087e-02 -8.05822551e-01 -6.69673860e-01 5.85084736e-01
5.06021619e-01 -5.88404417e-01 -3.23235691e-01 1.51704535e-01] | [6.7893476486206055, 2.7674167156219482] |
31a39b10-6f1d-4205-af5b-e04cb34ea353 | sleep-staging-based-on-serialized-dual | 2107.08442 | null | https://arxiv.org/abs/2107.08442v2 | https://arxiv.org/pdf/2107.08442v2.pdf | Sleep Staging Based on Multi Scale Dual Attention Network | Sleep staging plays an important role on the diagnosis of sleep disorders. In general, experts classify sleep stages manually based on polysomnography (PSG), which is quite time-consuming. Meanwhile, the acquisition process of multiple signals is much complex, which can affect the subject's sleep. Therefore, the use of single-channel electroencephalogram (EEG) for automatic sleep staging has become a popular research topic. In the literature, a large number of sleep staging methods based on single-channel EEG have been proposed with promising results and achieve the preliminary automation of sleep staging. However, the performance for most of these methods in the N1 stage do not satisfy the needs of the diagnosis. In this paper, we propose a deep learning model multi scale dual attention network(MSDAN) based on raw EEG, which utilizes multi-scale convolution to extract features in different waveforms contained in the EEG signal, connects channel attention and spatial attention mechanisms in series to filter and highlight key information, and uses soft thresholding to remove redundant information. Experiments were conducted using two datasets with 5-fold cross-validation and hold-out validation method. The final average accuracy, overall accuracy, macro F1 score and Cohen's Kappa coefficient of the model reach 96.70%, 91.74%, 0.8231 and 0.8723 on the Sleep-EDF dataset, 96.14%, 90.35%, 0.7945 and 0.8284 on the Sleep-EDFx dataset. Significantly, our model performed superiorly in the N1 stage, with F1 scores of 54.41% and 52.79% on the two datasets respectively. The results show the superiority of our network over the existing methods, reaching a new state-of-the-art. In particular, the proposed method achieves excellent results in the N1 sleep stage compared to other methods. | ['Wanquan Liu', 'Pingshu Zhang', 'Xiaodong Yuan', 'Zhimin Hu', 'Qi Zhang', 'Chonggang Lu', 'Huafeng Wang'] | 2021-07-18 | null | null | null | null | ['sleep-staging'] | ['medical'] | [-2.32832894e-01 -4.53492522e-01 2.18976274e-01 -3.36122543e-01
-1.80055514e-01 -1.12453863e-01 -1.49420857e-01 -1.19960211e-01
-6.14900410e-01 7.97363579e-01 -3.32240053e-02 -4.71196473e-02
-1.99354202e-01 -3.21135163e-01 1.13112792e-01 -7.67695606e-01
-5.57733513e-02 -2.31789686e-02 2.62520373e-01 -1.43226564e-01
3.16231579e-01 1.21009707e-01 -1.52428532e+00 5.05213737e-02
1.28720224e+00 1.28975809e+00 4.77938473e-01 2.98485816e-01
9.27841514e-02 1.88146248e-01 -1.00260162e+00 -1.92236458e-03
-8.24119821e-02 -5.71105778e-01 -4.53034818e-01 -2.07785532e-01
-2.90950179e-01 3.17476690e-02 -1.79843575e-01 1.34082818e+00
7.17046440e-01 1.37275234e-01 2.56009430e-01 -1.13991547e+00
-5.26448965e-01 1.81701019e-01 -6.57285273e-01 9.51582789e-01
1.55300885e-01 2.74950806e-02 5.97187936e-01 -5.38780689e-01
-2.46642038e-01 5.41539252e-01 6.97256088e-01 6.12201512e-01
-9.14206982e-01 -1.34900892e+00 -1.25712767e-01 6.22922957e-01
-1.57907641e+00 -3.26852292e-01 5.17214477e-01 -2.16609389e-01
9.41602230e-01 2.62766153e-01 1.13889742e+00 6.98287547e-01
8.05341184e-01 3.30752254e-01 1.44326258e+00 1.41072348e-02
1.76936820e-01 2.33680978e-01 4.09482896e-01 5.05178452e-01
3.61284465e-01 -3.60246837e-01 -4.76836652e-01 3.40731651e-01
5.31040549e-01 5.40744841e-01 -5.50566971e-01 5.70985556e-01
-9.86408651e-01 4.66220081e-01 5.53277791e-01 7.16675937e-01
-4.52958643e-01 -3.84233057e-01 3.32673192e-01 2.27357000e-01
5.54190993e-01 4.00060207e-01 -3.86703402e-01 -3.57211143e-01
-1.33520937e+00 -3.06588650e-01 4.75592196e-01 5.68537295e-01
4.95849520e-01 -3.40698957e-02 -1.89115867e-01 6.70659602e-01
2.37816632e-01 5.50079525e-01 1.11200464e+00 -4.84811425e-01
1.41767427e-01 7.76877403e-01 -4.93328609e-02 -8.70701134e-01
-9.50767457e-01 -7.97628164e-01 -1.28391910e+00 -1.97754875e-01
-2.02134758e-01 -3.76433581e-02 -7.26208627e-01 1.56777310e+00
-1.85103834e-01 2.46248916e-01 -2.73037553e-02 1.02943611e+00
1.07064760e+00 6.79195881e-01 1.29619166e-01 -5.81773937e-01
1.59494758e+00 -8.67713928e-01 -1.10154998e+00 -3.08404952e-01
-1.09961830e-01 -5.88928998e-01 1.19117808e+00 6.88784778e-01
-1.12876081e+00 -9.08170462e-01 -1.21791482e+00 1.86503038e-01
-1.66394919e-01 5.56948125e-01 3.98793697e-01 6.18881881e-01
-1.16610885e+00 4.53808397e-01 -1.12943864e+00 -5.02525568e-01
5.29293835e-01 9.63463247e-01 -2.03790158e-01 3.30197841e-01
-1.17616546e+00 8.35219920e-01 1.13185950e-01 3.92809212e-01
-7.37855434e-01 -5.09203076e-01 -3.33946556e-01 4.35397416e-01
-7.75519609e-02 -5.25007844e-01 9.47768271e-01 -8.03581715e-01
-1.27363884e+00 6.68199480e-01 -3.87360692e-01 -2.08994269e-01
-1.27699599e-01 -2.34451249e-01 -8.98320854e-01 3.00503820e-01
3.19066525e-01 3.91559035e-01 5.33657312e-01 -3.72754991e-01
-7.44261026e-01 -6.50322437e-01 -6.27612248e-02 1.53803661e-01
-5.23284197e-01 2.54626215e-01 -2.91590303e-01 -2.89539725e-01
7.59261623e-02 -7.31058061e-01 2.16764007e-02 -4.28127974e-01
-1.62624970e-01 -3.71764660e-01 5.73317349e-01 -8.14441323e-01
1.70999157e+00 -2.42247319e+00 3.54946554e-02 -8.49286169e-02
4.85586852e-01 4.00537848e-01 3.08559477e-01 -1.25939148e-02
-4.88800593e-02 -2.83529572e-02 -1.07574672e-01 -3.23258132e-01
-3.06772172e-01 -1.10222362e-02 1.61324367e-01 5.98580122e-01
-1.29362583e-01 6.74950957e-01 -6.38200402e-01 -3.55561256e-01
9.97400656e-02 3.28887761e-01 -3.22140038e-01 3.95371616e-01
7.05061615e-01 5.79832435e-01 -2.65651971e-01 4.55847204e-01
6.28761768e-01 -3.92436504e-01 -2.13728815e-01 -2.66797930e-01
-4.30169016e-01 2.29998037e-01 -9.36501086e-01 1.86384320e+00
-4.67312217e-01 6.97143257e-01 3.92175652e-02 -8.94118965e-01
8.58395457e-01 3.80962551e-01 3.32132548e-01 -9.33998585e-01
4.51586872e-01 2.36671552e-01 4.23948348e-01 -8.53282690e-01
1.10803396e-01 -5.08317947e-01 4.68731560e-02 3.11684370e-01
1.83906302e-01 3.01038295e-01 1.92107335e-01 -2.75839455e-02
1.10971522e+00 -5.27554393e-01 5.11841357e-01 -6.16733491e-01
6.99334323e-01 -4.95484322e-01 6.70721173e-01 3.36727500e-01
-3.03207517e-01 3.90297145e-01 2.03023061e-01 -2.15009600e-01
-3.00233066e-01 -7.27271616e-01 -2.43004709e-01 5.64231157e-01
5.34596980e-01 -5.16314566e-01 -8.11895430e-01 -4.03701007e-01
-3.86348695e-01 3.98854643e-01 -6.03038251e-01 -6.06501997e-01
-1.08276345e-01 -9.96447325e-01 3.92713904e-01 5.98392546e-01
1.03189051e+00 -1.27521622e+00 -7.72344172e-01 6.48764968e-02
-1.93186447e-01 -8.51040423e-01 -3.26165736e-01 1.67728871e-01
-8.93402040e-01 -1.08322036e+00 -7.21657515e-01 -7.92550862e-01
6.02536201e-01 4.06237692e-01 7.15528488e-01 1.95404917e-01
-2.42339402e-01 -5.48324436e-02 -2.76234299e-01 -3.53407413e-01
3.77052367e-01 2.55876005e-01 4.47946906e-01 1.70167699e-01
6.00863934e-01 -9.76124227e-01 -1.01074481e+00 4.78291541e-01
-5.66205263e-01 -5.81119955e-02 9.76608813e-01 6.52030587e-01
3.84966284e-01 2.93774486e-01 7.56979942e-01 -3.13591987e-01
8.34608197e-01 -5.60685635e-01 -4.32481498e-01 -2.43305247e-02
-8.57491612e-01 -4.52408314e-01 8.73186946e-01 -2.65618324e-01
-7.61951685e-01 -4.75153416e-01 -1.98226184e-01 -4.14542139e-01
-2.06224144e-01 2.76584625e-01 -2.13698506e-01 3.80241796e-02
2.40952343e-01 5.49155295e-01 -9.25017148e-02 -5.66001832e-01
-6.17254674e-01 1.06036603e+00 4.13762599e-01 1.90688208e-01
3.41622174e-01 1.36079118e-01 -1.92466959e-01 -9.94244516e-01
-9.01701093e-01 -6.86159372e-01 -4.29625660e-01 -2.34930287e-03
1.16282094e+00 -1.00580275e+00 -9.60426450e-01 5.76802075e-01
-8.32732201e-01 -1.07240498e-01 2.41551816e-01 8.56255949e-01
7.32160881e-02 2.39221871e-01 -3.68753344e-01 -7.70135164e-01
-9.70590651e-01 -1.27702641e+00 7.07587004e-01 7.97699988e-01
-3.24727416e-01 -6.64906561e-01 -1.14830822e-01 1.29270643e-01
6.61290050e-01 -3.63218606e-01 6.28130257e-01 -6.79069281e-01
-1.68675497e-01 -7.07383007e-02 -2.55370140e-01 3.71031582e-01
3.86257708e-01 -5.00615478e-01 -1.05745816e+00 -3.10600340e-01
4.95657057e-01 -8.16521887e-03 5.19362092e-01 5.26742935e-01
1.35171175e+00 5.26628830e-02 -2.80156732e-01 8.24802458e-01
1.30828667e+00 7.59505808e-01 7.41729319e-01 3.28314066e-01
3.32593292e-01 6.56662360e-02 3.19476873e-01 3.76015157e-01
3.84403914e-01 2.83434838e-01 2.45774195e-01 -5.36411107e-02
4.23702039e-02 3.11120450e-01 2.58853853e-01 1.10784626e+00
-1.60606489e-01 -2.85285916e-02 -6.55367970e-01 4.63782758e-01
-1.44033325e+00 -8.07500303e-01 -2.90851831e-01 1.99259472e+00
5.82877576e-01 2.81201541e-01 4.84359823e-02 2.96629727e-01
7.31570721e-01 -1.30398050e-01 -5.58949351e-01 -2.03820378e-01
1.06613018e-01 8.42689157e-01 1.03475012e-01 -1.02863804e-01
-9.42963660e-01 3.97577047e-01 5.34879827e+00 5.18708587e-01
-1.30594587e+00 5.25321364e-01 2.73161352e-01 -5.12946486e-01
4.24328297e-01 -4.17743117e-01 -7.59747982e-01 1.24071181e+00
1.30191123e+00 -1.56652153e-01 6.97431087e-01 5.13998449e-01
5.86152911e-01 -4.25089329e-01 -5.96822083e-01 1.65464258e+00
1.03821442e-01 -8.34940016e-01 -5.93337357e-01 3.05530988e-02
4.85678971e-01 -2.97488570e-02 -5.85714765e-02 4.39464808e-01
-6.68115973e-01 -1.01103783e+00 4.33285922e-01 7.08211601e-01
1.03143764e+00 -7.39237070e-01 1.15760064e+00 3.37031335e-01
-1.28964365e+00 -2.65283585e-01 -3.09789598e-01 -3.56910795e-01
-4.43074740e-02 5.78189552e-01 -1.20531775e-01 3.49594146e-01
1.21456015e+00 9.20904219e-01 -8.04485500e-01 1.33569837e+00
-4.85439986e-01 8.02920580e-01 -1.74060926e-01 -4.08926010e-01
-1.95799768e-02 -1.82837605e-01 1.26835167e-01 9.09761131e-01
5.04762769e-01 4.64932770e-01 -2.01757222e-01 7.59701252e-01
7.45742163e-03 -7.56233856e-02 -8.57970677e-03 1.89198539e-01
2.76973099e-01 1.59320533e+00 -9.40668285e-01 -2.32632294e-01
-3.92656505e-01 9.64044690e-01 -7.49298334e-02 1.73795208e-01
-9.71245706e-01 -1.00801933e+00 4.66752648e-01 -1.32017598e-01
3.63733284e-02 6.50901571e-02 -4.91359413e-01 -1.21415591e+00
1.17264658e-01 -6.29891694e-01 1.49322316e-01 -9.30256844e-01
-1.07270157e+00 8.80868018e-01 -1.39298573e-01 -1.09202456e+00
4.78726208e-01 -3.80274743e-01 -1.10063958e+00 1.02029538e+00
-1.27298450e+00 -5.17655849e-01 -8.03672075e-01 6.62149429e-01
6.37577236e-01 -1.80419326e-01 6.93178177e-01 7.15419233e-01
-1.03190446e+00 5.53385794e-01 -1.35719096e-02 -2.17168614e-01
6.71884894e-01 -1.18376005e+00 -1.04720287e-01 8.55825007e-01
-2.97122717e-01 9.92235243e-01 3.13562810e-01 -4.12398010e-01
-1.08028090e+00 -7.79176354e-01 7.80932486e-01 2.53997240e-02
4.16529328e-01 -2.42231399e-01 -8.81074011e-01 3.78639668e-01
3.89515549e-01 -2.28509814e-01 8.71227086e-01 1.59664571e-01
6.72709882e-01 -6.56422079e-01 -1.13590276e+00 2.31479600e-01
7.83825099e-01 -3.28960299e-01 -7.71360159e-01 -3.10895480e-02
2.15546116e-01 -1.65553689e-01 -8.05600464e-01 1.89211160e-01
6.11241877e-01 -1.12328005e+00 3.61564338e-01 -8.66647437e-02
2.18969196e-01 -3.92659992e-01 1.73533902e-01 -1.39139521e+00
-5.01658618e-01 -5.80937803e-01 1.11115836e-01 1.08235538e+00
2.01099470e-01 -8.17476690e-01 4.43928719e-01 4.42987889e-01
-6.65222168e-01 -9.74826276e-01 -8.34673762e-01 -3.78694057e-01
-6.13601744e-01 -1.27063721e-01 6.76293254e-01 4.73716110e-01
2.91629881e-02 6.59995437e-01 -1.20315515e-01 1.88775003e-01
2.20522135e-01 3.14191915e-02 2.98013330e-01 -1.46630144e+00
9.59430486e-02 -3.09044063e-01 -4.18919563e-01 -7.09902942e-01
5.27994148e-03 -7.42539465e-01 -1.62840813e-01 -1.76315320e+00
3.82687718e-01 -1.41686633e-01 -9.18146551e-01 5.66804647e-01
-3.19204569e-01 5.09158313e-01 -2.35407516e-01 2.33176678e-01
-7.33401775e-01 6.15219355e-01 1.09522128e+00 5.72459958e-02
-3.81071955e-01 2.30182260e-01 -9.21601713e-01 7.29303181e-01
1.17395842e+00 -4.28014398e-01 -5.63938737e-01 -3.51582676e-01
-4.85375375e-02 -1.03686266e-01 6.92220256e-02 -1.52003777e+00
3.96186322e-01 3.08559835e-01 8.41373801e-01 -5.20230353e-01
4.25311148e-01 -6.82419479e-01 1.62773848e-01 6.20737433e-01
1.39617696e-01 4.47394073e-01 3.43703032e-01 2.91999668e-01
-2.18208060e-01 -8.90861154e-02 7.21598864e-01 -1.75315216e-01
-7.75451481e-01 3.09737712e-01 -4.10650343e-01 1.05385017e-02
8.58110726e-01 -3.38369817e-01 -1.59470901e-01 -1.59264728e-02
-7.76516020e-01 2.24579617e-01 1.39922783e-01 1.97980240e-01
7.13653982e-01 -1.08060431e+00 -1.64280817e-01 7.21227288e-01
-1.16350748e-01 -2.12818414e-01 5.86618602e-01 1.57681811e+00
-2.73674965e-01 4.60518718e-01 -6.05423689e-01 -5.09075761e-01
-1.21939397e+00 1.95305899e-01 2.65302390e-01 -6.13319315e-02
-5.20345449e-01 8.49689066e-01 1.69471562e-01 3.87169272e-01
1.86854843e-02 -6.10730708e-01 -5.74004948e-01 2.22033128e-01
6.93060338e-01 5.54677486e-01 4.55075234e-01 -4.54854250e-01
-6.27510846e-01 6.62976384e-01 6.05974868e-02 2.57692218e-01
1.36656141e+00 -1.70890003e-01 -4.60581869e-01 4.75530714e-01
1.02525592e+00 4.07859199e-02 -7.08440006e-01 4.19176161e-01
-4.49180394e-01 -1.99227914e-01 1.49410591e-01 -9.58277404e-01
-1.28833985e+00 1.16773713e+00 1.03092992e+00 3.81902874e-01
1.66601896e+00 -2.06601098e-01 1.01608169e+00 1.11049332e-01
3.94800305e-01 -7.88501799e-01 -1.73855633e-01 1.63212940e-01
5.97257078e-01 -1.00596929e+00 5.89909293e-02 2.57637173e-01
-5.22971094e-01 1.10470104e+00 8.45696092e-01 -2.38644496e-01
7.58628726e-01 -1.14140645e-01 -4.16148566e-02 -3.76679480e-01
-4.61723834e-01 -1.16526559e-02 3.61980051e-01 1.76723123e-01
3.02466154e-01 6.40551746e-02 -6.34701610e-01 1.43735313e+00
-4.90381718e-01 1.67884663e-01 2.34671339e-01 5.99483430e-01
-5.87489605e-01 -6.40105784e-01 -2.74745703e-01 8.92163873e-01
-9.60168242e-01 -1.16874106e-01 7.16110021e-02 7.74045765e-01
4.75522906e-01 1.23508847e+00 1.42090812e-01 -7.00338364e-01
3.39936614e-01 7.52714742e-03 2.81712621e-01 -7.39838183e-01
-7.32464314e-01 1.04603626e-01 -3.32963645e-01 -4.68258947e-01
-5.58926105e-01 -3.86877030e-01 -1.26132274e+00 -7.54247233e-02
-3.97918940e-01 7.06818819e-01 4.91441369e-01 1.05111194e+00
6.08266294e-01 8.89449418e-01 5.30673444e-01 -7.74698079e-01
-9.48572829e-02 -1.43669808e+00 -9.73501384e-01 1.05830975e-01
4.11360383e-01 -9.24251497e-01 -4.74144876e-01 -1.81523427e-01] | [13.479604721069336, 3.4920172691345215] |
146d0f84-07b6-4158-9a80-be31577b8b90 | hsurf-net-normal-estimation-for-3d-point | 2210.07158 | null | https://arxiv.org/abs/2210.07158v1 | https://arxiv.org/pdf/2210.07158v1.pdf | HSurf-Net: Normal Estimation for 3D Point Clouds by Learning Hyper Surfaces | We propose a novel normal estimation method called HSurf-Net, which can accurately predict normals from point clouds with noise and density variations. Previous methods focus on learning point weights to fit neighborhoods into a geometric surface approximated by a polynomial function with a predefined order, based on which normals are estimated. However, fitting surfaces explicitly from raw point clouds suffers from overfitting or underfitting issues caused by inappropriate polynomial orders and outliers, which significantly limits the performance of existing methods. To address these issues, we introduce hyper surface fitting to implicitly learn hyper surfaces, which are represented by multi-layer perceptron (MLP) layers that take point features as input and output surface patterns in a high dimensional feature space. We introduce a novel space transformation module, which consists of a sequence of local aggregation layers and global shift layers, to learn an optimal feature space, and a relative position encoding module to effectively convert point clouds into the learned feature space. Our model learns hyper surfaces from the noise-less features and directly predicts normal vectors. We jointly optimize the MLP weights and module parameters in a data-driven manner to make the model adaptively find the most suitable surface pattern for various points. Experimental results show that our HSurf-Net achieves the state-of-the-art performance on the synthetic shape dataset, the real-world indoor and outdoor scene datasets. The code, data and pretrained models are publicly available. | ['Zhizhong Han', 'Yi Fang', 'Cheng Wang', 'Jin-San Cheng', 'Yu-Shen Liu', 'Qing Li'] | 2022-10-13 | null | null | null | null | ['surface-normals-estimation'] | ['computer-vision'] | [ 1.28974721e-01 -1.97303697e-01 8.68720636e-02 -6.08420014e-01
-7.30438292e-01 -7.98276588e-02 2.76833922e-01 -3.81970517e-02
-2.62264818e-01 2.17277110e-01 -3.54020745e-01 -3.31680626e-02
-2.14924470e-01 -1.29118204e+00 -1.26022470e+00 -6.16555095e-01
9.79079306e-02 7.07707047e-01 4.26325440e-01 2.44859196e-02
4.18392032e-01 1.09529841e+00 -1.86164796e+00 2.00226516e-01
1.07361782e+00 1.21489060e+00 2.48336032e-01 5.00737190e-01
-5.13474166e-01 -7.43477270e-02 -2.31348619e-01 5.84720038e-02
5.00215769e-01 5.13797164e-01 -2.85274267e-01 7.02380314e-02
6.74944520e-01 -1.75850317e-01 -1.70649678e-01 1.01618862e+00
2.62100518e-01 1.56295255e-01 8.29948723e-01 -1.18385732e+00
-9.92061555e-01 -1.85621992e-01 -6.28638864e-01 -4.15843129e-01
6.07882813e-03 8.90776888e-02 6.93295062e-01 -1.35925138e+00
2.39624918e-01 1.28479218e+00 1.00371683e+00 4.04506058e-01
-1.22756982e+00 -7.97114074e-01 2.12763309e-01 -1.85098916e-01
-1.74357080e+00 1.41044017e-02 1.08618248e+00 -5.20856142e-01
1.15591609e+00 3.26927841e-01 6.95397019e-01 4.77502048e-01
5.34922965e-02 4.82061356e-01 4.09409374e-01 -1.65599018e-01
2.02382654e-01 -1.09563001e-01 -2.27307662e-01 5.39805770e-01
2.29864299e-01 -4.05410508e-04 -6.03573918e-02 -4.44266379e-01
1.32343662e+00 6.04891956e-01 -2.55408973e-01 -7.56333113e-01
-1.15769565e+00 6.77086771e-01 9.94573712e-01 -1.44070499e-02
-5.18098712e-01 1.44667819e-01 -1.46847039e-01 9.51393247e-02
5.23690164e-01 2.67249167e-01 -7.51273930e-01 1.57525808e-01
-6.52569890e-01 4.64008629e-01 4.74719226e-01 1.18113267e+00
1.42673397e+00 -1.39162317e-01 7.03909695e-02 1.10603762e+00
8.51551175e-01 1.04849446e+00 2.83319116e-01 -8.02019954e-01
6.03721917e-01 1.14075613e+00 7.40247443e-02 -1.29156649e+00
-3.18131894e-01 -1.92826301e-01 -9.14671779e-01 4.25498158e-01
-8.86125937e-02 1.02723874e-01 -1.02525103e+00 1.20677531e+00
5.34486175e-01 7.99271822e-01 -1.88611999e-01 9.02410865e-01
8.07243943e-01 8.64360809e-01 -4.80679005e-01 2.87934601e-01
8.12658012e-01 -7.76574731e-01 -9.06928852e-02 -2.06528977e-01
3.80435854e-01 -6.75012946e-01 1.36440027e+00 2.73291022e-01
-1.00038588e+00 -5.98316550e-01 -1.12617135e+00 -1.61143124e-01
-3.11796099e-01 1.40118659e-01 2.81921715e-01 3.71434651e-02
-9.07380044e-01 8.53710651e-01 -1.09997451e+00 -4.05199304e-02
7.03886628e-01 6.63131475e-01 -1.47871733e-01 1.17080756e-01
-5.34578085e-01 3.66496474e-01 -4.84134723e-03 2.47109115e-01
-2.19889313e-01 -1.01431251e+00 -9.69521821e-01 7.37130567e-02
1.30395731e-03 -7.42499053e-01 1.06601477e+00 -5.08426666e-01
-1.56368780e+00 6.28887951e-01 -3.43395680e-01 2.86511760e-02
2.54171222e-01 -3.55298340e-01 -3.15354645e-01 -3.57920200e-01
1.24088153e-02 6.42807424e-01 8.22689295e-01 -1.46189332e+00
-6.98735058e-01 -6.30244315e-01 -4.87823159e-01 2.27770671e-01
-2.29993671e-01 -3.73508215e-01 -6.25083685e-01 -3.67240191e-01
6.87949300e-01 -7.24455833e-01 -5.07164121e-01 3.94083142e-01
-3.34651798e-01 -3.63937795e-01 1.00804102e+00 -8.76628458e-02
1.01407766e+00 -2.26807332e+00 -1.56362906e-01 7.68646836e-01
1.74852550e-01 -6.29466698e-02 -1.41711652e-01 5.55157475e-02
1.16664700e-01 -2.29574777e-02 -4.41971809e-01 -3.91006917e-01
-2.69586220e-02 3.74759823e-01 -5.23387074e-01 5.99906623e-01
3.98218244e-01 7.59618223e-01 -7.66550899e-01 -1.98225603e-01
5.93156099e-01 8.27972591e-01 -7.20396519e-01 3.92771035e-01
-4.55973834e-01 2.13611320e-01 -6.41293764e-01 7.64975488e-01
1.12038004e+00 -4.15815651e-01 -7.30403483e-01 -3.27483684e-01
-1.45513162e-01 -3.19753848e-02 -1.48361588e+00 1.70695925e+00
-3.58994633e-01 1.85350329e-01 -2.15501875e-01 -4.50411707e-01
1.57712662e+00 -1.93420365e-01 5.46968937e-01 -2.64530867e-01
-7.45141581e-02 4.58923548e-01 -4.19876814e-01 -3.59029651e-01
3.63956362e-01 9.85655934e-02 2.54534513e-01 -2.49168828e-01
-3.05148125e-01 -4.69463497e-01 -5.86472154e-01 -5.82251787e-01
8.45540047e-01 2.33975589e-01 -9.62616503e-02 2.10824888e-03
6.69018745e-01 -7.89905190e-02 6.75238013e-01 3.97136927e-01
2.39736050e-01 1.09648550e+00 6.16698898e-03 -8.01932275e-01
-1.13730788e+00 -1.35820580e+00 -5.90709567e-01 6.53450847e-01
3.97187650e-01 -1.91348165e-01 -4.84526128e-01 -4.54526424e-01
4.96328145e-01 5.24494171e-01 -3.93164158e-01 1.67844202e-02
-7.84068227e-01 -4.81128663e-01 1.17169999e-01 7.19576716e-01
3.25692981e-01 -1.06509900e+00 -3.34259570e-01 3.81876305e-02
4.32195097e-01 -8.45492184e-01 -4.35240299e-01 1.09384134e-01
-1.15839088e+00 -9.84258115e-01 -4.69649792e-01 -9.05785501e-01
1.09755421e+00 1.77646145e-01 8.04197252e-01 1.46425411e-01
1.33851608e-02 1.50197446e-01 1.69040456e-01 -5.46267092e-01
4.90656830e-02 1.06153272e-01 9.93070602e-02 2.09533516e-02
7.79231250e-01 -9.19285774e-01 -6.91085041e-01 4.37885314e-01
-7.32677281e-01 5.61364777e-02 6.08843505e-01 6.36159778e-01
1.30888426e+00 -3.07784855e-01 1.50292352e-01 -7.53264368e-01
3.21271271e-01 -3.86028618e-01 -9.42090869e-01 1.66908056e-02
-4.32863861e-01 7.34191686e-02 8.22184324e-01 -2.51229852e-01
-4.96412516e-01 4.36419010e-01 -3.74873489e-01 -1.02616739e+00
-4.04857457e-01 1.07387878e-01 -2.21538454e-01 -5.13136744e-01
7.26409256e-01 2.45122761e-01 -5.27505800e-02 -6.16303802e-01
3.35435420e-01 7.44207740e-01 6.15021229e-01 -6.62541687e-01
1.23230994e+00 6.29126549e-01 6.45551160e-02 -1.00425410e+00
-5.00696838e-01 -6.66591704e-01 -8.53105664e-01 -7.89938029e-03
6.00256979e-01 -8.10020983e-01 -7.57346153e-01 3.35948974e-01
-1.28742671e+00 -1.71128467e-01 -5.49308598e-01 2.92061508e-01
-6.96156502e-01 2.23121583e-01 -4.26267773e-01 -6.54684365e-01
-4.91674662e-01 -1.11133289e+00 1.46031845e+00 3.46107334e-01
1.94754392e-01 -7.92532504e-01 3.99650306e-01 -2.17826948e-01
2.32371613e-01 3.50312322e-01 7.40368247e-01 -4.57821041e-01
-1.06639910e+00 -3.46288532e-01 -3.57018679e-01 3.77963603e-01
2.37007365e-01 2.36840189e-01 -9.81738389e-01 -2.91242331e-01
-4.38351519e-02 5.40523604e-02 5.45262098e-01 4.32697207e-01
1.78746498e+00 -2.51424491e-01 -5.71694851e-01 1.36463308e+00
1.63088191e+00 -1.16954520e-02 5.59966564e-01 5.11418223e-01
1.05812001e+00 6.58812523e-02 4.02832866e-01 3.52883458e-01
6.12303913e-01 4.31655109e-01 6.99237049e-01 -7.56799281e-02
3.38064045e-01 -4.94863302e-01 -7.59320483e-02 1.04497957e+00
-7.64766484e-02 3.25576901e-01 -1.03548300e+00 3.60075742e-01
-1.90298617e+00 -4.16207701e-01 -1.71656400e-01 2.26760101e+00
6.06419444e-01 1.70933306e-01 -3.17029357e-01 -9.05882791e-02
5.19772112e-01 -1.28347529e-02 -8.31753969e-01 -1.16930105e-01
8.39286074e-02 1.18325040e-01 6.63178802e-01 4.31186408e-01
-1.12695098e+00 8.62504542e-01 5.65103579e+00 6.81912184e-01
-1.35941553e+00 -3.69457603e-01 3.05772603e-01 -2.07387600e-02
-5.15958846e-01 -2.55905271e-01 -8.29259098e-01 4.23818827e-01
4.87231851e-01 1.49289444e-01 5.15234768e-01 1.30259371e+00
5.26068313e-03 6.57240152e-01 -1.01079822e+00 1.35975277e+00
-6.61826357e-02 -1.56909311e+00 2.31756091e-01 2.92029995e-02
6.96051419e-01 6.75758421e-01 2.45835297e-02 2.20646337e-01
1.59276694e-01 -1.11171126e+00 4.64581043e-01 9.63560402e-01
8.61186385e-01 -6.73801899e-01 4.99450892e-01 5.12094080e-01
-1.42689466e+00 5.59902266e-02 -1.00760090e+00 -3.01810931e-02
-5.48261106e-02 5.76243460e-01 -8.57151091e-01 3.50348085e-01
8.60706866e-01 9.35355544e-01 -3.93192947e-01 1.42566979e+00
-1.35498457e-02 2.29565501e-01 -8.48627567e-01 -1.96743548e-01
6.36958238e-03 -3.41996849e-01 4.15547311e-01 9.52032626e-01
5.94280660e-01 -4.29461449e-02 3.39452565e-01 1.19763100e+00
-1.06872544e-02 2.30703533e-01 -5.62367857e-01 6.80734932e-01
8.20481718e-01 1.29839659e+00 -4.97706950e-01 1.55710399e-01
-4.95159060e-01 4.67926681e-01 7.17132390e-01 4.41940755e-01
-5.22479653e-01 -6.53964520e-01 9.27842379e-01 5.92952073e-01
3.64001513e-01 -3.61148596e-01 -8.08423877e-01 -1.02244675e+00
3.54259402e-01 -2.46899918e-01 -1.06871277e-01 -7.74088860e-01
-1.55820286e+00 6.09044135e-01 -2.08867848e-01 -1.49469066e+00
1.38245046e-01 -8.24246049e-01 -9.79847789e-01 9.42974329e-01
-1.71614385e+00 -1.07572544e+00 -6.42543197e-01 3.30485076e-01
3.80514830e-01 -1.89885631e-01 6.75987303e-01 2.55719125e-01
-4.55878347e-01 4.71955657e-01 2.40461528e-01 1.10321995e-02
3.30576807e-01 -1.27604949e+00 8.71225893e-01 3.71416718e-01
1.35717238e-03 6.05791330e-01 2.16022030e-01 -7.53057718e-01
-1.40766263e+00 -1.48524153e+00 4.99084026e-01 -5.54740846e-01
3.40324730e-01 -4.39594716e-01 -1.41700077e+00 7.10141003e-01
-6.09053612e-01 5.68744659e-01 3.64599019e-01 -1.07331304e-02
-1.67074744e-02 -2.68800437e-01 -1.20892394e+00 5.02579927e-01
1.19205356e+00 -2.83343643e-01 -6.44489467e-01 2.09588900e-01
1.05199695e+00 -8.26819539e-01 -1.06857538e+00 9.59635437e-01
2.63109416e-01 -6.30591989e-01 1.23920906e+00 -3.45939696e-01
1.91097692e-01 -6.38552606e-01 -3.38743895e-01 -1.32117236e+00
-6.50809348e-01 -2.81817108e-01 -3.31714034e-01 7.87485421e-01
4.56833154e-01 -7.36752331e-01 1.20330381e+00 6.88193858e-01
-4.59973216e-01 -1.42334187e+00 -9.82905149e-01 -4.68493462e-01
1.07070461e-01 -7.73973048e-01 1.33228815e+00 7.33666480e-01
-5.19251823e-01 6.25687167e-02 2.96224177e-01 6.78748012e-01
6.52084708e-01 1.62636966e-01 1.18668985e+00 -1.57428050e+00
2.10032061e-01 -3.85566413e-01 -6.34409308e-01 -1.48830450e+00
1.14346117e-01 -9.11226988e-01 2.12147653e-01 -1.63287246e+00
-3.96436930e-01 -9.66610968e-01 -1.55367106e-01 3.27432901e-01
-6.28171340e-02 1.07740033e-02 -8.54784921e-02 5.20391941e-01
-3.94052595e-01 9.74264264e-01 1.28287756e+00 -7.20436778e-03
-5.38280487e-01 3.55849862e-01 -4.09184992e-01 1.11821413e+00
7.31310546e-01 -3.07303905e-01 -2.05871046e-01 -8.08944702e-01
3.70508730e-01 -4.81049836e-01 3.07358980e-01 -1.33965111e+00
5.68340659e-01 -4.83401477e-01 8.39068472e-01 -1.08858800e+00
5.11045396e-01 -1.22887182e+00 -7.26044998e-02 2.91640200e-02
1.44545911e-02 2.30411962e-01 2.33046129e-01 6.54929817e-01
4.62801643e-02 -5.34939952e-03 5.80391765e-01 1.77517667e-01
-2.42375046e-01 1.04890060e+00 4.44108456e-01 -2.73632258e-01
9.52900887e-01 -6.04218304e-01 -1.72550194e-02 2.23195091e-01
-3.94838691e-01 4.77149487e-01 8.40241492e-01 5.15043199e-01
1.28132939e+00 -1.74053657e+00 -6.37794673e-01 1.03552926e+00
2.38189965e-01 1.15770113e+00 7.04074651e-02 2.18829617e-01
-8.05900872e-01 3.28392573e-02 1.35895208e-01 -1.26845825e+00
-6.11642599e-01 3.94569814e-01 4.96789277e-01 3.53141636e-01
-1.02997851e+00 7.83619583e-01 1.29863590e-01 -1.18412161e+00
3.21103871e-01 -1.14688587e+00 -1.20887890e-01 -8.34458292e-01
3.59377146e-01 3.60900342e-01 1.59885809e-01 -6.16429865e-01
-2.81902492e-01 1.22195387e+00 1.48569152e-01 3.66989434e-01
1.36945260e+00 1.83175281e-01 -1.37023181e-01 6.23600304e-01
1.44658077e+00 3.93215977e-02 -1.55027092e+00 -4.95015740e-01
-5.14051244e-02 -7.55260050e-01 -3.50804962e-02 -1.41748905e-01
-8.88986230e-01 6.73989058e-01 6.45296812e-01 4.04781736e-02
7.01139033e-01 -1.44813200e-02 1.05562127e+00 6.54547334e-01
2.86705077e-01 -9.43010926e-01 -2.39307106e-01 6.90371096e-01
1.06045866e+00 -1.12892997e+00 -1.25775442e-01 -4.96969521e-01
-3.73785533e-02 1.30802691e+00 1.05123830e+00 -7.42479622e-01
1.11768556e+00 4.39928651e-01 -6.77893236e-02 -3.67401421e-01
-4.75957125e-01 2.51790315e-01 5.97286224e-01 7.18923330e-01
3.20557989e-02 1.29731014e-01 4.99031276e-01 7.76395619e-01
-5.11335909e-01 7.75191560e-02 -1.71399891e-01 6.58552706e-01
-9.31521237e-01 -6.92002714e-01 -7.31012106e-01 8.72460008e-01
2.24209115e-01 1.71694681e-01 8.26185644e-02 4.10858005e-01
2.24881560e-01 1.68581262e-01 6.69999540e-01 -5.89059889e-01
9.92713869e-01 -2.30907440e-01 1.54103994e-01 -7.41157591e-01
-2.88606852e-01 -1.42072663e-01 -6.84070766e-01 -6.30158365e-01
1.29009306e-01 -5.58825254e-01 -1.58922195e+00 -1.27808213e-01
-2.30511412e-01 -4.65963371e-02 6.36246502e-01 6.31410420e-01
6.54909194e-01 4.03487444e-01 8.50487828e-01 -1.42763245e+00
-6.42642081e-01 -8.20767045e-01 -2.84746259e-01 2.94356465e-01
4.53411788e-01 -6.16342485e-01 -2.93531120e-01 -2.88565964e-01] | [8.218914985656738, -3.5113980770111084] |
16ad2cb8-6f76-4805-a736-d9596d934895 | deep-learning-techniques-for-blind-image | 2306.09426 | null | https://arxiv.org/abs/2306.09426v1 | https://arxiv.org/pdf/2306.09426v1.pdf | Deep learning techniques for blind image super-resolution: A high-scale multi-domain perspective evaluation | Despite several solutions and experiments have been conducted recently addressing image super-resolution (SR), boosted by deep learning (DL) techniques, they do not usually design evaluations with high scaling factors, capping it at 2x or 4x. Moreover, the datasets are generally benchmarks which do not truly encompass significant diversity of domains to proper evaluate the techniques. It is also interesting to remark that blind SR is attractive for real-world scenarios since it is based on the idea that the degradation process is unknown, and hence techniques in this context rely basically on low-resolution (LR) images. In this article, we present a high-scale (8x) controlled experiment which evaluates five recent DL techniques tailored for blind image SR: Adaptive Pseudo Augmentation (APA), Blind Image SR with Spatially Variant Degradations (BlindSR), Deep Alternating Network (DAN), FastGAN, and Mixture of Experts Super-Resolution (MoESR). We consider 14 small datasets from five different broader domains which are: aerial, fauna, flora, medical, and satellite. Another distinctive characteristic of our evaluation is that some of the DL approaches were designed for single-image SR but others not. Two no-reference metrics were selected, being the classical natural image quality evaluator (NIQE) and the recent transformer-based multi-dimension attention network for no-reference image quality assessment (MANIQA) score, to assess the techniques. Overall, MoESR can be regarded as the best solution although the perceptual quality of the created HR images of all the techniques still needs to improve. Supporting code: https://github.com/vsantjr/DL_BlindSR. Datasets: https://www.kaggle.com/datasets/valdivinosantiago/dl-blindsr-datasets. | ['Valdivino Alexandre de Santiago Júnior'] | 2023-06-15 | null | null | null | null | ['image-super-resolution', 'image-quality-assessment', 'super-resolution', 'no-reference-image-quality-assessment'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 2.26193950e-01 -4.01935935e-01 -5.65402433e-02 -4.47025523e-02
-1.00496686e+00 -4.47706819e-01 6.45538390e-01 -5.62771618e-01
-2.04001278e-01 7.65374064e-01 5.22205412e-01 -1.88245669e-01
-3.94401252e-01 -5.22035897e-01 -4.92489636e-01 -7.99414098e-01
-8.32855999e-02 -2.77063716e-02 6.64496645e-02 -6.06084049e-01
2.58005619e-01 6.31342649e-01 -1.87227929e+00 3.13237935e-01
1.05567551e+00 9.61288393e-01 2.59515196e-01 8.88708472e-01
1.50259912e-01 7.82453120e-01 -5.59865952e-01 -4.11070138e-01
5.91135263e-01 -6.40311003e-01 -7.63239563e-01 -2.99797058e-01
8.06096256e-01 -4.75238621e-01 -3.51124644e-01 1.30588293e+00
1.02203703e+00 5.23114987e-02 4.11550283e-01 -9.56699312e-01
-1.07544935e+00 4.03240234e-01 -7.47276485e-01 5.61646521e-01
5.83180189e-01 4.97826040e-01 8.02088082e-01 -7.52468944e-01
5.20826578e-01 1.44383359e+00 7.13117778e-01 6.42120063e-01
-1.17553914e+00 -8.09863865e-01 -2.20504910e-01 4.24405813e-01
-1.20009601e+00 -6.43626988e-01 4.50171769e-01 -2.28575587e-01
5.26320696e-01 4.82956916e-01 1.46674201e-01 1.63004756e+00
-9.04873386e-02 2.72725821e-01 1.90596044e+00 -2.51537561e-01
1.29638761e-01 1.99409470e-01 -1.21215321e-01 2.20684439e-01
9.55321416e-02 7.75411189e-01 -4.63779628e-01 1.78082496e-01
1.09859049e+00 -2.53767014e-01 -6.99801803e-01 3.94102074e-02
-1.22576869e+00 6.50334179e-01 7.64758885e-01 5.76041818e-01
-3.09763342e-01 -2.94871300e-01 2.20931724e-01 6.61120236e-01
4.01749939e-01 5.56740165e-01 -3.97713274e-01 1.93619639e-01
-9.62024570e-01 1.61496311e-01 3.40828896e-01 5.07543325e-01
4.96414363e-01 3.85052472e-01 -4.09619212e-01 9.99352336e-01
-5.00643626e-02 5.19889414e-01 6.80153131e-01 -1.13189971e+00
1.57332331e-01 1.75039947e-01 3.22245270e-01 -8.29338789e-01
-3.61469388e-01 -5.80218792e-01 -1.41196871e+00 7.68862426e-01
2.30975330e-01 7.95864314e-02 -1.08221102e+00 1.62482882e+00
-9.08425450e-03 4.27885912e-02 2.36896306e-01 1.41218209e+00
1.16732252e+00 5.25781631e-01 1.06524425e-02 -2.41786942e-01
1.29768586e+00 -9.44482863e-01 -7.25820124e-01 -6.62865043e-02
-2.19680369e-01 -7.93043554e-01 1.43798089e+00 6.76901221e-01
-1.17787004e+00 -1.14700925e+00 -1.05769920e+00 -5.48532754e-02
-4.74111527e-01 6.57569394e-02 3.58314008e-01 6.42863452e-01
-1.63150275e+00 7.05144227e-01 -2.09596410e-01 -4.93317068e-01
3.79918665e-01 9.38174650e-02 -3.90705645e-01 -2.89177239e-01
-1.37912846e+00 1.04685771e+00 1.29731521e-01 2.60557327e-02
-1.12840652e+00 -6.17795467e-01 -4.68784869e-01 -1.75550625e-01
6.55297115e-02 -7.62113154e-01 8.67374539e-01 -1.29033816e+00
-1.54471350e+00 9.48230565e-01 -9.07696725e-04 -5.28306186e-01
6.67067111e-01 -3.32135260e-01 -1.00665951e+00 2.31329873e-01
-7.30529577e-02 4.62937891e-01 1.26738012e+00 -1.69601870e+00
-4.21625108e-01 -4.26841557e-01 2.37339512e-01 2.85375595e-01
-1.87215041e-02 4.42808509e-01 -1.60626426e-01 -7.51080096e-01
-2.04310074e-01 -4.98147249e-01 -4.88487668e-02 7.79521710e-04
-2.05976069e-01 3.12407643e-01 7.16257572e-01 -1.17932463e+00
9.81655836e-01 -2.22784638e+00 2.07464933e-01 -3.57938468e-01
2.65353799e-01 6.40606284e-01 -5.19974589e-01 1.34934276e-01
-4.53995913e-01 1.31088927e-01 -5.22707760e-01 -2.50016809e-01
-1.63831756e-01 -1.29092440e-01 -3.29765737e-01 6.27075255e-01
1.88842520e-01 6.64930284e-01 -6.47103429e-01 -3.55524480e-01
3.99042398e-01 1.05224657e+00 -9.49444622e-02 4.15062696e-01
3.95507514e-02 8.66457880e-01 1.63054019e-02 6.98067963e-01
1.05688882e+00 -8.81415978e-02 -3.16234887e-01 -6.12111509e-01
-2.56470054e-01 -2.38284618e-01 -1.32933438e+00 1.66389608e+00
-6.32988453e-01 3.83204848e-01 3.05079013e-01 -6.47386611e-01
8.73487830e-01 3.72331440e-01 1.57825857e-01 -1.16840875e+00
-6.74331710e-02 2.39152476e-01 -1.66100219e-01 -4.34654564e-01
4.94701564e-01 -2.35330425e-02 4.06791121e-01 1.33122221e-01
7.59427398e-02 4.03978862e-03 3.23982909e-02 -2.36392282e-02
9.53733861e-01 2.15642542e-01 3.09390992e-01 -4.71919067e-02
8.11017334e-01 -2.86026359e-01 2.68748909e-01 5.65873027e-01
-3.53867024e-01 1.17367065e+00 1.52278408e-01 -2.18345731e-01
-1.26301527e+00 -1.23127151e+00 -3.37886751e-01 9.58378315e-01
2.81908661e-01 2.16146469e-01 -6.62303567e-01 -3.29249710e-01
-4.10286933e-01 5.44300437e-01 -5.93076706e-01 6.19763508e-02
-3.40804130e-01 -9.22584593e-01 4.58352774e-01 1.87383175e-01
8.56526196e-01 -1.23380649e+00 -4.86681134e-01 -2.73595423e-01
-2.35241875e-01 -1.16910720e+00 -2.03222647e-01 -2.36803159e-01
-5.79280496e-01 -1.04477763e+00 -1.20155454e+00 -3.71201128e-01
2.19824806e-01 3.99881929e-01 1.24175978e+00 -6.30788952e-02
-2.57697046e-01 2.98252553e-01 -6.33776128e-01 1.59342095e-01
-5.66866875e-01 -4.11700010e-01 2.00023696e-01 1.27621636e-01
-2.65864097e-02 -6.58657789e-01 -9.66142714e-01 3.90224218e-01
-1.03988886e+00 -1.43379852e-01 8.95733416e-01 8.12987447e-01
6.48720145e-01 2.54620910e-01 5.14085710e-01 -4.12473381e-01
5.67281902e-01 -3.07846695e-01 -4.89420682e-01 2.41418526e-01
-6.68591082e-01 -1.92312241e-01 6.92044139e-01 -4.52053279e-01
-1.19361937e+00 -4.93332267e-01 -2.74929494e-01 -5.62441170e-01
-4.83667105e-01 -4.77057621e-02 -3.14969778e-01 -2.99267501e-01
1.04882646e+00 2.51386702e-01 -9.29560065e-02 -7.17696667e-01
4.42826927e-01 8.21320236e-01 7.95789003e-01 -1.76026881e-01
9.56027448e-01 4.75939184e-01 -2.47122034e-01 -7.04401314e-01
-6.26780331e-01 -2.58009344e-01 -3.14704210e-01 -1.09050035e-01
8.82066786e-01 -1.22368085e+00 -5.33643782e-01 6.21069610e-01
-8.82395864e-01 -3.97201836e-01 -4.39735323e-01 2.88277864e-01
-5.17214239e-01 4.84763265e-01 -5.00000954e-01 -9.08809185e-01
-5.85591733e-01 -1.22467184e+00 9.66268897e-01 3.66154790e-01
2.33867735e-01 -6.12531543e-01 6.31032651e-03 4.97679710e-01
1.03148377e+00 4.25932884e-01 5.47985673e-01 -1.32840633e-01
-3.39660019e-01 2.97737747e-01 -8.15767765e-01 7.44359910e-01
1.02958903e-01 -2.33869657e-01 -1.36134589e+00 -6.38721526e-01
1.60118774e-01 -3.49459738e-01 8.77445340e-01 5.09793997e-01
1.17043364e+00 -2.90680021e-01 3.87967229e-01 9.45926368e-01
1.89987302e+00 -8.94125253e-02 1.07628226e+00 5.45792997e-01
4.86721277e-01 5.19433379e-01 5.41644514e-01 1.45579785e-01
1.48792416e-01 8.41600835e-01 7.42139339e-01 -4.55542535e-01
-8.79386485e-01 4.01732735e-02 7.57240891e-01 3.81381691e-01
-5.30903697e-01 -2.71077484e-01 -5.12293339e-01 4.67838377e-01
-1.22986245e+00 -9.57619309e-01 -1.51376024e-01 2.31212926e+00
7.37796605e-01 -2.51225531e-01 2.43200511e-01 4.63658333e-01
8.12594235e-01 3.51234883e-01 -6.92505777e-01 -2.35477194e-01
-8.55353177e-01 5.00757873e-01 5.33612669e-01 4.90165621e-01
-1.01360834e+00 7.25181997e-01 5.21757793e+00 8.51190865e-01
-1.09243643e+00 4.27156419e-01 7.05244720e-01 4.03514579e-02
-7.55366078e-03 -2.84473389e-01 -4.56852525e-01 4.13906276e-01
1.09487903e+00 1.06365494e-01 9.72572088e-01 4.24475938e-01
2.72189081e-01 -2.38860194e-02 -4.92507398e-01 1.07852340e+00
1.12855621e-01 -8.64425421e-01 9.11847800e-02 -1.70555070e-01
7.71501839e-01 2.50571340e-01 7.23573267e-01 6.19829148e-02
3.05923223e-01 -1.35787213e+00 4.29034561e-01 7.13769913e-01
1.46902990e+00 -5.60830891e-01 7.80950725e-01 -1.98943511e-01
-9.56992924e-01 -4.61877137e-01 -4.11207050e-01 4.64093447e-01
7.59678753e-03 4.68904078e-01 8.78785253e-02 9.76670444e-01
1.35324025e+00 6.24283195e-01 -8.95944595e-01 9.79721248e-01
-1.49459273e-01 3.74112159e-01 1.70794010e-01 7.53266215e-01
-2.24435646e-02 -1.35504872e-01 8.78557503e-01 1.12104726e+00
5.48576713e-01 -2.99186930e-02 -4.18215513e-01 9.31207299e-01
1.65063187e-01 2.90707052e-02 -3.58519971e-01 1.90295994e-01
1.17640011e-01 1.38337362e+00 -3.13474238e-01 -8.21691081e-02
-3.90973449e-01 1.18739188e+00 -2.63687700e-01 5.91457307e-01
-7.85463870e-01 -1.71696320e-01 7.44822323e-01 1.87041909e-01
2.82467455e-01 6.34361580e-02 -1.87889919e-01 -1.08492827e+00
2.25700513e-02 -1.54363930e+00 3.07035297e-01 -1.25063872e+00
-1.47180188e+00 1.16466248e+00 -2.24054605e-01 -1.49943721e+00
1.04900599e-01 -5.47828734e-01 -3.15057039e-01 1.27039933e+00
-2.10461283e+00 -1.25820267e+00 -8.10109973e-01 9.67501879e-01
5.71684420e-01 -3.70646149e-01 7.10288525e-01 5.21989405e-01
-5.07938325e-01 6.07977808e-01 1.14196196e-01 -1.75218403e-01
1.04881334e+00 -1.26880145e+00 2.65030026e-01 1.17378056e+00
-7.89155513e-02 2.36650199e-01 9.84566808e-01 -3.46023053e-01
-1.26317680e+00 -1.15427911e+00 2.59399563e-01 -3.62347484e-01
5.32070518e-01 1.06517740e-01 -1.13021779e+00 1.44863501e-01
3.64576846e-01 1.90116972e-01 8.81230757e-02 -3.60771179e-01
-6.80318713e-01 -2.71101654e-01 -1.61194801e+00 3.76274973e-01
1.15474153e+00 -4.78640020e-01 -2.51303852e-01 1.78169146e-01
9.61303055e-01 -2.72868365e-01 -1.10090756e+00 4.99121100e-01
3.62212896e-01 -1.61486387e+00 1.31223285e+00 -1.19336754e-01
6.83031201e-01 -5.25059283e-01 -5.05826235e-01 -1.35268176e+00
-3.57451171e-01 -4.15437013e-01 -2.46225148e-01 1.27265394e+00
1.59199923e-01 -7.34670341e-01 1.62061676e-01 -1.53243179e-02
3.71704251e-02 -1.53646693e-01 -8.88349771e-01 -8.81458640e-01
-5.97397238e-03 -1.45320937e-01 9.05984640e-01 9.89906132e-01
-8.81278634e-01 1.47973761e-01 -7.10329533e-01 5.29680729e-01
9.30774987e-01 -7.19535723e-02 5.56711733e-01 -1.07603920e+00
-3.71366680e-01 -5.97011626e-01 -2.02472731e-01 -3.26051891e-01
-2.18549281e-01 -4.77296591e-01 -2.81461656e-01 -1.68935275e+00
1.43719872e-03 -2.16667354e-01 -4.39684749e-01 1.61325902e-01
-5.61219305e-02 5.56377590e-01 6.70215487e-02 2.93166429e-01
-2.94985712e-01 4.02062863e-01 1.41140950e+00 -5.04646935e-02
-1.89697310e-01 -6.58745170e-02 -8.60848546e-01 3.63304049e-01
9.37016368e-01 4.39967290e-02 -2.27662101e-01 -4.73308921e-01
1.35668501e-01 2.25925326e-01 8.58208597e-01 -1.28899503e+00
3.50046577e-03 1.55358031e-01 4.38184440e-01 -1.66080594e-01
4.46124583e-01 -7.31821537e-01 4.92964536e-01 2.45144829e-01
-2.29692772e-01 -1.18201915e-02 1.59430429e-01 2.59600520e-01
-3.46158147e-01 2.09087282e-01 1.44304466e+00 -2.72790700e-01
-9.22213018e-01 4.09526527e-01 2.44588703e-01 -6.21625073e-02
4.54605520e-01 -2.57416248e-01 -7.03818440e-01 -3.87448341e-01
-5.75254023e-01 -2.21689120e-01 7.03335404e-01 5.23225546e-01
7.64301360e-01 -1.10514688e+00 -1.23361194e+00 3.44282016e-02
1.24017142e-01 -2.46429861e-01 8.37943614e-01 8.15624297e-01
-3.37027609e-01 1.18024632e-01 -6.01282656e-01 -2.59052038e-01
-1.19480097e+00 8.71297717e-01 5.16005635e-01 -2.15052322e-01
-8.24708164e-01 7.16537952e-01 8.11085105e-02 -3.73406678e-01
2.22118169e-01 4.76493724e-02 -6.65342987e-01 -4.63788137e-02
9.56157207e-01 5.72795033e-01 3.50660980e-02 -9.18969333e-01
-6.35468960e-02 7.61597514e-01 1.49972498e-01 -1.92185882e-02
1.46375155e+00 -4.43231493e-01 -2.81384885e-01 2.01213554e-01
9.65061247e-01 -2.57491082e-01 -1.06763136e+00 -2.91694671e-01
-2.55715728e-01 -6.37338161e-01 4.11340415e-01 -1.42514825e+00
-1.36115181e+00 8.25995803e-01 1.58650124e+00 1.92606956e-01
1.88949716e+00 -1.50508299e-01 5.70835173e-01 -3.11966300e-01
3.39066327e-01 -6.89032435e-01 -2.67803278e-02 7.32947886e-02
1.50249815e+00 -1.49211764e+00 6.12115487e-03 6.23629540e-02
-7.97985017e-01 7.46723473e-01 5.69074750e-01 -2.14770604e-02
2.98450887e-01 1.90095715e-02 2.51813412e-01 -2.50704959e-02
-3.60705554e-01 -5.83528757e-01 2.79233664e-01 1.05673957e+00
2.30649605e-01 -1.23016663e-01 -2.23608334e-02 3.48088503e-01
-2.90772051e-01 8.44721645e-02 6.00678980e-01 1.27573192e-01
-1.29128069e-01 -7.69214749e-01 -7.66140580e-01 2.59837449e-01
-5.78811586e-01 -2.59111255e-01 -1.35924131e-01 6.82307780e-01
2.36789390e-01 1.09269285e+00 -3.87224823e-01 -4.94708389e-01
3.92643094e-01 -5.24598181e-01 4.35116440e-01 -6.12674616e-02
-5.83617330e-01 -2.15533942e-01 -1.67231470e-01 -8.41847241e-01
-7.99036443e-01 -4.25746262e-01 -3.86038840e-01 -4.55699950e-01
1.25510432e-02 -1.19163223e-01 7.27332175e-01 3.79900604e-01
1.95686296e-01 6.27050161e-01 7.68883169e-01 -1.03122461e+00
-3.47750604e-01 -1.17529392e+00 -7.03945875e-01 3.14343721e-01
8.40823770e-01 -6.25387311e-01 -8.11279476e-01 -1.36609182e-01] | [11.192216873168945, -1.9543488025665283] |
7e76e706-91a6-4511-aa8e-e10ae8e19a06 | wsnet-learning-compact-and-efficient-networks | null | null | https://openreview.net/forum?id=H1I3M7Z0b | https://openreview.net/pdf?id=H1I3M7Z0b | WSNet: Learning Compact and Efficient Networks with Weight Sampling | We present a new approach and a novel architecture, termed WSNet, for learning compact and efficient deep neural networks. Existing approaches conventionally learn full model parameters independently and then compress them via \emph{ad hoc} processing such as model pruning or filter factorization. Alternatively, WSNet proposes learning model parameters by sampling from a compact set of learnable parameters, which naturally enforces {parameter sharing} throughout the learning process. We demonstrate that such a novel weight sampling approach (and induced WSNet) promotes both weights and computation sharing favorably. By employing this method, we can more efficiently learn much smaller networks with competitive performance compared to baseline networks with equal numbers of convolution filters. Specifically, we consider learning compact and efficient 1D convolutional neural networks for audio classification. Extensive experiments on multiple audio classification datasets verify the effectiveness of WSNet. Combined with weight quantization, the resulted models are up to \textbf{180$\times$} smaller and theoretically up to \textbf{16$\times$} faster than the well-established baselines, without noticeable performance drop. | ['Yingzhen Yang', 'Xiaojie Jin', 'Jianchao Yang', 'Ning Xu', 'Jiashi Feng', 'Shuicheng Yan'] | 2018-01-01 | null | null | null | iclr-2018-1 | ['paper-generation'] | ['natural-language-processing'] | [ 4.49484348e-01 5.99876642e-02 -1.28044754e-01 -5.74474216e-01
-5.78800678e-01 -1.78221554e-01 1.02855735e-01 3.01930495e-02
-1.03003836e+00 5.30643165e-01 -1.22508802e-01 -1.75996497e-01
-4.39088315e-01 -6.39692664e-01 -1.03513515e+00 -6.25531077e-01
-2.23331824e-01 3.45369160e-01 -5.59424385e-02 1.24446772e-01
1.18900105e-01 4.74612176e-01 -1.77148330e+00 2.85062134e-01
5.60937107e-01 1.56359017e+00 3.42005640e-01 5.58497667e-01
9.54895187e-03 7.78107524e-01 -5.78572452e-01 -3.73583287e-01
5.62961936e-01 1.90274358e-01 -8.24670672e-01 -1.69821188e-01
7.71528423e-01 -6.66461289e-01 -5.69485366e-01 9.41675782e-01
4.76784468e-01 4.36589330e-01 3.79708946e-01 -9.34122980e-01
-2.73283958e-01 1.23344564e+00 -2.90166825e-01 2.59158731e-01
-5.48301935e-01 9.45082605e-02 1.14361572e+00 -8.91150713e-01
1.06029492e-02 1.30399358e+00 8.89488280e-01 5.69144189e-01
-1.33360493e+00 -1.26563370e+00 5.32081127e-01 8.23486969e-02
-1.71350360e+00 -8.53633642e-01 7.48559713e-01 -1.68185476e-02
1.15769708e+00 2.70725101e-01 7.73344398e-01 8.89262974e-01
-3.31779033e-01 7.89071620e-01 5.73653579e-01 -5.19881248e-01
2.58563668e-01 -9.38453060e-03 1.75135493e-01 8.70390773e-01
3.39766234e-01 3.50455530e-02 -9.68698204e-01 -1.52809158e-01
8.44399333e-01 2.00474679e-01 -3.69798094e-01 -6.34222403e-02
-9.51700151e-01 6.70598686e-01 3.35188121e-01 -2.22275518e-02
-1.62548810e-01 7.99845397e-01 4.66138154e-01 4.30128515e-01
4.11494523e-01 3.60443860e-01 -7.56619930e-01 -1.35383189e-01
-1.22972286e+00 3.79748791e-01 5.65220594e-01 1.13131499e+00
7.78825700e-01 5.73548615e-01 -2.30280198e-02 9.65835035e-01
3.17158103e-01 3.92025560e-01 4.84304935e-01 -1.14710498e+00
5.59145331e-01 3.07114154e-01 -2.33341098e-01 -7.72062898e-01
-3.30047965e-01 -7.44098604e-01 -1.12806523e+00 -9.39616561e-02
1.54985681e-01 -3.14061165e-01 -8.11927736e-01 2.04828787e+00
8.27931315e-02 5.19010067e-01 -1.78175822e-01 5.26049793e-01
4.56199437e-01 5.31537831e-01 6.39743507e-02 -8.64700824e-02
1.17842352e+00 -9.48620200e-01 -5.08084118e-01 -1.89103603e-01
5.99650741e-01 -5.23736656e-01 1.15750933e+00 8.24108779e-01
-1.15031374e+00 -6.68978512e-01 -1.32298505e+00 -9.85814631e-02
-1.01316266e-01 2.34230399e-01 8.41244340e-01 7.82306731e-01
-1.08140397e+00 1.10513949e+00 -1.03763127e+00 3.38038445e-01
9.71874416e-01 9.54606593e-01 7.04015717e-02 -7.84870237e-02
-9.96103108e-01 2.62161821e-01 7.39067316e-01 6.97508976e-02
-9.89351332e-01 -1.20769167e+00 -6.34938478e-01 4.77488786e-01
3.43607336e-01 -6.18034303e-01 1.46667969e+00 -7.49309480e-01
-1.57371080e+00 2.76691884e-01 -6.89068660e-02 -1.07992554e+00
1.79276079e-01 -5.44343293e-01 -2.18511596e-01 2.68892944e-01
-7.02152014e-01 1.03463244e+00 1.25500095e+00 -8.46469283e-01
-6.71927154e-01 -1.39684409e-01 2.83485383e-01 -1.32779941e-01
-1.36399734e+00 4.43635993e-02 -2.62173653e-01 -9.98556793e-01
-5.37239276e-02 -7.79716969e-01 -3.42995226e-01 4.33313578e-01
-1.18130639e-01 -3.44530553e-01 6.02046728e-01 -2.80481040e-01
1.64645410e+00 -2.09961295e+00 -3.85170169e-02 4.14931267e-01
6.74113333e-01 5.49105704e-01 -2.58909911e-01 -7.71424696e-02
-1.41668255e-02 2.31985867e-01 -7.69912228e-02 -6.92869008e-01
3.18487912e-01 4.66140032e-01 -4.34823006e-01 2.39601105e-01
8.93926844e-02 5.91031849e-01 -5.17064512e-01 -3.69296312e-01
3.41624627e-03 7.84056962e-01 -1.06861508e+00 -6.85332641e-02
-3.16080749e-01 -2.50635743e-01 -3.39678377e-01 4.17319387e-01
6.66044176e-01 -2.40321726e-01 2.00705826e-01 -4.95813042e-01
2.38204971e-01 5.26748061e-01 -1.57662952e+00 1.80979347e+00
-5.48320353e-01 5.32632768e-01 1.83645621e-01 -1.31259680e+00
8.63064408e-01 3.17101538e-01 4.11459059e-01 -4.02779549e-01
2.57323831e-01 1.07987672e-01 -1.66532055e-01 2.51107737e-02
3.94267976e-01 2.82504223e-02 2.65574932e-01 5.85283399e-01
6.18138075e-01 7.57208094e-02 2.11167052e-01 3.11079230e-02
8.82912159e-01 -1.66265264e-01 -2.34754235e-01 -4.51969981e-01
2.41232425e-01 -6.30692720e-01 7.62365162e-01 8.03293407e-01
6.30215332e-02 1.69225708e-01 1.53061211e-01 -6.33317530e-01
-8.31123590e-01 -7.74387360e-01 -1.62804604e-01 1.51975286e+00
-3.61948609e-01 -9.93330419e-01 -9.78751659e-01 -2.75477350e-01
-8.93602744e-02 3.76160592e-01 -4.27005112e-01 -3.84602398e-01
-7.82927215e-01 -8.43239665e-01 9.83464181e-01 8.70564938e-01
4.25435156e-01 -7.45757222e-01 -8.15767884e-01 2.31577292e-01
2.43418366e-01 -9.49382305e-01 -4.91668046e-01 7.41427422e-01
-1.28489304e+00 -6.27234042e-01 -4.38448071e-01 -8.46534550e-01
6.58954680e-01 9.75880623e-02 1.10238743e+00 3.47154826e-01
-1.98764503e-01 -2.25475356e-02 -5.67756221e-02 -7.51977026e-01
1.46293238e-01 3.25051397e-01 5.14084280e-01 -4.97467890e-02
2.74915338e-01 -9.51041579e-01 -7.30408251e-01 -1.25083014e-01
-1.17304134e+00 8.16586316e-02 5.68468034e-01 8.60157967e-01
8.12225878e-01 3.01556289e-01 5.08333802e-01 -9.45058644e-01
3.68693084e-01 1.41801208e-01 -6.90666914e-01 1.20007887e-01
-9.15829718e-01 3.72347087e-01 7.63862014e-01 -7.84831643e-01
-7.58382499e-01 7.25139901e-02 -1.92412928e-01 -9.20288384e-01
6.87023904e-03 2.37320229e-01 -5.35673946e-02 -3.22285503e-01
6.34427130e-01 8.01408738e-02 -1.78398117e-01 -8.23517680e-01
3.79143208e-01 4.31488037e-01 3.33962709e-01 -8.46596420e-01
8.87047708e-01 3.43190849e-01 -1.77076876e-01 -6.36290789e-01
-8.20728779e-01 2.09754650e-02 -5.55475056e-01 2.15499878e-01
3.07517290e-01 -1.18153894e+00 -9.12766993e-01 3.81757051e-01
-9.29558933e-01 -5.17191291e-01 -6.18574977e-01 6.45035982e-01
-3.82785350e-01 -7.37901852e-02 -7.45199025e-01 -7.15154529e-01
-6.16676867e-01 -9.95134830e-01 8.09249759e-01 6.71886131e-02
-7.22434446e-02 -7.27950335e-01 -5.22296369e-01 1.79274172e-01
5.86268783e-01 -2.65541911e-01 1.28148651e+00 -6.11681819e-01
-5.40581286e-01 -8.37683976e-02 -2.30110630e-01 8.27320874e-01
-2.20252186e-01 -2.71776378e-01 -1.33491290e+00 -5.51098883e-01
-8.69853646e-02 -5.66377580e-01 1.26241612e+00 3.32697153e-01
1.99773777e+00 -5.92463911e-01 -2.24021986e-01 1.01993334e+00
1.30877590e+00 -4.26928885e-02 1.84744805e-01 -1.04217179e-01
6.82044685e-01 2.06998974e-01 7.43566677e-02 8.44417334e-01
3.40347812e-02 3.75202030e-01 3.85356843e-01 -3.05592734e-02
-8.73299465e-02 -9.58313569e-02 2.44889677e-01 1.11150360e+00
-1.75113052e-01 2.30603782e-03 -5.95319450e-01 3.31759602e-01
-1.43316507e+00 -6.40533447e-01 7.11930394e-01 2.01553464e+00
1.25029421e+00 3.93718749e-01 -1.30556434e-01 6.73791647e-01
2.85036445e-01 1.67733103e-01 -6.10063255e-01 -3.43635440e-01
7.81661347e-02 1.05735004e+00 5.73786855e-01 1.78139150e-01
-1.07773507e+00 7.53317714e-01 6.33450890e+00 1.21821105e+00
-1.13849986e+00 1.17106333e-01 7.19812334e-01 -8.64871264e-01
-2.01489955e-01 -2.29401097e-01 -1.15769398e+00 4.53777574e-02
1.40092695e+00 -2.61942334e-02 6.81763768e-01 7.74874210e-01
2.45819194e-03 5.24384975e-01 -1.18631577e+00 1.20167446e+00
-1.51191607e-01 -1.63631606e+00 3.17838997e-01 -6.21921532e-02
6.10522211e-01 -2.36912649e-02 2.81726211e-01 3.97619754e-01
3.07455719e-01 -1.10042071e+00 1.03171182e+00 4.76716280e-01
9.01978612e-01 -1.22483492e+00 3.17470431e-01 2.04125896e-01
-1.33911383e+00 -5.27131438e-01 -8.38547647e-01 -1.82347342e-01
-2.26787835e-01 7.07379580e-01 -5.23015082e-01 1.95782125e-01
1.04243457e+00 6.63358748e-01 -5.60287595e-01 9.11946654e-01
1.98342726e-01 1.06507158e+00 -6.39848530e-01 2.85489783e-02
1.65872887e-01 2.26102993e-01 9.86310020e-02 1.23350334e+00
2.73592502e-01 6.45109415e-02 1.84357896e-01 6.95624590e-01
-6.72180176e-01 -8.09033811e-02 -2.92108152e-02 -1.65212080e-01
8.03396046e-01 9.46469784e-01 -5.28725028e-01 -3.73611271e-01
-1.66134819e-01 5.53308189e-01 5.52881360e-01 3.43078613e-01
-7.89586902e-01 -6.47115409e-01 8.71666431e-01 -2.73523945e-02
7.65043318e-01 -2.37704605e-01 -1.72575012e-01 -8.56101155e-01
8.37071687e-02 -8.31142724e-01 2.04150587e-01 -1.97397262e-01
-8.45879078e-01 6.39726341e-01 1.19428813e-01 -1.02867317e+00
6.57807440e-02 -7.46611714e-01 -1.43353388e-01 5.03124475e-01
-1.69937623e+00 -9.32222962e-01 2.44507771e-02 6.64642215e-01
4.89866287e-01 -2.83572227e-01 1.01130378e+00 6.07415974e-01
-6.84290469e-01 1.28736675e+00 2.36211926e-01 6.79659396e-02
4.37683493e-01 -9.44534063e-01 2.58393049e-01 4.50563818e-01
5.58002949e-01 8.73201847e-01 2.21654132e-01 7.97066763e-02
-1.40982211e+00 -1.28113699e+00 5.47637999e-01 1.04150921e-01
5.49205601e-01 -7.55283594e-01 -7.94252336e-01 5.59175491e-01
9.76081267e-02 2.37820104e-01 9.13227201e-01 7.29890987e-02
-6.80718362e-01 -7.16219842e-01 -1.03713620e+00 4.95162606e-01
1.36815262e+00 -4.99767959e-01 -2.29329497e-01 3.61641884e-01
1.28798187e+00 -2.92348266e-01 -1.11068618e+00 5.60488462e-01
6.72023594e-01 -6.58543050e-01 1.22441983e+00 -7.29939640e-01
5.60818762e-02 1.00727357e-01 -4.98817235e-01 -8.48926961e-01
-1.82351395e-01 -9.51573670e-01 -7.09757149e-01 1.17510390e+00
3.83647650e-01 -3.51015598e-01 8.57930362e-01 3.98262888e-01
-2.79183447e-01 -1.10402966e+00 -1.04471767e+00 -6.90848172e-01
2.49394566e-01 -8.05086732e-01 8.68453503e-01 5.86738527e-01
-5.33906519e-01 -3.92427035e-02 -3.16947907e-01 6.32305071e-02
7.21472681e-01 -2.79063135e-01 3.41897696e-01 -1.48610222e+00
-5.08126020e-01 -5.79569817e-01 -2.52027214e-01 -1.38729680e+00
2.38359824e-01 -8.25192451e-01 -1.03145972e-01 -7.52591193e-01
-1.36068523e-01 -8.65132987e-01 -8.85722518e-01 7.98613787e-01
2.70362049e-01 4.40804183e-01 2.38209352e-01 -2.82786973e-02
-6.69586718e-01 5.71395159e-01 9.69335854e-01 -2.11450353e-01
-8.25353060e-03 -1.02665033e-02 -9.24708962e-01 8.43457878e-01
7.76279032e-01 -5.68278432e-01 -7.45789409e-01 -1.03491437e+00
1.67757735e-01 -4.94348973e-01 3.67062747e-01 -1.41775107e+00
3.04457694e-01 -1.47168580e-02 5.58914840e-01 -3.50439399e-01
5.63043237e-01 -8.51410449e-01 -2.24801943e-01 4.19906616e-01
-6.68029249e-01 2.31647934e-03 5.38819969e-01 6.37207329e-01
-3.33989263e-02 -3.37706268e-01 7.65177965e-01 -4.28154832e-03
-4.61447120e-01 5.92777193e-01 -2.11083367e-01 -8.23330320e-03
5.12766540e-01 -1.21442810e-01 5.08716563e-03 -2.44393684e-02
-7.02026367e-01 -4.82977415e-03 -3.06243658e-01 1.43906161e-01
7.72014976e-01 -1.32830560e+00 -4.32996064e-01 4.67664152e-01
-2.39585981e-01 4.26526666e-01 2.37042665e-01 6.15374327e-01
-2.40317941e-01 4.19274181e-01 4.27325293e-02 -4.98347282e-01
-1.18473959e+00 1.76695421e-01 3.22901845e-01 -2.87529528e-01
-6.77905738e-01 1.48222303e+00 -1.31196067e-01 -3.75986546e-01
8.65436614e-01 -9.09986079e-01 1.60495847e-01 -1.04967793e-02
8.10777783e-01 4.09222573e-01 3.29168290e-01 -1.07457593e-01
2.36124843e-02 3.38475376e-01 -4.03558910e-01 1.54242501e-01
1.73723757e+00 7.36062601e-02 1.08010463e-01 2.02284977e-01
1.25852299e+00 -4.83331770e-01 -1.68510771e+00 -7.40693569e-01
-5.30189648e-02 -3.18813473e-01 5.03814816e-01 -4.02150214e-01
-1.49444008e+00 1.09434235e+00 9.73535061e-01 -1.65598288e-01
1.43558848e+00 -4.68778491e-01 1.07316673e+00 9.72448230e-01
3.08730066e-01 -1.31303155e+00 2.06963748e-01 6.11890793e-01
5.59659362e-01 -6.85559511e-01 -1.22387065e-02 -8.13925639e-02
-6.00695908e-02 1.15918660e+00 5.27910352e-01 -2.68518090e-01
1.07598090e+00 7.54318297e-01 -2.26472110e-01 1.27181411e-01
-1.11244154e+00 2.00105593e-01 2.55991191e-01 5.91637135e-01
3.06248039e-01 -1.97742686e-01 1.97181001e-01 1.05455351e+00
-4.70041782e-01 1.17040090e-01 -1.70699246e-02 9.78616655e-01
-5.96272349e-01 -1.05911052e+00 -5.99689148e-02 7.81199872e-01
-4.84536827e-01 -4.25677687e-01 2.93576449e-01 4.55129117e-01
3.61792922e-01 7.24458039e-01 2.47299761e-01 -6.16051435e-01
1.83845252e-01 2.69942790e-01 6.06746972e-01 -4.09217507e-01
-7.51783252e-01 7.87044093e-02 -2.25834876e-01 -5.88140011e-01
-4.08758730e-01 -2.63248891e-01 -1.33534431e+00 -3.79124105e-01
-5.69382310e-01 -2.16793129e-03 5.46921074e-01 9.61426020e-01
3.84633720e-01 7.46780872e-01 4.05886948e-01 -1.06756997e+00
-1.04982793e+00 -8.68869126e-01 -5.44459045e-01 -7.75107518e-02
3.67331445e-01 -6.59698188e-01 -1.88233927e-01 2.41958514e-01] | [8.599371910095215, 3.2152457237243652] |
0b740e6f-710d-4726-ae8f-90c53352d8a0 | a-hybrid-deep-learning-model-for-arabic-text | 2009.01987 | null | https://arxiv.org/abs/2009.01987v1 | https://arxiv.org/pdf/2009.01987v1.pdf | A Hybrid Deep Learning Model for Arabic Text Recognition | Arabic text recognition is a challenging task because of the cursive nature of Arabic writing system, its joint writing scheme, the large number of ligatures and many other challenges. Deep Learning DL models achieved significant progress in numerous domains including computer vision and sequence modelling. This paper presents a model that can recognize Arabic text that was printed using multiple font types including fonts that mimic Arabic handwritten scripts. The proposed model employs a hybrid DL network that can recognize Arabic printed text without the need for character segmentation. The model was tested on a custom dataset comprised of over two million word samples that were generated using 18 different Arabic font types. The objective of the testing process was to assess the model capability in recognizing a diverse set of Arabic fonts representing a varied cursive styles. The model achieved good results in recognizing characters and words and it also achieved promising results in recognizing characters when it was tested on unseen data. The prepared model, the custom datasets and the toolkit for generating similar datasets are made publicly available, these tools can be used to prepare models for recognizing other font types as well as to further extend and enhance the performance of the proposed model. | ['Bassam Hammo', 'Mohammad Fasha', 'Jabir Widian', 'Nadim Obeid'] | 2020-09-04 | null | null | null | null | ['irregular-text-recognition'] | ['computer-vision'] | [ 6.01388663e-02 -5.70205629e-01 4.54363614e-01 -2.88201869e-01
-2.43029803e-01 -1.14499557e+00 8.15747917e-01 -3.01032513e-01
-3.60597998e-01 6.26318276e-01 -7.70428553e-02 -2.01129630e-01
2.62483299e-01 -6.33593559e-01 -3.19956869e-01 -7.25705206e-01
1.96653441e-01 9.14278805e-01 2.35749885e-01 -5.12383461e-01
6.74509346e-01 9.73578990e-01 -1.40302896e+00 6.41147017e-01
8.51212740e-01 5.22660613e-01 2.91700125e-01 1.12787163e+00
-2.67757028e-01 7.13947594e-01 -1.01977456e+00 -2.79816478e-01
2.51106024e-01 -4.62391019e-01 -4.90882397e-01 5.31701207e-01
3.52309912e-01 -4.05088753e-01 -2.17025191e-01 5.78946829e-01
7.42503941e-01 5.17785251e-02 1.00240564e+00 -7.58071840e-01
-1.15776527e+00 6.12773657e-01 -3.27093840e-01 -4.05967748e-03
3.72041434e-01 5.53555414e-02 4.38602567e-01 -1.00958252e+00
5.74532747e-01 1.28016603e+00 5.45178533e-01 4.66784894e-01
-6.17130220e-01 -2.28867263e-01 -2.50190377e-01 4.37087640e-02
-1.01930320e+00 -6.73916563e-02 4.67187732e-01 -3.46831381e-01
1.14078224e+00 2.16118529e-01 2.62008697e-01 1.12778819e+00
1.24799304e-01 9.37501729e-01 1.44187629e+00 -1.04663348e+00
1.52344478e-03 3.01377952e-01 5.56417823e-01 9.00172412e-01
-8.65957513e-02 -3.61012101e-01 -2.17513144e-01 1.56451911e-01
3.97636235e-01 -4.97636974e-01 1.11507960e-02 2.70784259e-01
-1.07654643e+00 8.15217018e-01 -8.92578885e-02 5.42559803e-01
-1.00244112e-01 -4.26187187e-01 5.06368756e-01 4.13652867e-01
-6.02424070e-02 2.78544188e-01 -4.00322765e-01 -2.92856872e-01
-8.33294868e-01 5.58342263e-02 1.10625803e+00 8.74852896e-01
8.13628286e-02 5.32251477e-01 1.33662269e-01 1.28820598e+00
4.49580342e-01 6.23063266e-01 1.12115180e+00 -1.12955630e-01
5.89616358e-01 6.70025706e-01 3.56398784e-02 -9.82850194e-01
-3.68813664e-01 2.83476770e-01 -5.22441328e-01 3.71646196e-01
5.88406622e-01 -4.59922820e-01 -1.56035924e+00 8.97755563e-01
-1.74185976e-01 -5.95087349e-01 3.20019871e-01 7.50665009e-01
8.65721047e-01 1.14719129e+00 -2.91786373e-01 1.90965965e-01
1.14622843e+00 -1.14578700e+00 -7.56514132e-01 -1.83122694e-01
4.37916726e-01 -1.33595932e+00 1.28233826e+00 8.28987479e-01
-7.49043465e-01 -5.82967043e-01 -1.34407115e+00 4.86205705e-02
-8.52839828e-01 8.38716328e-01 2.28760213e-01 1.14452481e+00
-8.28942060e-01 4.26748767e-02 -5.90731919e-01 -6.63830996e-01
1.63310245e-01 3.58906299e-01 -3.00518543e-01 -1.61617622e-01
-9.92896318e-01 1.28977501e+00 7.00429857e-01 3.73948485e-01
-8.07939947e-01 3.92726988e-01 -6.30533457e-01 -3.34916234e-01
-1.56681851e-01 5.08044422e-01 7.88574755e-01 -1.32152414e+00
-1.69931984e+00 8.15159559e-01 2.56832063e-01 -3.30082059e-01
9.04339373e-01 -1.68485701e-01 -9.16407466e-01 -3.91478911e-02
-3.59438092e-01 5.63112378e-01 9.47555125e-01 -1.12790060e+00
-3.32303584e-01 -2.12693557e-01 -5.77242851e-01 1.88099727e-01
-4.46977466e-01 2.18390524e-01 -3.00676167e-01 -9.61003125e-01
-1.20055169e-01 -1.11861539e+00 4.60988700e-01 -4.16928947e-01
-2.46201009e-01 -4.90483493e-02 1.24588442e+00 -1.14664423e+00
8.94562185e-01 -1.73020983e+00 2.76856963e-02 4.03815269e-01
-5.60442567e-01 9.26524341e-01 -5.27354002e-01 8.22141409e-01
2.15457752e-01 -1.08087650e-02 -3.42528552e-01 3.66140246e-01
3.28288525e-02 2.25801930e-01 -2.48170912e-01 2.45733231e-01
5.98415196e-01 6.24674797e-01 -3.22688043e-01 -1.57382622e-01
1.85735270e-01 3.68667156e-01 3.45469177e-01 6.86376393e-02
-6.19003832e-01 -1.39550105e-01 -8.88545513e-02 9.48066473e-01
7.37950146e-01 2.78716326e-01 1.85019717e-01 -6.42275214e-02
-4.85012420e-02 -4.24906939e-01 -1.24960053e+00 9.50299919e-01
-1.19957902e-01 1.09299850e+00 -3.79826456e-01 -6.69014454e-01
1.56416726e+00 1.71674386e-01 -3.75848532e-01 -5.84018469e-01
4.21779782e-01 5.64552963e-01 3.72962117e-01 -7.40556002e-01
6.45853579e-01 2.91773021e-01 7.11109191e-02 7.80863404e-01
-8.33684132e-02 -2.49109462e-01 7.89469779e-01 -1.16132952e-01
5.70998967e-01 3.93865198e-01 -9.82687026e-02 -1.34446383e-01
1.00801408e+00 2.61738271e-01 -2.25295305e-01 5.83112717e-01
-1.01238787e-02 6.17190540e-01 3.45447183e-01 -6.44482195e-01
-1.21272159e+00 -7.66491890e-01 -2.43380144e-01 1.17815697e+00
-4.09664810e-01 1.65841267e-01 -8.05028260e-01 -6.49262249e-01
-1.79773718e-01 8.38402152e-01 -4.73253250e-01 5.08994423e-02
-7.81205237e-01 -1.28794312e+00 1.14231241e+00 4.75071013e-01
8.75111282e-01 -1.63509560e+00 -4.10163790e-01 2.21645549e-01
6.72805309e-02 -9.00720596e-01 -2.12043002e-01 2.44674787e-01
-4.67831552e-01 -1.05181539e+00 -1.03777993e+00 -1.52076876e+00
9.16264296e-01 -2.88900405e-01 7.42903590e-01 -1.14768215e-01
-5.57216167e-01 2.68727064e-01 -7.01518238e-01 -7.36815095e-01
-1.06515646e+00 7.09934011e-02 -1.07322857e-01 2.74778134e-03
6.01569533e-01 2.89963782e-01 2.35995933e-01 2.57673889e-01
-1.20259464e+00 -2.84933209e-01 5.35891712e-01 9.68835652e-01
-1.99009236e-02 2.81634927e-02 7.12159812e-01 -7.68486857e-01
1.26801014e+00 -2.16831043e-01 -6.71807110e-01 6.42973959e-01
-3.35228920e-01 1.14947788e-01 8.12620223e-01 -7.10295141e-01
-1.10377717e+00 5.61289713e-02 -2.01976374e-01 5.05596220e-01
-3.28017980e-01 5.21794856e-01 -6.88939691e-02 -5.90510741e-02
7.68524945e-01 6.42353892e-01 2.25054339e-01 -4.47806001e-01
2.83477455e-01 1.39638615e+00 2.73779243e-01 -5.76423466e-01
4.07266885e-01 -8.11575353e-02 -3.23556751e-01 -1.18759489e+00
-1.98461354e-01 1.97572798e-01 -1.07491076e+00 -2.70743012e-01
7.27258682e-01 -5.05478144e-01 -3.39637548e-01 1.50208127e+00
-9.77270246e-01 -4.57431704e-01 4.82246637e-01 5.12545817e-02
-6.59370497e-02 6.54319108e-01 -8.65689874e-01 -7.84322679e-01
-4.25698668e-01 -1.22134101e+00 6.58475518e-01 4.39800858e-01
-1.11003943e-01 -1.18109000e+00 2.08399780e-02 2.95836926e-01
3.15238118e-01 2.80587047e-01 1.32048738e+00 -1.01249063e+00
-1.67270213e-01 -5.77484190e-01 -1.36311010e-01 6.96428835e-01
5.11103570e-01 5.65432012e-01 -7.64052629e-01 -2.53394723e-01
-3.45121115e-01 -9.13593888e-01 5.73818386e-01 -3.59187305e-01
6.46489441e-01 -1.03764445e-01 3.24989051e-01 1.16151087e-01
1.37204313e+00 7.66220808e-01 7.30656326e-01 7.00510502e-01
7.05014825e-01 4.56161380e-01 4.78478432e-01 2.90756226e-01
-1.85286045e-01 3.16297263e-01 8.92248601e-02 1.99523866e-01
-4.93366197e-02 3.01643133e-01 6.37302876e-01 1.05112410e+00
1.13540001e-01 -6.86584830e-01 -1.31193471e+00 3.42609942e-01
-1.38485980e+00 -7.89531708e-01 -3.62969905e-01 1.78269684e+00
1.05247915e+00 -3.22024487e-02 2.66992092e-01 4.04145211e-01
7.05821693e-01 -1.51595354e-01 -5.27854264e-01 -1.36035740e+00
-8.20925176e-01 2.84948081e-01 2.42887259e-01 5.44891417e-01
-1.14869332e+00 1.35402274e+00 6.77212954e+00 5.62429965e-01
-1.38094211e+00 -5.81006885e-01 6.28562152e-01 3.31752449e-01
2.74007618e-01 -5.74231803e-01 -1.10440171e+00 5.94067037e-01
1.05752671e+00 1.11993723e-01 4.80896890e-01 3.59877884e-01
2.80604828e-02 -1.75432399e-01 -7.47902334e-01 5.51046729e-01
8.33841383e-01 -1.15744829e+00 5.66035748e-01 -2.45345876e-01
7.69553065e-01 -7.32333362e-02 -4.11645733e-02 1.34991109e-01
5.02116203e-01 -1.35583949e+00 6.52217805e-01 6.28949940e-01
6.28272533e-01 -7.48651385e-01 8.89928162e-01 4.13124263e-01
-4.52495724e-01 -7.48330653e-02 -5.21741569e-01 1.18479319e-01
-2.57930130e-01 -2.14530095e-01 -1.25152564e+00 1.00124441e-01
3.11711431e-01 5.68745911e-01 -1.20112884e+00 7.68702924e-01
2.98361741e-02 6.88033998e-01 -3.77190828e-01 -6.65777683e-01
6.64998174e-01 -4.97508645e-01 3.41969728e-02 1.75721645e+00
3.60825598e-01 -4.56039101e-01 -3.39423940e-02 3.56275439e-01
1.25604570e-01 3.71060491e-01 -3.64210635e-01 -6.38296068e-01
2.62284845e-01 1.19500923e+00 -1.03452432e+00 -1.73956156e-01
-2.41379350e-01 1.21955740e+00 1.32012203e-01 2.95214385e-01
-7.58710325e-01 -8.02759707e-01 -1.91690475e-01 -5.23365617e-01
4.33317393e-01 -4.43315506e-01 -3.01145673e-01 -7.74575710e-01
3.83883603e-02 -1.64445853e+00 2.59356946e-01 -1.04811585e+00
-1.35840559e+00 1.02111638e+00 -3.74621928e-01 -1.06801355e+00
-3.37177932e-01 -1.54641306e+00 -3.96466881e-01 1.20056200e+00
-8.37027490e-01 -1.48292482e+00 -5.09295106e-01 4.39587116e-01
8.77987385e-01 -1.33449876e+00 1.00697112e+00 1.83448624e-02
-5.79657912e-01 6.59347594e-01 6.94611013e-01 6.56092763e-01
9.00699556e-01 -1.26343417e+00 1.78526640e-01 9.98622835e-01
4.23532933e-01 3.50852400e-01 3.53904456e-01 -6.76344931e-01
-1.30604756e+00 -9.78830278e-01 5.60685515e-01 -5.25976837e-01
6.00709319e-01 -4.02953446e-01 -8.84378374e-01 6.36840165e-01
5.02783000e-01 -5.44056296e-01 7.58985877e-01 -7.48153210e-01
-2.67671406e-01 1.76209331e-01 -1.18997312e+00 5.75660408e-01
1.64587274e-02 -6.36269376e-02 -7.80113339e-01 4.96392727e-01
-1.67999923e-01 -5.27108133e-01 -5.54728150e-01 -1.11648284e-01
6.14639997e-01 -6.41102135e-01 4.58846152e-01 -7.24861443e-01
5.10076582e-01 -3.35782528e-01 -1.62422299e-01 -1.14469278e+00
1.59468092e-02 -2.73424983e-01 -8.69783238e-02 1.48247933e+00
6.42853975e-01 -5.60261071e-01 4.94267434e-01 3.79804730e-01
9.15123373e-02 -3.10375422e-01 -3.53704393e-01 -6.13736808e-01
3.93864542e-01 1.07506409e-01 3.31370354e-01 1.01149035e+00
-4.03508127e-01 2.52326995e-01 -4.21642125e-01 -9.42659527e-02
9.53361318e-02 -1.95662126e-01 4.60707873e-01 -1.03976357e+00
-5.67624383e-02 -3.66541177e-01 -4.16736215e-01 -4.86217916e-01
2.94380952e-02 -9.87216353e-01 3.59679945e-02 -1.52852213e+00
-3.20189506e-01 -3.70395422e-01 3.35572392e-01 6.66771948e-01
9.17219296e-02 4.96262521e-01 4.52646434e-01 1.64246276e-01
5.95564395e-02 1.65865108e-01 1.00938821e+00 -3.11361253e-01
-1.79132789e-01 -8.31177905e-02 -2.13984162e-01 4.79302287e-01
1.15815675e+00 -1.12210587e-01 -3.00006181e-01 -6.98175013e-01
7.46326968e-02 -4.73865539e-01 -2.08883390e-01 -9.66072023e-01
-6.37870803e-02 1.65117860e-01 1.06216609e+00 -7.74670780e-01
6.74135536e-02 -6.40355885e-01 -2.73633301e-01 4.55357015e-01
-3.57370108e-01 6.08746409e-01 5.98992884e-01 2.24385694e-01
-1.11931317e-01 -8.38226199e-01 1.03872061e+00 -1.93163916e-01
-9.88722205e-01 -2.83933610e-01 -1.02601373e+00 -1.04604892e-01
1.14602995e+00 -5.66623628e-01 -3.31803501e-01 -1.02366321e-01
-7.59066522e-01 -1.12492569e-01 4.46538925e-01 8.99438858e-01
5.45557320e-01 -1.15630066e+00 -9.70813990e-01 4.06514943e-01
-7.01206997e-02 -5.18893063e-01 -4.88572299e-01 1.63317025e-01
-1.78214598e+00 5.38249969e-01 -9.35012043e-01 -3.22613657e-01
-1.52050757e+00 1.59210041e-01 5.15407085e-01 -1.88430527e-03
-3.66047502e-01 5.94596505e-01 -7.75853634e-01 -7.33885944e-01
4.14026856e-01 1.24467060e-01 -5.89560270e-01 2.07146123e-01
5.64530730e-01 5.25744259e-01 3.18638086e-01 -8.26257825e-01
-5.94887286e-02 5.54615676e-01 -5.75389624e-01 -3.53454113e-01
1.31545556e+00 1.90193683e-01 -3.14322293e-01 6.67846501e-01
7.75964081e-01 2.26976424e-01 -1.01362348e+00 9.72624049e-02
3.56727898e-01 -1.11955613e-01 -4.84561265e-01 -1.64482617e+00
-8.43027174e-01 8.68641376e-01 1.07779074e+00 3.70390266e-02
8.89616430e-01 -7.54753768e-01 4.63982165e-01 9.38883305e-01
4.07809764e-02 -1.55241799e+00 2.09731683e-01 1.08190656e+00
1.03899539e+00 -1.24912560e+00 -1.78485483e-01 2.08353296e-01
-1.03884697e+00 1.89921749e+00 7.95794606e-01 -2.45569885e-01
3.56518865e-01 5.67551553e-01 7.56291568e-01 1.94591194e-01
-2.73319662e-01 1.97397977e-01 2.91300684e-01 1.00616872e+00
1.03571153e+00 -1.91239640e-01 -6.40264690e-01 3.68315428e-01
-1.78128108e-01 -1.07928671e-01 1.15136611e+00 1.12289703e+00
-4.00764585e-01 -1.45150435e+00 -6.48929000e-01 5.53293645e-01
-3.80624771e-01 -1.67966366e-01 -1.10052681e+00 8.70834172e-01
-1.02201723e-01 9.64463592e-01 -9.38641727e-02 -1.15840696e-01
7.10123852e-02 4.90106314e-01 4.70179468e-01 -2.91179359e-01
-7.39456713e-01 -1.08570412e-01 1.99440405e-01 4.11170304e-01
-1.45582139e-01 -6.90097213e-01 -1.18213153e+00 -1.36024326e-01
-1.39130488e-01 -2.87171062e-02 8.01440537e-01 8.42547059e-01
-1.43345390e-02 1.56893596e-01 4.03519928e-01 -6.62243724e-01
-6.56700015e-01 -1.38045728e+00 -5.57352245e-01 3.74237895e-01
-5.13733318e-03 -4.68349308e-02 -1.03092700e-01 5.66650331e-01] | [11.852977752685547, 2.5889556407928467] |
e04e4120-af24-4dbc-8d50-418db24d3019 | ev-mgrflownet-motion-guided-recurrent-network | 2305.07853 | null | https://arxiv.org/abs/2305.07853v1 | https://arxiv.org/pdf/2305.07853v1.pdf | EV-MGRFlowNet: Motion-Guided Recurrent Network for Unsupervised Event-based Optical Flow with Hybrid Motion-Compensation Loss | Event cameras offer promising properties, such as high temporal resolution and high dynamic range. These benefits have been utilized into many machine vision tasks, especially optical flow estimation. Currently, most existing event-based works use deep learning to estimate optical flow. However, their networks have not fully exploited prior hidden states and motion flows. Additionally, their supervision strategy has not fully leveraged the geometric constraints of event data to unlock the potential of networks. In this paper, we propose EV-MGRFlowNet, an unsupervised event-based optical flow estimation pipeline with motion-guided recurrent networks using a hybrid motion-compensation loss. First, we propose a feature-enhanced recurrent encoder network (FERE-Net) which fully utilizes prior hidden states to obtain multi-level motion features. Then, we propose a flow-guided decoder network (FGD-Net) to integrate prior motion flows. Finally, we design a hybrid motion-compensation loss (HMC-Loss) to strengthen geometric constraints for the more accurate alignment of events. Experimental results show that our method outperforms the current state-of-the-art (SOTA) method on the MVSEC dataset, with an average reduction of approximately 22.71% in average endpoint error (AEE). To our knowledge, our method ranks first among unsupervised learning-based methods. | ['Zheng Fang', 'Chenming Hu', 'Delei Kong', 'Kuanxu Hou', 'XinJie Huang', 'Hao Zhuang'] | 2023-05-13 | null | null | null | null | ['event-based-optical-flow', 'motion-compensation'] | ['computer-vision', 'computer-vision'] | [-6.54754639e-02 -3.76565754e-01 -3.23513746e-01 -1.95788950e-01
-2.09191963e-01 -9.60293859e-02 4.90595430e-01 -3.67533505e-01
-4.37377572e-01 6.97574139e-01 5.44241965e-01 -6.36205077e-02
9.26957279e-02 -6.61856830e-01 -6.88921511e-01 -5.02124846e-01
1.15819194e-03 -2.97897398e-01 6.13728940e-01 1.42424792e-01
3.56970817e-01 3.88995260e-01 -1.25237560e+00 4.44294810e-02
9.14451122e-01 9.77167130e-01 2.88688451e-01 6.62949085e-01
5.63556999e-02 1.52908516e+00 -1.32020220e-01 -2.92571694e-01
3.12458366e-01 -5.56382060e-01 -6.70218945e-01 1.21800997e-03
7.78312683e-01 -1.00328147e+00 -9.73038375e-01 6.21491551e-01
5.06047487e-01 4.23558414e-01 2.37818509e-01 -1.28238177e+00
-3.07095647e-01 1.72991142e-01 -5.82119405e-01 5.40697277e-01
1.34247437e-01 4.94874179e-01 1.01526105e+00 -9.21230018e-01
9.54902887e-01 1.11467493e+00 3.51458430e-01 4.50525761e-01
-9.41955030e-01 -6.97064638e-01 1.82301626e-01 6.79330349e-01
-1.05069900e+00 -6.01714134e-01 9.60456789e-01 -5.60216904e-01
1.16294873e+00 -3.07909608e-01 8.28800738e-01 1.00511706e+00
2.27798074e-01 1.00194335e+00 5.05118966e-01 -8.27721730e-02
-1.62068021e-03 -4.67650443e-01 -3.00979108e-01 1.05081391e+00
1.03513554e-01 2.81784236e-01 -7.97444522e-01 4.25394177e-01
1.14356470e+00 1.72956839e-01 -5.52014649e-01 -2.56662041e-01
-1.44526124e+00 6.66001081e-01 6.05268180e-01 3.85983326e-02
-4.79470313e-01 3.85670930e-01 5.29351652e-01 -1.04266889e-01
3.41706127e-01 1.45784438e-01 -2.09360600e-01 -4.47056234e-01
-9.92719531e-01 -7.19844922e-02 4.44639087e-01 8.69212449e-01
9.21120644e-01 3.75000179e-01 -3.96561861e-01 5.34475029e-01
5.38234770e-01 3.17368925e-01 3.46003056e-01 -1.53347766e+00
5.97253799e-01 5.31539440e-01 1.10535383e-01 -1.28842330e+00
-3.23672056e-01 -3.75872999e-01 -8.69714200e-01 1.19845040e-01
4.31232423e-01 -2.26656765e-01 -6.04213595e-01 1.76295114e+00
1.89211637e-01 1.00796056e+00 -5.36787212e-02 1.18402994e+00
6.28890097e-01 8.24405372e-01 -1.19285487e-01 -3.67879510e-01
8.57463062e-01 -1.46151376e+00 -6.91154659e-01 -1.86127231e-01
5.28623879e-01 -8.26582074e-01 8.32415760e-01 2.31497154e-01
-1.19839096e+00 -6.73776329e-01 -1.06982565e+00 -2.67866522e-01
2.65313745e-01 1.65405452e-01 6.29024684e-01 1.69650927e-01
-9.45275009e-01 5.01562357e-01 -1.28156674e+00 -2.42888182e-01
6.14766121e-01 2.27654830e-01 -2.23652571e-01 -2.48721894e-02
-9.97769952e-01 4.90496337e-01 2.20491648e-01 3.50125611e-01
-1.07613087e+00 -7.43228078e-01 -1.19837999e+00 1.01624921e-01
4.54606414e-01 -1.05632198e+00 1.02308500e+00 -6.80333912e-01
-1.81579041e+00 3.23602021e-01 -6.17155612e-01 -6.33012652e-01
5.39934993e-01 -4.68439579e-01 -2.29736477e-01 6.17322147e-01
8.71980786e-02 9.69668508e-01 6.09647036e-01 -7.98872948e-01
-7.41398156e-01 1.70064136e-01 1.69343621e-01 1.30236214e-02
-5.52532136e-01 -8.64515677e-02 -5.76223075e-01 -7.76443601e-01
-1.67396113e-01 -7.77803838e-01 -2.32745901e-01 3.55811596e-01
-3.11784148e-01 -5.67137264e-02 9.63321209e-01 -4.90509272e-01
1.56355000e+00 -1.96604133e+00 7.92794824e-02 -2.87946522e-01
3.98590118e-01 5.48856020e-01 -1.08748309e-01 2.27056190e-01
2.29963720e-01 -1.23198748e-01 -1.89374715e-01 -6.95291460e-01
-3.38931084e-01 2.80065686e-01 -4.10977393e-01 4.68927205e-01
4.67638522e-01 9.75159287e-01 -1.09580529e+00 -8.04430187e-01
8.40897143e-01 7.54097402e-01 -9.55253959e-01 3.02201301e-01
-1.38408363e-01 7.30271339e-01 -4.00980860e-01 3.46428871e-01
5.20235240e-01 -4.82400328e-01 -1.91145733e-01 -4.72298771e-01
-4.57682699e-01 3.29237282e-01 -1.15937114e+00 2.15371156e+00
-4.07150507e-01 9.95324314e-01 -4.33595061e-01 -6.01965129e-01
7.16052473e-01 2.08895564e-01 8.67502630e-01 -5.91026366e-01
7.53190666e-02 1.87166452e-01 -1.58844776e-02 -5.46924531e-01
6.38253093e-01 2.40710884e-01 5.00113249e-01 2.38003567e-01
1.88376546e-01 5.34417450e-01 4.08270508e-01 3.71837050e-01
1.11185789e+00 5.38374782e-01 -4.68202643e-02 1.45859405e-01
8.50040972e-01 -3.66157085e-01 1.22096622e+00 3.98104340e-01
-5.81834495e-01 8.38104367e-01 3.94066870e-01 -5.14463842e-01
-9.45751965e-01 -8.68932009e-01 1.20908447e-01 5.39674640e-01
4.51588094e-01 -6.27844036e-01 -4.25832182e-01 -5.53154886e-01
-4.49578792e-01 2.29981378e-01 -2.08040610e-01 -1.35674581e-01
-1.10618246e+00 -4.52054799e-01 4.12432790e-01 6.81556642e-01
9.77831662e-01 -1.06942832e+00 -8.55213761e-01 5.59802830e-01
-5.22355676e-01 -1.66067266e+00 -8.86648774e-01 -7.14780629e-01
-9.68673170e-01 -9.71381605e-01 -6.44719064e-01 -6.82989240e-01
4.30629849e-01 3.33769709e-01 8.75102758e-01 2.66578384e-02
-2.78177112e-01 1.47054598e-01 -3.10553849e-01 1.68561220e-01
9.97560769e-02 2.00194284e-01 -1.64065525e-01 4.33162451e-01
1.37736633e-01 -7.50879526e-01 -1.28065014e+00 4.52617466e-01
-8.72264683e-01 2.43760273e-01 4.72627997e-01 7.65849233e-01
4.15866941e-01 -3.75133514e-01 3.53191257e-01 -4.20329690e-01
-7.32406601e-02 -2.66780615e-01 -5.52513003e-01 -8.01216736e-02
-4.70095336e-01 1.54480580e-02 7.72509277e-01 -2.44638190e-01
-1.29605496e+00 3.64530124e-02 -4.25438173e-02 -9.71625507e-01
1.32491499e-01 2.21785769e-01 -1.02760762e-01 2.11730879e-02
2.10921779e-01 2.31142104e-01 -1.07434385e-01 -1.74847633e-01
2.06953794e-01 2.97857106e-01 7.34742939e-01 -3.50994140e-01
4.78786469e-01 8.97369921e-01 3.03188145e-01 -6.03554189e-01
-8.83127630e-01 -5.89088857e-01 -5.49760103e-01 -4.21285331e-01
1.09281182e+00 -1.09012520e+00 -9.23639357e-01 8.11588228e-01
-1.38412547e+00 -4.24533278e-01 -4.10006456e-02 1.00729394e+00
-6.54448748e-01 6.19793177e-01 -9.37658489e-01 -6.11622453e-01
-3.04795057e-01 -1.33216023e+00 8.27390730e-01 4.34371352e-01
1.71275303e-01 -1.11367118e+00 1.01216085e-01 3.66802007e-01
4.60828394e-01 3.75350893e-01 1.54546276e-01 8.35230649e-02
-1.21819484e+00 4.06223357e-01 -5.28089106e-01 2.87804693e-01
2.52943128e-01 2.27592155e-01 -8.15988064e-01 -2.80726016e-01
-3.04775506e-01 -1.32302031e-01 1.18081987e+00 6.29880846e-01
1.02845180e+00 -1.61518708e-01 -1.72534063e-01 1.12160158e+00
1.32120073e+00 3.66525799e-02 9.52824652e-01 3.59182537e-01
1.16619551e+00 3.28248620e-01 5.49136162e-01 6.69438422e-01
6.35118544e-01 7.20711946e-01 6.12220824e-01 -4.41288613e-02
-5.21301448e-01 -4.28143144e-01 7.10035801e-01 9.04688776e-01
-2.78939009e-01 -4.24354464e-01 -6.00123882e-01 7.20536113e-01
-2.16079164e+00 -1.13900125e+00 -2.15358973e-01 1.93087876e+00
5.77866912e-01 2.46943444e-01 2.31023468e-02 -1.17975757e-01
7.53259540e-01 6.30956352e-01 -7.19651699e-01 5.65588661e-02
-1.26536518e-01 8.75265151e-02 3.90540540e-01 6.82067633e-01
-1.15679383e+00 1.13422108e+00 4.71811771e+00 5.23933113e-01
-1.23308635e+00 -2.82881614e-02 3.95240128e-01 -1.82216942e-01
-3.05273086e-02 2.74882972e-01 -8.88447821e-01 6.47924304e-01
7.50477433e-01 3.56894583e-02 7.91433528e-02 3.59119594e-01
8.25473905e-01 2.19765659e-02 -8.67206216e-01 9.48925376e-01
7.79970586e-02 -1.57517338e+00 1.37803778e-01 1.57228619e-01
8.55689943e-01 1.11804746e-01 -1.75589830e-01 -1.14952624e-01
8.79014060e-02 -5.66848218e-01 5.80536544e-01 8.03516626e-01
6.04853809e-01 -6.72254384e-01 5.40296018e-01 1.48877986e-02
-1.55327773e+00 2.31612455e-02 -3.00801069e-01 -1.55774176e-01
8.76151860e-01 7.71518409e-01 -3.75420570e-01 6.54133439e-01
5.13388872e-01 1.74534917e+00 -2.47709185e-01 1.29072988e+00
-3.09504926e-01 7.88297892e-01 -2.35841617e-01 3.90056252e-01
4.09623355e-01 -1.68453261e-01 8.45701754e-01 1.08124495e+00
2.13490695e-01 -7.24775419e-02 3.18654507e-01 8.43714893e-01
-1.44430459e-01 -9.74882692e-02 -2.94118762e-01 1.52528256e-01
3.95318240e-01 1.25114465e+00 -4.64964777e-01 -3.32787514e-01
-7.40526855e-01 1.17692447e+00 2.66110986e-01 3.38745743e-01
-9.29134607e-01 -3.87509227e-01 9.18804884e-01 9.81041864e-02
5.02640605e-01 -5.08823574e-01 1.95510760e-01 -1.65265036e+00
1.88868031e-01 -2.35165954e-01 3.17159027e-01 -7.74158895e-01
-1.09606779e+00 3.50273788e-01 -2.98891485e-01 -1.64529276e+00
-3.52727532e-01 -3.80592406e-01 -7.09734261e-01 6.38303697e-01
-2.04177737e+00 -1.22092545e+00 -5.12572527e-01 7.44466960e-01
6.41748667e-01 -1.09622560e-01 1.00969024e-01 7.31555700e-01
-1.01751339e+00 3.68769616e-01 -3.26059520e-01 5.92409372e-01
1.07660115e+00 -9.01972711e-01 2.85200655e-01 1.40609062e+00
3.30145247e-02 5.78607917e-01 2.03946024e-01 -5.08446157e-01
-1.24799728e+00 -1.43055844e+00 8.08991194e-01 -1.64093822e-01
6.39148533e-01 2.33724732e-02 -8.28784585e-01 7.72853017e-01
2.17509434e-01 6.15070581e-01 3.46546978e-01 -5.82192302e-01
-2.59520441e-01 -3.31850737e-01 -6.42359376e-01 5.42542756e-01
1.31143868e+00 -4.09219772e-01 -1.77393049e-01 -2.12021545e-01
9.17665839e-01 -4.46337938e-01 -8.65487635e-01 5.19701958e-01
5.89505434e-01 -1.33706748e+00 9.61552083e-01 -3.23701143e-01
8.92926574e-01 -6.66880131e-01 1.21050976e-01 -8.61279964e-01
-2.45239720e-01 -9.83041465e-01 -5.94493270e-01 1.30389547e+00
-1.08100094e-01 -6.07122302e-01 9.18165088e-01 3.16790849e-01
-4.45151120e-01 -7.58288443e-01 -6.98824823e-01 -6.00597620e-01
-3.22120219e-01 -4.27875549e-01 2.49528632e-01 9.87802327e-01
-4.13778692e-01 2.55244732e-01 -6.90462351e-01 6.75583333e-02
6.04251564e-01 1.58807456e-01 8.59859765e-01 -8.64114344e-01
-3.04363370e-01 -5.33039570e-01 -6.35343850e-01 -1.70475233e+00
2.03045428e-01 -6.29340470e-01 -1.92638576e-01 -1.54316854e+00
1.11087210e-01 -9.96233672e-02 -2.73237467e-01 3.04123878e-01
-3.13484281e-01 1.80270746e-01 4.15534794e-01 3.49163741e-01
-8.91817689e-01 7.99555957e-01 1.59666455e+00 2.32872322e-01
-3.70143324e-01 -3.56127471e-01 -2.04931766e-01 8.12106550e-01
6.76881969e-01 -2.37802207e-01 -4.62031662e-01 -6.93057179e-01
2.85615772e-02 2.01989681e-01 6.24562681e-01 -1.12721431e+00
6.03913069e-01 -1.85589418e-01 2.26712465e-01 -6.78725481e-01
2.10874811e-01 -5.54209709e-01 -1.24833271e-01 4.69590962e-01
-1.19182207e-01 6.54312000e-02 1.32028083e-03 6.76510036e-01
-4.44875032e-01 1.64717972e-01 6.71167254e-01 1.33966357e-01
-1.02934873e+00 8.21997225e-01 -2.22982645e-01 3.37541401e-01
8.66971791e-01 -1.98523268e-01 -4.19810444e-01 -3.34109843e-01
-2.51756638e-01 3.96507114e-01 4.16443795e-01 4.21984196e-01
8.13758790e-01 -1.30886245e+00 -7.79104412e-01 2.00111955e-01
-1.74575616e-02 2.64945835e-01 3.73295873e-01 1.08304894e+00
-7.73009777e-01 3.73401105e-01 -3.20101708e-01 -8.35785985e-01
-8.15536857e-01 1.93158403e-01 2.75571913e-01 -3.84810865e-01
-8.90470922e-01 6.66024983e-01 3.02091330e-01 1.29159585e-01
7.74313733e-02 -2.58002132e-01 -1.05481856e-01 -1.30980566e-01
5.66564500e-01 7.81403303e-01 -2.68960387e-01 -6.95547700e-01
-2.86335409e-01 7.41832733e-01 -4.29285206e-02 -5.19303381e-02
1.16187799e+00 -3.09733987e-01 -5.88169247e-02 9.54669192e-02
1.44904757e+00 -6.94189966e-02 -1.96867919e+00 -3.11454087e-01
-2.74892300e-01 -7.75411606e-01 2.27122709e-01 -2.16029435e-01
-1.67328000e+00 1.20974123e+00 3.36189091e-01 -4.20450568e-01
1.29931545e+00 -5.08765161e-01 1.44577706e+00 1.45702153e-01
1.79082721e-01 -6.46331012e-01 4.11298603e-01 5.60144544e-01
3.49851578e-01 -1.31130314e+00 -8.66490304e-02 -5.22488058e-01
-4.81151521e-01 1.14060950e+00 9.63996351e-01 -2.16381282e-01
6.04415834e-01 -2.28213295e-02 -1.54292777e-01 1.21119238e-01
-9.21751559e-01 -3.35798949e-01 2.56568700e-01 2.79721588e-01
3.55033755e-01 -4.85909164e-01 -1.68240428e-01 7.06178918e-02
1.15186721e-01 4.80426788e-01 6.16997182e-01 7.68197596e-01
-3.23562264e-01 -1.05855083e+00 1.26725107e-01 1.64481327e-01
-4.23506379e-01 -1.88909784e-01 2.13537559e-01 4.95332360e-01
6.68939203e-02 9.16385829e-01 2.70593643e-01 -3.47690403e-01
2.62464434e-01 -4.28375602e-01 3.64683509e-01 -2.80332059e-01
-2.43645281e-01 1.88996628e-01 -1.14033073e-01 -9.62561011e-01
-9.58784938e-01 -6.49203718e-01 -1.43903065e+00 -4.14686471e-01
-6.75449595e-02 -3.12081605e-01 1.59266710e-01 8.11280012e-01
7.19411910e-01 5.26707649e-01 7.85391748e-01 -8.91340971e-01
2.55614758e-01 -7.35615492e-01 -7.81805217e-02 3.47746730e-01
5.71852982e-01 -6.93317354e-01 -4.17866856e-01 3.38130474e-01] | [8.883075714111328, -1.5517407655715942] |
ebf60239-c6b0-43fa-8cd2-34a59062add7 | improving-event-temporal-relation | null | null | https://aclanthology.org/2022.ccl-1.76 | https://aclanthology.org/2022.ccl-1.76.pdf | Improving Event Temporal Relation Classification via Auxiliary Label-Aware Contrastive Learning | “Event Temporal Relation Classification (ETRC) is crucial to natural language understanding. In recent years, the mainstream ETRC methods may not take advantage of lots of semantic information contained in golden temporal relation labels, which is lost by the discrete one-hot labels. To alleviate the loss of semantic information, we propose learning Temporal semantic information of the golden labels by Auxiliary Contrastive Learning (TempACL). Different from traditional contrastive learning methods, which further train the PreTrained Language Model (PTLM) with unsupervised settings before fine-tuning on target tasks, we design a supervised contrastive learning framework and make three improvements. Firstly, we design a new data augmentation method that generates augmentation data via matching templates established by us with golden labels. Secondly, we propose patient contrastive learning and design three patient strategies. Thirdly we design a label-aware contrastive learning loss function. Extensive experimental results show that our TempACL effectively adapts contrastive learning to supervised learning tasks which remain a challenge in practice. TempACL achieves new state-of-the-art results on TB-Dense and MATRES and outperforms the baseline model with up to 5.37%F1 on TB-Dense and 1.81%F1 on MATRES.” | ['Li Lishuang', 'Sun Tiesen'] | null | null | null | null | ccl-2022-10 | ['temporal-relation-classification', 'relation-classification'] | ['natural-language-processing', 'natural-language-processing'] | [ 3.07506621e-01 1.64422631e-01 -4.10570055e-01 -6.17008030e-01
-9.30474102e-01 -1.35038584e-01 6.73213482e-01 -3.90274711e-02
-6.92955434e-01 6.50241017e-01 4.24284667e-01 -5.60800396e-02
-1.63706720e-01 -6.69712782e-01 -6.78758323e-01 -5.66364110e-01
1.36649702e-03 6.69739604e-01 2.10671529e-01 -2.49869168e-01
-3.99665952e-01 -2.72347685e-02 -1.00475061e+00 6.95182562e-01
8.86189282e-01 1.14137328e+00 2.56465804e-02 2.21252888e-01
-1.33800596e-01 1.35599959e+00 -3.76719058e-01 -3.51727694e-01
1.32178128e-01 -5.86552978e-01 -1.18117416e+00 5.46619408e-02
5.41242361e-02 -2.64132600e-02 -6.68776989e-01 7.69354284e-01
4.32598233e-01 2.62817979e-01 4.80339587e-01 -1.23455906e+00
-5.91100991e-01 7.32922018e-01 -6.05887413e-01 4.49890733e-01
1.55424988e-02 4.40678149e-02 1.06073296e+00 -9.27710593e-01
6.75464153e-01 1.32320368e+00 7.14452863e-01 6.91471219e-01
-1.18408418e+00 -7.40372837e-01 6.11589909e-01 4.48303133e-01
-1.24478078e+00 -2.28782669e-01 9.10430849e-01 -2.90093988e-01
7.73761094e-01 2.94782281e-01 4.48301375e-01 1.33067381e+00
7.36027882e-02 1.08363986e+00 1.39086473e+00 -2.71766782e-01
-1.05643243e-01 2.05070358e-02 2.66565114e-01 7.46326923e-01
-6.91535532e-01 3.65018189e-01 -7.81735122e-01 1.23475321e-01
5.10889709e-01 2.45146960e-01 -1.20792784e-01 -5.89018799e-02
-1.46053398e+00 7.45330930e-01 7.27447629e-01 4.22749698e-01
-2.44981289e-01 1.49086118e-01 6.56471252e-01 3.47792476e-01
8.94446909e-01 1.60944879e-01 -7.54029512e-01 7.62367696e-02
-6.66013181e-01 -1.79167837e-01 1.98073670e-01 9.78626370e-01
4.98464465e-01 -3.79389465e-01 -6.06391490e-01 1.06166208e+00
3.11369486e-02 1.51365757e-01 6.37838840e-01 -6.76408529e-01
5.83015501e-01 7.15429425e-01 -3.70194644e-01 -5.91277122e-01
-7.06657588e-01 -8.13870966e-01 -9.88863170e-01 -3.67751002e-01
1.33521274e-01 -2.23060995e-02 -1.24767125e+00 1.97579217e+00
1.81081325e-01 3.28255057e-01 -4.09852900e-02 6.61368370e-01
9.45656359e-01 6.43999994e-01 3.71995300e-01 -4.28801507e-01
1.35622227e+00 -1.51607156e+00 -1.13197994e+00 -4.13559705e-01
9.38270628e-01 -5.34817636e-01 1.39057481e+00 2.90588498e-01
-8.96938264e-01 -4.24942702e-01 -8.33563983e-01 -2.43495166e-01
-3.49355400e-01 1.34235471e-01 8.81095767e-01 7.84272179e-02
-7.98471510e-01 4.30282563e-01 -1.04487967e+00 -1.85070768e-01
5.88352323e-01 3.15483570e-01 -2.32786804e-01 -1.44697443e-01
-1.59800136e+00 9.19659436e-01 4.36412036e-01 1.86745390e-01
-9.28576648e-01 -9.81480539e-01 -9.66108143e-01 -2.88749188e-01
8.19244742e-01 -6.68750644e-01 1.48558056e+00 -8.85352552e-01
-1.28621256e+00 1.06825650e+00 -2.14502990e-01 -4.41117913e-01
7.32963502e-01 -4.55985248e-01 -5.45040727e-01 1.17141269e-01
3.47077340e-01 7.67540812e-01 5.97791016e-01 -1.03399265e+00
-4.70188707e-01 -1.99043825e-01 -1.18144326e-01 2.68712699e-01
-3.42624336e-01 1.54955015e-01 -5.07073522e-01 -1.14238918e+00
1.45208910e-01 -1.04894805e+00 -3.97283792e-01 5.08962423e-02
-3.76593679e-01 -4.17833298e-01 8.35992217e-01 -5.84718287e-01
1.27314162e+00 -2.08857465e+00 -3.77163664e-02 -2.51782499e-02
2.19436914e-01 2.27855563e-01 -6.39639720e-02 6.24721013e-02
-3.23193580e-01 -1.68958500e-01 -5.88047862e-01 -6.28971398e-01
-2.42457569e-01 3.93972844e-01 -2.97248036e-01 2.45808229e-01
2.85634667e-01 1.18348169e+00 -1.23579049e+00 -7.29382932e-01
7.99826831e-02 4.02449876e-01 -4.36207324e-01 2.02206805e-01
-3.90649855e-01 7.93188572e-01 -5.48460722e-01 5.53819418e-01
5.21297991e-01 -6.65747285e-01 1.05577558e-01 -3.04112315e-01
1.72023043e-01 7.25340486e-01 -5.35347164e-01 2.11635876e+00
-5.41497827e-01 2.24707380e-01 -6.12387955e-01 -1.19975984e+00
7.79440761e-01 4.51161385e-01 7.61967301e-01 -1.07245147e+00
1.22453444e-01 7.12844357e-02 -2.90895641e-01 -3.82325500e-01
3.20893675e-02 -5.27959466e-01 -2.25393131e-01 3.31615776e-01
-5.19452011e-03 1.58737615e-01 -2.07074583e-01 2.51576900e-01
1.14105272e+00 2.28682190e-01 1.35989040e-01 -6.79422170e-02
5.51297724e-01 8.97284690e-03 9.72187757e-01 2.70223945e-01
-2.53093034e-01 5.53368270e-01 4.93512511e-01 -5.67568898e-01
-5.20941734e-01 -1.04459786e+00 -1.84095219e-01 1.11195111e+00
4.89562415e-02 -5.58654666e-01 -3.56665105e-01 -1.39330435e+00
-4.11678016e-01 6.60048664e-01 -8.98584306e-01 -5.03596067e-01
-8.89902711e-01 -1.06339109e+00 4.10021544e-01 8.21259856e-01
9.08838391e-01 -1.21887100e+00 -8.71913135e-02 2.54046679e-01
-5.53986430e-01 -1.41988814e+00 -8.71263742e-01 3.31864923e-01
-8.70754659e-01 -8.78161788e-01 -6.11644208e-01 -9.54779983e-01
5.20911932e-01 3.66378129e-02 1.27342057e+00 -2.13927731e-01
-6.16765767e-02 5.82663342e-02 -4.78444993e-01 -1.66243389e-01
-1.15096875e-01 1.45735070e-01 -1.71315894e-01 7.42552802e-02
3.34152222e-01 -6.09363019e-01 -6.46544456e-01 3.96907389e-01
-6.07202291e-01 3.77033710e-01 5.91443598e-01 1.02900195e+00
9.41331148e-01 9.93116125e-02 7.00522184e-01 -1.32928061e+00
2.06327319e-01 -5.52284181e-01 -2.52248257e-01 3.62846941e-01
-8.29541504e-01 2.44921103e-01 4.81228471e-01 -6.14096224e-01
-1.26137137e+00 -7.91895855e-03 -8.64470527e-02 -5.18170416e-01
1.67991146e-01 4.97411638e-01 -1.04486495e-01 4.17899817e-01
4.16681588e-01 9.57872197e-02 -2.08198115e-01 -6.56008065e-01
4.94852602e-01 2.15227142e-01 7.36068726e-01 -6.58353746e-01
4.99284387e-01 5.66167235e-01 -5.04993647e-02 -1.12930208e-01
-1.67472923e+00 -4.64348525e-01 -6.10596895e-01 5.09503074e-02
1.09399092e+00 -9.79944408e-01 -5.16184330e-01 6.04785144e-01
-1.01126289e+00 -6.87908649e-01 -5.43063402e-01 5.71346760e-01
-5.40182829e-01 3.81413884e-02 -8.79448116e-01 -2.81769186e-01
-4.05282646e-01 -8.52367461e-01 1.01828921e+00 -2.23653257e-01
-3.86986844e-02 -1.08088875e+00 1.41598359e-01 5.46270609e-01
1.70306429e-01 2.34298885e-01 8.73166144e-01 -6.07787967e-01
-3.91481787e-01 1.84815466e-01 -3.15470278e-01 4.71811831e-01
3.05761904e-01 -7.92342961e-01 -9.81738508e-01 -2.09025759e-02
2.31672421e-01 -4.03938711e-01 1.15357137e+00 2.99862444e-01
1.47003388e+00 -2.70811737e-01 -6.24220788e-01 6.53489649e-01
1.01122415e+00 3.42225283e-01 5.36172807e-01 2.28506222e-01
1.02835572e+00 6.54082477e-01 9.30296481e-01 1.30668879e-01
6.51703715e-01 8.68232191e-01 1.05313890e-01 -3.03814441e-01
-4.47670013e-01 -4.20698702e-01 2.14004740e-01 9.89340425e-01
2.50528187e-01 -3.47740613e-02 -1.03725755e+00 6.41213596e-01
-2.15500784e+00 -5.80957413e-01 6.67533278e-02 1.92655766e+00
1.33444047e+00 3.17307860e-01 2.53108796e-02 7.08650500e-02
6.67020857e-01 2.06521332e-01 -5.42420149e-01 1.18303791e-01
-1.05531767e-01 4.45295811e-01 2.89595187e-01 3.12225938e-01
-1.51858544e+00 1.21983957e+00 5.65901279e+00 8.96033466e-01
-1.12565923e+00 6.05879426e-01 8.89616311e-01 -4.21163253e-02
-1.72638759e-01 -8.31214990e-03 -5.72252333e-01 4.73277271e-01
9.96524990e-01 1.34160548e-01 3.50736529e-02 4.73584086e-01
2.59926438e-01 2.38192022e-01 -1.40979683e+00 1.09856677e+00
-9.63006690e-02 -1.27436113e+00 -1.15191281e-01 -1.34204388e-01
8.28584433e-01 -7.03274235e-02 1.41313881e-01 6.24875426e-01
3.11482131e-01 -1.06864560e+00 6.46113396e-01 5.64328134e-01
7.97954321e-01 -6.08313680e-01 7.32070386e-01 2.31408358e-01
-1.35416520e+00 1.25978719e-02 1.60137117e-01 9.04926285e-02
3.52681786e-01 6.84119165e-01 -6.05822384e-01 6.42036021e-01
7.34537721e-01 1.18436837e+00 -3.96314442e-01 7.12852895e-01
-5.76477349e-01 7.53294230e-01 -1.09469898e-01 3.93220007e-01
3.61447871e-01 2.04072535e-01 3.34595531e-01 1.10621846e+00
-1.29513606e-01 2.43541971e-01 4.85934168e-01 7.63967931e-01
-3.25764686e-01 6.64623529e-02 -2.37682685e-01 1.60490908e-02
3.62290949e-01 9.36904728e-01 -6.13577485e-01 -4.08687234e-01
-4.65825796e-01 1.18233073e+00 4.12703604e-01 1.70287013e-01
-1.02312756e+00 1.16869889e-01 1.78909883e-01 1.51434049e-01
-5.82532175e-02 -2.76812948e-02 -3.05287719e-01 -1.14114940e+00
9.06624272e-02 -7.32261360e-01 9.97194052e-01 -5.27855456e-01
-1.51605952e+00 7.09821284e-01 1.07937366e-01 -1.26848459e+00
-6.26365319e-02 -3.38743836e-01 -3.27054501e-01 6.16564214e-01
-1.55334544e+00 -1.56130338e+00 -1.50222078e-01 7.31471837e-01
7.97669768e-01 -1.88440718e-02 6.78840101e-01 6.92041874e-01
-5.86499155e-01 8.67153466e-01 -3.61294836e-01 2.47875616e-01
9.54358518e-01 -1.28507257e+00 3.30649287e-01 7.57405043e-01
3.19923684e-02 3.13917100e-01 2.31950700e-01 -6.26431406e-01
-6.86646998e-01 -1.54765308e+00 1.17633116e+00 -3.30347568e-01
7.11617291e-01 -4.75298464e-01 -9.53589797e-01 1.07602954e+00
1.15206264e-01 4.69075501e-01 6.07202053e-01 2.58534998e-01
-6.30451918e-01 -3.25913310e-01 -1.01211011e+00 4.59471822e-01
1.46546745e+00 -8.15861166e-01 -7.23033965e-01 7.17454791e-01
1.41994202e+00 -3.38087291e-01 -8.75266373e-01 8.74617994e-01
2.47437246e-02 -3.34847957e-01 9.95753527e-01 -7.16353536e-01
3.17068636e-01 -2.00806096e-01 5.11898287e-02 -1.09067833e+00
-1.18447609e-01 -7.49209166e-01 -1.58604309e-01 1.24224734e+00
5.34510493e-01 -6.76468551e-01 6.47065103e-01 4.43223149e-01
-4.67801064e-01 -9.04204488e-01 -8.40179145e-01 -9.22512829e-01
1.66259930e-01 -5.01171529e-01 3.04868311e-01 1.38965976e+00
-8.61748978e-02 7.32328653e-01 -5.99057376e-01 -8.95526111e-02
5.17230093e-01 8.17038398e-03 1.19159512e-01 -1.06558156e+00
-3.42084229e-01 -1.71532869e-01 6.64951801e-02 -1.13250649e+00
3.66002619e-01 -1.21815264e+00 6.83475211e-02 -1.44798756e+00
4.55878228e-01 -9.14126813e-01 -7.42249489e-01 1.03247118e+00
-3.20619553e-01 1.39487147e-01 -1.04854822e-01 2.20665410e-01
-7.91518807e-01 9.86609101e-01 1.44258773e+00 -2.89839000e-01
-2.44027957e-01 -1.94272436e-02 -5.48752964e-01 7.34283447e-01
5.94588995e-01 -8.19887280e-01 -7.75891721e-01 -3.64407897e-01
1.24301985e-01 7.55463867e-03 3.74104202e-01 -6.74723148e-01
2.26119384e-01 -9.08164680e-02 2.01598415e-03 -8.34832549e-01
3.09560925e-01 -7.24804878e-01 -1.02487884e-01 4.58356470e-01
-7.18114972e-01 8.09336640e-03 1.10557452e-01 5.62262833e-01
-3.05623055e-01 2.48627096e-01 7.97700584e-01 -9.91822779e-02
-8.38974297e-01 6.05791867e-01 1.91933457e-02 4.26760137e-01
8.60705256e-01 3.85991454e-01 -4.31826800e-01 -1.14586398e-01
-1.11758208e+00 5.03633440e-01 -2.61888444e-01 5.28192699e-01
4.49973762e-01 -1.51644289e+00 -6.88575685e-01 -1.70143783e-01
1.92935452e-01 3.23217034e-01 3.13106745e-01 1.14438331e+00
-4.39065658e-02 3.75472784e-01 2.80849785e-01 -5.78669310e-01
-1.02625096e+00 9.53286350e-01 4.05327410e-01 -8.02060306e-01
-9.95589137e-01 1.10488331e+00 6.49772346e-01 -5.00116408e-01
2.22582743e-01 -3.48937333e-01 -1.39234975e-01 3.31625715e-02
3.14962566e-01 1.55130014e-01 2.13370949e-01 -5.05287230e-01
-6.09521985e-01 3.90886635e-01 -5.70023417e-01 -1.84147865e-01
1.28659868e+00 -7.80622438e-02 1.51693001e-02 5.46254277e-01
1.34424233e+00 -2.41625890e-01 -1.13719761e+00 -9.32996869e-01
3.22136462e-01 -1.20241538e-01 1.67043149e-01 -1.16926599e+00
-1.26751852e+00 7.70448387e-01 6.36747718e-01 -3.22441608e-01
1.45257437e+00 4.29907471e-01 1.13739967e+00 2.36039117e-01
2.06984207e-01 -9.79418039e-01 6.02162540e-01 5.34367859e-01
9.39556003e-01 -1.23794484e+00 -2.20812231e-01 -7.99054980e-01
-7.48159766e-01 4.99886483e-01 6.53191090e-01 7.34190494e-02
8.87754261e-01 2.53442943e-01 2.09218964e-01 -4.06015873e-01
-8.86722982e-01 -3.33536744e-01 5.75842440e-01 2.47220650e-01
4.85179663e-01 3.97067554e-02 -4.89761889e-01 8.57111216e-01
-1.62621424e-01 -1.92175861e-02 -6.56480268e-02 9.46962416e-01
1.42966345e-01 -1.37166548e+00 2.47905150e-01 3.21881115e-01
-5.62791348e-01 -3.74806494e-01 -1.73239201e-01 8.08351576e-01
3.17820907e-01 8.69804919e-01 1.40083097e-02 -4.80151951e-01
4.27566200e-01 1.25513017e-01 2.98789680e-01 -7.43810415e-01
-6.16605818e-01 2.38149136e-01 3.24894518e-01 -5.47458112e-01
-6.60282731e-01 -6.32853329e-01 -1.61444330e+00 1.09859720e-01
-1.28100127e-01 3.29089314e-02 3.28853615e-02 1.25341702e+00
8.00531209e-02 1.05965233e+00 6.42724752e-01 -1.41179234e-01
-2.94502228e-01 -9.29315925e-01 -2.17218220e-01 5.07937610e-01
2.98313230e-01 -8.21802676e-01 -1.74909055e-01 1.09208994e-01] | [9.556543350219727, 3.2574658393859863] |
cb544604-0a15-4f46-b1cd-5a6d2ad3b21c | an-efficient-deep-learning-approach-using | 2109.02629 | null | https://arxiv.org/abs/2109.02629v1 | https://arxiv.org/pdf/2109.02629v1.pdf | An Efficient Deep Learning Approach Using Improved Generative Adversarial Networks for Incomplete Information Completion of Self-driving | Autonomous driving is the key technology of intelligent logistics in Industrial Internet of Things (IIoT). In autonomous driving, the appearance of incomplete point clouds losing geometric and semantic information is inevitable owing to limitations of occlusion, sensor resolution, and viewing angle when the Light Detection And Ranging (LiDAR) is applied. The emergence of incomplete point clouds, especially incomplete vehicle point clouds, would lead to the reduction of the accuracy of autonomous driving vehicles in object detection, traffic alert, and collision avoidance. Existing point cloud completion networks, such as Point Fractal Network (PF-Net), focus on the accuracy of point cloud completion, without considering the efficiency of inference process, which makes it difficult for them to be deployed for vehicle point cloud repair in autonomous driving. To address the above problem, in this paper, we propose an efficient deep learning approach to repair incomplete vehicle point cloud accurately and efficiently in autonomous driving. In the proposed method, an efficient downsampling algorithm combining incremental sampling and one-time sampling is presented to improves the inference speed of the PF-Net based on Generative Adversarial Network (GAN). To evaluate the performance of the proposed method, a real dataset is used, and an autonomous driving scene is created, where three incomplete vehicle point clouds with 5 different sizes are set for three autonomous driving situations. The improved PF-Net can achieve the speedups of over 19x with almost the same accuracy when compared to the original PF-Net. Experimental results demonstrate that the improved PF-Net can be applied to efficiently complete vehicle point clouds in autonomous driving. | ['Francesco Piccialli', 'Gang Mei', 'Jingzhi Tu'] | 2021-09-01 | null | null | null | null | ['point-cloud-completion'] | ['computer-vision'] | [-5.49162775e-02 -1.15937874e-01 2.50839442e-01 -1.93597317e-01
-3.67770493e-01 -3.32707852e-01 4.04632717e-01 -2.87823230e-01
-9.02285054e-02 6.62287772e-01 -6.06290102e-01 -5.48161685e-01
-7.27043301e-02 -1.11813664e+00 -1.09963000e+00 -8.00776124e-01
2.93440670e-01 6.00606024e-01 3.62106949e-01 -2.02284366e-01
1.77665964e-01 8.11440170e-01 -1.68129528e+00 -1.73976019e-01
1.24191594e+00 1.16816711e+00 7.32282341e-01 2.47911483e-01
-4.71668005e-01 3.56752902e-01 -5.77561915e-01 -3.57072383e-01
5.07536650e-01 3.77223223e-01 -2.18253527e-02 7.32212737e-02
1.29590228e-01 -6.78249061e-01 -6.73101902e-01 1.27404857e+00
2.57092088e-01 -1.07824072e-01 3.61715376e-01 -1.77002430e+00
-5.16235352e-01 1.77637324e-01 -4.42495912e-01 -1.26248226e-01
-3.81566346e-01 4.93908912e-01 5.23939580e-02 -9.70712245e-01
2.71001667e-01 1.41683936e+00 7.35763788e-01 3.08483541e-01
-5.88617444e-01 -1.19924796e+00 -5.63122407e-02 5.59561789e-01
-1.55936146e+00 -1.34964228e-01 9.62686419e-01 -2.48511851e-01
4.66415167e-01 -1.89184304e-02 6.24964774e-01 6.50667727e-01
6.16855562e-01 4.67350841e-01 7.40231037e-01 2.09191278e-01
2.53372610e-01 7.85022378e-02 -1.07411548e-01 4.06925529e-01
6.02389812e-01 5.07869840e-01 2.10582674e-01 1.36815965e-01
7.85577536e-01 7.09563792e-01 2.22595736e-01 -2.93732584e-01
-1.02146792e+00 7.34115481e-01 8.11862290e-01 -1.42825142e-01
-4.77198362e-01 3.28640997e-01 3.80283356e-01 8.98444653e-02
3.05276394e-01 -2.65180767e-01 -2.92788386e-01 3.51532511e-02
-5.08640885e-01 4.41514343e-01 3.86380345e-01 1.61008036e+00
1.00382888e+00 4.76034671e-01 2.25039244e-01 3.34244698e-01
2.57398009e-01 1.03536963e+00 2.12091446e-01 -1.03255332e+00
5.61336637e-01 6.81833863e-01 2.07781389e-01 -1.14340460e+00
-1.84313983e-01 -2.71003753e-01 -1.15735650e+00 4.86238331e-01
-2.20364690e-01 -4.54206616e-02 -1.04302621e+00 1.21703768e+00
4.65627402e-01 5.93517840e-01 2.73295671e-01 7.89171636e-01
8.73620450e-01 9.21391308e-01 3.69714573e-02 -9.73326117e-02
1.16340744e+00 -6.97891831e-01 -7.45435596e-01 -1.00302041e-01
3.74365270e-01 -7.11702287e-01 8.13162744e-01 1.77949637e-01
-7.18204916e-01 -9.98651564e-01 -1.26170385e+00 -3.66836637e-02
-4.16258484e-01 -2.17615202e-01 5.33508718e-01 2.63685852e-01
-6.59360111e-01 4.37391460e-01 -6.89709008e-01 1.17585417e-02
7.50712991e-01 4.16626662e-01 -8.59794244e-02 -6.30298376e-01
-1.14542818e+00 7.38208354e-01 3.18057030e-01 2.82257020e-01
-1.06668437e+00 -9.03951168e-01 -5.32707870e-01 1.71658620e-02
2.96971351e-01 -8.38704765e-01 1.00176907e+00 -4.89814341e-01
-1.19267046e+00 1.02853097e-01 -1.15235247e-01 -6.34223580e-01
4.91206527e-01 -9.44654867e-02 -6.11172199e-01 -6.05314001e-02
2.34250724e-01 8.43705952e-01 8.88277292e-01 -1.50607371e+00
-9.01519597e-01 -6.73631549e-01 -4.38732132e-02 9.25584510e-03
3.06762695e-01 -5.52320659e-01 -1.23684637e-01 -1.22470796e-01
1.34178773e-01 -1.05938530e+00 -2.86319464e-01 6.89370632e-02
-2.57484615e-01 -5.23887515e-01 1.72334981e+00 -4.95304704e-01
2.76583999e-01 -2.37488627e+00 -4.66383308e-01 -1.29660502e-01
2.48903841e-01 3.36908430e-01 -2.19518319e-02 6.65887073e-02
2.20453098e-01 -3.65818329e-02 -2.20207065e-01 -1.20298281e-01
-2.97994036e-02 6.52408540e-01 -6.44942582e-01 3.45350415e-01
2.29813218e-01 1.09910774e+00 -7.39014268e-01 -5.00061750e-01
7.82928050e-01 6.51946425e-01 -1.51179820e-01 -7.51599222e-02
-2.78516322e-01 5.08597493e-01 -6.87470794e-01 7.44045794e-01
1.55862904e+00 1.99536026e-01 -5.23437142e-01 -2.19878510e-01
-2.30826467e-01 -1.76126286e-01 -9.83882010e-01 1.37924898e+00
-5.68988264e-01 3.86190981e-01 6.20252006e-02 -6.93796873e-01
1.22480881e+00 1.28552988e-01 6.36166275e-01 -8.93371999e-01
6.02144673e-02 2.87805438e-01 -2.15612620e-01 -4.75070387e-01
4.64354932e-01 -1.86904758e-01 -6.48984388e-02 -2.52501547e-01
-4.12060469e-01 -5.51294804e-01 -1.74229756e-01 -4.57093157e-02
9.40609038e-01 2.77179498e-02 -4.50567991e-01 2.23554187e-02
6.51023984e-01 3.05057257e-01 7.32386410e-01 3.13228697e-01
-2.08404467e-01 2.21607178e-01 -1.56749561e-01 -8.54231656e-01
-1.36312795e+00 -1.26313782e+00 -3.24474901e-01 9.10233706e-02
8.83328676e-01 2.35694811e-01 -5.37059903e-01 -4.59696293e-01
4.35357392e-01 9.97753859e-01 9.60029885e-02 -4.01633263e-01
-7.66047418e-01 -2.86049873e-01 3.09546381e-01 4.65609401e-01
9.85397398e-01 -1.09540761e+00 -3.19594532e-01 2.89922833e-01
-1.76556483e-02 -1.40373445e+00 8.66777748e-02 -3.24213266e-01
-1.01953101e+00 -9.46592450e-01 -1.92748919e-01 -7.04938591e-01
6.91411197e-01 8.74055028e-01 8.12433541e-01 1.50430843e-01
2.62289625e-02 -1.57578453e-01 -1.19048201e-01 -1.03331721e+00
-4.61037576e-01 -8.22007060e-02 2.01165184e-01 -2.58327484e-01
7.17018902e-01 -6.55478895e-01 -6.54868364e-01 4.70861852e-01
-9.73532736e-01 -8.88343006e-02 8.01320314e-01 5.51817298e-01
7.03795969e-01 6.06387198e-01 7.11342990e-01 -4.41768199e-01
1.95094824e-01 -5.40460348e-01 -1.04798782e+00 -4.40601915e-01
-7.54852712e-01 -4.93285745e-01 9.73459303e-01 -1.26354158e-01
-7.11471200e-01 1.21138923e-01 -2.44910747e-01 -1.08434689e+00
-1.08565085e-01 -1.83801875e-02 -3.95639122e-01 -3.37515086e-01
4.54049334e-02 4.05884594e-01 1.53547868e-01 -2.24594638e-01
3.76739532e-01 7.82718599e-01 7.39349544e-01 -9.53862444e-02
1.31284583e+00 6.82637691e-01 4.76229638e-01 -7.32507348e-01
-2.15724558e-01 -2.40736976e-01 -3.63771915e-01 -2.82697976e-01
6.74568295e-01 -1.05611789e+00 -1.08947027e+00 4.01169658e-01
-1.51197195e+00 2.84923345e-01 -4.28679109e-01 5.70968032e-01
-6.20874822e-01 3.32401007e-01 -2.13901490e-01 -5.95086515e-01
-2.71710843e-01 -1.50626612e+00 1.17857635e+00 1.59339920e-01
6.68359518e-01 -4.88711387e-01 -4.80748028e-01 4.08644319e-01
3.31271291e-01 2.14874879e-01 8.52968872e-01 -1.84384748e-01
-1.42610502e+00 -3.59188795e-01 -4.65841323e-01 4.02405322e-01
5.01079932e-02 -1.13210991e-01 -7.96042085e-01 -3.12696964e-01
3.00342709e-01 1.92949161e-01 5.44684887e-01 3.33140433e-01
1.51797378e+00 -2.23261699e-01 -6.35508478e-01 7.90094137e-01
1.57876444e+00 5.48058212e-01 9.70609128e-01 2.49229312e-01
9.40422356e-01 3.42597216e-01 9.35842216e-01 1.78812370e-01
4.67310965e-01 4.21212792e-01 1.20063591e+00 3.91116366e-02
-8.25576186e-02 -3.94090563e-01 7.90604502e-02 9.66311932e-01
1.54280961e-01 -1.62454601e-02 -4.96273607e-01 5.42803526e-01
-1.65053070e+00 -9.01547551e-01 -3.28602910e-01 1.95467794e+00
9.49881524e-02 1.96263924e-01 -3.38675290e-01 2.47741207e-01
9.93947685e-01 -5.61570786e-02 -9.90603745e-01 -2.34360263e-01
2.30542272e-01 -1.95509687e-01 9.55480278e-01 2.64232099e-01
-7.45654523e-01 7.98195302e-01 4.99786997e+00 9.28402841e-01
-1.17005205e+00 3.46727699e-01 2.45265022e-01 2.34934002e-01
-3.28981280e-01 1.97905377e-02 -9.35997009e-01 1.01053655e+00
7.69783616e-01 -1.54547870e-01 6.14011288e-01 1.21356714e+00
3.86914790e-01 3.88312966e-01 -6.50293171e-01 1.17585492e+00
-2.52178878e-01 -1.38640130e+00 1.73183560e-01 1.92987010e-01
6.33927763e-01 3.96154135e-01 1.02315612e-01 4.16559726e-01
3.67509574e-01 -8.00768793e-01 5.55489898e-01 4.95802999e-01
8.24286759e-01 -1.20629859e+00 9.94144142e-01 6.83631063e-01
-1.23503602e+00 -3.50145191e-01 -9.11638439e-01 8.67434293e-02
3.98108512e-01 8.56266141e-01 -1.02695656e+00 8.38948071e-01
8.43608677e-01 5.70505023e-01 -1.05498187e-01 8.82518888e-01
8.99037123e-02 1.80266678e-01 -3.97460163e-01 -8.85402411e-02
2.57508487e-01 -4.13660526e-01 8.76424134e-01 4.20893639e-01
6.62684202e-01 -8.17003939e-03 3.05494547e-01 1.11263657e+00
-1.13120042e-01 -3.48838449e-01 -1.12835991e+00 3.16250145e-01
8.15894902e-01 1.34903514e+00 -2.81652272e-01 -2.59023935e-01
-3.60812187e-01 3.71732682e-01 -1.97482765e-01 2.29370266e-01
-1.03100502e+00 -5.44434845e-01 7.75482893e-01 4.07614052e-01
4.09377068e-01 -4.46805716e-01 -3.58248413e-01 -5.29397964e-01
1.93102434e-01 -2.60592997e-01 -3.42292428e-01 -1.10629141e+00
-1.29347599e+00 4.64363158e-01 -8.31257403e-02 -1.51679170e+00
5.99502213e-03 -5.04574418e-01 -5.79394937e-01 7.93984532e-01
-1.91922688e+00 -1.41743433e+00 -8.75674844e-01 5.97672403e-01
7.32167363e-01 -2.60808468e-01 2.75507808e-01 5.33281744e-01
-2.53893971e-01 7.30617344e-02 2.74874151e-01 -3.51586431e-01
1.43393546e-01 -6.74416721e-01 6.58030808e-01 7.83433259e-01
-4.90778595e-01 2.51921594e-01 6.45493567e-01 -8.87746453e-01
-1.75756919e+00 -1.86045575e+00 4.07392949e-01 -3.26880723e-01
3.29385638e-01 -1.97936103e-01 -8.34704638e-01 7.27715194e-01
-1.33552507e-01 2.34270096e-01 -2.59543180e-01 -7.23518014e-01
2.18936607e-01 -8.26743484e-01 -1.57037103e+00 3.89701664e-01
9.99362171e-01 -1.06177747e-01 -3.06778997e-01 5.99892259e-01
1.52293742e+00 -3.86577964e-01 -6.41756535e-01 9.16893065e-01
1.58390686e-01 -6.82860255e-01 1.04977250e+00 -7.64732882e-02
2.57706046e-01 -7.54916072e-01 -2.20288187e-01 -1.20244586e+00
-3.81946474e-01 -3.30326945e-01 3.28322239e-02 1.06757414e+00
-2.19437107e-01 -9.69403923e-01 8.58374119e-01 1.05523825e-01
-8.47464681e-01 -6.74316883e-01 -1.29814637e+00 -8.02844226e-01
8.14545006e-02 -4.54888880e-01 1.50265086e+00 5.66040754e-01
-8.86850119e-01 1.49646446e-01 1.47735653e-02 5.93405783e-01
7.58317828e-01 2.00599823e-02 1.17653382e+00 -1.55133235e+00
5.27256489e-01 1.68209039e-02 -7.45816708e-01 -1.06234944e+00
2.15631559e-01 -8.17095518e-01 7.46406764e-02 -1.66906583e+00
-2.81928003e-01 -8.29629481e-01 1.31145853e-03 -1.55445384e-02
1.83816835e-01 6.11625575e-02 2.26940051e-01 4.25910294e-01
-2.06247777e-01 7.26902485e-01 1.57398856e+00 -5.33897281e-01
1.44864187e-01 2.58933395e-01 -3.05268943e-01 6.22353971e-01
8.07680309e-01 -4.60382015e-01 -6.68443501e-01 -6.32855177e-01
-1.32271543e-01 1.20934121e-01 7.98674464e-01 -1.30540729e+00
3.21557581e-01 -8.68037567e-02 5.10094047e-01 -1.42575538e+00
4.78732049e-01 -1.45010138e+00 3.97749841e-01 7.17543483e-01
6.74890161e-01 4.82757181e-01 2.95629889e-01 7.52338350e-01
-1.14282787e-01 -9.48012471e-02 5.89294553e-01 -2.25637779e-01
-8.24791789e-01 8.28217804e-01 4.90973964e-02 -5.50385714e-01
1.45536649e+00 -4.37868953e-01 -4.07881767e-01 -1.16815448e-01
-7.19219223e-02 3.41856420e-01 4.36218560e-01 6.51217401e-01
9.46204841e-01 -1.69723594e+00 -6.68124557e-01 4.52024639e-01
8.07289481e-02 8.15522730e-01 5.87676883e-01 5.03028870e-01
-8.92491996e-01 6.42848790e-01 -1.58630177e-01 -1.01088846e+00
-6.91055179e-01 1.06184554e+00 1.17386132e-01 9.98677164e-02
-8.64633441e-01 1.75603032e-01 3.14684898e-01 -7.20615387e-01
-1.85673371e-01 -4.72811043e-01 1.55383632e-01 -8.28596592e-01
1.56117260e-01 5.84445119e-01 2.67322093e-01 -6.32326603e-01
-1.06151320e-01 6.81302071e-01 5.97925335e-02 4.83189821e-01
1.03865838e+00 -2.75554091e-01 -7.32141212e-02 2.23994896e-01
1.04275501e+00 -2.46228337e-01 -1.34508979e+00 1.08062990e-01
-6.83625877e-01 -6.64880455e-01 1.27055258e-01 -3.64690512e-01
-1.22145653e+00 9.74972665e-01 7.92675555e-01 2.33614117e-01
7.94739723e-01 -2.76468903e-01 1.52848697e+00 3.27016920e-01
9.19096112e-01 -6.52961552e-01 -3.98764789e-01 3.22710544e-01
7.07553267e-01 -1.19169641e+00 -7.70675093e-02 -6.51720941e-01
-2.95211107e-01 8.49042177e-01 7.31126487e-01 -4.43668365e-01
6.28834724e-01 2.57868320e-01 -5.66465966e-02 -1.87564597e-01
-4.14793938e-01 2.22701877e-01 -3.57039988e-01 8.54818761e-01
-7.16486990e-01 1.83092743e-01 8.74435604e-02 3.35885406e-01
-4.06350225e-01 3.00052375e-01 2.83603519e-01 7.34833002e-01
-5.06453693e-01 -7.20321894e-01 -3.99916530e-01 5.55899084e-01
-1.28992677e-01 2.00855881e-01 3.62190038e-01 8.97245347e-01
5.88457644e-01 1.00355136e+00 5.11985302e-01 -5.76813996e-01
4.83541608e-01 -3.68212730e-01 1.77275524e-01 -4.25916202e-02
1.21677034e-01 -4.20283467e-01 -5.11750817e-01 -4.46142107e-01
-2.24808604e-02 -3.62679631e-01 -1.60258067e+00 -7.61440635e-01
-4.30042535e-01 9.40853078e-03 1.23151243e+00 8.96153271e-01
6.76024914e-01 6.73781931e-01 1.02034414e+00 -8.99324059e-01
-7.52062142e-01 -8.47617984e-01 -5.17095566e-01 3.18822384e-01
2.23932445e-01 -9.18973386e-01 -3.93013716e-01 -2.68682986e-01] | [8.15957260131836, -2.763535976409912] |
eb30ea09-28f4-41b4-bb9d-ab0e19327cbc | shuffle-qudio-accelerate-distributed-vqe-with | 2209.12454 | null | https://arxiv.org/abs/2209.12454v1 | https://arxiv.org/pdf/2209.12454v1.pdf | Shuffle-QUDIO: accelerate distributed VQE with trainability enhancement and measurement reduction | The variational quantum eigensolver (VQE) is a leading strategy that exploits noisy intermediate-scale quantum (NISQ) machines to tackle chemical problems outperforming classical approaches. To gain such computational advantages on large-scale problems, a feasible solution is the QUantum DIstributed Optimization (QUDIO) scheme, which partitions the original problem into $K$ subproblems and allocates them to $K$ quantum machines followed by the parallel optimization. Despite the provable acceleration ratio, the efficiency of QUDIO may heavily degrade by the synchronization operation. To conquer this issue, here we propose Shuffle-QUDIO to involve shuffle operations into local Hamiltonians during the quantum distributed optimization. Compared with QUDIO, Shuffle-QUDIO significantly reduces the communication frequency among quantum processors and simultaneously achieves better trainability. Particularly, we prove that Shuffle-QUDIO enables a faster convergence rate over QUDIO. Extensive numerical experiments are conducted to verify that Shuffle-QUDIO allows both a wall-clock time speedup and low approximation error in the tasks of estimating the ground state energy of molecule. We empirically demonstrate that our proposal can be seamlessly integrated with other acceleration techniques, such as operator grouping, to further improve the efficacy of VQE. | ['DaCheng Tao', 'Yuxuan Du', 'Yang Qian'] | 2022-09-26 | null | null | null | null | ['distributed-optimization'] | ['methodology'] | [ 3.09441648e-02 -2.62542427e-01 1.56132206e-01 4.82410081e-02
-1.00621235e+00 -5.74216127e-01 8.46004263e-02 3.08562636e-01
-6.58138335e-01 9.16778445e-01 -4.72516268e-01 -6.24950171e-01
-2.57010698e-01 -1.23211253e+00 -8.14737022e-01 -1.24132502e+00
-1.29971221e-01 3.83459836e-01 6.44011945e-02 -4.64652568e-01
4.78812009e-01 3.55388969e-01 -1.21612370e+00 -9.51216593e-02
1.18044555e+00 7.60152936e-01 1.50633335e-01 5.94836712e-01
2.01857716e-01 3.29439372e-01 -3.20447534e-01 -2.02769890e-01
5.24982095e-01 -7.04050422e-01 -9.47766304e-01 -5.58535218e-01
-1.58621430e-01 1.05177857e-01 -4.71623719e-01 1.53096151e+00
9.60598290e-01 5.00140190e-01 9.80627760e-02 -1.00839400e+00
-2.46042281e-01 6.87420189e-01 -5.13993859e-01 8.87110904e-02
3.32051814e-01 3.67725074e-01 1.13399339e+00 -5.02082884e-01
6.53636158e-01 9.93619919e-01 4.82681096e-01 2.01426595e-01
-1.60457754e+00 -8.35473955e-01 -6.95473194e-01 4.53560799e-01
-1.97397065e+00 -2.02151328e-01 4.68820065e-01 1.31631643e-01
1.16831255e+00 4.01225537e-01 5.62559783e-01 2.83685595e-01
6.47375405e-01 2.38943800e-01 1.33548009e+00 -3.96087319e-01
7.61377573e-01 -2.70007044e-01 1.69453412e-01 8.59666526e-01
2.92187095e-01 3.52984041e-01 -8.93898487e-01 -5.66542983e-01
3.27346504e-01 -1.77796006e-01 -1.62320435e-01 -7.69668818e-02
-1.24144089e+00 8.26270878e-01 7.22187877e-01 8.34683776e-02
-5.22203028e-01 4.51834649e-01 5.62290728e-01 5.31756938e-01
3.03972196e-02 8.06815624e-01 -1.79663643e-01 -2.54060388e-01
-7.47947693e-01 3.89350921e-01 8.65096569e-01 6.98584735e-01
1.35752344e+00 -3.58806163e-01 -1.36893049e-01 1.19791366e-01
-2.49429062e-01 7.59787798e-01 2.09670931e-01 -1.13579738e+00
2.54538417e-01 2.37988517e-01 2.18213409e-01 -8.53147864e-01
-3.95925939e-01 -6.61488771e-02 -1.12514067e+00 -3.97028923e-02
9.74034891e-02 -3.25302362e-01 -5.63732028e-01 1.57707310e+00
7.41142273e-01 -6.14020266e-02 8.86611044e-02 8.76773536e-01
3.26478243e-01 1.14182413e+00 -1.10511258e-01 -6.25774741e-01
1.34209061e+00 -8.24558556e-01 -6.26727045e-01 3.32329363e-01
1.05746710e+00 -6.45691872e-01 5.63351333e-01 4.06658590e-01
-9.46713924e-01 -3.05681199e-01 -1.04090703e+00 -2.97153424e-02
-3.70062798e-01 -5.29059231e-01 1.07910836e+00 8.99535656e-01
-9.79411840e-01 1.23864567e+00 -1.14979351e+00 -7.27577077e-04
4.29600216e-02 9.11639750e-01 -3.06938499e-01 1.40571684e-01
-1.28725123e+00 6.93963230e-01 8.21697712e-01 2.80450910e-01
-7.35180616e-01 -4.67698097e-01 -4.54892427e-01 1.40641674e-01
5.80194175e-01 -7.58192480e-01 8.57965112e-01 -2.21493632e-01
-1.99752510e+00 3.70632201e-01 -4.16912287e-01 -3.90832692e-01
-2.67095380e-02 4.48193520e-01 1.13260411e-01 1.17053114e-01
-3.06732580e-03 1.95936486e-01 4.19580400e-01 -4.40023363e-01
-2.62769252e-01 -4.99846816e-01 1.28037795e-01 1.85735166e-01
-1.68403134e-01 -1.90503940e-01 -7.15309605e-02 1.98222756e-01
3.95014942e-01 -1.23149037e+00 -7.69726753e-01 -7.00191915e-01
-5.37969112e-01 -5.54556131e-01 2.17970610e-01 4.41535551e-04
1.24373055e+00 -1.98065722e+00 6.63173735e-01 4.11657900e-01
4.65423882e-01 1.51966497e-01 1.90772917e-02 8.52227151e-01
1.07437395e-01 4.90154065e-02 -3.18269730e-01 -8.45958740e-02
3.30992013e-01 3.13567400e-01 -2.30781790e-02 6.74590945e-01
-3.20491374e-01 1.16707718e+00 -9.42179918e-01 -4.10766512e-01
4.06560954e-03 6.88779494e-03 -7.80782342e-01 5.52864224e-02
-5.08225663e-03 4.95215207e-01 -5.89748561e-01 4.18234736e-01
1.05716872e+00 -4.70621198e-01 4.62599009e-01 -2.19962835e-01
-4.39827591e-01 4.18427348e-01 -1.45391142e+00 1.84740567e+00
-2.33215213e-01 1.58443041e-02 2.91219085e-01 -1.21689284e+00
4.39478099e-01 1.71875253e-01 4.67046171e-01 -9.15992618e-01
2.98366010e-01 6.40386641e-01 1.94100454e-01 -2.15941876e-01
7.18496025e-01 -4.81875002e-01 -3.62532794e-01 5.61327815e-01
-1.16158329e-01 -4.95803416e-01 4.06850398e-01 2.76968896e-01
1.30996513e+00 -7.20268041e-02 4.55819517e-01 -5.69551706e-01
6.02567136e-01 4.18641940e-02 7.45001733e-01 1.05029500e+00
-4.67855513e-01 -5.51957488e-02 5.07735968e-01 -2.00931862e-01
-9.67492580e-01 -8.61543596e-01 -2.34616220e-01 9.88319099e-01
7.27838635e-01 -9.44820940e-01 -8.51407886e-01 -2.87057281e-01
-2.71066055e-02 3.19001764e-01 -3.77002001e-01 -1.45481750e-01
-4.92332131e-01 -1.43025804e+00 4.91923749e-01 -3.97790596e-02
5.44555664e-01 -7.71258056e-01 -4.52793032e-01 3.40831906e-01
-7.71665573e-02 -7.84794509e-01 -4.35028493e-01 6.14965856e-01
-6.95451200e-01 -6.22603357e-01 -1.63299799e-01 -3.23927701e-01
4.90297139e-01 4.70441669e-01 6.90949082e-01 -1.04779392e-01
-3.46604645e-01 -3.82495672e-01 -2.88284659e-01 9.03822705e-02
-4.08019185e-01 2.03957811e-01 2.38702461e-01 -2.35475495e-01
4.16167766e-01 -8.21343362e-01 -8.50029647e-01 -4.51527610e-02
-9.86157417e-01 2.71252189e-02 5.46166897e-01 1.08154774e+00
8.36925626e-01 3.30686659e-01 8.19767267e-02 -7.91838646e-01
5.01927733e-01 -2.81663984e-01 -1.02464259e+00 1.06525525e-01
-7.11516976e-01 6.13461196e-01 9.98620808e-01 1.60524584e-02
-3.74959409e-01 1.79134101e-01 -3.03022772e-01 1.95753407e-02
3.87557626e-01 7.40444183e-01 8.88287053e-02 -6.70748413e-01
6.14248991e-01 4.76945907e-01 -1.69764072e-01 -1.32561162e-01
4.47002053e-01 7.50380993e-01 7.40355700e-02 -7.63811290e-01
9.09001648e-01 4.13782209e-01 6.37924910e-01 -8.89245749e-01
-5.32542706e-01 -4.50345367e-01 -2.78013289e-01 3.70718986e-01
9.45436239e-01 -7.09791839e-01 -1.59130538e+00 2.76337057e-01
-1.11947405e+00 -1.34323418e-01 -1.87197596e-01 5.65406501e-01
-3.68616998e-01 6.99726284e-01 -8.38550150e-01 -8.95094097e-01
-4.92498785e-01 -1.59389746e+00 1.05778754e+00 2.54924715e-01
2.25435212e-01 -6.02175653e-01 2.36522764e-01 4.12337363e-01
3.76704395e-01 1.92360744e-01 7.23446310e-01 -2.55806893e-01
-1.01802063e+00 -6.65615499e-02 -1.85172975e-01 -9.68784615e-02
-3.67622226e-01 -3.73366058e-01 -5.78425348e-01 -6.77561343e-01
-2.30277365e-04 -6.23667538e-01 6.15482211e-01 4.48202668e-03
1.10407066e+00 2.52449960e-02 -2.28497565e-01 8.69542897e-01
1.61138415e+00 1.46367013e-01 5.48634708e-01 1.36378974e-01
5.08663714e-01 -8.47122371e-02 3.91708761e-01 5.19895792e-01
3.05359900e-01 7.71005511e-01 3.42733741e-01 1.13056749e-01
4.80398804e-01 -6.96173832e-02 4.68863904e-01 1.55908895e+00
-2.46302485e-01 1.60672460e-02 -6.85778499e-01 -6.76000565e-02
-1.82035112e+00 -8.24901879e-01 -1.96935385e-01 2.34155154e+00
8.74132335e-01 -1.45121455e-01 -8.47304538e-02 6.26119450e-02
5.71809530e-01 1.56279817e-01 -6.38885200e-01 -6.60543680e-01
1.14284441e-01 8.51245403e-01 9.36439812e-01 7.02119887e-01
-7.63578892e-01 1.07160151e+00 5.64696217e+00 1.37635899e+00
-1.07794845e+00 6.10878766e-01 2.44096667e-01 1.88352317e-01
-1.72488883e-01 3.73294115e-01 -6.42679334e-01 3.16927016e-01
1.33111382e+00 -4.46790189e-01 1.04237700e+00 6.04795396e-01
2.39674538e-01 -2.62759417e-01 -9.37527716e-01 1.21126044e+00
-4.47677732e-01 -1.43900561e+00 -2.62106895e-01 4.21143144e-01
9.09128308e-01 2.85217285e-01 -1.51993439e-01 3.08205873e-01
1.63207322e-01 -1.00066602e+00 3.12986076e-01 -8.19640458e-02
6.37980700e-01 -9.49210286e-01 8.85681152e-01 4.93235230e-01
-1.22757602e+00 2.05732569e-01 -7.20581293e-01 -4.32147950e-01
1.50362417e-01 5.64561427e-01 -3.68869275e-01 1.14758408e+00
3.25963706e-01 1.54737473e-01 -1.78676262e-01 7.14594603e-01
-1.10653892e-01 4.04551327e-01 -6.75213099e-01 -6.24295056e-01
4.43801880e-01 -1.07488382e+00 4.24469411e-01 6.43937469e-01
2.60575831e-01 4.65473652e-01 2.53393680e-01 1.06758881e+00
-2.13237152e-01 2.47011051e-01 -2.63499498e-01 -4.15535718e-01
6.12847865e-01 1.25959373e+00 -6.53169453e-01 -4.86524493e-01
1.76890753e-02 1.31322885e+00 4.12978351e-01 2.39132419e-01
-9.81287062e-01 -9.30589855e-01 4.84714895e-01 -3.50468397e-01
3.03880781e-01 -5.61968505e-01 -1.21924840e-01 -1.35315108e+00
-1.63558543e-01 -9.14013982e-01 -3.92419770e-02 -2.52940118e-01
-9.27865446e-01 4.88072395e-01 -4.51101393e-01 -6.29931509e-01
1.82079002e-01 -5.24720550e-01 -2.53479600e-01 9.47527468e-01
-1.43167627e+00 -4.95962739e-01 -6.96708113e-02 4.50274497e-01
-1.43142089e-01 5.50832510e-01 1.12753499e+00 2.97711611e-01
-9.89072800e-01 5.85005939e-01 7.93737054e-01 -3.52338225e-01
3.56974721e-01 -1.37263036e+00 2.40388393e-01 7.43935943e-01
3.92914638e-02 9.50487435e-01 8.80037785e-01 -3.60382795e-01
-2.37228656e+00 -6.87911987e-01 7.26369858e-01 -2.24260256e-01
8.17210674e-01 -5.00632226e-01 -5.76123655e-01 1.41629785e-01
9.93041396e-02 4.65880148e-02 7.30159461e-01 2.20302746e-01
-5.03553674e-02 -3.05412382e-01 -8.78700912e-01 3.45203459e-01
9.50144947e-01 -8.56817126e-01 -1.86398819e-01 9.70190585e-01
6.26459718e-01 -6.30082667e-01 -1.17167020e+00 2.62997597e-02
2.31229544e-01 -1.03119373e+00 7.43824065e-01 -3.77916455e-01
2.04025805e-01 -4.95128840e-01 -1.84977785e-01 -9.90874887e-01
-1.71462819e-01 -1.56395400e+00 1.11099645e-01 5.23450196e-01
3.08830589e-01 -1.03257227e+00 7.99162328e-01 4.91663784e-01
5.62353916e-02 -6.30143046e-01 -1.52154207e+00 -8.85108650e-01
3.73458326e-01 -2.09963471e-01 5.42585671e-01 7.35944331e-01
4.88098323e-01 4.33762133e-01 -3.96645159e-01 4.08386111e-01
7.68085301e-01 5.60861409e-01 9.17736053e-01 -5.27928174e-01
-7.53501892e-01 -1.54036984e-01 -3.99868906e-01 -1.27693784e+00
-8.53986815e-02 -1.19480503e+00 2.41671950e-01 -9.46935236e-01
5.02993226e-01 -4.15219903e-01 -4.80978400e-01 2.59576172e-01
-4.97771859e-01 2.91096151e-01 2.26296589e-01 2.49109164e-01
-1.01799905e+00 7.85758793e-01 1.33438337e+00 8.34074765e-02
-3.92252147e-01 -4.65776950e-01 -2.55967915e-01 1.27784461e-01
4.17711437e-01 -7.52651036e-01 5.60457036e-02 -1.66215852e-01
5.99657655e-01 4.28256333e-01 -5.72611094e-02 -9.93722320e-01
5.52890897e-01 -2.23538548e-01 -4.94788677e-01 -3.18207800e-01
3.12699258e-01 -1.82276011e-01 2.92534173e-01 9.86521363e-01
1.37779757e-01 8.34769905e-02 -1.65254667e-01 6.00078106e-01
-1.83652177e-01 -1.61892548e-01 7.34539628e-01 -1.27792001e-01
-2.50467479e-01 4.83737677e-01 -3.39686811e-01 -2.78035969e-01
1.09097767e+00 1.83687195e-01 -2.05745474e-01 9.18453857e-02
-5.40702462e-01 3.77550930e-01 5.21264851e-01 -5.88262677e-01
1.56178117e-01 -1.04023659e+00 -4.13984865e-01 1.57229245e-01
-1.73402146e-01 -1.40232136e-02 3.86047006e-01 1.34986973e+00
-9.86261904e-01 7.38053977e-01 1.90333918e-01 -5.76509953e-01
-9.37051356e-01 7.16191947e-01 2.97959775e-01 -4.51546222e-01
-3.86184931e-01 1.02256608e+00 -1.10497758e-01 -5.09487569e-01
-3.58165592e-01 -1.21508479e-01 7.21338153e-01 -4.12642181e-01
3.09420556e-01 6.14081800e-01 1.37642115e-01 -4.02130514e-01
-5.00932515e-01 4.60859716e-01 5.32382689e-02 -1.31521985e-01
1.15185678e+00 -5.24438918e-02 -8.17540288e-01 9.57453474e-02
1.36689031e+00 1.73524007e-01 -5.88256598e-01 -1.33965045e-01
-1.65589258e-01 -1.70339286e-01 1.33432835e-01 -1.53273985e-01
-5.17103136e-01 8.66421878e-01 4.92922097e-01 4.10581261e-01
9.88641202e-01 -3.05310100e-01 1.35255623e+00 1.06835508e+00
1.02292073e+00 -1.10091138e+00 -4.62888032e-01 4.69887316e-01
1.30529171e-02 -1.17664206e+00 3.22505474e-01 -5.01436710e-01
-1.22048482e-01 9.74261045e-01 1.44695327e-01 -1.74901560e-01
3.87959301e-01 -1.33319423e-01 -4.47438300e-01 -3.78478557e-01
-5.14383793e-01 -2.44294778e-01 -2.45492533e-01 -2.76266336e-01
1.39523983e-01 4.85165745e-01 -9.53990817e-01 2.26964265e-01
-3.12647700e-01 -7.38108605e-02 4.70365196e-01 1.07156980e+00
-4.03045177e-01 -1.62346804e+00 -3.22706550e-01 8.68319049e-02
-2.55091012e-01 -2.84149349e-01 -1.03487140e-02 4.05595392e-01
3.40875834e-01 9.24570918e-01 -4.24137414e-01 -4.57114309e-01
2.67495960e-02 3.47412750e-02 5.45952559e-01 -5.31666934e-01
-8.01202893e-01 -1.91135481e-02 -2.50531882e-01 -9.92785454e-01
-2.96701312e-01 -2.02871427e-01 -1.47064447e+00 -8.88916373e-01
-7.57630229e-01 9.83536780e-01 6.31620526e-01 1.02883065e+00
7.16436923e-01 4.59463239e-01 8.63812566e-01 -8.42175543e-01
-1.23631108e+00 -5.81126571e-01 -8.04910719e-01 2.25398555e-01
-3.24444398e-02 -3.96055877e-01 -4.92823660e-01 -7.14494586e-01] | [5.583877086639404, 4.932258605957031] |
468f3dcf-f1b3-4c18-8baa-9787a5d942b2 | bringing-blurry-alive-at-high-frame-rate-with | 1903.06531 | null | https://arxiv.org/abs/1903.06531v3 | https://arxiv.org/pdf/1903.06531v3.pdf | High Frame Rate Video Reconstruction based on an Event Camera | Event-based cameras measure intensity changes (called `events') with microsecond accuracy under high-speed motion and challenging lighting conditions. With the `active pixel sensor' (APS), the `Dynamic and Active-pixel Vision Sensor' (DAVIS) allows the simultaneous output of intensity frames and events. However, the output images are captured at a relatively low frame rate and often suffer from motion blur. A blurred image can be regarded as the integral of a sequence of latent images, while events indicate changes between the latent images. Thus, we are able to model the blur-generation process by associating event data to a latent sharp image. Based on the abundant event data alongside a low frame rate, easily blurred images, we propose a simple yet effective approach to reconstruct high-quality and high frame rate sharp videos. Starting with a single blurred frame and its event data from DAVIS, we propose the Event-based Double Integral (EDI) model and solve it by adding regularization terms. Then, we extend it to multiple Event-based Double Integral (mEDI) model to get more smooth results based on multiple images and their events. Furthermore, we provide a new and more efficient solver to minimize the proposed energy model. By optimizing the energy function, we achieve significant improvements in removing blur and the reconstruction of a high temporal resolution video. The video generation is based on solving a simple non-convex optimization problem in a single scalar variable. Experimental results on both synthetic and real datasets demonstrate the superiority of our mEDI model and optimization method compared to the state-of-the-art. | ['Miaomiao Liu', 'Xin Yu', 'Cedric Scheerlinck', 'Richard Hartley', 'Liyuan Pan', 'Yuchao Dai'] | 2019-03-12 | null | null | null | null | ['video-reconstruction'] | ['computer-vision'] | [ 4.81470168e-01 -5.34311295e-01 3.50874275e-01 -2.18623459e-01
-6.51402295e-01 -2.90283233e-01 2.87458062e-01 -2.68692583e-01
-5.73679030e-01 7.17443049e-01 1.86050721e-02 3.53428930e-01
-1.34424776e-01 -5.22884190e-01 -9.25694168e-01 -9.64528799e-01
9.41172838e-02 -2.40174368e-01 4.97196794e-01 3.40754777e-01
8.34599882e-02 2.66132146e-01 -1.49719787e+00 1.58247158e-01
1.10640538e+00 1.24490333e+00 5.88934302e-01 6.58618212e-01
1.28342509e-01 1.07005012e+00 -4.25426006e-01 -1.60191163e-01
2.33059257e-01 -3.97636831e-01 -2.98806071e-01 4.49515134e-01
3.78844917e-01 -6.46684051e-01 -4.13720161e-01 1.23764598e+00
3.29346925e-01 2.52172798e-01 1.54888064e-01 -1.01333690e+00
-6.80847287e-01 -5.83455674e-02 -9.93948102e-01 4.52703923e-01
5.81670344e-01 3.81070524e-01 3.86216640e-01 -9.16690230e-01
6.72831774e-01 1.08065021e+00 5.67505002e-01 2.66468227e-01
-1.40143454e+00 -3.43039125e-01 5.81350476e-02 4.29723859e-01
-1.36825025e+00 -5.59669018e-01 8.28744650e-01 -3.09054881e-01
7.06381679e-01 3.68766278e-01 7.42419064e-01 9.73770678e-01
2.85663873e-01 6.43120766e-01 1.15171158e+00 -8.86291489e-02
2.64372885e-01 -3.05059642e-01 -5.47764376e-02 6.68854356e-01
1.76127017e-01 1.39131576e-01 -6.29916966e-01 3.97926681e-02
1.13537502e+00 4.76789296e-01 -9.42841649e-01 5.56354448e-02
-1.48150504e+00 3.05346191e-01 2.18606755e-01 1.19382754e-01
-7.75900483e-01 1.71945110e-01 1.43333180e-02 -6.50658086e-02
5.44431806e-01 1.70436967e-02 -1.69098347e-01 -2.70380646e-01
-1.13442981e+00 1.61053866e-01 6.00110471e-01 7.04979360e-01
7.96443224e-01 4.16198857e-02 -3.63069415e-01 6.40665174e-01
3.51428866e-01 6.16485596e-01 3.14667553e-01 -1.32317901e+00
2.87867516e-01 3.17252904e-01 6.35201931e-01 -1.00762618e+00
4.04940210e-02 -6.10862933e-02 -1.07775021e+00 3.98135900e-01
4.98721927e-01 -1.66173026e-01 -7.77599096e-01 1.54935861e+00
3.25411201e-01 1.00779617e+00 -7.34423697e-02 1.33456028e+00
5.10888219e-01 1.16258514e+00 -2.85718650e-01 -1.15771127e+00
1.30584264e+00 -7.63391912e-01 -1.21253526e+00 -1.05492197e-01
-3.06715220e-01 -8.68292809e-01 8.13795030e-01 5.74115992e-01
-1.60631645e+00 -5.61584353e-01 -9.12905872e-01 -2.18555704e-01
1.22850351e-01 -5.91857992e-02 4.01827604e-01 -2.11087018e-02
-8.87366354e-01 5.65757632e-01 -9.87326741e-01 1.29494593e-01
2.87269473e-01 6.19560331e-02 -2.53701717e-01 -1.99931726e-01
-9.32627380e-01 6.55788362e-01 1.36964411e-01 2.31166109e-01
-8.61699462e-01 -9.24249768e-01 -7.44857728e-01 -3.57706137e-02
5.39591253e-01 -8.14684510e-01 1.03506660e+00 -9.61892664e-01
-1.67208338e+00 6.38082922e-01 -4.35276777e-01 -3.59250516e-01
6.15609169e-01 -4.12730277e-01 -1.94445327e-01 5.18494725e-01
-7.70967081e-02 2.73540467e-01 1.18495834e+00 -1.19035554e+00
-6.15568995e-01 -2.90005412e-02 -4.18975204e-02 2.12276042e-01
-3.53096426e-02 2.21352026e-01 -7.30016708e-01 -7.08420515e-01
-8.25084746e-02 -6.14440918e-01 -4.79897521e-02 3.04851353e-01
1.53103739e-01 2.03948587e-01 9.22159076e-01 -8.04278493e-01
1.33329117e+00 -2.33699393e+00 3.52155328e-01 -6.07902348e-01
3.23774666e-01 2.41196081e-01 1.63570076e-01 -6.41650558e-02
-2.35866942e-02 -4.03174192e-01 -5.53583145e-01 -5.11768878e-01
-3.98781657e-01 2.95084357e-01 -3.42396587e-01 5.33304274e-01
1.81188911e-01 8.29162478e-01 -1.18117476e+00 -4.71688062e-01
7.24420071e-01 8.26056242e-01 -2.88870335e-01 4.87933189e-01
-1.19095184e-01 6.89309955e-01 -2.71647811e-01 4.81205016e-01
1.04193890e+00 -4.23588037e-01 -2.39562467e-01 -6.85115933e-01
-6.39540493e-01 -2.77942210e-01 -1.44226611e+00 1.92400420e+00
-3.53560209e-01 5.81678629e-01 3.46668035e-01 -7.43610382e-01
5.81350148e-01 3.66507053e-01 7.74618208e-01 -5.63653350e-01
4.75077033e-02 7.91926980e-02 -5.45711935e-01 -8.55543733e-01
3.19377840e-01 -2.52449274e-01 2.48417720e-01 1.09157018e-01
-1.19389661e-01 -1.44013437e-03 2.64447123e-01 9.70703214e-02
1.10440242e+00 2.57310480e-01 4.77828346e-02 8.18468705e-02
6.70503676e-01 -4.26566720e-01 7.83342600e-01 5.47302365e-01
-3.18030685e-01 9.37748492e-01 7.54367262e-02 -3.63696367e-01
-1.03870428e+00 -1.26769555e+00 -2.18389541e-01 3.26891184e-01
6.25242591e-01 -2.03770176e-01 -7.81190217e-01 -1.49788961e-01
-2.68965691e-01 4.80435252e-01 -3.16928625e-01 -1.62574137e-03
-6.44172192e-01 -1.00067580e+00 -8.32645744e-02 2.86798835e-01
8.41498375e-01 -6.46055698e-01 -7.99393594e-01 4.71531808e-01
-5.61942697e-01 -1.43037295e+00 -8.34852934e-01 -1.61798909e-01
-6.85403168e-01 -9.50462103e-01 -9.54188287e-01 -4.73952055e-01
5.90217471e-01 5.29253662e-01 9.15006757e-01 -3.90891105e-01
-4.02629524e-01 4.68486905e-01 -1.84051722e-01 -6.75771460e-02
-5.97340763e-02 -9.75788832e-01 2.35754810e-02 8.17448974e-01
-1.92016825e-01 -5.97076416e-01 -1.00489414e+00 9.08273533e-02
-1.33351696e+00 4.52048242e-01 4.75392759e-01 6.57275617e-01
7.65720189e-01 1.99454904e-01 6.44713864e-02 -3.13763440e-01
2.67522842e-01 -3.38628501e-01 -8.78685892e-01 1.66019350e-01
-3.88368934e-01 -1.65715083e-01 6.35085523e-01 -7.75058568e-01
-1.68286908e+00 1.21965960e-01 2.52641827e-01 -1.02350068e+00
6.61258027e-02 2.00411737e-01 4.17441800e-02 -2.18735129e-01
3.81404340e-01 3.91333908e-01 2.66318000e-03 -3.60963255e-01
2.51976907e-01 5.29820442e-01 8.78357112e-01 -3.76934528e-01
6.46452546e-01 9.90291774e-01 -8.95860139e-03 -7.31661975e-01
-9.11258042e-01 -4.73325938e-01 -1.89799666e-01 -5.75139701e-01
1.13133240e+00 -1.14293158e+00 -9.19244409e-01 9.69818950e-01
-1.40432334e+00 -1.95864812e-01 -3.65985394e-01 6.69277847e-01
-6.09397173e-01 7.28821874e-01 -1.02582800e+00 -7.95683205e-01
-2.42421180e-01 -1.13375354e+00 1.19333887e+00 6.71324193e-01
2.76596636e-01 -9.80748177e-01 -2.85476670e-02 3.41499507e-01
3.99824530e-01 4.20820624e-01 4.10763882e-02 6.78340733e-01
-1.22152543e+00 4.15844880e-02 -5.02638936e-01 4.86500829e-01
2.99575657e-01 1.07235350e-01 -8.61406446e-01 -1.75572544e-01
8.33141685e-01 1.85912371e-01 7.22206354e-01 8.20425034e-01
1.27242458e+00 -3.55275065e-01 -1.17027663e-01 1.04896927e+00
1.93058920e+00 3.00790071e-01 8.26271713e-01 6.50792494e-02
7.45026231e-01 1.46740213e-01 7.13467777e-01 6.21137857e-01
1.88008785e-01 8.66619468e-01 3.93979371e-01 -1.50999859e-01
-1.71293497e-01 2.28054255e-01 5.73891759e-01 7.37389326e-01
-5.43586612e-01 -1.23980142e-01 -4.52015758e-01 5.12344897e-01
-2.13937569e+00 -1.30284083e+00 -5.04104316e-01 2.40864682e+00
1.16603661e+00 -2.03032017e-01 -1.81379616e-01 4.45386302e-03
7.99915552e-01 2.30434760e-01 -5.23229420e-01 1.23941354e-01
-2.85902351e-01 5.91105185e-02 4.70672786e-01 6.35101676e-01
-9.40789223e-01 4.52229112e-01 5.81729174e+00 7.45452046e-01
-1.24684370e+00 1.97514728e-01 4.54258800e-01 -4.33621228e-01
-4.31584194e-02 -3.50960307e-02 -6.30971074e-01 1.19593072e+00
9.36832607e-01 -4.67844635e-01 8.96917522e-01 2.24912360e-01
8.74874294e-01 -4.27284539e-01 -7.82881021e-01 1.51302779e+00
2.27922276e-02 -1.27081358e+00 -1.81274444e-01 -1.40106142e-01
7.84004569e-01 -2.87768394e-01 -1.14103310e-01 -3.86093616e-01
-2.18942642e-01 -5.03789008e-01 7.75941133e-01 1.19910347e+00
8.27368438e-01 -4.27599460e-01 3.42853427e-01 2.55231857e-01
-1.18775225e+00 1.74446646e-02 -3.14771265e-01 -6.95659071e-02
7.84082830e-01 1.13801646e+00 1.13809511e-01 5.83124101e-01
9.69371021e-01 1.13995075e+00 -1.47070795e-01 1.10127389e+00
-5.46191931e-02 2.80053556e-01 -2.63469040e-01 5.75971842e-01
-1.34215593e-01 -8.16824913e-01 8.29871416e-01 1.11946142e+00
5.26785731e-01 5.76095641e-01 -6.59407973e-02 1.12163067e+00
1.99860007e-01 -5.93683362e-01 -2.99264848e-01 2.60184944e-01
1.20391548e-01 1.44466686e+00 -4.39266980e-01 -4.50019628e-01
-7.94114053e-01 1.64935458e+00 -1.20920157e-02 6.63925588e-01
-1.25170779e+00 -9.18351412e-02 7.84902930e-01 -5.92802651e-02
2.81779438e-01 -2.83509761e-01 -8.65765065e-02 -1.67663324e+00
3.55268568e-01 -5.67817926e-01 1.91186126e-02 -1.29261827e+00
-1.21134496e+00 1.91499680e-01 -2.58961730e-02 -1.30609095e+00
1.08913437e-01 -3.42678905e-01 -4.93936330e-01 9.24969733e-01
-1.74858725e+00 -6.87496901e-01 -7.64183521e-01 9.08175349e-01
6.27865255e-01 5.51609337e-01 3.25302213e-01 6.16270185e-01
-6.40802860e-01 -1.63461372e-01 2.98247695e-01 -3.04667473e-01
7.84554362e-01 -1.21363282e+00 -1.17459983e-01 1.16252303e+00
-3.26818824e-01 2.67477691e-01 7.88069367e-01 -5.92257202e-01
-1.65297925e+00 -1.04448152e+00 5.64822495e-01 -1.69475421e-01
6.35635614e-01 -1.81724783e-02 -1.17949605e+00 5.62127948e-01
2.28963122e-01 3.07801217e-01 1.92306042e-01 -8.35481822e-01
3.12601894e-01 -4.67184663e-01 -1.08048868e+00 3.07246566e-01
9.96855438e-01 -4.24808681e-01 -5.86700797e-01 2.27432176e-01
7.84134507e-01 -7.26305783e-01 -9.51354802e-01 4.41822559e-01
2.30184913e-01 -1.15942669e+00 1.25714266e+00 2.02300593e-01
6.13671243e-01 -7.38358557e-01 2.74845690e-01 -1.00714111e+00
-4.12234634e-01 -9.39891100e-01 -6.88712358e-01 1.25358534e+00
-3.87506604e-01 -5.38622379e-01 2.99527615e-01 7.05639660e-01
-1.34004772e-01 -6.30426466e-01 -8.52983594e-01 -6.98226869e-01
-8.58947635e-01 -2.36637369e-01 9.86077860e-02 8.17828476e-01
-3.20544958e-01 1.91736087e-01 -6.07024789e-01 4.45146501e-01
1.23710716e+00 1.40193745e-01 2.77275592e-01 -9.07235384e-01
-4.32507008e-01 -7.50320032e-02 -2.28079021e-01 -1.24291384e+00
-2.90289283e-01 -2.75398672e-01 2.41493717e-01 -1.29810441e+00
4.34299886e-01 5.11681922e-02 -2.68157870e-01 -1.59051195e-01
-6.76878810e-01 1.79977611e-01 2.58618414e-01 4.35320467e-01
-6.78683341e-01 4.39950019e-01 1.37234521e+00 1.17068544e-01
-8.39752331e-02 -2.61732668e-01 -2.28496343e-01 7.33312845e-01
1.70297548e-01 -1.69241905e-01 -2.70597428e-01 -6.48516476e-01
1.19482227e-01 4.08529073e-01 7.31906712e-01 -9.85955954e-01
4.45503294e-01 -3.14528972e-01 3.74218345e-01 -3.59157294e-01
5.24564862e-01 -8.59378994e-01 7.28398323e-01 1.22554459e-01
1.05145819e-01 -2.26050735e-01 8.09083059e-02 8.03062558e-01
-4.56872046e-01 -1.49188027e-01 1.19048679e+00 -2.62481064e-01
-6.73331499e-01 4.42378163e-01 -1.99155688e-01 -1.31764799e-01
1.21787763e+00 -2.89867222e-01 -3.28929543e-01 -2.36035362e-01
-6.02959514e-01 7.34979361e-02 7.89703250e-01 8.73876065e-02
8.42158139e-01 -1.33171856e+00 -5.73576689e-01 1.60627514e-01
-3.97417217e-01 3.36596996e-01 6.33773327e-01 1.09935820e+00
-6.58345103e-01 -7.25089200e-03 -5.24097569e-02 -8.64843190e-01
-1.12428606e+00 7.05537319e-01 3.53957117e-01 -2.64884550e-02
-7.67068744e-01 7.33419836e-01 5.35589278e-01 5.49800098e-01
4.63427566e-02 -4.55903143e-01 5.53248776e-03 -7.49066621e-02
8.72604787e-01 6.46316230e-01 -3.13426852e-01 -4.41517383e-01
-1.60722822e-01 7.71703541e-01 2.58120030e-01 -1.35617003e-01
1.43021274e+00 -6.48306847e-01 -1.91181406e-01 5.60909092e-01
1.14276564e+00 -1.92237627e-02 -2.00877762e+00 -1.63043886e-01
-5.75442851e-01 -8.14383507e-01 3.60996038e-01 -4.49550837e-01
-1.11308110e+00 6.40473247e-01 7.32464671e-01 2.66493976e-01
1.71758199e+00 -3.29195499e-01 1.12704313e+00 -4.53857332e-01
1.34627149e-01 -9.21254575e-01 2.07404822e-01 1.20288454e-01
8.36382627e-01 -1.23582006e+00 -7.14039663e-03 -5.95567882e-01
-3.15189391e-01 1.06226194e+00 3.07296097e-01 -1.02138653e-01
3.05481672e-01 3.58661979e-01 -1.31636336e-01 8.62265825e-02
-7.52933800e-01 -1.15172744e-01 1.65778846e-01 2.15463877e-01
1.43266261e-01 -2.88081586e-01 -3.78124535e-01 1.94901198e-01
5.56959748e-01 6.40175045e-01 6.80886567e-01 6.94191635e-01
-3.03023726e-01 -6.42378747e-01 -6.43811762e-01 1.13339968e-01
-5.39015412e-01 -6.83510005e-02 3.32928419e-01 8.44720230e-02
3.74565899e-01 9.28244889e-01 2.03808650e-01 1.93584621e-01
3.72945189e-01 -1.60803273e-01 6.79664075e-01 -2.39610761e-01
-8.67638960e-02 3.25680494e-01 -4.89087820e-01 -9.50675964e-01
-7.72344053e-01 -8.79305899e-01 -1.10963786e+00 -3.27230841e-01
-2.85276175e-01 -2.42012292e-01 5.12297332e-01 7.77610481e-01
3.80271167e-01 6.86955810e-01 7.31369972e-01 -1.04364479e+00
-1.38956606e-01 -6.47323668e-01 -6.42483234e-01 8.24076355e-01
6.55170679e-01 -4.42763567e-01 -7.66784966e-01 7.01134205e-01] | [10.939847946166992, -2.1599936485290527] |
8d5d6dba-2f25-447d-bcf4-2e5c99b56863 | proximity-forest-2-0-a-new-effective-and | 2304.05800 | null | https://arxiv.org/abs/2304.05800v2 | https://arxiv.org/pdf/2304.05800v2.pdf | Proximity Forest 2.0: A new effective and scalable similarity-based classifier for time series | Time series classification (TSC) is a challenging task due to the diversity of types of feature that may be relevant for different classification tasks, including trends, variance, frequency, magnitude, and various patterns. To address this challenge, several alternative classes of approach have been developed, including similarity-based, features and intervals, shapelets, dictionary, kernel, neural network, and hybrid approaches. While kernel, neural network, and hybrid approaches perform well overall, some specialized approaches are better suited for specific tasks. In this paper, we propose a new similarity-based classifier, Proximity Forest version 2.0 (PF 2.0), which outperforms previous state-of-the-art similarity-based classifiers across the UCR benchmark and outperforms state-of-the-art kernel, neural network, and hybrid methods on specific datasets in the benchmark that are best addressed by similarity-base methods. PF 2.0 incorporates three recent advances in time series similarity measures -- (1) computationally efficient early abandoning and pruning to speedup elastic similarity computations; (2) a new elastic similarity measure, Amerced Dynamic Time Warping (ADTW); and (3) cost function tuning. It rationalizes the set of similarity measures employed, reducing the eight base measures of the original PF to three and using the first derivative transform with all similarity measures, rather than a limited subset. We have implemented both PF 1.0 and PF 2.0 in a single C++ framework, making the PF framework more efficient. | ['Geoffrey I. Webb', 'Mahsa Salehi', 'Chang Wei Tan', 'Matthieu Herrmann'] | 2023-04-12 | null | null | null | null | ['time-series-classification', 'dynamic-time-warping'] | ['time-series', 'time-series'] | [-3.93172838e-02 -6.59301579e-01 -1.62001491e-01 -1.33914977e-01
-4.82788891e-01 -6.07846439e-01 7.74341226e-01 4.35882360e-01
-3.56292337e-01 4.90688622e-01 3.74075510e-02 -3.74519110e-01
-7.88577497e-01 -7.63536572e-01 -6.43542707e-02 -6.39834881e-01
-8.64068329e-01 4.07551616e-01 7.16670752e-01 -6.19656324e-01
5.80127180e-01 8.65669012e-01 -1.87162042e+00 2.85727531e-01
7.89049268e-01 1.60370171e+00 -2.42132112e-01 4.37434167e-01
-6.34548366e-02 4.53644902e-01 -4.95396584e-01 2.70531196e-02
3.09556752e-01 -4.75379340e-02 -4.22345161e-01 -7.18768537e-01
-5.12871742e-02 2.63120502e-01 -1.78937644e-01 5.48507869e-01
5.67857802e-01 5.93054652e-01 8.74009013e-01 -1.53951514e+00
-5.64762354e-01 3.82542640e-01 -5.12461364e-01 7.43795991e-01
5.07318795e-01 -2.89601773e-01 7.11203516e-01 -7.97117054e-01
3.61381412e-01 8.51221323e-01 1.31835496e+00 -1.45763680e-01
-1.12134004e+00 -6.56977892e-01 -2.10152641e-01 7.77475953e-01
-1.52659690e+00 -5.71861528e-02 9.11062598e-01 -4.25712913e-01
1.55354428e+00 5.58326483e-01 8.51610601e-01 9.38948512e-01
3.51266563e-01 4.53350842e-01 1.39430952e+00 -5.65864146e-01
4.25038010e-01 -1.00771435e-01 4.08907175e-01 2.21915126e-01
-5.86896911e-02 2.90334642e-01 -1.40524447e-01 -8.06936264e-01
4.35796469e-01 1.58585772e-01 -1.13196954e-01 -3.04077953e-01
-1.20399475e+00 8.80172372e-01 -2.41304506e-02 8.54812801e-01
-4.02243406e-01 -2.71458268e-01 8.80058110e-01 7.60670125e-01
7.49763191e-01 4.47892487e-01 -6.89313531e-01 -4.21311170e-01
-1.24998200e+00 5.09388983e-01 1.02143443e+00 4.97708768e-01
3.89822751e-01 2.41736174e-02 -2.97366291e-01 9.34155047e-01
-2.97673255e-01 2.14784846e-01 1.09741592e+00 -5.60784101e-01
1.17210612e-01 5.10870755e-01 -2.12902561e-01 -1.30763185e+00
-7.50477850e-01 -4.02896225e-01 -9.20692146e-01 7.19715580e-02
2.27181658e-01 3.13306123e-01 -7.34003067e-01 1.38409626e+00
3.41528058e-01 4.58497345e-01 -2.83096153e-02 4.79144692e-01
5.46824574e-01 8.05814922e-01 -3.69102061e-01 -3.75386834e-01
1.11698854e+00 -8.55763733e-01 -4.45231795e-01 3.11516225e-01
5.54102182e-01 -1.01277113e+00 6.99620008e-01 5.22636414e-01
-8.65636706e-01 -6.82089508e-01 -1.14693427e+00 4.15698081e-01
-9.63501155e-01 -3.39172304e-01 5.95001936e-01 4.69189554e-01
-1.17969942e+00 1.24694443e+00 -8.45268369e-01 -5.57213843e-01
-5.48734441e-02 3.33875299e-01 -2.45932430e-01 3.79991710e-01
-1.24845171e+00 1.14108920e+00 4.61850613e-01 -4.51986730e-01
7.94130936e-02 -9.26621914e-01 -6.22020304e-01 6.07104190e-02
5.29335476e-02 -2.53208190e-01 1.02885342e+00 -6.20334506e-01
-1.42852986e+00 5.16455054e-01 2.38079891e-01 -6.88600361e-01
3.78872335e-01 -3.21083739e-02 -9.45653796e-01 1.10836156e-01
-2.65617426e-02 1.22340344e-01 7.87014544e-01 -4.29232329e-01
-5.75471044e-01 -1.93102121e-01 -6.47921503e-01 -2.88806796e-01
-4.33757186e-01 3.65033448e-01 1.82614207e-01 -1.13764286e+00
-8.10704306e-02 -9.94827449e-01 -1.01944983e-01 -1.01467043e-01
2.31937975e-01 -6.01342678e-01 1.11191726e+00 -6.39802039e-01
1.73395479e+00 -2.15164185e+00 -9.30206403e-02 5.12665272e-01
-3.27103496e-01 5.51189542e-01 -1.13280915e-01 8.75894129e-01
-6.47270143e-01 -2.00061113e-01 -2.26690844e-01 4.96549867e-02
-1.85967572e-02 1.88206092e-01 -4.31541055e-01 4.78784144e-01
3.39500494e-02 5.66614151e-01 -6.95312262e-01 -4.20728177e-01
2.49925703e-01 4.31316793e-01 -1.94510564e-01 1.94438286e-02
3.52215439e-01 -9.73594263e-02 -1.79003537e-01 7.97920704e-01
4.62355465e-01 8.31203386e-02 -3.64712149e-01 -3.34102452e-01
-5.29921412e-01 -6.48186728e-02 -1.39942265e+00 1.34643269e+00
-2.78754383e-01 6.01615310e-01 -4.54860091e-01 -1.49633968e+00
1.35326815e+00 4.19011712e-01 1.02726567e+00 -5.26559174e-01
1.06473388e-02 5.08173525e-01 5.74825741e-02 -4.90925610e-01
5.28809011e-01 1.97666705e-01 3.52149382e-02 3.93884361e-01
4.12255228e-02 -1.27657130e-01 4.51604426e-01 -3.48920584e-01
1.37757492e+00 1.30597427e-01 6.99347615e-01 -4.47403193e-01
6.18693650e-01 1.53641641e-01 3.85655463e-01 4.12216842e-01
-4.66738284e-01 6.01731122e-01 1.06071874e-01 -7.98855662e-01
-1.21200073e+00 -9.93663132e-01 -3.39760214e-01 1.09500515e+00
-3.19935411e-01 -3.97064120e-01 -2.06370696e-01 -2.35267043e-01
3.03612977e-01 6.68622255e-01 -6.62417412e-01 -2.61920065e-01
-7.32031226e-01 -7.60018349e-01 6.28675640e-01 6.20788693e-01
3.99889290e-01 -9.96474147e-01 -7.17411280e-01 4.89714414e-01
-4.44854200e-02 -7.38887727e-01 -4.95740443e-01 4.66014236e-01
-1.20255673e+00 -9.31668520e-01 -5.99502206e-01 -5.27068496e-01
-1.26049981e-01 3.37363958e-01 1.04055882e+00 -4.20152873e-01
-4.22750711e-01 4.06130791e-01 -7.61722863e-01 -4.59286660e-01
-9.02051330e-02 3.76939215e-02 1.26645461e-01 -1.16425171e-01
4.46379900e-01 -1.06443059e+00 -5.08701384e-01 4.97841120e-01
-8.43282342e-01 -6.33692205e-01 2.95077682e-01 9.20575857e-01
5.09755492e-01 2.34180838e-01 6.15175128e-01 -1.11957788e-01
1.11756301e+00 -7.46803880e-01 -3.08716714e-01 2.50283718e-01
-9.53648508e-01 -1.04918517e-01 8.00982237e-01 -1.08671069e+00
-5.13047814e-01 -2.65345633e-01 -3.96649092e-02 -8.53109181e-01
-3.28948162e-02 6.64005399e-01 6.96114779e-01 -3.43809068e-01
9.17213202e-01 4.74086165e-01 3.02739501e-01 -5.03804445e-01
3.57836373e-02 7.95021236e-01 5.51217675e-01 -6.25340581e-01
6.00599527e-01 3.77516389e-01 1.12908937e-01 -7.36195624e-01
-3.88628751e-01 -9.55509186e-01 -6.75719500e-01 -1.21531427e-01
3.28280687e-01 -4.25951660e-01 -4.22417760e-01 4.00266230e-01
-8.55366707e-01 2.43594497e-02 -4.36110288e-01 6.18162096e-01
-5.65389991e-01 4.26760554e-01 -4.18729335e-01 -7.63964057e-01
-7.64952481e-01 -6.44609094e-01 8.30822587e-01 7.33678713e-02
-6.71586335e-01 -1.07756126e+00 5.93077302e-01 -2.56446600e-01
9.88424003e-01 8.49969506e-01 7.87545800e-01 -1.20319045e+00
4.40633297e-01 -5.38517416e-01 -4.71256897e-02 3.97941679e-01
8.08182731e-02 4.09169525e-01 -5.34815967e-01 -4.18346077e-01
9.61161256e-02 1.80476196e-02 6.29969537e-01 4.09736603e-01
1.05883026e+00 -6.28116205e-02 -4.50876355e-01 6.77778304e-01
1.28407943e+00 7.24803269e-01 4.28030193e-01 8.22031021e-01
-6.01109788e-02 5.68325758e-01 7.54021108e-01 6.47228777e-01
2.47574508e-01 8.79570603e-01 1.21711433e-01 -1.03510469e-01
3.01067501e-01 4.27774101e-01 3.36310506e-01 9.86482441e-01
-5.67531824e-01 2.48263747e-01 -9.55152392e-01 5.52302599e-01
-2.11393762e+00 -1.39498377e+00 -2.14195207e-01 2.12058616e+00
7.41115093e-01 1.53147921e-01 5.87532043e-01 7.46873438e-01
7.22915471e-01 2.22270429e-01 -2.89206296e-01 -8.66189122e-01
-4.79525141e-02 7.42279708e-01 2.29728371e-01 -2.36973122e-01
-1.30332899e+00 3.71125758e-01 6.68047857e+00 1.15041125e+00
-1.32615638e+00 5.28325327e-02 9.77489278e-02 1.13811441e-01
2.79510975e-01 -2.12147400e-01 -4.81484532e-01 6.33849382e-01
1.24354398e+00 -6.29647732e-01 3.26908112e-01 1.14810741e+00
3.14855278e-02 1.66660801e-01 -8.86552274e-01 1.04254162e+00
8.31848159e-02 -1.13346624e+00 -3.91648650e-01 -2.07961529e-01
4.73573893e-01 1.52671918e-01 -7.26980641e-02 4.86948669e-01
-1.08430430e-01 -7.47825980e-01 6.53851926e-01 6.32168412e-01
6.06118917e-01 -7.57133782e-01 8.56889367e-01 6.39349669e-02
-1.74277341e+00 -2.95244098e-01 -1.69919580e-01 -2.76515156e-01
1.54019250e-02 7.18986750e-01 -3.32002044e-01 8.41169238e-01
1.10404074e+00 7.81577170e-01 -5.24797678e-01 1.25987399e+00
5.74415624e-01 5.65059066e-01 -6.48725390e-01 -9.99267101e-02
3.39070648e-01 -1.46687016e-01 6.01665258e-01 1.35829723e+00
8.53493929e-01 -2.85914689e-02 2.62248278e-01 4.21023101e-01
9.82090056e-01 2.15107128e-01 -5.61347425e-01 1.85375571e-01
6.90365672e-01 1.21154559e+00 -8.10947835e-01 -3.34782273e-01
-4.92072433e-01 6.92758143e-01 2.58197164e-04 2.78264899e-02
-9.12633896e-01 -8.94362330e-01 6.59400165e-01 2.67725624e-03
4.14769471e-01 -4.25408632e-01 4.78712464e-04 -9.34743583e-01
1.04025885e-01 -8.66891325e-01 8.73470545e-01 -6.48309529e-01
-1.59720492e+00 9.01203692e-01 5.70914090e-01 -1.75482714e+00
-5.10483623e-01 -6.18931890e-01 -8.48852336e-01 7.14555860e-01
-1.28576314e+00 -7.96824694e-01 -2.66640931e-01 8.61039162e-01
3.03768456e-01 -3.51796895e-01 9.20083582e-01 4.24238116e-01
-2.68967181e-01 4.76072729e-01 4.59672481e-01 -2.44337887e-01
8.62001896e-01 -1.10390818e+00 4.63839382e-01 4.52182055e-01
-1.38468757e-01 3.70306283e-01 6.64765060e-01 -3.77511859e-01
-1.06417203e+00 -8.85987520e-01 8.90455306e-01 -6.00342378e-02
1.17981386e+00 1.39121339e-01 -1.16476572e+00 1.47148356e-01
1.69566576e-03 2.63021946e-01 8.20464313e-01 -7.55878389e-02
-5.68938553e-01 -3.67421895e-01 -1.21105683e+00 3.55009884e-01
9.56913412e-01 -4.55329865e-01 -9.86105382e-01 1.48990497e-01
4.53770787e-01 -1.10445417e-01 -1.44996643e+00 6.66250587e-01
8.31362367e-01 -1.14661193e+00 1.36049449e+00 -3.85910273e-01
1.01062812e-01 -1.71691701e-01 -2.24770963e-01 -1.46390700e+00
-6.88959658e-01 -6.73836708e-01 -3.94920379e-01 1.14715374e+00
1.15848832e-01 -1.03977692e+00 2.27823675e-01 1.34103805e-01
-3.46070051e-01 -9.96842206e-01 -1.16028953e+00 -1.47793078e+00
-5.15181795e-02 -5.08640826e-01 8.02293897e-01 1.31357539e+00
2.72500068e-01 -2.09579825e-01 -1.31430611e-01 -3.46488237e-01
3.37551266e-01 3.83502156e-01 4.15036201e-01 -1.59187818e+00
-2.15696692e-01 -8.28675926e-01 -8.04830670e-01 -2.03015119e-01
-7.69253373e-02 -8.24370861e-01 -5.58836699e-01 -1.04908133e+00
-2.74931312e-01 -5.09879112e-01 -4.81271923e-01 5.95264792e-01
6.19627424e-02 1.25342170e-02 -1.83646921e-02 5.12077451e-01
-8.15415531e-02 2.95388311e-01 7.42505789e-01 5.08505031e-02
-4.75361198e-01 1.67893276e-01 -1.26336411e-01 5.64491570e-01
9.67983246e-01 -4.13334638e-01 -2.48021454e-01 3.22218001e-01
-1.20296814e-01 1.29355922e-01 2.26659283e-01 -1.35480809e+00
4.12332028e-01 -2.04798117e-01 3.27669591e-01 -8.34589779e-01
1.56225085e-01 -8.23442161e-01 5.12324810e-01 7.36599207e-01
-2.33581029e-02 8.36399078e-01 2.27236107e-01 3.36561382e-01
-5.08517504e-01 -1.00490652e-01 8.40989828e-01 1.10910937e-01
-8.10425222e-01 1.23432159e-01 -1.97319835e-01 -1.17342815e-01
1.13937759e+00 -6.88413739e-01 -3.43876481e-01 -1.21951543e-01
-7.28126943e-01 -1.30488858e-01 2.01398298e-01 5.32227039e-01
4.11977887e-01 -1.57444787e+00 -7.48717189e-01 1.44779250e-01
1.99818790e-01 -6.63144350e-01 -6.74330145e-02 1.09337103e+00
-4.62174267e-01 3.85848492e-01 -6.17156148e-01 -6.18396521e-01
-1.34778321e+00 8.98278236e-01 1.59128651e-01 -5.82468212e-01
-5.84630966e-01 2.96057343e-01 -5.47318339e-01 -3.05429518e-01
7.09646344e-02 -4.99327600e-01 -4.44293976e-01 4.12623823e-01
5.15736580e-01 1.00630820e+00 4.18373495e-01 -5.12132168e-01
-4.81438011e-01 1.03029370e+00 1.09973371e-01 7.17190802e-02
1.59920096e+00 2.51915962e-01 -1.74321532e-01 5.31906724e-01
1.26146996e+00 -3.78203839e-01 -4.79410738e-01 -3.07637781e-01
4.26241010e-01 -3.36774528e-01 -1.21357776e-01 -5.97538054e-01
-6.54089332e-01 4.42806154e-01 7.50195324e-01 8.41824114e-01
1.61792386e+00 -3.71002853e-01 7.67342985e-01 1.93461359e-01
3.53628397e-01 -1.22424781e+00 -2.04434283e-02 6.62058175e-01
1.04713237e+00 -8.34988832e-01 1.46088928e-01 -2.33897626e-01
-3.46738905e-01 1.58502984e+00 2.51177520e-01 -3.49341720e-01
1.03807890e+00 4.97369796e-01 -2.92882323e-01 1.00488693e-01
-9.12327826e-01 -2.31012866e-01 6.06956601e-01 6.80806339e-01
4.90234882e-01 -5.63231744e-02 -6.96053565e-01 5.39539456e-01
-2.90460497e-01 5.44131063e-02 -2.13310905e-02 1.23606598e+00
-3.12141031e-01 -9.91751075e-01 -6.68175995e-01 6.18911564e-01
-4.86461103e-01 3.09807882e-02 -2.02603623e-01 9.81279314e-01
1.58724368e-01 7.94391751e-01 2.86680043e-01 -6.25953734e-01
6.28909945e-01 1.37622222e-01 3.35174471e-01 -9.94764119e-02
-1.10606015e+00 -3.42777744e-02 8.36453587e-02 -6.28170967e-01
-4.77501929e-01 -7.91468561e-01 -8.64320934e-01 -4.25505400e-01
-3.58886003e-01 1.70238584e-01 6.76318169e-01 7.04357386e-01
4.10797030e-01 2.40512639e-01 8.05447042e-01 -1.11886692e+00
-7.64507055e-01 -1.03228259e+00 -5.95472813e-01 4.10874337e-01
1.20101580e-02 -9.00825262e-01 -6.18527651e-01 -2.01685190e-01] | [7.251235008239746, 3.2935171127319336] |
d4e8751e-64a0-426d-b4a3-49893fc34006 | how-to-design-a-three-stage-architecture-for | 2106.03932 | null | https://arxiv.org/abs/2106.03932v2 | https://arxiv.org/pdf/2106.03932v2.pdf | How to Design a Three-Stage Architecture for Audio-Visual Active Speaker Detection in the Wild | Successful active speaker detection requires a three-stage pipeline: (i) audio-visual encoding for all speakers in the clip, (ii) inter-speaker relation modeling between a reference speaker and the background speakers within each frame, and (iii) temporal modeling for the reference speaker. Each stage of this pipeline plays an important role for the final performance of the created architecture. Based on a series of controlled experiments, this work presents several practical guidelines for audio-visual active speaker detection. Correspondingly, we present a new architecture called ASDNet, which achieves a new state-of-the-art on the AVA-ActiveSpeaker dataset with a mAP of 93.5% outperforming the second best with a large margin of 4.7%. Our code and pretrained models are publicly available. | ['Gerhard Rigoll', 'Maja Taseska', 'Okan Köpüklü'] | 2021-06-07 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Kopuklu_How_To_Design_a_Three-Stage_Architecture_for_Audio-Visual_Active_Speaker_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Kopuklu_How_To_Design_a_Three-Stage_Architecture_for_Audio-Visual_Active_Speaker_ICCV_2021_paper.pdf | iccv-2021-1 | ['audio-visual-active-speaker-detection'] | ['computer-vision'] | [ 2.34691337e-01 1.55983418e-01 -2.94423811e-02 -4.16693598e-01
-1.48829985e+00 -3.95214736e-01 7.04227567e-01 6.66781068e-02
-2.69040406e-01 5.34263346e-03 3.88418436e-01 -4.34897952e-02
2.76420176e-01 -9.71511900e-02 -5.08373141e-01 -7.16561019e-01
-3.58366221e-01 2.57495791e-01 6.33521318e-01 1.50957301e-01
1.24238901e-01 5.51758349e-01 -1.81647217e+00 7.25091100e-01
2.68720806e-01 1.21231639e+00 -9.28441714e-03 1.13147032e+00
-1.58097073e-01 1.01621330e+00 -8.05586934e-01 -2.45330855e-01
-1.12833045e-01 -4.94600087e-01 -6.10248327e-01 1.18735515e-01
7.25673079e-01 -3.43210399e-01 -1.80163488e-01 6.86110139e-01
9.24059927e-01 1.73798427e-01 3.07416797e-01 -1.15293634e+00
-9.76663008e-02 1.01155913e+00 -5.29098034e-01 6.38380766e-01
5.61450183e-01 2.46767253e-01 9.90657270e-01 -1.27104342e+00
1.75076678e-01 1.42917502e+00 5.06571531e-01 4.71094251e-01
-9.85702038e-01 -9.04049456e-01 3.18165898e-01 3.62881809e-01
-1.61453140e+00 -1.41227019e+00 7.75354087e-01 -4.86741334e-01
9.90628541e-01 3.75706345e-01 8.19355369e-01 1.05138588e+00
-2.21943468e-01 9.22915041e-01 6.16404057e-01 -5.53711653e-01
3.01236749e-01 2.30874896e-01 2.21290395e-01 4.07336771e-01
-5.18608570e-01 1.04521446e-01 -1.34669173e+00 -2.15149164e-01
3.15878540e-01 -6.77788079e-01 -1.11916155e-01 -2.09364310e-01
-8.78720760e-01 7.22934425e-01 9.41015258e-02 2.20714286e-01
-2.28944093e-01 1.79874361e-01 3.77710789e-01 9.34503227e-03
5.07625937e-01 -1.05668046e-01 1.58817351e-01 -3.57865125e-01
-1.35399032e+00 4.23740111e-02 7.63858259e-01 6.71398044e-01
2.49630585e-01 2.89574504e-01 -1.55557737e-01 9.58296299e-01
9.17573154e-01 4.44620192e-01 2.18114734e-01 -8.08860302e-01
2.61550486e-01 2.44310826e-01 -8.45035389e-02 -6.05080307e-01
-8.87499750e-02 -5.31074643e-01 -2.80832171e-01 3.83963943e-01
3.72026652e-01 1.08142428e-01 -8.60605955e-01 1.60013175e+00
4.77035612e-01 4.75059599e-01 -1.56664953e-01 7.90756404e-01
1.18260419e+00 8.74674022e-01 2.05756158e-01 -4.35922354e-01
1.48405612e+00 -1.16218007e+00 -8.64366710e-01 -5.08489370e-01
-7.26450458e-02 -9.04773116e-01 8.18452239e-01 3.85955185e-01
-1.45042455e+00 -7.62073874e-01 -1.20374453e+00 4.57655415e-02
1.28189743e-01 1.89096898e-01 3.23655725e-01 8.95555675e-01
-1.41444433e+00 -2.96356469e-01 -9.74659324e-01 -3.67073685e-01
3.00064653e-01 3.85223299e-01 -1.51137888e-01 4.15115923e-01
-9.92391706e-01 4.91514564e-01 -2.39155963e-01 2.07833469e-01
-1.56005788e+00 -6.84974015e-01 -7.56105304e-01 2.47895375e-01
1.91205293e-01 -3.87365371e-01 1.54776359e+00 -1.12585592e+00
-1.74367344e+00 1.10877061e+00 -5.41100979e-01 -6.81622446e-01
7.09770083e-01 -3.14076215e-01 -6.76284134e-01 3.39583904e-01
-2.26148203e-01 6.78588927e-01 1.17982996e+00 -1.19625247e+00
-8.02831590e-01 -3.19837868e-01 -3.78058702e-02 2.07161844e-01
-1.38936877e-01 7.41165638e-01 -9.31847632e-01 -5.64303100e-01
7.97573179e-02 -7.23842025e-01 2.24312201e-01 2.02328578e-01
-2.21789494e-01 -2.68662393e-01 7.08112121e-01 -7.83948123e-01
1.20862985e+00 -2.63423347e+00 -4.93810177e-02 -3.21649872e-02
2.21600339e-01 2.01721564e-01 -1.23392105e-01 1.96137846e-01
-1.93616137e-01 -2.79899061e-01 1.13785096e-01 -8.31894457e-01
-5.02528735e-02 -5.89133739e-01 -5.16962945e-01 5.31373739e-01
-1.79084707e-02 4.14668322e-01 -5.15296102e-01 -4.64449853e-01
1.76215485e-01 8.40313733e-01 -3.39134425e-01 4.65835899e-01
3.04702669e-01 2.09856108e-01 1.70477033e-02 5.20316362e-01
5.77913165e-01 7.12332204e-02 -1.35614192e-02 5.51240072e-02
-3.31192523e-01 6.97676301e-01 -1.46728194e+00 1.69092321e+00
-1.89141810e-01 9.98356938e-01 7.35424280e-01 -3.81798685e-01
8.96989167e-01 6.91771030e-01 3.28656465e-01 -4.82936710e-01
1.48969376e-03 8.66609737e-02 1.39555246e-01 -4.29998040e-01
4.08698052e-01 1.12370618e-01 2.93190837e-01 4.02781129e-01
2.06582040e-01 3.16896856e-01 1.26995414e-01 4.66468096e-01
9.22037601e-01 -3.44975710e-01 1.32703960e-01 -2.67880797e-01
8.78283262e-01 -5.40601969e-01 4.47074503e-01 8.03941131e-01
-7.06874728e-01 7.50961483e-01 4.52778280e-01 -2.45492086e-01
-5.62052667e-01 -1.23551333e+00 -4.16917400e-03 1.49236584e+00
-1.19086422e-01 -5.29857099e-01 -8.75115216e-01 -2.45715246e-01
-3.47209781e-01 5.86890221e-01 -6.95380926e-01 4.24940214e-02
-6.44552231e-01 -2.62848318e-01 6.36095285e-01 6.52956843e-01
2.41969585e-01 -1.03690886e+00 -4.63075250e-01 1.38527647e-01
-2.73559868e-01 -1.12774909e+00 -6.39454603e-01 6.49399087e-02
-4.92723167e-01 -7.26642787e-01 -6.69269383e-01 -8.43773663e-01
1.61609903e-01 2.79129356e-01 1.08573043e+00 -2.62595624e-01
-1.15358800e-01 4.00155842e-01 -1.27807707e-01 -5.05415559e-01
-7.09866941e-01 -7.83353746e-02 -6.81847483e-02 3.20644945e-01
2.12149203e-01 -3.59606624e-01 -6.92234278e-01 2.97744840e-01
-2.73684055e-01 -4.03267331e-02 2.90382385e-01 3.19353223e-01
4.27361935e-01 -3.59667510e-01 3.92244518e-01 -2.15652794e-01
2.57934511e-01 2.40503456e-02 -6.48615539e-01 8.66379440e-02
-2.07204651e-02 -5.37339807e-01 -6.65823594e-02 -6.61508381e-01
-1.04270101e+00 5.30902684e-01 -4.35165673e-01 -4.50049788e-01
-1.39372915e-01 -3.00836153e-02 -4.75727260e-01 1.80620357e-01
6.59779847e-01 2.75748014e-01 9.02628750e-02 -5.33543348e-01
1.91161260e-01 8.58778596e-01 6.87047839e-01 2.88235699e-03
3.86386633e-01 5.83773673e-01 -5.42791724e-01 -1.15467966e+00
-6.65550530e-01 -6.29096389e-01 -5.70275486e-01 -6.76958561e-01
9.20465946e-01 -1.45943880e+00 -7.90600836e-01 7.31489837e-01
-1.06357479e+00 -2.56978959e-01 -1.62686601e-01 5.69115162e-01
-2.28122339e-01 2.15560228e-01 -4.97711241e-01 -1.33401418e+00
-4.76741165e-01 -1.34135544e+00 1.00763011e+00 4.46318574e-02
-3.19941819e-01 -5.10932803e-01 2.32467011e-01 6.82363808e-01
5.20213068e-01 -3.50701988e-01 1.99940100e-01 -8.37554276e-01
-5.45508564e-01 -1.28159314e-01 1.78590328e-01 2.99608409e-01
-1.73932150e-01 3.30912173e-01 -1.70011795e+00 -3.75539660e-01
5.24335951e-02 -1.39873311e-01 9.51564908e-01 6.21279120e-01
7.47978091e-01 -6.09754100e-02 -2.83186972e-01 4.51177150e-01
5.59242785e-01 4.38605279e-01 5.31707227e-01 7.17902556e-02
3.67484212e-01 6.57759964e-01 3.60315025e-01 3.88927966e-01
2.78365970e-01 8.97725165e-01 4.65073854e-01 -1.37644500e-01
-4.84518439e-01 6.99276105e-03 8.00539374e-01 7.92560697e-01
2.36167580e-01 -3.01277667e-01 -1.06290579e+00 5.91537952e-01
-1.46981788e+00 -1.10341895e+00 -4.26801108e-02 2.17709708e+00
5.46147466e-01 3.28106016e-01 5.91999769e-01 4.12387282e-01
9.76502955e-01 4.53306139e-01 -3.96801889e-01 -2.81734705e-01
-4.76703830e-02 1.21809160e-02 -2.79515445e-01 8.06349218e-01
-1.40648508e+00 7.82886684e-01 7.38449907e+00 7.15581894e-01
-1.34337366e+00 1.45139441e-01 5.33245921e-01 -4.77490664e-01
1.71500444e-01 -2.80036986e-01 -1.07500541e+00 3.60817999e-01
1.27363992e+00 -4.00238857e-03 2.30426580e-01 9.21574831e-01
4.04609174e-01 -8.06390718e-02 -1.06907248e+00 1.17198896e+00
5.00584602e-01 -1.05714202e+00 -1.81621149e-01 2.85972375e-02
8.02122951e-02 1.50238350e-01 -1.55532388e-02 1.95438698e-01
-3.74163948e-02 -8.59693527e-01 1.40933383e+00 1.65036470e-01
6.43415809e-01 -7.52968669e-01 2.76043266e-01 8.81071240e-02
-1.50903893e+00 -8.00111815e-02 6.69260919e-02 1.99634135e-01
3.43527883e-01 4.31401521e-01 -7.53467143e-01 6.49830922e-02
9.55790758e-01 3.15923661e-01 -5.27752042e-01 1.15707195e+00
-3.60047758e-01 1.01019692e+00 -3.59208047e-01 3.00506890e-01
-1.33275479e-01 5.72805882e-01 8.69654238e-01 1.47654629e+00
1.05682008e-01 -1.79373428e-01 -1.09976240e-01 5.87992251e-01
9.10651088e-02 6.78748712e-02 -2.28825007e-02 3.75663266e-02
6.42773092e-01 1.34450245e+00 -6.18160307e-01 -2.85612375e-01
-3.73622835e-01 6.28879189e-01 1.07592642e-01 2.29625568e-01
-1.05797815e+00 -1.91536397e-01 7.04640210e-01 3.19607049e-01
4.72940803e-01 5.03956005e-02 6.39743060e-02 -7.05220520e-01
-2.79766358e-02 -9.59818363e-01 5.43026090e-01 -7.19440818e-01
-7.80914068e-01 8.36868465e-01 -3.54055852e-01 -1.09622908e+00
-4.40210372e-01 -2.22846538e-01 -7.74984598e-01 9.95272398e-01
-1.09705198e+00 -1.23556125e+00 -4.63073820e-01 5.80621481e-01
6.77849174e-01 -3.55335981e-01 8.70324433e-01 4.46897238e-01
-7.09052801e-01 8.26617956e-01 -3.54202896e-01 2.91871190e-01
8.10993433e-01 -9.79698479e-01 6.81846082e-01 1.09086466e+00
3.76343578e-01 4.96289760e-01 7.02682853e-01 -1.01485476e-01
-1.29827023e+00 -6.77804470e-01 1.03073728e+00 -3.12685847e-01
5.88228762e-01 -9.48184907e-01 -7.77552545e-01 6.88769281e-01
5.14609396e-01 -1.54592827e-01 1.00434780e+00 2.39960611e-01
-5.41360617e-01 -4.24628645e-01 -8.28112543e-01 7.56263584e-02
6.64595783e-01 -7.59451389e-01 -5.22188663e-01 -8.08163136e-02
5.17296612e-01 -5.23035526e-01 -4.01112407e-01 1.40438169e-01
8.90258789e-01 -1.18985140e+00 1.06267536e+00 -1.01755828e-01
7.98689127e-02 -3.69362295e-01 1.26066338e-02 -7.94669092e-01
-4.21831578e-01 -7.08113372e-01 -4.87389565e-01 1.67840958e+00
5.33544660e-01 -1.54661864e-01 4.98393297e-01 3.59017044e-01
-2.33086810e-01 -3.99519742e-01 -1.22180927e+00 -5.20331144e-01
-4.47331846e-01 -8.19929898e-01 2.26290554e-01 6.17222905e-01
-9.29462761e-02 5.02024055e-01 -3.22108030e-01 2.91797459e-01
4.84063596e-01 -1.30547091e-01 9.18664277e-01 -1.05709088e+00
-4.13315386e-01 -4.38470393e-01 -5.03098309e-01 -1.20698524e+00
2.45192442e-02 -4.22260046e-01 2.26524726e-01 -1.43912411e+00
3.78435254e-01 2.20602855e-01 -3.51914734e-01 3.70058656e-01
-2.26352606e-02 3.56913209e-01 9.17313322e-02 3.41996253e-01
-7.71585345e-01 3.96100283e-01 5.87432027e-01 -2.63150066e-01
-4.93292212e-01 1.11222066e-01 -5.96670747e-01 7.07845688e-01
3.54938895e-01 -3.92959952e-01 -3.44076514e-01 -3.77545029e-01
-1.32725000e-01 1.18399799e-01 2.77791798e-01 -1.11313617e+00
4.97233450e-01 3.27645749e-01 3.35062414e-01 -9.58450496e-01
8.63318145e-01 -5.48789918e-01 7.41273314e-02 3.71108860e-01
-5.46327293e-01 -1.89795673e-01 4.52323645e-01 3.30412060e-01
-4.72095639e-01 1.33838058e-01 1.18740690e+00 2.96671182e-01
-6.08746827e-01 -8.74850228e-02 -7.00683713e-01 3.59393992e-02
8.45620155e-01 -9.26577449e-02 -2.99981117e-01 -6.40684605e-01
-9.84830678e-01 -5.53709269e-03 -7.68738007e-03 6.34600222e-01
3.28745723e-01 -1.02442181e+00 -9.03077841e-01 3.42500985e-01
1.11149244e-01 -4.13901776e-01 6.41190350e-01 8.82500708e-01
-3.72991771e-01 3.07989866e-01 2.17050970e-01 -1.00537741e+00
-1.87920725e+00 1.94619149e-01 5.64879596e-01 1.03515282e-01
-5.79387784e-01 1.36723530e+00 4.46781009e-01 2.62222648e-01
9.30956721e-01 -5.91006279e-02 -3.87185156e-01 4.61648077e-01
1.17867029e+00 3.89775842e-01 2.08038941e-01 -1.20638967e+00
-9.14292216e-01 2.35917270e-01 -1.65368333e-01 -6.54298007e-01
1.04776645e+00 -2.99198747e-01 1.80162802e-01 6.94026470e-01
9.43085551e-01 3.91373217e-01 -1.37280011e+00 -1.73246011e-01
-2.91000336e-01 -4.03484285e-01 4.52987581e-01 -8.03472817e-01
-1.09217215e+00 1.05364645e+00 1.07557356e+00 2.63008863e-01
1.13889158e+00 3.48236382e-01 1.71912730e-01 -1.59768611e-01
-1.22323021e-01 -7.62834489e-01 1.60200685e-01 3.69101465e-01
1.08225560e+00 -9.45719242e-01 -1.62460402e-01 -4.87665534e-01
-6.96067035e-01 9.14177775e-01 3.63737673e-01 3.48819673e-01
7.79966593e-01 6.41922772e-01 6.20399356e-01 -4.13733423e-02
-9.96498227e-01 -7.74026811e-02 5.21120071e-01 3.76986474e-01
7.78471231e-01 -1.62229896e-01 3.54762822e-01 6.56989098e-01
-2.41240159e-01 -3.93893301e-01 -4.55611199e-02 6.82075620e-01
-5.68554997e-01 -6.55666828e-01 -5.13490558e-01 -3.40901107e-01
-6.68889999e-01 -1.59885567e-02 -6.89439178e-01 4.92712885e-01
-2.45278999e-01 1.43677890e+00 4.17622924e-01 -2.48198330e-01
4.35109735e-01 2.77954549e-01 2.70192146e-01 -4.61033672e-01
-9.76340115e-01 8.92189145e-01 1.48890585e-01 -6.46332085e-01
-5.05400836e-01 -9.28662241e-01 -9.40837502e-01 -1.78868949e-01
-2.19572067e-01 1.24730095e-01 7.67413974e-01 5.22952914e-01
3.81320685e-01 6.64804578e-01 6.60338521e-01 -8.53472471e-01
-1.57467589e-01 -1.36319649e+00 -5.98953545e-01 -3.24796140e-02
4.99950349e-01 -3.33496660e-01 -7.65954673e-01 2.35270277e-01] | [14.421168327331543, 5.19271993637085] |
6ab562d2-42c0-46e9-bacd-a74efbcbd0a0 | template-based-approach-to-zero-shot-intent | 2206.10914 | null | https://arxiv.org/abs/2206.10914v1 | https://arxiv.org/pdf/2206.10914v1.pdf | Template-based Approach to Zero-shot Intent Recognition | The recent advances in transfer learning techniques and pre-training of large contextualized encoders foster innovation in real-life applications, including dialog assistants. Practical needs of intent recognition require effective data usage and the ability to constantly update supported intents, adopting new ones, and abandoning outdated ones. In particular, the generalized zero-shot paradigm, in which the model is trained on the seen intents and tested on both seen and unseen intents, is taking on new importance. In this paper, we explore the generalized zero-shot setup for intent recognition. Following best practices for zero-shot text classification, we treat the task with a sentence pair modeling approach. We outperform previous state-of-the-art f1-measure by up to 16\% for unseen intents, using intent labels and user utterances and without accessing external sources (such as knowledge bases). Further enhancement includes lexicalization of intent labels, which improves performance by up to 7\%. By using task transferring from other sentence pair tasks, such as Natural Language Inference, we gain additional improvements. | ['Irina Piontkovskaya', 'Andrey Bout', 'Valentin Malykh', 'Ekaterina Artemova', 'Pavel Burnyshev', 'Dmitry Lamanov'] | 2022-06-22 | null | null | null | null | ['intent-recognition', 'sentence-pair-modeling'] | ['natural-language-processing', 'natural-language-processing'] | [ 4.02055532e-01 3.49523902e-01 -8.37332234e-02 -8.06045175e-01
-6.51447296e-01 -3.51081401e-01 7.80031919e-01 2.76793063e-01
-8.74972939e-01 8.06666017e-01 4.22046751e-01 -4.03154731e-01
2.32302189e-01 -6.12359762e-01 -2.84173012e-01 -2.43345633e-01
9.81645510e-02 7.30742574e-01 1.11329854e-01 -6.71959519e-01
2.91662335e-01 -2.74517894e-01 -1.60339415e+00 4.29017931e-01
7.18583643e-01 7.77152956e-01 3.95668000e-01 1.01784658e+00
-3.44439745e-01 9.67257619e-01 -6.69638216e-01 -5.14959872e-01
-1.89404398e-01 -3.49405140e-01 -1.20706987e+00 -1.06019683e-01
3.24927211e-01 -8.13782275e-01 -2.09571317e-01 5.71852386e-01
5.65179288e-01 6.04539156e-01 5.59430897e-01 -1.03779626e+00
-8.64809275e-01 6.27167881e-01 7.14342743e-02 1.89279497e-01
7.55365610e-01 1.46930721e-02 1.19296813e+00 -9.10049856e-01
5.25580227e-01 1.03508437e+00 5.00944316e-01 9.57556129e-01
-1.07345116e+00 -4.04555321e-01 2.41241217e-01 3.95220816e-01
-8.62315059e-01 -9.13813055e-01 5.31778991e-01 -3.22314918e-01
1.65907168e+00 2.75678366e-01 -7.97470286e-02 1.57788253e+00
-3.45577672e-02 9.37938988e-01 5.50843477e-01 -9.18410778e-01
4.17477906e-01 4.35389549e-01 6.80731595e-01 5.33438146e-01
-1.79475591e-01 -2.96715736e-01 -5.89989007e-01 -2.83620477e-01
1.70643963e-02 2.14199260e-01 -2.48014495e-01 -6.71186820e-02
-9.64061618e-01 9.79226232e-01 1.26220599e-01 4.10629928e-01
-1.95286676e-01 -3.09233934e-01 7.66820252e-01 6.05807602e-01
7.01854587e-01 6.04331493e-01 -6.72674239e-01 -6.37384951e-01
-7.12160349e-01 -2.88641918e-02 1.28845215e+00 1.16768003e+00
8.55959594e-01 -3.37244868e-01 -4.18396890e-01 1.10822558e+00
2.70371497e-01 3.19421798e-01 9.64576423e-01 -5.80364466e-01
6.00705266e-01 4.92520511e-01 1.64737478e-01 -3.85985106e-01
-3.92044514e-01 5.66169843e-02 -5.69244266e-01 -3.64290267e-01
1.11468844e-01 -4.01178271e-01 -9.45574343e-01 1.79212117e+00
7.18416274e-02 1.26652345e-01 4.21553046e-01 3.97711456e-01
8.57293189e-01 6.93888366e-01 1.65663332e-01 -3.36676270e-01
1.53279161e+00 -9.62899625e-01 -8.64185691e-01 -5.64518988e-01
1.12717044e+00 -6.29176855e-01 1.23561311e+00 9.38680694e-02
-6.20677114e-01 -6.18378818e-01 -1.01412809e+00 -3.76883179e-01
-6.32690430e-01 -1.87912211e-01 4.75986868e-01 5.74432492e-01
-1.12343645e+00 5.34152091e-01 -6.79909348e-01 -8.78218114e-01
4.05527689e-02 3.71069521e-01 -2.36604020e-01 -1.27967834e-01
-1.52495277e+00 1.07070196e+00 3.48747343e-01 -2.54101872e-01
-5.10914981e-01 -6.75876677e-01 -1.19921756e+00 2.10874856e-01
5.85086405e-01 -6.14402652e-01 1.88972747e+00 -3.49896520e-01
-1.74005103e+00 6.67379379e-01 -2.96731919e-01 -5.79392433e-01
9.20451060e-02 -3.61250371e-01 -3.35596174e-01 -2.27648482e-01
1.20445669e-01 6.00708187e-01 6.84850514e-01 -7.10885346e-01
-6.92904770e-01 -3.25851291e-01 2.86438704e-01 4.25187171e-01
-8.55492353e-01 3.80574390e-02 -1.89287931e-01 -1.18982263e-01
-4.58220810e-01 -7.49172509e-01 3.34470272e-02 -2.78270692e-01
-1.07018843e-01 -5.25053561e-01 1.06280077e+00 -6.74176633e-01
1.18735993e+00 -2.03841424e+00 1.32592232e-03 -4.78618890e-01
1.33285120e-01 3.90671998e-01 -2.00876091e-02 8.29736888e-01
2.57646441e-01 -7.32114986e-02 -1.80049062e-01 -9.31841433e-01
2.18851268e-01 4.23963964e-01 -4.84422088e-01 -1.17055401e-01
1.73959568e-01 8.65265489e-01 -1.06917369e+00 -4.02366996e-01
5.04548311e-01 1.90543279e-01 -5.06669343e-01 6.56154871e-01
-1.45386502e-01 8.20209309e-02 -1.89036086e-01 1.64718390e-01
6.35392517e-02 -4.09996122e-01 1.37129307e-01 1.01980321e-01
6.87161759e-02 6.74410403e-01 -5.86533785e-01 2.01407409e+00
-1.03284371e+00 6.14206433e-01 -8.79683420e-02 -8.11566889e-01
9.97961283e-01 7.06059396e-01 1.41608387e-01 -5.74337304e-01
5.81721403e-02 -1.10865802e-01 -9.42611024e-02 -6.64273381e-01
7.60726452e-01 -3.10720891e-01 -4.02386636e-01 9.04490769e-01
7.23542869e-01 1.46898240e-01 2.13663876e-01 4.40251142e-01
1.23639059e+00 -1.62728906e-01 6.92211151e-01 -5.88102033e-04
3.42675745e-01 -2.22693622e-01 8.41742102e-03 9.17405367e-01
-2.37284184e-01 3.69920045e-01 1.71408415e-01 -4.00451630e-01
-9.00321305e-01 -6.71713352e-01 -9.65879336e-02 1.88128114e+00
-8.11281130e-02 -4.92847800e-01 -5.46351671e-01 -8.77018988e-01
-1.35298938e-01 1.39819944e+00 -5.70889831e-01 -4.17345345e-01
-4.42826152e-01 -4.53539073e-01 3.29152852e-01 6.40454948e-01
3.97498071e-01 -1.17958534e+00 -7.34252810e-01 3.80446792e-01
-3.27120543e-01 -1.12722838e+00 -5.16439676e-01 5.62510729e-01
-6.13521636e-01 -6.49965584e-01 -6.68716550e-01 -7.36612439e-01
3.90466183e-01 2.79703498e-01 1.08951771e+00 -9.68400016e-02
-1.49016574e-01 5.50835669e-01 -7.97331035e-01 -3.74258548e-01
-5.48099160e-01 3.36385250e-01 2.08904907e-01 -1.14279836e-01
7.89449751e-01 -4.92584169e-01 -2.07997069e-01 1.45930111e-01
-7.16516912e-01 1.40681550e-01 4.75328207e-01 1.15733182e+00
-4.17772591e-01 -5.04164875e-01 8.91307175e-01 -1.14561605e+00
9.88329589e-01 -4.55861419e-01 6.04358725e-02 4.65228409e-01
-7.14858294e-01 2.68613368e-01 7.31249988e-01 -3.22535902e-01
-1.54077125e+00 -3.69747877e-01 -3.94070506e-01 -1.50811404e-01
-4.77128536e-01 2.84875274e-01 9.53556001e-02 3.50892007e-01
7.66912580e-01 1.73110053e-01 7.44170249e-02 -4.31259096e-01
6.24598324e-01 1.53736126e+00 3.92991006e-01 -3.04975957e-01
1.09158128e-01 1.29726216e-01 -8.92896056e-01 -1.09383595e+00
-9.80958879e-01 -9.93877172e-01 -7.04827905e-01 4.59472947e-02
7.89278209e-01 -6.60733640e-01 -5.45662582e-01 2.78435141e-01
-1.43757796e+00 -4.01004314e-01 -2.95869648e-01 3.20700586e-01
-5.83626151e-01 4.57396448e-01 -8.20727050e-01 -1.03574491e+00
-6.59761727e-01 -9.02803719e-01 1.33378327e+00 1.56065390e-01
-8.05619061e-01 -1.19702780e+00 2.86265492e-01 4.96713102e-01
5.35912216e-01 -6.90685630e-01 8.34777355e-01 -1.41828799e+00
1.53196275e-01 -3.98328125e-01 -3.83849931e-03 3.40088993e-01
3.98756623e-01 -5.81935287e-01 -1.31566703e+00 -3.51626456e-01
4.82499480e-01 -7.83297360e-01 6.07959449e-01 -9.03857797e-02
5.81738532e-01 -3.67602736e-01 -4.21300530e-01 2.83307731e-02
9.19484258e-01 5.05999684e-01 3.93620968e-01 3.81963030e-02
3.25211316e-01 8.09179068e-01 5.16069651e-01 6.25645101e-01
4.66722369e-01 7.88969994e-01 -7.74687976e-02 1.88400790e-01
1.38021916e-01 -2.07947612e-01 2.70879775e-01 9.70424056e-01
2.73095101e-01 -5.44558108e-01 -8.21519196e-01 3.35393697e-01
-2.02660728e+00 -9.00090694e-01 5.86843908e-01 2.16549802e+00
9.92560804e-01 1.68064699e-01 -1.45532429e-01 -2.39709780e-01
5.59092760e-01 3.12215924e-01 -5.60285985e-01 -6.11546874e-01
5.56275010e-01 2.72151470e-01 1.97952725e-02 7.38743007e-01
-1.15610695e+00 9.72019017e-01 6.15044785e+00 4.50447053e-01
-9.72681642e-01 3.61138731e-01 5.07513523e-01 -4.87784594e-02
-7.91334435e-02 -9.94328484e-02 -1.01055443e+00 4.83772665e-01
1.27122307e+00 -3.37001562e-01 2.94488549e-01 1.09808648e+00
-1.27870873e-01 -2.83212233e-02 -1.49078262e+00 8.16638052e-01
5.69420755e-01 -8.82930100e-01 -2.37256810e-01 -2.34248430e-01
3.00969660e-01 3.85244861e-02 -3.33483905e-01 1.15766025e+00
3.31417680e-01 -6.32875979e-01 1.11023244e-02 3.98523599e-01
6.84478760e-01 -2.50870347e-01 1.02504516e+00 6.22053862e-01
-8.79396200e-01 -4.55296114e-02 -3.87584925e-01 -3.56272966e-01
4.04479623e-01 1.46049708e-01 -1.31283832e+00 4.96082991e-01
3.69124115e-01 7.58146763e-01 -3.46055835e-01 3.10701013e-01
-1.14245728e-01 4.30784464e-01 -1.51895732e-01 -4.72275257e-01
9.87768248e-02 6.14311323e-02 1.87867001e-01 1.23486137e+00
6.49843141e-02 1.14132367e-01 2.09761053e-01 5.10238767e-01
-9.22701061e-02 1.10695831e-01 -9.77420032e-01 -4.20531109e-02
3.42077613e-01 1.22497308e+00 -3.38098526e-01 -7.07631528e-01
-7.74351239e-01 1.46411562e+00 6.30158603e-01 2.59632707e-01
-5.85409403e-01 -6.23959839e-01 5.93305469e-01 -3.51051480e-01
2.46693030e-01 -1.43283457e-01 5.43985851e-02 -1.27621400e+00
-8.27792566e-03 -5.39592206e-01 3.97241443e-01 -6.40375078e-01
-1.46192026e+00 7.37502158e-01 1.74922734e-01 -8.73758376e-01
-9.81178820e-01 -6.76724792e-01 -8.35602343e-01 6.80788636e-01
-1.20123458e+00 -8.49155605e-01 -1.62669808e-01 2.44070023e-01
1.25713277e+00 -1.37442037e-01 1.33452153e+00 2.10365713e-01
-4.12839592e-01 5.94375491e-01 -7.78939296e-03 9.46285054e-02
1.08989727e+00 -1.25665295e+00 7.25946546e-01 4.03378665e-01
1.51794508e-01 7.67871141e-01 7.74802327e-01 -6.12165153e-01
-1.19300032e+00 -8.41258943e-01 1.30145931e+00 -5.87676287e-01
7.48450398e-01 -8.52779269e-01 -1.06751621e+00 8.89157951e-01
4.55630183e-01 -4.51572537e-01 1.06293631e+00 7.20671415e-01
-2.82941043e-01 1.80473551e-01 -9.44696546e-01 8.44629467e-01
1.08512366e+00 -7.51309335e-01 -1.21115482e+00 5.13906062e-01
1.18661904e+00 -1.83806613e-01 -6.97201252e-01 2.29205549e-01
4.03729498e-01 -7.55860627e-01 7.35357821e-01 -9.52575147e-01
2.46227607e-01 4.29216146e-01 -1.97663292e-01 -1.33515728e+00
-1.13482937e-01 -5.68006158e-01 -4.60757703e-01 1.09485626e+00
5.15100241e-01 -6.96117759e-01 6.17727399e-01 1.06421196e+00
-3.78945649e-01 -6.44877017e-01 -9.70509648e-01 -6.64815366e-01
-2.48363495e-01 -4.06153142e-01 2.59153098e-01 8.63649368e-01
8.74231279e-01 1.21595204e+00 -6.43811107e-01 -3.63142252e-01
2.67491281e-01 -8.27112421e-02 7.59919882e-01 -1.17214346e+00
-4.13638026e-01 -8.21099579e-02 -3.58840734e-01 -1.33093333e+00
3.43126416e-01 -7.52564251e-01 3.64537358e-01 -1.42546511e+00
1.47276372e-01 2.34115180e-02 -1.23895824e-01 6.83786809e-01
-4.72290069e-01 -1.71648473e-01 7.79863521e-02 2.06785932e-01
-9.80161548e-01 9.31685627e-01 6.74672663e-01 -1.66716218e-01
-3.30468982e-01 2.31258087e-02 -6.99643850e-01 5.20831823e-01
7.73261011e-01 -2.20805898e-01 -6.75165415e-01 -2.16268480e-01
-1.87862903e-01 2.38556229e-03 -1.80428237e-01 -9.70991135e-01
5.13370454e-01 2.91039437e-01 -1.57934144e-01 -2.67980754e-01
8.01228225e-01 -6.54456317e-01 -5.51750064e-01 3.66943002e-01
-7.23377228e-01 -2.21741349e-01 1.21118933e-01 6.38504922e-01
-7.61630610e-02 -7.63495564e-01 3.18101972e-01 -2.26355016e-01
-1.04493892e+00 -8.09369329e-03 -5.89823842e-01 1.20218188e-01
1.11732471e+00 -4.79219705e-02 -5.90312958e-01 -7.30313480e-01
-5.77964842e-01 3.82824898e-01 1.99628949e-01 6.16407990e-01
5.17619014e-01 -9.35514688e-01 -4.05640215e-01 3.85670960e-01
6.20994508e-01 -2.58179188e-01 3.65951568e-01 5.96863210e-01
1.11587003e-01 7.78737903e-01 7.12827817e-02 -4.48096216e-01
-1.23503149e+00 6.20074809e-01 -6.47867694e-02 -2.81752259e-01
-6.45421326e-01 7.72583306e-01 1.02438636e-01 -7.81522453e-01
4.12057966e-01 -1.86266840e-01 -2.89681137e-01 1.24529637e-01
8.65836263e-01 2.19302043e-01 3.00479472e-01 -2.28041038e-01
-2.61585206e-01 -9.58872680e-03 -7.70476162e-01 -2.64453739e-01
1.16987240e+00 -3.15271497e-01 1.82900578e-01 9.87864912e-01
1.32572460e+00 -4.50982898e-01 -8.90769899e-01 -4.47526127e-01
1.79693371e-01 -1.97084621e-01 -1.54402524e-01 -7.21569419e-01
-7.19739497e-02 8.26744378e-01 3.05050880e-01 4.82774079e-01
6.94866896e-01 2.55106926e-01 9.19902682e-01 1.05596161e+00
5.56282878e-01 -1.11501718e+00 1.79709941e-01 8.93919408e-01
6.46656632e-01 -1.69742858e+00 -3.52204055e-01 -2.84830816e-02
-8.89504671e-01 9.58902836e-01 7.22143292e-01 3.76733214e-01
5.38146675e-01 1.67270303e-01 2.35620424e-01 -1.00591369e-01
-1.23699749e+00 -1.47350132e-01 8.72547254e-02 5.34387589e-01
6.82734013e-01 -3.46386246e-02 -4.09635827e-02 4.35602635e-01
-1.01939186e-01 -4.46721688e-02 2.74938941e-01 1.23458874e+00
-6.69308603e-01 -1.20050216e+00 9.67694819e-02 9.25979674e-01
5.88127337e-02 -5.06621838e-01 -3.27790827e-01 4.37475085e-01
-3.76129717e-01 1.26423943e+00 3.30661654e-01 -5.70285559e-01
2.68209457e-01 8.88556004e-01 2.37393919e-02 -1.34143901e+00
-4.74097371e-01 -4.97741848e-01 4.33825493e-01 -2.53390640e-01
-1.92375004e-01 -5.06178021e-01 -9.26738262e-01 -9.58941728e-02
-6.43508554e-01 2.67397791e-01 6.39369130e-01 1.22518742e+00
5.36814630e-01 4.34910417e-01 5.15894055e-01 -7.37456322e-01
-1.00193334e+00 -1.62366855e+00 -2.28660405e-01 4.29547071e-01
1.80360883e-01 -5.87155282e-01 -4.37829077e-01 3.59701365e-02] | [12.34579849243164, 7.620105266571045] |
3d6f6ab7-173e-43b1-868b-0e993f309f44 | a-survey-on-geocoding-algorithms-and-datasets | null | null | https://openreview.net/forum?id=-koTfmSDsM | https://openreview.net/pdf?id=-koTfmSDsM | A Survey on Geocoding: Algorithms and Datasets for Toponym Resolution | Geocoding, the task of converting unstructured text to structured spatial data, has recently seen progress thanks to a variety of new datasets, evaluation metrics, and machine-learning algorithms. We provide a survey to review, organize and analyze recent work on geocoding (also known as toponym resolution) where the text is matched to geospatial coordinates and/or ontologies. We summarize the findings of this research and suggest some promising directions for future work. | ['Anonymous'] | 2021-07-17 | null | https://openreview.net/forum?id=p_ibGwUc8_C | https://openreview.net/pdf?id=p_ibGwUc8_C | acl-arr-may-2021-5 | ['toponym-resolution'] | ['natural-language-processing'] | [ 1.29932240e-01 3.39262813e-01 -4.34410632e-01 -6.15407407e-01
-1.00856400e+00 -5.82313716e-01 1.09913945e+00 5.51434934e-01
-4.15300906e-01 9.25732970e-01 1.13690960e+00 -2.95151830e-01
-4.11842585e-01 -1.30344605e+00 -1.67339876e-01 -2.27339014e-01
-4.28641319e-01 4.32125390e-01 1.44040033e-01 -1.74429923e-01
5.71881413e-01 3.20562989e-01 -1.66581929e+00 3.93470377e-01
1.17700362e+00 8.22263122e-01 -8.27773064e-02 2.04652101e-01
-6.30626976e-01 9.55774069e-01 -4.53270167e-01 -5.07324159e-01
2.22174659e-01 9.05668139e-02 -7.82962322e-01 -7.97112584e-01
7.32096255e-01 -2.80522853e-02 -4.29851949e-01 1.23088825e+00
4.38752890e-01 -2.19377712e-03 6.52588964e-01 -1.31918120e+00
-1.28894877e+00 8.69991541e-01 -4.26297605e-01 2.61187077e-01
8.76322508e-01 -7.16683269e-01 8.42899561e-01 -1.00119424e+00
7.71727860e-01 1.09446299e+00 1.20805991e+00 -1.34816751e-01
-6.41633093e-01 -7.14306653e-01 -1.52846366e-01 -1.23835199e-01
-2.11325955e+00 -5.51194608e-01 -2.46450622e-02 -6.65327907e-01
1.15306711e+00 2.82608539e-01 4.02753055e-01 3.75813991e-01
2.27474257e-01 -1.79696716e-02 1.01182640e+00 -4.75607246e-01
3.43416542e-01 -3.52203548e-01 -1.49929807e-01 5.14696419e-01
6.72979474e-01 4.35816310e-02 -4.30158168e-01 -5.39321721e-01
5.42137504e-01 5.40655013e-03 1.48756474e-01 -1.70099735e-01
-1.28609645e+00 8.37205172e-01 5.76561868e-01 4.90875214e-01
-2.36958325e-01 2.84316063e-01 9.20168087e-02 8.94477144e-02
8.86465788e-01 6.23097479e-01 -3.94678973e-02 -3.57148170e-01
-1.35650063e+00 3.57090890e-01 8.02739739e-01 1.39729524e+00
7.93191195e-01 -7.99273625e-02 2.38408476e-01 7.80712903e-01
5.19071668e-02 6.33820176e-01 3.55311126e-01 -9.05203342e-01
1.29990196e+00 5.74844122e-01 4.34206784e-01 -1.47594714e+00
-4.88643408e-01 4.14963275e-01 -5.31747937e-01 -1.55781046e-01
4.01864685e-02 -1.47841796e-01 -7.51149297e-01 1.02675498e+00
-3.99804227e-02 1.85933456e-01 4.93494272e-02 3.76320988e-01
1.01919603e+00 6.80430830e-01 4.97352183e-01 3.69995505e-01
1.20924985e+00 -3.74371588e-01 -6.99948430e-01 -1.13202356e-01
7.56040215e-01 -4.27918792e-01 7.71721125e-01 -3.75131875e-01
-7.06719220e-01 -7.83872679e-02 -6.66833341e-01 -4.29304391e-01
-1.48163748e+00 -3.64479907e-02 8.48537505e-01 7.67011344e-01
-1.51500940e+00 3.72072756e-01 -6.99527979e-01 -8.76584232e-01
4.39494073e-01 9.60244387e-02 -7.96652913e-01 -8.47580656e-02
-1.57393098e+00 1.09486175e+00 5.68388820e-01 -3.62184495e-01
2.34215960e-01 -9.02158141e-01 -1.13470602e+00 7.40468279e-02
-1.15735859e-01 -5.74027359e-01 7.52112269e-01 -3.90613219e-03
-7.29509115e-01 1.01668334e+00 -1.62842840e-01 -4.34479564e-01
2.47183874e-01 -5.86878695e-02 -1.05439842e+00 -7.88266510e-02
8.42629910e-01 6.34914458e-01 9.01641138e-03 -6.62238002e-01
-1.30594301e+00 -5.53029716e-01 -8.78744721e-02 1.58887163e-01
-5.63372791e-01 5.87582111e-01 -1.98348880e-01 -1.13578212e+00
1.99541658e-01 -6.18017972e-01 -2.48254791e-01 -1.46825984e-01
6.98471069e-03 -1.81753308e-01 2.51059085e-01 -9.19110298e-01
1.93578565e+00 -2.06739068e+00 -4.45582122e-01 3.27315956e-01
2.29461506e-01 -2.29811430e-01 8.28779340e-02 1.11624765e+00
-1.09557584e-01 7.78007746e-01 -4.11606044e-01 8.77921134e-02
5.28395772e-02 3.52527015e-03 -7.63897896e-01 4.30075556e-01
-2.46992633e-01 9.54378724e-01 -1.22442794e+00 -6.97249413e-01
2.21727461e-01 2.43617699e-01 -2.47296810e-01 -4.69132990e-01
5.00812590e-01 -1.19556196e-01 -5.84594667e-01 9.51865375e-01
6.75927699e-01 4.63568605e-02 -1.43945575e-01 1.84126407e-01
-8.13141942e-01 4.06507641e-01 -1.06263161e+00 1.50262952e+00
-5.07559061e-01 7.25763500e-01 -4.64159369e-01 -4.72122610e-01
9.16637182e-01 2.44829029e-01 7.46221006e-01 -8.19302917e-01
-4.49903041e-01 4.28031802e-01 -8.91751409e-01 -5.02706528e-01
1.06977379e+00 3.74811351e-01 -7.73239791e-01 2.79485524e-01
-3.62285525e-01 -3.43744010e-01 1.89358741e-01 1.59743249e-01
1.24037981e+00 3.97345349e-02 9.60012734e-01 -4.79075849e-01
1.14523135e-01 5.88469803e-01 6.12317510e-02 8.74396265e-01
4.54416387e-02 5.34170389e-01 -2.62258444e-02 -7.27558672e-01
-1.33631277e+00 -1.04432547e+00 -5.04865289e-01 1.19185853e+00
1.92281436e-02 -7.24479079e-01 -6.33179545e-01 -1.92864180e-01
6.41697884e-01 6.90717816e-01 -9.07169044e-01 3.78664553e-01
-4.62134063e-01 -6.44461751e-01 1.01430035e+00 8.40314269e-01
2.31864884e-01 -7.00192273e-01 -3.72130513e-01 5.13900071e-02
-1.84816629e-01 -1.02259326e+00 -1.35810021e-02 -4.46048044e-02
-9.38390195e-01 -8.15588832e-01 -6.27824962e-01 -7.28997052e-01
6.08249724e-01 6.38559222e-01 1.10058379e+00 -1.18195333e-01
1.78220913e-01 3.09061944e-01 -6.60705626e-01 -2.11185500e-01
2.11841673e-01 5.51852703e-01 -2.11959444e-02 -7.23017275e-01
7.34323621e-01 -6.00903034e-01 -1.94294602e-01 1.57040164e-01
-9.94040012e-01 -2.34058890e-02 6.40361905e-02 1.16013996e-01
4.98915523e-01 2.49678772e-02 5.04071593e-01 -8.69010746e-01
6.74216151e-01 -1.03902566e+00 -8.87134433e-01 5.31604350e-01
-5.48181951e-01 -9.69175398e-02 4.39096630e-01 5.85948348e-01
-9.85424697e-01 1.50266141e-01 -6.14608862e-02 6.23484552e-01
-3.87440287e-02 1.11568773e+00 -2.91234534e-02 -4.58166927e-01
8.65343273e-01 -2.48329371e-01 -7.62589395e-01 -5.16718745e-01
8.29405665e-01 1.06691813e+00 8.53551865e-01 -5.64425409e-01
8.88759851e-01 9.14959431e-01 -2.94354022e-01 -7.52668917e-01
-6.36465847e-01 -7.19306111e-01 -1.03384161e+00 7.43033886e-02
7.12128580e-01 -1.25664592e+00 1.32951483e-01 5.82588185e-03
-9.21216607e-01 -1.07157297e-01 -1.22677423e-01 4.37174857e-01
-4.51641530e-01 1.24779753e-01 -1.03700079e-01 -4.83860105e-01
-7.51318559e-02 -2.98438042e-01 1.40923297e+00 1.35983795e-01
-3.96181345e-01 -1.25989366e+00 4.56795663e-01 5.09311706e-02
6.34254277e-01 6.30977750e-01 7.21866727e-01 -6.01586163e-01
-1.78942278e-01 -5.35046339e-01 -5.29503345e-01 -8.17177892e-01
5.44479728e-01 1.08691268e-01 -6.43581688e-01 1.11811861e-01
-6.59455359e-01 2.08717257e-01 7.41961360e-01 1.68736860e-01
1.26592445e+00 -5.24028540e-01 -8.19258809e-01 1.03089750e+00
1.67628598e+00 8.59012455e-02 8.57220650e-01 8.75174820e-01
6.03425622e-01 6.04637206e-01 6.77721858e-01 5.36223114e-01
9.75654006e-01 6.56118512e-01 1.74833879e-01 1.47511169e-01
8.57304558e-02 -6.81603432e-01 -3.02955449e-01 5.88357091e-01
-3.71844113e-01 -3.09399039e-01 -1.75376844e+00 7.88128853e-01
-1.65110147e+00 -1.37083793e+00 -2.80639350e-01 1.97537529e+00
4.68197167e-01 -6.17511034e-01 -1.10416107e-01 -1.14116333e-01
1.04175079e+00 4.07296658e-01 -3.41535546e-02 5.71893007e-02
-3.99560958e-01 3.07571739e-01 1.26831734e+00 4.71579403e-01
-1.36224687e+00 1.41781783e+00 8.20753193e+00 4.88455087e-01
-9.79025781e-01 1.43444344e-01 9.31890085e-02 1.44528538e-01
-4.93006855e-01 1.71939239e-01 -7.03273892e-01 6.29006922e-01
8.79473567e-01 -8.54334235e-01 3.92885983e-01 1.10327804e+00
2.68629193e-01 -2.07214937e-01 -6.22361422e-01 8.83887172e-01
5.79736196e-02 -1.73239434e+00 1.90605715e-01 1.28639802e-01
1.07850373e+00 4.63855296e-01 2.65127271e-02 -4.07096744e-02
1.10144830e+00 -1.10124230e+00 8.79538417e-01 5.26472986e-01
1.40699780e+00 -3.82869333e-01 7.42610216e-01 -2.00666547e-01
-1.74905467e+00 -3.34865749e-01 -6.05862856e-01 -3.79082471e-01
3.53696167e-01 4.67636853e-01 -3.15005749e-01 8.55093479e-01
1.07701516e+00 1.04082167e+00 -8.78074646e-01 1.39015484e+00
-2.12948889e-01 1.74437970e-01 -2.78482169e-01 2.34257787e-01
3.51933271e-01 -1.38931379e-01 6.07306771e-02 1.30058694e+00
1.03426063e+00 3.71089965e-01 3.21798003e-03 2.39845932e-01
-2.83901505e-02 3.04082990e-01 -1.08451664e+00 -1.09547623e-01
1.34736645e+00 7.74059236e-01 -7.19096482e-01 -3.77261698e-01
-6.11257017e-01 4.37435269e-01 3.15724373e-01 3.50062639e-01
-7.05654621e-01 -9.82727051e-01 5.28994024e-01 4.46280748e-01
2.53142626e-03 -4.58664030e-01 -7.17126608e-01 -1.31315207e+00
-5.00320829e-02 -1.17868990e-01 7.83615589e-01 -1.17479157e+00
-1.19175923e+00 5.15810311e-01 4.69679207e-01 -1.31512678e+00
-4.38476562e-01 -2.37384334e-01 -3.30562502e-01 8.05393577e-01
-1.43520224e+00 -1.10646284e+00 -7.41598964e-01 4.19975817e-01
-1.38293758e-01 -2.47476637e-01 9.87534046e-01 5.16878545e-01
7.14598894e-02 2.83197671e-01 7.19771326e-01 4.23982561e-01
9.33946490e-01 -1.23018885e+00 1.31459153e+00 7.27448165e-01
4.50631119e-02 7.23594069e-01 3.94449383e-01 -1.10906124e+00
-9.15495574e-01 -1.53361464e+00 1.61783981e+00 -8.45848322e-01
1.17119920e+00 -3.11128795e-01 -4.77438927e-01 9.01909113e-01
-4.45202664e-02 1.04127705e-01 7.57808745e-01 3.17472480e-02
-4.67065364e-01 -1.80598497e-01 -1.36609709e+00 5.24132550e-01
1.44134021e+00 -7.91456938e-01 -9.90263224e-01 1.21305294e-01
5.42530000e-01 -3.96573365e-01 -1.23451328e+00 5.53369224e-02
6.25821650e-01 -6.20900691e-01 9.79728878e-01 -3.77828330e-01
2.60468900e-01 -3.84076327e-01 -4.48738217e-01 -1.39900172e+00
-6.37396097e-01 -3.01207691e-01 3.39697123e-01 1.22210336e+00
2.78976053e-01 -9.11294818e-01 6.40434384e-01 9.19373393e-01
-2.95685917e-01 1.40454903e-01 -1.17737138e+00 -8.05887580e-01
3.62321168e-01 -3.37670475e-01 1.51398575e+00 1.47316670e+00
5.76651812e-01 -5.68838120e-01 -2.29372233e-01 1.37429968e-01
2.70871788e-01 -1.84612840e-01 5.14588416e-01 -1.66588569e+00
1.06508577e+00 -3.70466471e-01 -8.89843047e-01 -5.21906257e-01
1.75839975e-01 -8.95106375e-01 -2.47778088e-01 -2.10977578e+00
-1.66247651e-01 -8.44302654e-01 5.79663254e-02 7.79718518e-01
2.70511545e-02 4.28852588e-01 -1.37904510e-01 5.17397463e-01
-5.55436432e-01 1.52265374e-03 2.91563213e-01 -1.81422621e-01
-1.08630627e-01 -3.11992317e-01 -9.09744918e-01 5.82815707e-01
6.71214163e-01 -8.54733706e-01 -9.14286450e-02 -9.25416648e-01
9.20298398e-01 2.09343452e-02 2.83615775e-02 -1.20174885e+00
6.48030281e-01 -6.00642622e-01 1.77903458e-01 -7.11352408e-01
-1.95924491e-01 -1.01916540e+00 5.57536960e-01 1.13946296e-01
-3.88111144e-01 8.26825500e-01 -8.62721652e-02 4.82876629e-01
-4.53557312e-01 -6.53455928e-02 2.98707366e-01 -1.80447459e-01
-1.25384367e+00 3.79575908e-01 -5.79402447e-01 3.39647114e-01
8.45596492e-01 -5.64461946e-01 -6.68405712e-01 -3.33598286e-01
-4.81441110e-01 2.68100291e-01 1.00066018e+00 7.50779986e-01
2.60199875e-01 -1.74752688e+00 -6.83788538e-01 -3.88469368e-01
7.78473258e-01 -5.01459122e-01 -1.45271301e-01 2.63460487e-01
-9.66119885e-01 9.29844022e-01 -3.50636721e-01 5.02714328e-02
-5.52653193e-01 5.21652043e-01 1.10469796e-01 7.70967156e-02
-6.01565123e-01 8.31228793e-02 -4.69698489e-01 -4.69529778e-01
-1.89877264e-02 -4.51964825e-01 -5.33628345e-01 3.86763096e-01
8.24558735e-01 7.55572200e-01 3.16305131e-01 -1.03348112e+00
-6.54641509e-01 6.77747607e-01 7.29435325e-01 -1.72449380e-01
1.67748857e+00 -3.57367754e-01 -3.96844983e-01 2.29385361e-01
9.72830892e-01 3.67870033e-01 -4.32733983e-01 -1.54385716e-01
8.84443521e-01 -8.50920558e-01 -1.98064029e-01 -5.22121131e-01
-7.67988682e-01 4.38381404e-01 4.64203119e-01 3.55798334e-01
7.15651274e-01 1.51610509e-01 3.93887013e-01 5.70064127e-01
9.69220281e-01 -1.39059794e+00 -6.56680822e-01 8.01619053e-01
7.28783727e-01 -9.71682966e-01 3.18848610e-01 -3.47326636e-01
-2.31626421e-01 9.92085576e-01 1.95121467e-01 -6.79903850e-02
1.13859832e+00 5.41702092e-01 -7.88924843e-02 -3.72596413e-01
-1.36027053e-01 -3.97727221e-01 1.30646555e-02 1.22996700e+00
6.28410101e-01 1.57417849e-01 -1.28557310e-01 4.50791657e-01
-8.02993298e-01 2.74499089e-01 3.80297393e-01 9.07553375e-01
-6.29478514e-01 -9.67900217e-01 -6.62850678e-01 7.89766014e-01
-4.13440883e-01 -6.76337421e-01 -4.78129923e-01 9.93041337e-01
-1.21312086e-02 8.57936382e-01 5.13498902e-01 -1.91111445e-01
3.24301720e-01 -1.18153438e-01 -1.56643957e-01 -8.76803637e-01
-5.36982000e-01 -8.31748605e-01 2.35803545e-01 -5.65825641e-01
-5.07066369e-01 -8.47784460e-01 -1.16512656e+00 -7.18901932e-01
-3.00507285e-02 4.49134916e-01 7.38083959e-01 6.30499005e-01
6.45039439e-01 -2.15595052e-01 1.99379236e-01 -7.05982447e-01
1.40409112e-01 -7.13230848e-01 -6.25795722e-01 7.41745979e-02
3.32023539e-02 -7.40209877e-01 -9.47736055e-02 5.70204109e-02] | [9.362208366394043, 9.147689819335938] |
12f45325-b9af-4b2c-86ad-6f61877e4c32 | working-hard-to-know-your-neighbors-margins | 1705.10872 | null | http://arxiv.org/abs/1705.10872v4 | http://arxiv.org/pdf/1705.10872v4.pdf | Working hard to know your neighbor's margins: Local descriptor learning loss | We introduce a novel loss for learning local feature descriptors which is
inspired by the Lowe's matching criterion for SIFT. We show that the proposed
loss that maximizes the distance between the closest positive and closest
negative patch in the batch is better than complex regularization methods; it
works well for both shallow and deep convolution network architectures.
Applying the novel loss to the L2Net CNN architecture results in a compact
descriptor -- it has the same dimensionality as SIFT (128) that shows
state-of-art performance in wide baseline stereo, patch verification and
instance retrieval benchmarks. It is fast, computing a descriptor takes about 1
millisecond on a low-end GPU. | ['Jiri Matas', 'Anastasiya Mishchuk', 'Dmytro Mishkin', 'Filip Radenovic'] | 2017-05-30 | working-hard-to-know-your-neighbors-margins-1 | http://papers.nips.cc/paper/7068-working-hard-to-know-your-neighbors-margins-local-descriptor-learning-loss | http://papers.nips.cc/paper/7068-working-hard-to-know-your-neighbors-margins-local-descriptor-learning-loss.pdf | neurips-2017-12 | ['patch-matching'] | ['computer-vision'] | [-2.00225145e-01 -4.39168304e-01 -5.10008335e-01 -6.16742551e-01
-9.59136009e-01 -3.62677932e-01 5.54079413e-01 3.00609350e-01
-7.91307926e-01 1.69591695e-01 1.47307232e-01 1.84269510e-02
-3.22992086e-01 -7.46073544e-01 -9.15415227e-01 -5.90645611e-01
-2.69279867e-01 5.79157583e-02 3.97894561e-01 -1.14214949e-01
4.89483833e-01 1.07930148e+00 -1.66423249e+00 2.77831733e-01
-7.71139339e-02 1.69743264e+00 -1.84895426e-01 4.64028627e-01
4.36692327e-01 5.80348492e-01 -2.12331325e-01 -3.02139968e-01
8.91954601e-01 4.28066671e-01 -8.42371166e-01 -4.10445273e-01
1.62523735e+00 -7.09786713e-01 -7.51815557e-01 9.40237999e-01
7.06716537e-01 1.40727237e-01 3.81518006e-01 -1.00403619e+00
-7.25164652e-01 -2.36273627e-03 -4.09429550e-01 1.58194721e-01
4.18151915e-01 7.16022849e-02 1.24403965e+00 -1.07718253e+00
8.25106740e-01 1.42294884e+00 1.35256457e+00 1.80551335e-01
-1.20514905e+00 -4.75337327e-01 -1.32427618e-01 2.17344448e-01
-1.73268890e+00 -3.99378687e-01 2.83940017e-01 -1.44182533e-01
1.57908034e+00 -7.02927262e-02 7.86261559e-01 6.58385515e-01
7.14582384e-01 8.25756669e-01 4.89214212e-01 -2.12128773e-01
-9.53352377e-02 -3.85536015e-01 2.44150266e-01 8.34305346e-01
1.06065169e-01 3.69776070e-01 -7.42710292e-01 -4.39406633e-01
9.49602485e-01 1.80232048e-01 6.58218712e-02 -5.25110662e-01
-1.19221413e+00 1.03668070e+00 8.44851196e-01 1.77774541e-02
-3.39374393e-01 7.54618347e-01 7.48196006e-01 7.34289527e-01
4.16563839e-01 6.97490051e-02 -4.66520160e-01 -7.01179877e-02
-9.59677696e-01 4.33630764e-01 7.45307624e-01 1.00003600e+00
1.02823389e+00 -2.09104151e-01 -1.20812267e-01 6.92649066e-01
2.11071879e-01 6.45206094e-01 1.84358940e-01 -9.28705931e-01
1.83569361e-02 4.03492033e-01 5.18818013e-03 -1.40227723e+00
-3.14451635e-01 -1.58334091e-01 -7.94763446e-01 4.59608436e-01
1.91982836e-01 2.91543663e-01 -7.58689582e-01 1.38703144e+00
4.20371033e-02 7.38190264e-02 -4.28941905e-01 1.02219760e+00
1.03030944e+00 2.84797877e-01 -2.09427834e-01 7.40848780e-01
1.02064848e+00 -1.01551092e+00 -1.48710981e-01 -1.19389512e-01
5.91106832e-01 -1.05364788e+00 6.94075346e-01 2.88796723e-01
-1.14901090e+00 -7.94718206e-01 -1.33273649e+00 -7.62950897e-01
-5.28458178e-01 2.86163539e-01 1.00106907e+00 2.56114721e-01
-1.56197751e+00 9.77002263e-01 -8.04778576e-01 -6.91750109e-01
3.30495209e-01 8.51573050e-01 -7.53129780e-01 -2.57455468e-01
-8.45012248e-01 7.34907269e-01 -1.23757958e-01 8.29530135e-02
-6.66292131e-01 -8.45727324e-01 -1.10469544e+00 2.52903879e-01
-3.21775824e-01 -7.55804062e-01 1.01969492e+00 -7.65320957e-01
-1.25708401e+00 1.23163021e+00 -1.65727809e-01 -7.18315959e-01
3.91337395e-01 -2.44142115e-01 3.44775990e-02 1.48157313e-01
1.43744111e-01 1.29714811e+00 8.95253241e-01 -3.16085815e-01
-6.06136143e-01 -3.78370792e-01 -2.70558652e-02 -1.57380268e-01
-9.57223549e-02 7.52879679e-02 -3.45070183e-01 -4.20214564e-01
2.13825598e-01 -8.27925861e-01 3.23618017e-02 7.84301579e-01
5.41219860e-03 -6.06057167e-01 8.36489558e-01 -1.66902721e-01
7.07940340e-01 -2.43872976e+00 -5.80197632e-01 6.14380777e-01
1.76419280e-02 4.00516093e-01 -5.30713141e-01 5.01616895e-01
-7.34595656e-02 -2.46410087e-01 5.76637805e-01 -2.15860277e-01
3.53367597e-01 -7.73110762e-02 -4.38711405e-01 1.04179692e+00
2.27062538e-01 1.02661550e+00 -6.24051809e-01 -3.15229088e-01
4.09078360e-01 9.04227138e-01 -6.54171705e-01 -9.40346196e-02
2.69648969e-01 -3.82913679e-01 -2.16135174e-01 7.43503869e-01
1.09647548e+00 -2.73187041e-01 -3.08295518e-01 -5.12663901e-01
-2.02816829e-01 3.20691288e-01 -1.16039371e+00 2.01962066e+00
-3.87765229e-01 1.23306596e+00 3.38727683e-02 -7.41866708e-01
1.12324286e+00 -1.34836614e-01 4.38908517e-01 -8.12625468e-01
-9.07691778e-04 4.42039311e-01 -3.98771375e-01 -7.20086768e-02
4.29492861e-01 2.82768935e-01 5.76254204e-02 1.32355422e-01
3.45365256e-01 -1.58456042e-01 -6.94723129e-02 4.62747961e-02
1.30926812e+00 -9.63074267e-02 2.12890759e-01 -6.19227767e-01
4.91075516e-01 -1.25967383e-01 3.37347180e-01 1.14359534e+00
-2.56870180e-01 4.95250046e-01 4.45121139e-01 -1.26042938e+00
-9.84922051e-01 -1.04844308e+00 -3.70549738e-01 1.08187842e+00
2.36068815e-01 -5.74203253e-01 -2.94996887e-01 -1.88212261e-01
6.61373734e-01 -3.56394917e-01 -5.69233716e-01 -2.43922323e-01
-5.16245306e-01 2.32726857e-01 8.16698492e-01 6.89173341e-01
7.38677621e-01 -1.05708671e+00 -6.27165556e-01 8.56949612e-02
5.11816263e-01 -9.00901377e-01 -7.85666943e-01 2.28684306e-01
-8.47035468e-01 -1.20800745e+00 -4.33124155e-01 -1.34968579e+00
5.19740283e-01 4.09709334e-01 1.30347097e+00 3.31320047e-01
-6.92138910e-01 5.50368965e-01 8.63191560e-02 4.04945649e-02
4.28243399e-01 -1.91840101e-02 -8.13384354e-02 -2.96732843e-01
1.01418829e+00 -3.49331558e-01 -9.04758215e-01 1.27837688e-01
-7.88346231e-01 -7.77837634e-01 5.87489665e-01 1.24002624e+00
8.40054691e-01 -3.25684518e-01 -1.68759167e-01 -1.01045735e-01
5.72126389e-01 1.14655107e-01 -9.42948341e-01 1.44903779e-01
-4.28820789e-01 6.90408945e-02 5.97495854e-01 -3.48855525e-01
-2.64262915e-01 3.50344062e-01 -1.53257385e-01 -5.01620948e-01
-1.18417315e-01 1.78724498e-01 5.52174985e-01 -1.11273074e+00
4.75450516e-01 1.18832596e-01 9.05511156e-02 -3.45009983e-01
2.81997561e-01 4.75023717e-01 4.97060329e-01 -4.18655515e-01
5.41972339e-01 8.85807931e-01 3.09834331e-01 -9.76234138e-01
-6.02376401e-01 -9.04802144e-01 -8.47180247e-01 2.45557025e-01
4.75555927e-01 -1.07121539e+00 -1.38839602e+00 6.58010483e-01
-1.35745835e+00 -1.20902427e-01 -4.20073301e-01 6.19329333e-01
-7.58237720e-01 2.76551813e-01 -8.30927014e-01 -5.75122237e-03
-7.30880737e-01 -1.19722819e+00 1.62276924e+00 2.12295242e-02
5.00685722e-02 -1.07570958e+00 2.79461861e-01 -4.71248806e-01
8.19267094e-01 2.98210792e-02 6.14452720e-01 -4.68518436e-01
-9.55809832e-01 -7.63532579e-01 -7.43587971e-01 5.93907058e-01
-2.47562721e-01 1.12605765e-01 -7.27525771e-01 -7.12210059e-01
-2.90700167e-01 -8.47896278e-01 1.31053913e+00 7.47823596e-01
1.06714511e+00 -3.31211388e-02 -3.36597413e-01 1.23164856e+00
1.83248746e+00 -4.07340676e-01 7.16651261e-01 5.65009356e-01
2.65109688e-01 6.27911463e-02 6.12245321e-01 5.11702895e-01
1.19610004e-01 6.90703392e-01 3.45871836e-01 -3.73494476e-01
-3.05264741e-01 -1.69149205e-01 2.05595925e-01 3.84233087e-01
3.06676090e-01 2.70517617e-01 -6.28258824e-01 6.54699445e-01
-1.91029191e+00 -9.53080356e-01 1.20468386e-01 2.27727938e+00
3.76867712e-01 2.24987185e-03 -1.71798021e-01 -5.19339085e-01
5.07692814e-01 4.03829575e-01 -3.70766491e-01 -6.83193207e-01
-2.95406550e-01 8.46574247e-01 1.01613510e+00 7.18389988e-01
-1.52241087e+00 1.18992996e+00 7.66042519e+00 1.05972314e+00
-1.27674806e+00 -1.65302843e-01 1.36640251e-01 -8.97845812e-03
1.39537722e-01 -9.93963107e-02 -1.06116319e+00 -8.46902952e-02
6.90344155e-01 3.23939212e-02 3.26127112e-01 1.06399739e+00
-6.71073198e-02 -3.43395025e-02 -1.45363474e+00 1.43691957e+00
2.06904113e-01 -1.53510988e+00 1.50979504e-01 1.40258357e-01
6.48904920e-01 7.72312164e-01 2.97222048e-01 5.28074652e-02
8.69167596e-02 -9.81376231e-01 4.79956210e-01 4.62326497e-01
8.16688716e-01 -8.55283022e-01 9.02031243e-01 -2.31827870e-01
-1.49734545e+00 3.61967646e-02 -9.96745527e-01 8.45239013e-02
-4.14296269e-01 4.10378963e-01 -5.03397226e-01 -8.85744914e-02
7.83010721e-01 1.04308808e+00 -7.20118821e-01 1.43128109e+00
1.81461558e-01 -2.65849143e-01 -6.04851246e-01 -1.06513910e-02
8.64024460e-01 8.16353261e-02 2.88843274e-01 1.54605865e+00
1.35773703e-01 -4.53654289e-01 2.53211915e-01 6.71733856e-01
-1.82023123e-01 2.88590752e-02 -9.77048695e-01 3.26686740e-01
3.08569610e-01 1.12361908e+00 -4.21947449e-01 -1.51373416e-01
-4.99300748e-01 9.40190375e-01 4.20052767e-01 6.90451860e-02
-4.56679195e-01 -1.05470991e+00 1.07821596e+00 -1.11984953e-01
7.06231415e-01 -1.58994228e-01 1.23638526e-01 -7.90160239e-01
3.35246950e-01 -6.49408519e-01 7.52707049e-02 -4.67890173e-01
-1.58451569e+00 4.37700272e-01 -4.73839730e-01 -1.37552643e+00
-2.48319417e-01 -1.17231488e+00 -4.44440395e-01 7.90433824e-01
-1.83218336e+00 -1.16321123e+00 -5.16794622e-01 1.03972220e+00
1.03445590e-01 -2.07188800e-01 8.48169923e-01 4.94264752e-01
3.45270663e-01 1.00068283e+00 2.24621907e-01 3.63104045e-01
1.03233016e+00 -9.96004045e-01 9.78437245e-01 3.60685766e-01
-1.83276851e-02 7.85892010e-01 1.30415842e-01 -2.04709426e-01
-1.68974710e+00 -8.16262841e-01 1.09611595e+00 -1.26770232e-02
8.02624762e-01 -2.20030591e-01 -5.37002146e-01 7.14357555e-01
2.84074455e-01 7.40587950e-01 2.82075912e-01 2.35884245e-02
-9.07258987e-01 -5.39489985e-01 -1.32893217e+00 1.64545044e-01
9.36891496e-01 -1.04529822e+00 -2.10642174e-01 7.77008712e-01
3.89712423e-01 -4.95146900e-01 -9.37395155e-01 2.97265381e-01
9.97217715e-01 -1.20119441e+00 1.34035301e+00 -6.16869926e-01
-6.91023022e-02 2.06199475e-02 -3.26904863e-01 -7.48427391e-01
-5.75845420e-01 -6.59603715e-01 2.86264628e-01 4.39060628e-01
-5.70261106e-02 -7.71689773e-01 7.99893856e-01 2.36613795e-01
-3.34174633e-02 -9.44294214e-01 -1.13358712e+00 -1.13845718e+00
-8.53236474e-04 -2.15392664e-01 3.98363113e-01 4.64231968e-01
-2.36290410e-01 -2.28898197e-01 -1.61804974e-01 -7.84396902e-02
6.94797635e-01 1.01580679e-01 8.96496058e-01 -1.26801777e+00
1.85883060e-01 -4.88638043e-01 -1.32502317e+00 -1.60233045e+00
5.11454225e-01 -9.14963186e-01 -7.34800473e-02 -1.20693624e+00
1.86488107e-01 -2.78122365e-01 -3.83685112e-01 6.55821145e-01
6.80772126e-01 5.63579500e-01 9.61171761e-02 5.36201149e-02
-7.70912886e-01 3.79303664e-01 8.45021486e-01 -6.00642681e-01
1.97788432e-01 -2.88243622e-01 -5.37545905e-02 6.22729659e-01
3.04585785e-01 -5.17906487e-01 1.78699434e-01 -6.46068931e-01
2.70127267e-01 -4.21367735e-01 7.13798225e-01 -1.12447941e+00
5.99911451e-01 4.33998108e-01 4.90511686e-01 -8.40064228e-01
4.91543412e-01 -8.49108815e-01 -3.29104900e-01 6.00947440e-01
-4.59658116e-01 4.28182662e-01 2.58127987e-01 2.83034533e-01
-6.78003967e-01 -2.35116079e-01 8.77871215e-01 -8.91021863e-02
-8.25660169e-01 5.91521442e-01 1.27185568e-01 -1.14611447e-01
6.07793152e-01 -4.09123778e-01 -4.60350901e-01 -3.73172045e-01
-3.91878515e-01 -1.18040647e-02 6.01841033e-01 3.70427310e-01
8.87592673e-01 -1.79023504e+00 -6.22714341e-01 6.77628398e-01
2.03171205e-02 -4.31646347e-01 -2.19942089e-02 7.55977511e-01
-1.18025696e+00 9.91603315e-01 -3.48987639e-01 -9.19539154e-01
-1.33059597e+00 3.40725869e-01 5.28441668e-01 -9.41259786e-02
-7.96700597e-01 1.06030130e+00 -1.06066927e-01 -3.67265165e-01
6.74657345e-01 -3.74210268e-01 5.02440214e-01 -2.38601714e-01
6.64801180e-01 4.22460824e-01 4.95630950e-01 -6.22296214e-01
-6.77799225e-01 1.02195501e+00 -2.61110336e-01 3.75941843e-01
1.42412531e+00 2.52786785e-01 -5.08000314e-01 -1.07820228e-01
2.07931590e+00 -2.39946038e-01 -1.15124047e+00 -2.73345262e-01
-1.15453191e-01 -9.04771984e-01 3.94341677e-01 -2.14114681e-01
-1.11261714e+00 8.01247358e-01 1.15887105e+00 -7.41293877e-02
7.73364902e-01 -7.38602430e-02 1.05000830e+00 1.17479730e+00
5.97272933e-01 -1.11749113e+00 -6.34204550e-03 8.95642817e-01
9.47680354e-01 -1.22353232e+00 2.67450541e-01 3.16656455e-02
5.16230166e-02 1.42939067e+00 3.10493857e-01 -1.08371735e+00
1.09942329e+00 2.36331135e-01 6.04107566e-02 -4.60112423e-01
-5.85134387e-01 -1.28623396e-01 6.26271188e-01 5.21379232e-01
4.35947001e-01 -1.72480017e-01 -1.57237694e-01 -4.33390170e-01
6.78779036e-02 2.42069401e-02 1.87680242e-04 1.01904273e+00
-2.73531675e-01 -9.64381874e-01 -6.65475503e-02 4.03775483e-01
-4.09079760e-01 -3.20953995e-01 -4.00263846e-01 9.13771987e-01
-1.72572639e-02 5.11568069e-01 4.26053256e-01 -2.66596764e-01
3.78682882e-01 -3.18669200e-01 8.39157879e-01 -3.11108768e-01
-9.00777280e-01 6.70253634e-02 -4.17335868e-01 -1.39779699e+00
-5.31620443e-01 -3.84892732e-01 -8.58818471e-01 -7.21151710e-01
-2.90583819e-01 -1.89499959e-01 7.13805914e-01 5.31212509e-01
6.83252096e-01 -3.72483104e-01 6.16346002e-01 -1.36918247e+00
-9.41257477e-01 -7.22801626e-01 -7.94261575e-01 3.74990851e-01
7.60733902e-01 -6.02274597e-01 -6.38399959e-01 -3.31001014e-01] | [8.26047134399414, -1.898010015487671] |
c6a18bbe-b035-4aae-95c0-167b82753afa | attention-based-context-aware-reasoning-for | null | null | http://openaccess.thecvf.com/content_CVPR_2020/html/Cooray_Attention-Based_Context_Aware_Reasoning_for_Situation_Recognition_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Cooray_Attention-Based_Context_Aware_Reasoning_for_Situation_Recognition_CVPR_2020_paper.pdf | Attention-Based Context Aware Reasoning for Situation Recognition | Situation Recognition (SR) is a fine-grained action recognition task where the model is expected to not only predict the salient action of the image, but also predict values of all associated semantic roles of the action. Predicting semantic roles is very challenging: a vast variety of possibilities can be the match for a semantic role. Existing work has focused on dependency modelling architectures to solve this issue. Inspired by the success achieved by query-based visual reasoning (e.g., Visual Question Answering), we propose to address semantic role prediction as a query-based visual reasoning problem. However, existing query-based reasoning methods have not considered handling of inter-dependent queries which is a unique requirement of semantic role prediction in SR. Therefore, to the best of our knowledge, we propose the first set of methods to address inter-dependent queries in query-based visual reasoning. Extensive experiments demonstrate the effectiveness of our proposed method which achieves outstanding performance on Situation Recognition task. Furthermore, leveraging query inter-dependency, our methods improve upon a state-of-the-art method that answers queries separately. Our code: https://github.com/thilinicooray/context-aware-reasoning-for-sr
| [' Wei Lu', ' Ngai-Man Cheung', 'Thilini Cooray'] | 2020-06-01 | null | null | null | cvpr-2020-6 | ['grounded-situation-recognition', 'situation-recognition', 'fine-grained-action-recognition'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 4.90903735e-01 9.86075178e-02 -3.23686719e-01 -4.94335145e-01
-7.07820892e-01 -5.03949285e-01 7.87730575e-01 7.04311728e-02
-1.24548256e-01 4.23182368e-01 6.26030803e-01 -3.16222459e-01
-6.50292039e-02 -6.45192146e-01 -6.69462204e-01 -2.36769497e-01
3.88158560e-01 2.90766656e-01 7.87637115e-01 -3.89594376e-01
4.91797864e-01 4.11218733e-01 -2.07876754e+00 9.89486158e-01
4.28066283e-01 9.78810430e-01 2.40105197e-01 6.05566680e-01
-2.30516165e-01 1.77884054e+00 -4.72877115e-01 -2.96336502e-01
-8.16292390e-02 -5.37357330e-01 -1.28826129e+00 1.56697318e-01
4.12200332e-01 -2.88798988e-01 -3.57643604e-01 8.12654912e-01
2.18215063e-01 2.48827949e-01 4.11676645e-01 -1.69208825e+00
-6.96602285e-01 3.31460685e-01 -2.61741549e-01 4.13821459e-01
8.94344270e-01 -2.65062619e-02 1.31484509e+00 -9.08504367e-01
7.70516336e-01 1.44716084e+00 -1.44008860e-01 6.72838330e-01
-7.79176414e-01 -3.33497524e-01 5.18424332e-01 9.39266145e-01
-1.14574540e+00 -5.40416777e-01 1.03081036e+00 -3.11536252e-01
1.15905344e+00 4.81716067e-01 5.39523542e-01 8.39337051e-01
-3.55711669e-01 1.38181591e+00 1.19224763e+00 -4.84362334e-01
2.49205679e-01 -2.69566327e-02 2.02936515e-01 5.65440416e-01
-2.87989557e-01 -2.48520851e-01 -9.65251207e-01 -4.18199785e-02
7.15685785e-01 1.31371662e-01 -2.01469839e-01 -5.30383408e-01
-1.47444773e+00 4.57841426e-01 4.38325077e-01 3.99942547e-01
-2.72636235e-01 3.60464424e-01 3.78711760e-01 3.53330612e-01
1.24501809e-02 3.01308036e-01 -3.43016148e-01 -2.82483101e-01
-4.82835710e-01 2.99406469e-01 7.46157110e-01 8.97631526e-01
6.43965244e-01 -2.52092719e-01 -4.80240911e-01 7.17965662e-01
3.25666338e-01 1.12472437e-01 2.49221891e-01 -1.14477479e+00
3.51488948e-01 1.11811411e+00 1.86962113e-01 -9.63318110e-01
-1.16905369e-01 1.21565811e-01 -2.81218439e-01 7.98659697e-02
3.42543989e-01 6.12558663e-01 -8.76107395e-01 1.59409809e+00
5.87745726e-01 1.57188401e-01 1.96640477e-01 1.11671555e+00
1.03786695e+00 4.61432070e-01 5.60560822e-01 -5.83329834e-02
1.55505300e+00 -1.18390071e+00 -6.19739830e-01 -3.18601698e-01
3.93955588e-01 -6.36801720e-01 1.22571397e+00 1.06556967e-01
-9.11224961e-01 -5.05904675e-01 -7.97170520e-01 -3.38174313e-01
-4.97344792e-01 6.38389960e-02 7.20338762e-01 1.36423573e-01
-9.81079340e-01 9.75306928e-02 -5.41082680e-01 -6.59981966e-01
4.50378656e-01 -1.89604964e-02 -5.01581788e-01 -2.87040323e-01
-1.18927503e+00 9.11895394e-01 3.58911544e-01 -6.78612515e-02
-1.05346119e+00 -4.25197780e-01 -7.60998011e-01 -5.28053418e-02
1.04615390e+00 -5.98050594e-01 1.47485209e+00 -1.11649013e+00
-1.02320659e+00 1.03615510e+00 -7.53815830e-01 -2.70726562e-01
2.04621077e-01 -1.05915420e-01 -5.18555105e-01 4.70131099e-01
3.57671946e-01 6.75610006e-01 7.90348589e-01 -1.40113914e+00
-8.12286258e-01 -5.87039113e-01 8.49564731e-01 6.36209548e-01
-8.17507356e-02 3.59456837e-01 -6.70402825e-01 -4.15267766e-01
1.61426380e-01 -5.64848602e-01 -1.00816011e-01 2.89956123e-01
-1.45118624e-01 -5.21857381e-01 9.10002410e-01 -4.22626764e-01
1.24258912e+00 -2.27471876e+00 -1.64358772e-03 -2.05719769e-01
1.28694281e-01 1.06439829e-01 -6.44077361e-02 7.58765221e-01
-8.29928145e-02 9.48325247e-02 -1.79352418e-01 -8.09866562e-02
9.06337276e-02 4.89116937e-01 -5.59033453e-01 1.01811953e-01
2.07437247e-01 1.06536627e+00 -9.27971125e-01 -8.55006814e-01
2.52379745e-01 2.68083245e-01 -2.96196461e-01 3.89498025e-01
-5.36812067e-01 3.51796389e-01 -7.25287974e-01 1.02602386e+00
1.77476317e-01 -4.94884759e-01 2.63460040e-01 -3.27555478e-01
9.75674391e-02 1.76318929e-01 -1.09850359e+00 1.57937777e+00
-1.55150190e-01 2.45119900e-01 -1.56067044e-01 -1.13899016e+00
7.36983955e-01 2.60905892e-01 4.45382714e-01 -9.24508095e-01
-3.31338316e-01 3.91547866e-02 -2.03609124e-01 -6.16908193e-01
4.51118022e-01 -2.10204363e-01 -1.98669657e-01 4.00338948e-01
-3.25989991e-01 -5.98425306e-02 2.78008908e-01 4.33269620e-01
1.08804595e+00 6.08578444e-01 7.75088429e-01 1.87502295e-01
8.64686966e-01 3.15948814e-01 5.32878935e-01 7.31614351e-01
-5.46345115e-01 4.92548406e-01 7.32022405e-01 -5.20761311e-01
-4.98921037e-01 -9.67405856e-01 3.84805918e-01 1.56502342e+00
6.06440425e-01 -5.79416633e-01 -3.50227326e-01 -9.39247310e-01
1.47352712e-02 6.71709657e-01 -6.08791769e-01 8.42049643e-02
-5.74688435e-01 -9.21096355e-02 3.22351605e-01 8.06526065e-01
5.78174293e-01 -1.54485393e+00 -8.96576285e-01 -4.29332405e-02
-2.60954499e-01 -1.32856214e+00 -8.69610459e-02 -1.90344974e-01
-5.56825340e-01 -1.66916084e+00 -1.21944726e-01 -6.95807040e-01
5.98902404e-01 5.93495905e-01 1.24225485e+00 2.90785342e-01
-1.12848766e-01 9.66713309e-01 -6.17006183e-01 -3.46617281e-01
-2.79944748e-01 -5.99461973e-01 -4.61663693e-01 2.63488531e-01
5.87338686e-01 -3.34725380e-01 -8.32074881e-01 6.35034382e-01
-1.02710462e+00 3.30611318e-01 5.03493369e-01 4.46969926e-01
9.75057423e-01 7.52651989e-02 4.81623918e-01 -8.18108618e-01
4.83015507e-01 -3.54381621e-01 -2.75619328e-01 7.84366488e-01
-2.44700536e-01 4.95274700e-02 3.70609224e-01 -8.91004279e-02
-1.36152732e+00 2.45575562e-01 8.77034944e-03 -3.70634764e-01
-5.31492531e-01 3.55688661e-01 -2.40378842e-01 1.76750079e-01
4.00761843e-01 4.87048119e-01 -2.31554136e-01 -2.59171218e-01
5.90356469e-01 4.66664255e-01 4.23315525e-01 -5.69851041e-01
4.32601601e-01 8.13095152e-01 1.37719408e-01 -6.01903260e-01
-1.08492637e+00 -9.25630987e-01 -5.47415853e-01 -5.92270315e-01
1.10297787e+00 -9.72382843e-01 -7.98197865e-01 2.57653922e-01
-1.14269400e+00 -2.41110682e-01 -2.25379050e-01 -5.00203818e-02
-8.72625291e-01 5.83499730e-01 -1.48433700e-01 -9.88969088e-01
-4.81568910e-02 -9.77549434e-01 1.23734224e+00 1.43376980e-02
-1.90677732e-01 -6.33655250e-01 -2.26750419e-01 9.81727421e-01
3.00464571e-01 1.63482338e-01 8.11095297e-01 -6.27412856e-01
-9.24372554e-01 1.41794905e-01 -4.80110019e-01 5.61170420e-03
-1.81657006e-03 -3.39847505e-01 -8.38575959e-01 1.64501727e-01
-7.25065097e-02 -5.70823133e-01 7.64790416e-01 -6.96919784e-02
1.34245515e+00 -9.55644622e-02 -2.33923331e-01 -5.06326817e-02
1.33380723e+00 3.37560982e-01 8.28983903e-01 2.99746096e-01
7.24692285e-01 7.30376840e-01 1.29009962e+00 4.02729183e-01
8.83032799e-01 8.65301549e-01 6.68426096e-01 4.32685316e-02
-3.39042723e-01 -4.94391680e-01 1.68262675e-01 -4.33486663e-02
-7.89157674e-02 -5.42373620e-02 -9.65666234e-01 6.22297823e-01
-2.35289717e+00 -1.29562807e+00 -2.96142519e-01 1.78142977e+00
6.42488658e-01 -1.30218044e-01 1.49508551e-01 1.43122673e-01
4.98267293e-01 5.70995629e-01 -6.92468464e-01 -1.51459485e-01
1.25114903e-01 -1.55855298e-01 -1.20058797e-01 2.62303531e-01
-1.20469701e+00 1.28893006e+00 5.70910120e+00 7.57040918e-01
-6.87634408e-01 1.94659814e-01 3.72916728e-01 -3.74575369e-02
-4.08245504e-01 2.73680031e-01 -7.50538707e-01 2.83910669e-02
3.95713151e-01 -3.54764611e-02 3.01672459e-01 1.00284576e+00
1.35820538e-01 -4.36487973e-01 -1.15901947e+00 1.11407328e+00
4.81975168e-01 -1.21385133e+00 3.60134989e-01 -2.78591841e-01
3.11806351e-01 -4.86359298e-01 -2.76084185e-01 3.86676997e-01
8.22540000e-02 -9.18269575e-01 9.12406623e-01 6.79642916e-01
4.75299358e-01 -4.81928200e-01 3.26282978e-01 3.69506150e-01
-1.46955645e+00 -3.79027098e-01 -1.71170741e-01 -2.61960536e-01
1.55320361e-01 1.99988544e-01 -6.97896123e-01 4.80526626e-01
7.99324274e-01 9.67215836e-01 -6.81964874e-01 6.32040560e-01
-6.42922223e-01 3.59252572e-01 1.30817428e-01 -2.07121465e-02
-5.44403568e-02 2.86950082e-01 4.13852841e-01 8.60291183e-01
-6.34067059e-02 3.94551575e-01 2.93646574e-01 6.31021619e-01
7.77308792e-02 1.17360719e-01 -6.75993025e-01 -1.33671030e-01
3.58365327e-01 9.31253135e-01 -6.23716235e-01 -4.23160523e-01
-6.56731427e-01 1.13893950e+00 5.34642279e-01 3.73787045e-01
-7.40452528e-01 2.75279153e-02 6.31423354e-01 3.34182769e-01
3.57248098e-01 -3.87162827e-02 -7.73733109e-02 -1.21974671e+00
3.68610263e-01 -9.44930971e-01 9.95186508e-01 -1.27163935e+00
-1.06592667e+00 3.03122520e-01 2.02542663e-01 -1.28050995e+00
-2.30048358e-01 -6.32729828e-01 -3.35667461e-01 5.13073146e-01
-1.81268525e+00 -1.67266846e+00 -3.53959084e-01 1.01794624e+00
8.08686018e-01 4.29009274e-02 7.33573973e-01 -7.68282712e-02
-5.92520088e-02 -2.56615542e-02 -7.95532823e-01 -2.37586629e-02
5.79833508e-01 -1.24239111e+00 8.63796622e-02 9.90467250e-01
3.12254548e-01 4.64753360e-01 4.72300828e-01 -6.20387137e-01
-1.57910979e+00 -9.03842032e-01 1.18653977e+00 -7.48377025e-01
7.46066570e-01 -1.83136106e-01 -8.59204531e-01 7.01380730e-01
2.09801793e-02 7.43384138e-02 7.59653449e-01 -1.05485000e-01
-6.27488434e-01 -1.74027294e-01 -9.42888856e-01 7.04578757e-01
1.49326193e+00 -8.94509435e-01 -9.96858180e-01 2.70155966e-01
7.19242454e-01 -2.50773698e-01 -6.64363503e-01 3.58291686e-01
5.31098068e-01 -1.31601262e+00 1.23746502e+00 -7.20251143e-01
5.56134045e-01 -7.55966604e-01 -4.83098298e-01 -6.29215717e-01
-1.01029903e-01 -9.15444344e-02 -3.16857517e-01 9.95866418e-01
1.80905506e-01 -3.97552043e-01 7.03689516e-01 8.69535148e-01
-3.55536602e-02 -6.83691025e-01 -9.23158765e-01 -4.99326587e-01
-6.58365607e-01 -6.71242774e-01 5.93615949e-01 7.77283728e-01
1.38293266e-01 2.22819760e-01 -4.15665746e-01 2.31613204e-01
3.85299951e-01 5.80120742e-01 7.80939162e-01 -8.92785549e-01
-4.61334050e-01 -3.05147290e-01 -5.76883316e-01 -1.05767548e+00
3.01523268e-01 -7.02663600e-01 2.61968784e-02 -2.15843177e+00
3.71776342e-01 -1.00356601e-01 -4.29392517e-01 9.68065262e-01
-3.65958691e-01 1.47453457e-01 5.23605347e-01 4.26558912e-01
-1.31139863e+00 3.42541456e-01 1.40702367e+00 -3.50788571e-02
3.89035270e-02 -1.20244734e-01 -9.48653817e-01 6.20761096e-01
7.97610760e-01 -1.47205397e-01 -8.87427568e-01 -3.47765535e-01
3.95280659e-01 3.78530771e-01 8.53315532e-01 -9.49396074e-01
3.94250661e-01 -6.62798345e-01 4.14405167e-02 -5.14835835e-01
5.35156131e-01 -9.32745337e-01 1.19138218e-01 8.56531486e-02
-5.08476496e-01 -1.46399096e-01 -1.29217893e-01 7.73598731e-01
-6.38662219e-01 1.29058287e-02 3.81753832e-01 -4.32448268e-01
-1.73855603e+00 1.20119877e-01 -3.58636081e-01 -5.57765327e-02
1.33651030e+00 -3.24586123e-01 -6.37560129e-01 -5.18878281e-01
-7.59217918e-01 3.52943689e-01 3.95353705e-01 6.77664280e-01
1.04005969e+00 -1.25981188e+00 -4.76186842e-01 -9.74936560e-02
7.45114207e-01 -2.04407424e-01 4.44285482e-01 7.34729946e-01
-1.05200194e-01 4.57777560e-01 -1.21997356e-01 -3.32566172e-01
-1.65427852e+00 7.32626319e-01 2.20412955e-01 -3.04876417e-01
-3.43867064e-01 6.77815139e-01 4.03662622e-01 -1.81725085e-01
1.27684802e-01 4.63265553e-02 -5.13465703e-01 -3.94024141e-03
6.52473867e-01 5.62471002e-02 -1.88044876e-01 -9.45020854e-01
-8.02744985e-01 5.34943879e-01 6.47812560e-02 -3.32360193e-02
1.05382347e+00 -1.02639601e-01 -3.30823064e-01 3.99309784e-01
7.81684160e-01 -3.87778342e-01 -9.88780737e-01 -2.32289046e-01
1.04753397e-01 -7.73239732e-01 -2.12598965e-01 -1.10599554e+00
-7.20764101e-01 6.88062668e-01 2.12383106e-01 2.04450250e-01
1.43512225e+00 5.62116206e-01 1.84133083e-01 3.80566746e-01
6.40476525e-01 -9.95941281e-01 5.97383916e-01 3.69734794e-01
1.31805420e+00 -1.38612270e+00 -1.44562507e-02 -6.99292541e-01
-1.15286863e+00 8.92594516e-01 1.04156315e+00 2.68679857e-01
4.60695207e-01 -2.45545283e-01 1.82176828e-01 -5.57830632e-01
-1.01828671e+00 -7.24182487e-01 3.87567341e-01 6.08859301e-01
3.37998241e-01 1.06232151e-01 -2.72006065e-01 2.05054000e-01
3.80235106e-01 2.18603700e-01 2.64062017e-01 1.45118940e+00
-4.51594591e-01 -1.19751227e+00 -2.93718666e-01 2.61307240e-01
-2.16608360e-01 -4.85857464e-02 -7.16179788e-01 6.57960951e-01
-1.33443341e-01 1.29922080e+00 -2.01410994e-01 -8.56830329e-02
5.69818437e-01 2.64406532e-01 3.42330635e-01 -6.24758601e-01
-3.68642062e-01 -3.21065545e-01 4.30649519e-01 -1.14730656e+00
-9.64753449e-01 -5.96471608e-01 -1.48004830e+00 2.60510206e-01
3.82220715e-01 -2.28165805e-01 3.64769995e-01 9.66468751e-01
4.77481753e-01 2.84513205e-01 2.56308973e-01 -2.68434048e-01
-2.68877149e-01 -4.91590142e-01 -3.40239048e-01 8.93885672e-01
1.29787311e-01 -7.17227101e-01 -1.08792856e-01 2.01706648e-01] | [10.32751178741455, 1.2205402851104736] |
11335599-abe4-41b0-a0cf-629c015b0e8e | topologically-aware-deformation-fields-for | 2205.06267 | null | https://arxiv.org/abs/2205.06267v2 | https://arxiv.org/pdf/2205.06267v2.pdf | Topologically-Aware Deformation Fields for Single-View 3D Reconstruction | We present a framework for learning 3D object shapes and dense cross-object 3D correspondences from just an unaligned category-specific image collection. The 3D shapes are generated implicitly as deformations to a category-specific signed distance field and are learned in an unsupervised manner solely from unaligned image collections and their poses without any 3D supervision. Generally, image collections on the internet contain several intra-category geometric and topological variations, for example, different chairs can have different topologies, which makes the task of joint shape and correspondence estimation much more challenging. Because of this, prior works either focus on learning each 3D object shape individually without modeling cross-instance correspondences or perform joint shape and correspondence estimation on categories with minimal intra-category topological variations. We overcome these restrictions by learning a topologically-aware implicit deformation field that maps a 3D point in the object space to a higher dimensional point in the category-specific canonical space. At inference time, given a single image, we reconstruct the underlying 3D shape by first implicitly deforming each 3D point in the object space to the learned category-specific canonical space using the topologically-aware deformation field and then reconstructing the 3D shape as a canonical signed distance field. Both canonical shape and deformation field are learned end-to-end in an inverse-graphics fashion using a learned recurrent ray marcher (SRN) as a differentiable rendering module. Our approach, dubbed TARS, achieves state-of-the-art reconstruction fidelity on several datasets: ShapeNet, Pascal3D+, CUB, and Pix3D chairs. Result videos and code at https://shivamduggal4.github.io/tars-3D/ | ['Deepak Pathak', 'Shivam Duggal'] | 2022-05-12 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Duggal_Topologically-Aware_Deformation_Fields_for_Single-View_3D_Reconstruction_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Duggal_Topologically-Aware_Deformation_Fields_for_Single-View_3D_Reconstruction_CVPR_2022_paper.pdf | cvpr-2022-1 | ['single-view-3d-reconstruction'] | ['computer-vision'] | [-9.89300907e-02 -1.30275026e-01 6.30066022e-02 -6.13207996e-01
-7.86987841e-01 -9.74924207e-01 6.98025525e-01 -2.07687721e-01
-4.59483825e-02 8.25379938e-02 -3.69094796e-02 -6.21055141e-02
-5.14084613e-03 -8.13595414e-01 -1.23078096e+00 -3.95737052e-01
7.98677951e-02 1.19948506e+00 2.67015040e-01 4.70887944e-02
2.20375359e-01 8.79315555e-01 -1.33352923e+00 9.47889537e-02
7.62030721e-01 9.28191423e-01 5.26756458e-02 4.59084302e-01
-2.88628608e-01 -1.90849647e-01 -9.71680731e-02 -2.96488792e-01
7.48942912e-01 2.82108653e-02 -7.45983899e-01 5.37383318e-01
1.21736383e+00 -2.27906823e-01 -1.27566829e-01 9.84502912e-01
1.74529985e-01 4.85512353e-02 9.46098387e-01 -1.10150135e+00
-8.09852064e-01 -1.08788498e-01 -6.98880315e-01 -5.08977711e-01
3.14102203e-01 3.29757929e-01 8.40946019e-01 -1.32590914e+00
1.02921927e+00 1.59192312e+00 5.57618439e-01 4.56328005e-01
-1.77848101e+00 -5.70197761e-01 2.83707023e-01 -2.22709864e-01
-1.37517071e+00 -8.46716634e-04 1.10141635e+00 -7.88991749e-01
6.86635256e-01 1.30691424e-01 7.83639491e-01 9.01495218e-01
-2.42581472e-01 5.58962703e-01 1.07031608e+00 -6.75047785e-02
1.88106000e-01 -2.51922339e-01 -9.35209468e-02 7.95658946e-01
1.37036115e-01 6.56068102e-02 -1.53914958e-01 -1.33725181e-01
1.42783153e+00 3.02633435e-01 -1.33450478e-01 -1.20564890e+00
-1.45150590e+00 5.63872039e-01 8.26242149e-01 2.53212382e-03
-1.62720308e-01 2.44086221e-01 7.78903514e-02 8.53036493e-02
5.93591630e-01 1.42942414e-01 -5.79175174e-01 1.45376757e-01
-4.39696997e-01 5.46482563e-01 7.43665636e-01 1.26005232e+00
1.20338976e+00 -8.79498869e-02 3.10516983e-01 6.96029723e-01
4.30864066e-01 7.09791362e-01 -1.62659585e-01 -1.24910367e+00
4.08608139e-01 8.70946169e-01 -3.68538871e-02 -1.01784670e+00
-4.71657142e-02 -2.25936413e-01 -8.76593888e-01 4.54664767e-01
4.88638073e-01 5.35938144e-01 -1.27865279e+00 1.56712246e+00
7.19001234e-01 2.76172161e-01 -4.45115000e-01 1.01422417e+00
7.62309968e-01 4.38621312e-01 -4.52694260e-02 3.59871507e-01
1.32346249e+00 -6.90893292e-01 4.93975282e-02 -1.00845695e-02
2.77639687e-01 -7.23244131e-01 1.26619828e+00 1.26718730e-02
-1.09812641e+00 -4.44093227e-01 -6.90270901e-01 -3.24070483e-01
-3.11694592e-01 -8.25810805e-02 3.39812964e-01 1.21257678e-01
-7.79022276e-01 5.01769185e-01 -9.40055192e-01 -3.49217981e-01
6.99882030e-01 1.71258330e-01 -5.85298240e-01 -2.40197867e-01
-4.35171276e-01 6.32444799e-01 7.01762885e-02 -9.63046849e-02
-9.20602739e-01 -1.22921908e+00 -9.17397022e-01 -4.54312027e-01
3.12310517e-01 -1.12832379e+00 9.45803463e-01 -6.24188840e-01
-1.14936376e+00 1.49560165e+00 1.58881128e-03 3.49425264e-02
7.42813706e-01 3.26542407e-02 1.59365430e-01 -1.39157608e-01
1.62725344e-01 6.71488822e-01 8.39178145e-01 -1.73928761e+00
-6.18340224e-02 -7.88431525e-01 1.92633376e-01 3.30722153e-01
3.79530728e-01 -5.05170465e-01 -6.94715142e-01 -7.90452063e-01
7.23477483e-01 -1.07607031e+00 -1.06874131e-01 7.66616881e-01
-4.09469426e-01 -2.65736222e-01 1.15357232e+00 -5.33743560e-01
2.83871323e-01 -2.09617519e+00 5.14122248e-01 3.29833001e-01
1.70743883e-01 -1.21965721e-01 -3.53358805e-01 -6.88273683e-02
1.19556198e-02 3.54030401e-01 -6.69628918e-01 -5.51375806e-01
6.74178153e-02 4.09581929e-01 -1.61009580e-01 6.15589917e-01
2.93240368e-01 9.98886526e-01 -1.15438724e+00 -3.01649868e-01
4.40919518e-01 7.95634270e-01 -8.90974045e-01 3.62267822e-01
-4.75874484e-01 9.18793738e-01 -5.63975871e-01 8.58702660e-01
1.04307532e+00 -1.19325712e-01 -1.05405062e-01 -5.94590247e-01
-4.40469421e-02 1.91628233e-01 -1.35849512e+00 2.29678917e+00
-6.10984683e-01 2.27865744e-02 2.46181160e-01 -9.42924559e-01
9.97264087e-01 -7.45168328e-02 5.15792251e-01 -2.28804678e-01
-8.39328095e-02 3.28298420e-01 -4.31345701e-01 -1.02228366e-01
1.18265219e-01 -7.33160228e-02 -1.21407717e-01 3.46040487e-01
1.70798972e-01 -8.75621915e-01 -3.37416261e-01 9.74969268e-02
6.57577038e-01 6.15739226e-01 -1.82599083e-01 -2.55706489e-01
3.20741504e-01 -8.69036242e-02 4.54585493e-01 2.22170383e-01
2.53300786e-01 9.71401691e-01 1.37033433e-01 -6.48705900e-01
-1.37613726e+00 -1.77550304e+00 -5.27470112e-01 5.00522137e-01
3.85636836e-01 -1.45267472e-01 -5.20500839e-01 -7.75010765e-01
5.68085372e-01 3.59421283e-01 -5.60667455e-01 1.76808000e-01
-7.09140837e-01 -2.87396746e-04 5.53848557e-02 3.97611499e-01
5.30134320e-01 -8.14050734e-01 -1.86665490e-01 2.49577053e-02
-1.40644088e-01 -1.05776405e+00 -9.51674283e-01 -1.82105824e-01
-1.11370897e+00 -1.23953605e+00 -5.98110974e-01 -8.27452064e-01
1.17502868e+00 3.29458117e-01 1.24109244e+00 9.66720805e-02
-4.54370975e-01 6.88171446e-01 1.63215380e-02 -1.94796603e-02
-2.32520878e-01 -1.80416942e-01 8.63732174e-02 1.23956963e-01
2.86283866e-02 -1.05128562e+00 -7.30357170e-01 5.21443486e-01
-6.84983432e-01 2.01593950e-01 2.12339967e-01 6.48735285e-01
1.28235066e+00 -3.51738572e-01 -5.21880342e-03 -8.29964519e-01
-1.09132238e-01 -2.67271429e-01 -8.88817251e-01 -7.27463216e-02
-1.69972807e-01 6.55538291e-02 3.42415035e-01 -5.05422413e-01
-8.16466093e-01 4.11419272e-01 1.08571067e-01 -9.31142271e-01
-4.10763919e-01 7.67294019e-02 -5.87989867e-01 -1.59944430e-01
5.39504230e-01 1.56469777e-01 1.50940239e-01 -9.21351194e-01
6.38211370e-01 2.35017076e-01 7.89653182e-01 -1.02448905e+00
1.41547346e+00 6.83423996e-01 1.58940196e-01 -6.53950691e-01
-7.65158296e-01 -4.42483425e-01 -1.32602346e+00 -1.47480890e-01
9.55641747e-01 -1.08430183e+00 -5.87255776e-01 6.35963023e-01
-1.27733600e+00 -4.71786767e-01 -4.78737622e-01 2.34348536e-01
-8.20851207e-01 2.99617648e-01 -2.14987308e-01 -2.63286948e-01
-1.93001956e-01 -1.11578012e+00 1.51801610e+00 -2.76630700e-01
-6.47041276e-02 -9.22363460e-01 -1.07231669e-01 4.97058660e-01
9.40490514e-02 7.87716806e-01 1.06696224e+00 7.53354132e-02
-1.23051560e+00 -1.44805938e-01 -2.78186202e-01 3.22901279e-01
2.14153111e-01 -5.69627844e-02 -6.30543172e-01 -3.08306813e-01
-2.80429214e-01 -2.06456333e-01 5.67536891e-01 2.22841293e-01
1.55852926e+00 -2.21990481e-01 -1.64809316e-01 1.20295858e+00
1.54577136e+00 -2.87471145e-01 3.59393924e-01 -2.51228102e-02
1.20361602e+00 4.09378916e-01 2.83624321e-01 1.46956980e-01
6.95575714e-01 8.64016354e-01 8.56241703e-01 9.61130634e-02
-4.16429222e-01 -6.00864291e-01 5.41195422e-02 8.03088188e-01
-3.34692508e-01 3.79305780e-01 -7.36572802e-01 5.15721738e-01
-1.55798995e+00 -7.21975446e-01 -1.75954551e-01 2.50240517e+00
9.60522175e-01 -3.52592804e-02 1.07912570e-01 -3.41938853e-01
5.52985907e-01 5.86265214e-02 -9.19067204e-01 4.34769280e-02
-1.25481812e-02 1.40104994e-01 2.93650776e-01 7.07842350e-01
-8.70418131e-01 9.21723127e-01 4.55944681e+00 5.37643194e-01
-1.09829986e+00 3.69944908e-02 4.32933837e-01 4.68065292e-02
-6.57447278e-01 2.20339626e-01 -5.33564925e-01 2.26757810e-01
1.71219960e-01 -2.54849847e-02 5.61710358e-01 8.99757504e-01
8.26637610e-04 3.03900003e-01 -1.34982872e+00 1.14975536e+00
-1.21912852e-01 -1.50299084e+00 3.37889940e-01 3.86796683e-01
8.21458876e-01 4.48711105e-02 -2.32894029e-02 2.45169597e-03
3.28197062e-01 -8.54162276e-01 1.05896771e+00 5.41357458e-01
1.34562659e+00 -5.06500602e-01 2.67278999e-02 3.62087905e-01
-1.43964052e+00 4.58525747e-01 -3.74447614e-01 2.60231763e-01
1.46310702e-01 4.88225788e-01 -6.81273997e-01 5.18714190e-01
9.04039800e-01 1.03642535e+00 -3.19559634e-01 9.39451039e-01
-1.85837939e-01 2.89489061e-01 -5.44290245e-01 5.25363326e-01
8.35534278e-03 -5.95733464e-01 8.64718854e-01 8.78919363e-01
3.29802722e-01 3.59251738e-01 4.45645958e-01 1.43333638e+00
-4.44988459e-01 -2.72360981e-01 -7.28180647e-01 3.63131344e-01
6.14692271e-01 1.21311545e+00 -6.51768029e-01 -1.76964715e-01
-1.77340627e-01 9.72350359e-01 2.96127230e-01 2.66218096e-01
-5.67432582e-01 1.49649918e-01 1.06054223e+00 6.97262287e-01
3.06234598e-01 -6.13367379e-01 -5.12268484e-01 -1.38216341e+00
2.37853050e-01 -4.19284612e-01 7.46176913e-02 -8.54488075e-01
-1.57034302e+00 2.87432998e-01 1.94438010e-01 -1.47781670e+00
1.19513884e-01 -4.95792925e-01 -6.85540974e-01 1.03201342e+00
-1.22181070e+00 -1.47405922e+00 -5.44654429e-01 7.09619224e-01
4.72355098e-01 1.36384711e-01 8.26037049e-01 6.21953160e-02
4.68151346e-02 3.98786724e-01 -1.56479686e-01 2.05566421e-01
6.03634775e-01 -1.47011626e+00 6.61462069e-01 3.60659748e-01
7.56180510e-02 4.57805037e-01 2.16053009e-01 -6.02155983e-01
-1.78686559e+00 -1.47978759e+00 5.29926658e-01 -1.05096138e+00
3.35387379e-01 -8.15172374e-01 -1.09371448e+00 7.96025753e-01
-3.41560066e-01 6.43173754e-01 1.19881351e-02 -9.67175961e-02
-8.88578773e-01 -7.32567459e-02 -1.45230114e+00 6.39229238e-01
1.83832920e+00 -6.96583271e-01 -5.48785388e-01 4.03827637e-01
7.37627685e-01 -7.73183167e-01 -1.18790960e+00 3.65608424e-01
4.66376603e-01 -6.10189795e-01 1.41709912e+00 -4.40246165e-01
2.92628497e-01 -7.49228776e-01 -3.48775774e-01 -1.35044968e+00
-2.66053885e-01 -4.04065192e-01 3.00501171e-03 1.01104259e+00
9.12770350e-03 -4.77482259e-01 8.86402369e-01 7.18529224e-01
-3.17350537e-01 -7.72571266e-01 -1.07329357e+00 -7.97709644e-01
3.40666085e-01 -4.92399752e-01 8.29413950e-01 1.01302242e+00
-7.80220866e-01 1.78033695e-01 6.53357506e-02 2.70607233e-01
1.21864998e+00 5.63749433e-01 1.16303575e+00 -1.45474160e+00
3.47079746e-02 -4.38204318e-01 -6.93027914e-01 -1.35344791e+00
3.06978554e-01 -1.36184514e+00 -6.03227466e-02 -1.37522185e+00
1.00221917e-01 -8.63037646e-01 2.81387597e-01 5.28827369e-01
3.04756761e-01 3.75791818e-01 3.50345403e-01 2.41273358e-01
-2.16197819e-01 6.88680530e-01 1.79594553e+00 -3.12164336e-01
-1.50361404e-01 -2.00615227e-01 -3.78792286e-01 8.19608331e-01
2.35485569e-01 -2.58108675e-01 -3.03328604e-01 -7.59508312e-01
-2.09463999e-01 -7.97124580e-02 8.96608710e-01 -5.87467670e-01
-7.00069815e-02 -3.76535445e-01 4.64851499e-01 -9.47444439e-01
4.96814758e-01 -1.13505292e+00 4.77439314e-01 1.46325812e-01
-8.43394846e-02 -1.61138624e-01 4.63113636e-02 6.38535559e-01
1.64698645e-01 3.45850468e-01 9.15073037e-01 -3.45984101e-01
-5.69785416e-01 1.04692841e+00 5.46056807e-01 2.85488933e-01
8.25765312e-01 -5.74356377e-01 8.37850496e-02 -6.98451325e-02
-8.40917051e-01 1.45175084e-01 1.16743100e+00 7.10344493e-01
8.10758531e-01 -1.76777506e+00 -7.71798074e-01 5.39535105e-01
2.31220484e-01 1.06093383e+00 3.36102188e-01 4.86517102e-01
-4.53131199e-01 -1.18742540e-01 -1.83546811e-01 -1.29627919e+00
-9.91080403e-01 2.45632112e-01 6.37841523e-01 9.97359306e-02
-8.68651211e-01 7.88838506e-01 6.25375330e-01 -1.29001272e+00
1.60028234e-01 -4.78534937e-01 3.89041960e-01 -3.59395266e-01
6.37817308e-02 2.50676256e-02 1.53904215e-01 -9.81363595e-01
-4.69762504e-01 1.38319135e+00 2.46506691e-01 1.24616697e-01
1.53606522e+00 1.25296324e-01 -2.31020436e-01 3.29573393e-01
1.61490881e+00 -1.40819177e-01 -1.70667636e+00 -5.32443404e-01
-3.89746815e-01 -9.40427423e-01 -1.95291415e-01 -6.83036685e-01
-1.29522657e+00 7.24309623e-01 3.85767967e-01 -3.43475759e-01
7.13128984e-01 4.49944407e-01 6.54876471e-01 7.84038156e-02
7.93821871e-01 -5.41839540e-01 3.09579611e-01 6.42721653e-01
1.49906683e+00 -1.11206877e+00 -8.20302293e-02 -6.17063522e-01
-2.15854153e-01 1.01159894e+00 5.87855816e-01 -6.59652650e-01
1.12667859e+00 -7.50407204e-02 -1.32064223e-01 -3.41726810e-01
-3.74167413e-01 8.73754546e-02 7.42333949e-01 9.32713389e-01
1.62538707e-01 3.27040881e-01 2.07733750e-01 1.83213234e-01
-2.87083268e-01 -2.62104660e-01 1.81622431e-01 6.96065426e-01
-5.68766631e-02 -1.16180336e+00 -4.90730852e-01 3.60348016e-01
1.96436733e-01 2.90340275e-01 -3.17557216e-01 7.66918302e-01
2.72829413e-01 2.51398116e-01 4.83897150e-01 -2.49815896e-01
8.11074197e-01 -1.15963824e-01 8.37441981e-01 -9.03499305e-01
-1.38130322e-01 9.90863442e-02 -3.51620346e-01 -7.69131660e-01
-3.49171668e-01 -8.58053029e-01 -1.46338940e+00 -5.17300963e-01
1.38938889e-01 -2.79265493e-01 8.26704621e-01 5.46851158e-01
5.78988135e-01 -2.98188161e-03 6.70314014e-01 -1.58752167e+00
-5.70473731e-01 -5.65832078e-01 -4.85893995e-01 1.03736603e+00
2.57868290e-01 -9.23367858e-01 -4.16185230e-01 3.02873224e-01] | [8.349109649658203, -3.2173285484313965] |
3499325d-35d0-4f6e-964d-c17ac89c3305 | learning-first-order-rules-with | 2204.13570 | null | https://arxiv.org/abs/2204.13570v1 | https://arxiv.org/pdf/2204.13570v1.pdf | Learning First-Order Rules with Differentiable Logic Program Semantics | Learning first-order logic programs (LPs) from relational facts which yields intuitive insights into the data is a challenging topic in neuro-symbolic research. We introduce a novel differentiable inductive logic programming (ILP) model, called differentiable first-order rule learner (DFOL), which finds the correct LPs from relational facts by searching for the interpretable matrix representations of LPs. These interpretable matrices are deemed as trainable tensors in neural networks (NNs). The NNs are devised according to the differentiable semantics of LPs. Specifically, we first adopt a novel propositionalization method that transfers facts to NN-readable vector pairs representing interpretation pairs. We replace the immediate consequence operator with NN constraint functions consisting of algebraic operations and a sigmoid-like activation function. We map the symbolic forward-chained format of LPs into NN constraint functions consisting of operations between subsymbolic vector representations of atoms. By applying gradient descent, the trained well parameters of NNs can be decoded into precise symbolic LPs in forward-chained logic format. We demonstrate that DFOL can perform on several standard ILP datasets, knowledge bases, and probabilistic relation facts and outperform several well-known differentiable ILP models. Experimental results indicate that DFOL is a precise, robust, scalable, and computationally cheap differentiable ILP model. | ['Hanpin Wang', 'Yongzhi Cao', 'Katsumi Inoue', 'Kun Gao'] | 2022-04-28 | null | null | null | null | ['inductive-logic-programming'] | ['methodology'] | [ 1.43322244e-01 7.26059377e-01 -5.73011398e-01 -8.11784625e-01
-3.91653478e-01 -6.50581241e-01 5.32739401e-01 -6.00514002e-02
5.32832593e-02 7.52159774e-01 2.52968848e-01 -6.63606286e-01
-5.84877014e-01 -1.00113750e+00 -1.51173723e+00 -3.01725060e-01
-4.46886152e-01 9.74882722e-01 -1.34760216e-01 -3.71805996e-01
3.57500277e-02 5.28330863e-01 -1.32383180e+00 8.60273600e-01
9.14744794e-01 1.21348572e+00 -1.90947458e-01 4.63523597e-01
-4.72495258e-01 1.54103553e+00 -1.50432095e-01 -9.16926026e-01
5.65254018e-02 1.52673766e-01 -1.20784461e+00 -7.23484635e-01
3.51019651e-01 -2.69659698e-01 -5.37045896e-01 1.35307574e+00
-4.35298592e-01 1.37740865e-01 7.99611568e-01 -1.38333809e+00
-1.29007292e+00 1.46949947e+00 6.57169968e-02 3.06617767e-02
5.04415691e-01 -2.19582215e-01 1.69478118e+00 -1.01452076e+00
5.16442657e-01 1.86157405e+00 6.21182382e-01 2.94258863e-01
-1.45840323e+00 -2.61050403e-01 2.31366709e-01 4.97711897e-01
-1.14036024e+00 -8.29480439e-02 9.36048508e-01 -3.71920407e-01
1.08160818e+00 4.54094797e-01 4.57613766e-01 9.86846089e-01
2.88041651e-01 1.05216992e+00 7.69041419e-01 -4.38872814e-01
1.89937577e-01 1.90647140e-01 5.77853858e-01 1.36149359e+00
2.56049216e-01 1.89985365e-01 -6.40072107e-01 -2.44367748e-01
8.34608495e-01 -1.26418859e-01 -1.25979662e-01 -3.59117597e-01
-1.49692702e+00 9.45363402e-01 7.93999374e-01 -6.07635826e-02
-3.88381362e-01 7.01508403e-01 5.80810070e-01 4.09594834e-01
2.28073411e-02 7.44058251e-01 -7.72958934e-01 1.65289983e-01
-5.64824976e-02 3.12947422e-01 1.09727252e+00 1.08650577e+00
5.23172319e-01 1.16628468e-01 -4.37757105e-01 5.74822783e-01
5.93513131e-01 6.63540304e-01 1.11067742e-01 -1.02985489e+00
6.77443385e-01 8.37879956e-01 -1.74850821e-01 -1.54034841e+00
-2.36942366e-01 2.08163410e-02 -5.90449810e-01 -3.41909826e-01
1.65628850e-01 1.24284156e-01 -4.59083021e-01 1.57972825e+00
1.11049667e-01 2.92948484e-01 4.79512781e-01 6.52759790e-01
9.69680667e-01 9.60164845e-01 1.30019244e-02 8.85997862e-02
1.40301979e+00 -5.36144376e-01 -6.34647071e-01 1.39076042e-03
7.11935937e-01 1.47268757e-01 1.20490181e+00 5.32189608e-01
-9.58126366e-01 -2.36015439e-01 -8.96334589e-01 -4.04539198e-01
-3.95305604e-01 4.76022601e-01 1.25160050e+00 1.84912264e-01
-6.75298929e-01 6.94605291e-01 -8.21929693e-01 4.04461622e-01
5.07354140e-01 5.55172741e-01 -4.49956357e-01 1.51557878e-01
-1.48759675e+00 9.36577499e-01 1.11192966e+00 6.11891150e-01
-7.96659470e-01 -9.51265872e-01 -1.27222013e+00 1.62945062e-01
6.08671129e-01 -6.19895339e-01 1.08435154e+00 -7.75555730e-01
-1.75878954e+00 7.51979053e-01 -1.67917579e-01 -6.84849560e-01
-1.00600414e-01 -2.67414451e-01 -3.91489238e-01 5.87437227e-02
-1.59894571e-01 4.54096168e-01 7.08572626e-01 -1.38052058e+00
-5.12736678e-01 -1.20839082e-01 6.21400535e-01 9.57349315e-02
1.21901192e-01 -9.46177691e-02 -2.62993127e-02 -3.59674156e-01
3.64969283e-01 -7.68793344e-01 2.21723422e-01 -1.54142991e-01
-9.27626789e-01 -7.07869112e-01 2.91483849e-01 -7.53648996e-01
9.99068081e-01 -1.94600034e+00 5.73536634e-01 6.06592417e-01
4.01733637e-01 -1.41141675e-02 8.92868415e-02 -1.00001939e-01
-2.51437664e-01 8.03166404e-02 -1.50491714e-01 2.82346517e-01
5.24393976e-01 7.75238454e-01 -1.14924705e+00 1.46214843e-01
6.95029080e-01 1.27998042e+00 -1.34415102e+00 -3.86072785e-01
-6.59734756e-02 2.32706711e-01 -7.65377700e-01 1.83388069e-01
-9.51650202e-01 -8.39716941e-02 -5.16621768e-01 6.66261554e-01
2.90875256e-01 -3.92702073e-02 3.95697236e-01 -6.84430122e-01
3.73718441e-01 5.65031469e-01 -1.06411493e+00 1.12794805e+00
-4.25922245e-01 5.66320777e-01 -7.88249075e-01 -1.40158725e+00
1.09847057e+00 1.10317029e-01 -1.20839961e-01 -1.95993915e-01
2.23768100e-01 3.14356953e-01 -1.05933309e-01 -7.62048006e-01
4.22572881e-01 -3.16819012e-01 -1.62320599e-01 -8.66442099e-02
2.78579324e-01 -1.78070232e-01 1.60166565e-02 1.32625192e-01
7.83506453e-01 3.66732538e-01 1.71975061e-01 -1.69504210e-01
7.58316398e-01 -9.94578674e-02 5.95451832e-01 6.36688888e-01
4.60531116e-01 -3.24627727e-01 1.24546456e+00 -7.86036253e-01
-4.38742757e-01 -1.55114532e+00 -1.56536117e-01 1.27869010e+00
-1.30072773e-01 -1.63451478e-01 -2.40394562e-01 -8.59994829e-01
3.88558984e-01 1.20244658e+00 -7.00580418e-01 -2.37482861e-01
-6.70253217e-01 -3.45158130e-01 9.38553393e-01 7.22023845e-01
4.82885510e-01 -1.23372447e+00 2.63673812e-02 -7.70217627e-02
-1.51252061e-01 -1.07499933e+00 -8.84873793e-02 4.91219074e-01
-9.03820395e-01 -1.12567770e+00 3.30581367e-01 -1.01479173e+00
8.87469828e-01 -6.09877229e-01 1.18100536e+00 -1.36815473e-01
3.82578284e-01 1.14369325e-01 3.35291661e-02 -3.40471119e-01
-5.07282734e-01 -3.50256771e-01 3.30300689e-01 1.69543043e-01
3.88046324e-01 -3.91823858e-01 3.07935655e-01 -3.22188050e-01
-6.79618657e-01 1.09282807e-01 4.84421283e-01 1.22199893e+00
9.41542566e-01 2.41591245e-01 3.69257778e-01 -1.40986109e+00
7.90196896e-01 -6.61946952e-01 -9.38801467e-01 7.33858109e-01
-2.81895787e-01 9.00387108e-01 1.00211728e+00 -3.88980269e-01
-1.15410495e+00 -4.84557673e-02 2.97755539e-01 -7.32290447e-01
1.99418336e-01 1.18291819e+00 -1.53108358e-01 -3.92079838e-02
8.39369476e-01 2.58362740e-01 -2.61823595e-01 -2.66436078e-02
7.71384776e-01 2.08555609e-01 9.34351444e-01 -1.51317203e+00
9.44653094e-01 2.30779603e-01 3.97318691e-01 -3.64379883e-01
-1.38351846e+00 3.01410198e-01 -5.51612735e-01 2.08627716e-01
5.46512008e-01 -4.74012673e-01 -1.30817723e+00 -6.73717856e-02
-1.50106144e+00 -2.65858710e-01 -3.70764822e-01 4.46112722e-01
-7.49397218e-01 7.67688677e-02 -7.09260225e-01 -8.08513045e-01
-2.10203618e-01 -1.02491176e+00 7.23410189e-01 -1.27743468e-01
-2.62250185e-01 -1.34456801e+00 -2.67337888e-01 1.19760901e-01
-2.80268341e-01 3.32318425e-01 1.74318743e+00 -8.02921474e-01
-8.16716075e-01 -1.65919453e-01 -3.59689355e-01 5.80095470e-01
-2.49174386e-01 1.99873954e-01 -6.68189406e-01 4.77012962e-01
-1.20276093e-01 -4.23384398e-01 5.71090817e-01 2.03325704e-01
1.49641275e+00 -1.03243184e+00 -2.65476584e-01 8.44725370e-01
1.28397810e+00 1.01299152e-01 5.77929854e-01 1.11791059e-01
1.04240406e+00 4.46527541e-01 4.18088645e-01 -9.19211954e-02
8.01686823e-01 1.80593774e-01 3.68234217e-01 5.98330855e-01
4.59816426e-01 -7.52736986e-01 5.92502296e-01 8.18443775e-01
-4.61700976e-01 4.53774333e-01 -1.02199244e+00 3.26304525e-01
-2.00316072e+00 -1.07054210e+00 -9.81481299e-02 1.78212059e+00
1.48277986e+00 4.14366364e-01 -4.35017616e-01 7.99602643e-02
2.16584504e-01 -2.59505928e-01 -4.55582470e-01 -7.76551843e-01
-8.52042437e-02 2.68624216e-01 3.74776214e-01 7.96434522e-01
-1.00998199e+00 1.04629409e+00 5.77296591e+00 4.69358355e-01
-9.08926189e-01 -2.56873459e-01 3.39280009e-01 2.71405548e-01
-7.92759776e-01 4.95135784e-02 -7.16441095e-01 3.03447783e-01
8.26128125e-01 -1.00402877e-01 1.04908180e+00 1.15264404e+00
-3.32414627e-01 2.25192010e-01 -1.91764188e+00 1.07265711e+00
-7.41570219e-02 -1.84121120e+00 4.32972610e-01 -5.20068109e-01
6.30986094e-01 -2.17105702e-01 1.89163890e-02 8.94932210e-01
6.69711530e-01 -1.30873144e+00 1.00668085e+00 1.15899837e+00
6.82958364e-01 -7.76915133e-01 7.38839328e-01 4.91630770e-02
-9.48720872e-01 -4.71627235e-01 -4.98701662e-01 -8.95496979e-02
-2.72787750e-01 6.08114600e-01 -9.56726551e-01 4.67147112e-01
4.94339049e-01 7.66988218e-01 -2.79405802e-01 1.60279199e-01
-8.37037444e-01 5.77211797e-01 -3.26312661e-01 -3.13851506e-01
1.66338578e-01 -2.42212757e-01 3.26433331e-01 1.08726346e+00
-1.76070496e-01 2.35133618e-01 -5.90479150e-02 1.59125662e+00
-4.21498597e-01 -4.83039111e-01 -6.68127298e-01 -2.86455870e-01
5.58541954e-01 9.67007577e-01 -1.04855947e-01 -3.74671608e-01
-8.47232267e-02 5.16187608e-01 9.04195428e-01 4.25869554e-01
-9.78949785e-01 -4.65219885e-01 5.38497329e-01 -4.73674744e-01
2.02888504e-01 -2.49808677e-03 -6.07547343e-01 -1.26908088e+00
2.11495101e-01 -9.89808321e-01 4.30899799e-01 -8.08933854e-01
-1.44626153e+00 4.16216284e-01 5.13008118e-01 -5.99738598e-01
-1.84138536e-01 -1.22700453e+00 -4.05991584e-01 7.55941808e-01
-1.55071747e+00 -1.20391774e+00 2.48149425e-01 5.82298219e-01
-7.00896084e-02 -3.88461292e-01 1.04850185e+00 -1.84578121e-01
-5.45158327e-01 6.12160921e-01 -1.79708466e-01 4.86502320e-01
-2.20343709e-01 -1.48916733e+00 -5.47932014e-02 4.28018451e-01
2.61161834e-01 1.22603536e+00 5.98027349e-01 -5.83781242e-01
-2.06160712e+00 -1.21881413e+00 1.12012458e+00 -6.94963098e-01
1.03568923e+00 -1.71060577e-01 -9.76207972e-01 1.50976586e+00
-4.11485702e-01 2.20390275e-01 3.25247943e-01 5.19107103e-01
-7.86671400e-01 -3.64385635e-01 -1.01179171e+00 8.32178116e-01
1.02663839e+00 -9.68507707e-01 -1.30470741e+00 5.96467912e-01
9.95132744e-01 -8.63422513e-01 -1.10444748e+00 4.13655430e-01
3.86973113e-01 -2.18925506e-01 1.22435009e+00 -1.23477960e+00
7.92130351e-01 -4.11819428e-01 -3.83359998e-01 -1.20234108e+00
-3.43002200e-01 -4.28812027e-01 -9.20433462e-01 8.25065196e-01
6.25097632e-01 -8.44920218e-01 3.65682125e-01 7.74724960e-01
-2.07340226e-01 -1.18971336e+00 -7.14635551e-01 -7.20901906e-01
-7.94090629e-02 -6.51027739e-01 7.96062469e-01 1.10008013e+00
5.64196467e-01 4.59589273e-01 3.05283535e-02 6.35847926e-01
5.24122894e-01 2.18866333e-01 3.26891035e-01 -1.20731044e+00
-4.41684604e-01 -4.15295780e-01 -5.08616507e-01 -9.51051950e-01
8.95431340e-01 -1.69527626e+00 1.21610545e-01 -1.27353036e+00
-8.69745165e-02 -5.86004376e-01 -2.84325004e-01 1.01610899e+00
-7.01284595e-03 -5.52773595e-01 -8.68080556e-02 -1.82094164e-02
-2.76033252e-01 6.13119543e-01 1.07338154e+00 -4.86092776e-01
-1.06559314e-01 -3.97023976e-01 -7.36116230e-01 1.09229827e+00
6.06325209e-01 -3.15696090e-01 -7.02737749e-01 -6.87836528e-01
9.35641587e-01 1.78951636e-01 8.53916824e-01 -5.46827257e-01
3.01580727e-01 -7.10070014e-01 1.23249769e-01 -4.60735649e-01
2.70147681e-01 -7.90710628e-01 -3.82852890e-02 2.47464672e-01
-7.47865736e-01 -1.29910558e-01 5.93957119e-02 5.38855612e-01
-9.20704082e-02 -3.31216097e-01 2.12813482e-01 -2.45634746e-02
-9.01676118e-01 6.21823408e-02 2.19312519e-01 1.80226073e-01
7.46008992e-01 1.85205042e-01 -3.66586626e-01 8.76783729e-02
-7.39912450e-01 2.64047742e-01 -5.66486597e-01 1.94998428e-01
1.03599930e+00 -1.58691502e+00 -4.09752160e-01 1.95228338e-01
4.37015593e-02 4.48624730e-01 -2.57079363e-01 7.73024142e-01
-7.51798391e-01 8.66970122e-01 8.02094191e-02 -3.46106172e-01
-5.84153891e-01 5.55364907e-01 4.81872439e-01 -3.24594110e-01
-5.97077131e-01 1.24548602e+00 -4.27844636e-02 -1.10601640e+00
5.68995774e-01 -1.32962215e+00 -2.23679289e-01 -3.72411728e-01
3.92173916e-01 1.69718087e-01 -2.55361080e-01 -3.67243469e-01
-3.26709747e-01 -1.96008720e-02 6.16426803e-02 3.20764810e-01
1.43823802e+00 5.86509526e-01 -6.63277328e-01 7.93434441e-01
1.05272770e+00 -2.93309242e-01 -8.70765686e-01 -5.13110280e-01
3.85270506e-01 -1.51950568e-01 -7.93038309e-02 -6.49202764e-01
-8.81956935e-01 4.84722048e-01 -3.31513435e-01 3.14497575e-02
6.18549585e-01 2.44794071e-01 3.58027786e-01 1.28188598e+00
4.36114371e-01 -8.84352088e-01 -5.08563519e-02 1.15875113e+00
1.34544873e+00 -9.50876296e-01 -1.70421943e-01 -6.48518384e-01
-6.89388871e-01 1.43176973e+00 4.19620275e-01 -4.01676506e-01
6.83366358e-01 2.51380149e-02 -5.81380427e-01 -5.57872713e-01
-8.81323636e-01 2.88901657e-01 6.58759832e-01 6.85552120e-01
1.77866772e-01 4.82392162e-01 1.90588117e-01 1.14966714e+00
-6.72223210e-01 2.41711035e-01 1.56191200e-01 4.61718768e-01
-1.11452706e-01 -5.55475593e-01 -2.87996292e-01 7.36432135e-01
3.49168666e-02 -3.92362446e-01 -1.91926509e-02 4.89332259e-01
1.25781208e-01 3.67257297e-01 -8.47540423e-02 -4.35603112e-01
4.79272604e-01 3.45607430e-01 7.61951447e-01 -7.04658806e-01
-1.38973087e-01 -1.19314897e+00 3.62639844e-01 -5.67028880e-01
-2.54796464e-02 -5.23487568e-01 -1.90510249e+00 -4.65185881e-01
-4.53474447e-02 -1.75738961e-01 3.46220702e-01 1.29561400e+00
1.62961259e-01 6.60032630e-01 3.85138631e-01 -3.71445805e-01
-9.94297802e-01 -5.33067822e-01 -5.01928806e-01 3.12692344e-01
3.16962868e-01 -9.44785655e-01 -1.90831855e-01 3.32450688e-01] | [9.15035343170166, 7.556769371032715] |
014f797d-91e1-4d25-bc5d-47aad3aea580 | infogan-interpretable-representation-learning | 1606.03657 | null | http://arxiv.org/abs/1606.03657v1 | http://arxiv.org/pdf/1606.03657v1.pdf | InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets | This paper describes InfoGAN, an information-theoretic extension to the
Generative Adversarial Network that is able to learn disentangled
representations in a completely unsupervised manner. InfoGAN is a generative
adversarial network that also maximizes the mutual information between a small
subset of the latent variables and the observation. We derive a lower bound to
the mutual information objective that can be optimized efficiently, and show
that our training procedure can be interpreted as a variation of the Wake-Sleep
algorithm. Specifically, InfoGAN successfully disentangles writing styles from
digit shapes on the MNIST dataset, pose from lighting of 3D rendered images,
and background digits from the central digit on the SVHN dataset. It also
discovers visual concepts that include hair styles, presence/absence of
eyeglasses, and emotions on the CelebA face dataset. Experiments show that
InfoGAN learns interpretable representations that are competitive with
representations learned by existing fully supervised methods. | ['Pieter Abbeel', 'John Schulman', 'Ilya Sutskever', 'Yan Duan', 'Rein Houthooft', 'Xi Chen'] | 2016-06-12 | infogan-interpretable-representation-learning-1 | http://papers.nips.cc/paper/6399-infogan-interpretable-representation-learning-by-information-maximizing-generative-adversarial-nets | http://papers.nips.cc/paper/6399-infogan-interpretable-representation-learning-by-information-maximizing-generative-adversarial-nets.pdf | neurips-2016-12 | ['unsupervised-image-classification', 'unsupervised-mnist'] | ['computer-vision', 'methodology'] | [ 2.79914290e-01 5.98382354e-01 -7.54866526e-02 -5.38090527e-01
-2.75171220e-01 -9.12224829e-01 8.73509169e-01 -8.25081050e-01
2.02252910e-01 5.84924698e-01 3.68813634e-01 -2.40036231e-02
-7.90698677e-02 -7.62595117e-01 -7.95743525e-01 -9.03179884e-01
6.11868873e-03 8.11433613e-01 -7.92503417e-01 -1.00523531e-01
1.27862729e-02 5.80565572e-01 -1.31110406e+00 8.03077370e-02
3.39530617e-01 6.52463675e-01 -2.93239594e-01 8.39668989e-01
2.41942167e-01 9.12058532e-01 -7.62185156e-01 -4.45361584e-01
4.88802522e-01 -7.79432237e-01 -5.41050494e-01 4.66885060e-01
5.06313384e-01 -3.50745171e-01 -7.50137150e-01 1.07169962e+00
3.96918595e-01 -1.42484620e-01 9.40711558e-01 -1.58264887e+00
-1.44351685e+00 5.52843332e-01 -3.75417978e-01 8.81066471e-02
1.19412862e-01 7.46367872e-02 1.05209124e+00 -5.99051356e-01
6.63830101e-01 1.55356956e+00 1.05462953e-01 7.87495673e-01
-1.68879795e+00 -9.17607367e-01 -1.08710177e-01 -5.07424250e-02
-1.19385242e+00 -4.91321206e-01 1.04452777e+00 -4.95185673e-01
5.27078509e-01 3.21779341e-01 7.05487669e-01 1.81686234e+00
1.65588588e-01 8.58652532e-01 1.37541497e+00 -3.33692133e-01
3.78617197e-01 7.82096758e-03 -1.52952015e-01 1.02979922e+00
3.43189538e-01 5.22874057e-01 -8.14691424e-01 -1.55174688e-01
1.00972116e+00 1.00153029e-01 -1.17196582e-01 -7.35851109e-01
-1.19030154e+00 1.18860900e+00 2.78236389e-01 -2.22479865e-01
2.19211783e-04 3.07229519e-01 -1.83151335e-01 4.93235022e-01
4.40840840e-01 7.01333582e-01 -3.92601132e-01 2.47512534e-01
-5.12288392e-01 -3.00755091e-02 1.13658488e+00 9.26516652e-01
7.39904284e-01 6.22874916e-01 2.82130502e-02 3.90410215e-01
5.35812974e-01 1.08869660e+00 4.70781714e-01 -1.10655117e+00
1.52753908e-02 5.80045879e-01 -3.41539770e-01 -8.18206847e-01
-6.18133098e-02 -2.13892654e-01 -9.02109623e-01 6.83882952e-01
1.38029262e-01 -4.57627863e-01 -1.14505315e+00 1.99380696e+00
-2.97627121e-01 2.20884308e-01 3.50335240e-01 7.96300232e-01
8.50190759e-01 2.55727947e-01 -2.72141784e-01 1.33713499e-01
1.24763095e+00 -6.44895256e-01 -8.01012337e-01 -6.54675126e-01
-2.23185822e-01 -5.24018407e-01 4.96243298e-01 3.41148883e-01
-9.53235030e-01 -3.46212745e-01 -1.39784360e+00 -9.99910086e-02
-3.24474931e-01 9.30252597e-02 1.06647277e+00 8.82570326e-01
-9.57169473e-01 4.91849929e-01 -9.56224322e-01 2.44629439e-02
5.38434863e-01 5.04246175e-01 -6.25844896e-01 -2.95978151e-02
-8.01246703e-01 6.33398294e-01 2.33250529e-01 -6.82887957e-02
-1.38595593e+00 -3.29040349e-01 -1.16615796e+00 -2.12991629e-02
2.37009197e-01 -8.73118401e-01 8.27576518e-01 -1.25010026e+00
-1.68229687e+00 1.19707811e+00 -9.20147356e-03 -2.41806835e-01
1.14731491e-01 -1.19377382e-01 -5.20546734e-01 7.64999986e-02
-1.52057484e-01 6.54988825e-01 1.27820599e+00 -1.29217267e+00
3.06378424e-01 -7.11958170e-01 1.34987548e-01 8.71582255e-02
-1.26826823e-01 -3.74239534e-01 -1.01574332e-01 -8.97520304e-01
1.46808237e-01 -1.32798302e+00 -5.45808636e-02 1.55212125e-02
-8.37981999e-01 2.80979842e-01 8.49392056e-01 -5.46761572e-01
1.55297592e-01 -2.14225030e+00 7.41488457e-01 1.39410555e-01
5.52029133e-01 -2.80087799e-01 -3.81776661e-01 8.26630145e-02
-4.58897650e-01 2.27709740e-01 -8.97125006e-02 -5.25032401e-01
1.73043042e-01 7.13963747e-01 -3.56988609e-01 5.85701883e-01
2.99222589e-01 1.10726571e+00 -8.21532726e-01 7.89556839e-03
5.71130924e-02 5.74353993e-01 -4.93665785e-01 4.03445691e-01
-1.64374322e-01 4.90501821e-01 -1.76739022e-01 6.09680235e-01
4.66049105e-01 -3.52135688e-01 4.71082628e-01 9.22066644e-02
5.70385456e-01 5.29986136e-02 -8.35500717e-01 1.91641855e+00
-2.31091157e-01 1.04302156e+00 -1.56927019e-01 -7.35349774e-01
1.05110145e+00 3.76305908e-01 1.83597445e-01 -3.46508890e-01
3.72027785e-01 -4.43312943e-01 -4.54658344e-02 -3.40757996e-01
3.26857679e-02 -2.49681592e-01 -2.32814342e-01 6.66298985e-01
6.35563433e-01 -2.69721568e-01 -2.57205248e-01 3.44842643e-01
1.07070756e+00 1.59965098e-01 3.47076386e-01 -1.35728166e-01
-1.45849407e-01 -5.87828338e-01 5.79494655e-01 7.15713263e-01
1.59453869e-01 7.77361214e-01 7.99117565e-01 -4.45361555e-01
-8.72898161e-01 -1.62644446e+00 2.87189931e-01 7.10175455e-01
-2.96713915e-02 -3.10962856e-01 -5.11768699e-01 -8.46452594e-01
-4.22546417e-02 7.63026357e-01 -1.16333127e+00 -4.95882839e-01
-8.00569654e-02 -6.79619193e-01 6.04108930e-01 5.53966403e-01
3.45822901e-01 -1.00012815e+00 -1.91056103e-01 -5.21324217e-01
6.22267574e-02 -1.14599645e+00 -1.87977746e-01 4.77391779e-01
-7.02283204e-01 -1.35834193e+00 -2.46070772e-01 -4.30047303e-01
9.55754280e-01 -3.35496031e-02 1.27550077e+00 -3.25993657e-01
-3.54761511e-01 4.00168300e-01 -1.56974733e-01 -5.64638972e-01
-5.79677224e-01 -2.20217019e-01 6.07886277e-02 1.28426850e-01
4.96300071e-01 -8.78789604e-01 -3.10544461e-01 1.62922606e-01
-8.47587168e-01 2.48661458e-01 6.64719164e-01 8.22989166e-01
3.25271964e-01 -9.65397507e-02 1.75968617e-01 -1.08936584e+00
3.02384377e-01 -5.27664661e-01 -5.03579319e-01 1.75271839e-01
-4.90261942e-01 7.23084569e-01 3.60110372e-01 -5.48765838e-01
-8.87966514e-01 3.59869391e-01 2.94745296e-01 -7.60045230e-01
-3.79930317e-01 -4.09412831e-02 -5.93136489e-01 9.24221333e-03
6.43750489e-01 1.56271845e-01 1.19042389e-01 -3.78374368e-01
7.57632852e-01 1.30767629e-01 7.89557695e-01 -4.02883708e-01
1.15602815e+00 6.44317627e-01 9.57075357e-02 -6.20103419e-01
-1.01871645e+00 2.51407981e-01 -7.49955535e-01 2.16250136e-01
1.02500939e+00 -1.07428813e+00 -7.00874090e-01 2.39432752e-01
-9.82601643e-01 -2.16438174e-01 -6.57012522e-01 3.79024088e-01
-7.66877353e-01 1.14227775e-02 -5.02002001e-01 -5.17432392e-01
-2.05356441e-02 -8.45384657e-01 1.09122741e+00 2.22282469e-01
-3.71222585e-01 -1.09072673e+00 2.22547978e-01 4.17006254e-01
-8.88288021e-02 8.33320379e-01 9.67398047e-01 -7.84642041e-01
-7.48241127e-01 -1.21572450e-01 1.01626880e-01 5.25912762e-01
4.03129160e-01 -1.56480342e-01 -1.20685267e+00 -2.34092727e-01
2.66583636e-02 -4.15134519e-01 1.00331116e+00 1.63578123e-01
1.00350833e+00 -7.18303740e-01 1.93490833e-02 1.07541406e+00
1.18886161e+00 1.73989117e-01 6.93351805e-01 -1.48172215e-01
9.26564038e-01 2.03523695e-01 -3.52653593e-01 4.28462774e-01
1.40395895e-01 2.25982666e-01 6.99584126e-01 -1.28978789e-01
-2.03574330e-01 -3.37576926e-01 5.49465001e-01 6.25310779e-01
-2.14095682e-01 -3.51345062e-01 -4.55582321e-01 2.73069412e-01
-1.41957343e+00 -9.94350910e-01 3.39142025e-01 1.78333640e+00
7.30268598e-01 -3.97340246e-02 -2.66706973e-01 1.38140187e-01
3.86618227e-01 4.30332094e-01 -8.37624013e-01 -4.60379899e-01
-4.10284311e-01 6.71210647e-01 3.73478472e-01 4.06976610e-01
-1.03692794e+00 7.96866298e-01 7.68786621e+00 8.05798769e-02
-8.17225814e-01 -1.25073781e-02 4.55489427e-01 -3.27348679e-01
-5.90110183e-01 -7.12213442e-02 -3.85789931e-01 3.04645121e-01
7.89326489e-01 -2.09654331e-01 9.94919300e-01 8.53522360e-01
-4.39311683e-01 4.48590189e-01 -1.69396901e+00 1.04158163e+00
7.22653151e-01 -1.03859091e+00 1.42069623e-01 4.80769575e-01
1.03560030e+00 -8.74958262e-02 5.88754296e-01 8.79114717e-02
1.05154419e+00 -1.53537941e+00 5.93032002e-01 6.63619995e-01
8.69004786e-01 -7.99722493e-01 3.50145757e-01 2.90963352e-01
-4.29153353e-01 1.00445911e-01 -2.54871547e-01 -1.36629313e-01
-3.67632329e-01 3.01189214e-01 -9.01646614e-01 2.10770696e-01
2.56434172e-01 8.25140655e-01 -7.63309360e-01 2.96080321e-01
-9.81742382e-01 4.76167083e-01 2.60342713e-02 2.68851310e-01
-2.32592180e-01 -2.17197955e-01 7.32954264e-01 6.68994308e-01
1.96384609e-01 1.32715553e-01 -1.29444391e-01 1.37120438e+00
-5.51074445e-01 -6.12326801e-01 -1.02162075e+00 -4.54291552e-01
1.62387639e-01 1.19834805e+00 -5.81440330e-01 -1.38029814e-01
-8.99902433e-02 1.40044391e+00 1.72059253e-01 5.47278583e-01
-5.51265359e-01 1.25162571e-03 1.07327628e+00 -4.22233582e-01
4.23885047e-01 -2.74525642e-01 -3.55832547e-01 -1.51050627e+00
-4.69385386e-01 -1.01521552e+00 6.14681561e-03 -1.04375648e+00
-1.48947012e+00 5.65222800e-01 -2.51756102e-01 -9.73993421e-01
-6.19602323e-01 -7.80604482e-01 -4.12656307e-01 7.08572745e-01
-1.00774586e+00 -1.14116395e+00 -3.22523803e-01 8.28873694e-01
3.94081116e-01 -6.39560342e-01 1.45070934e+00 -3.54518026e-01
-5.82895279e-01 5.08396626e-01 1.06163621e-01 6.55371308e-01
4.47067767e-01 -1.62941730e+00 6.30024433e-01 8.59719396e-01
8.54183078e-01 6.29450560e-01 7.27063596e-01 -4.08365071e-01
-1.58329523e+00 -1.00332546e+00 3.79231691e-01 -7.16832697e-01
4.35206085e-01 -7.30728507e-01 -3.34501386e-01 1.26094520e+00
5.70939004e-01 1.81181505e-01 1.13531375e+00 3.27097297e-01
-9.18234706e-01 2.64960706e-01 -1.00850379e+00 6.44813716e-01
1.02279127e+00 -8.72569740e-01 -8.89364421e-01 3.42513502e-01
6.38000488e-01 -4.67640877e-01 -6.24432802e-01 -7.16039389e-02
6.46858931e-01 -9.65938449e-01 1.17155552e+00 -9.69375134e-01
6.79554045e-01 9.72347707e-02 -3.34254384e-01 -1.66705501e+00
-3.35518628e-01 -8.06217730e-01 -5.43553293e-01 9.00248230e-01
1.34400517e-01 -5.13682306e-01 8.06976140e-01 4.34162885e-01
3.47070843e-01 -2.88265973e-01 -6.18369699e-01 -7.39928186e-01
-1.41255677e-01 -3.39034498e-02 6.46981776e-01 1.13685441e+00
-4.69649106e-01 6.87780082e-01 -7.78443396e-01 2.99193054e-01
1.09621346e+00 2.57360190e-01 7.00191975e-01 -1.52991605e+00
-7.57490337e-01 -4.97024395e-02 -7.84426153e-01 -7.16743410e-01
6.66650414e-01 -1.14317501e+00 -2.46573254e-01 -1.09746921e+00
4.02638108e-01 1.33016095e-01 -2.81890243e-01 7.97810555e-01
1.35185465e-01 5.86593390e-01 2.83825576e-01 -5.20486608e-02
-2.15192571e-01 5.90785623e-01 1.36850905e+00 -5.66937625e-01
2.22893313e-01 -1.37456536e-01 -1.17391980e+00 9.87733603e-01
7.29135394e-01 -7.06898332e-01 -6.09011889e-01 -4.53017563e-01
4.11505103e-01 1.98002249e-01 7.91452885e-01 -6.39984548e-01
-1.33808315e-01 -7.75941536e-02 9.33802128e-01 -2.34252274e-01
6.83253884e-01 -6.75234020e-01 2.47210488e-01 1.83854952e-01
-6.05947733e-01 -3.44648585e-02 -3.58972661e-02 8.34735692e-01
1.84005312e-05 1.25530511e-01 6.30280435e-01 -2.43556976e-01
-4.44760591e-01 2.27838963e-01 -2.92724431e-01 1.55068189e-01
8.32256496e-01 8.50168690e-02 -5.44960499e-01 -5.90010941e-01
-1.05566049e+00 -2.09106296e-01 4.50934619e-01 7.42483079e-01
7.44527817e-01 -1.62122667e+00 -6.78609550e-01 7.06783175e-01
-7.47928545e-02 -2.99556762e-01 -1.71395212e-01 2.14824766e-01
-2.09884495e-01 1.14731871e-01 -6.79551959e-01 -2.68540263e-01
-1.32579207e+00 5.13383329e-01 2.11256355e-01 8.05707276e-03
-4.90133107e-01 9.00064111e-01 4.54083920e-01 -3.73156488e-01
-7.95549899e-02 -4.65016477e-02 2.29855049e-02 -1.40531972e-01
4.29363817e-01 1.22695781e-01 -3.01594079e-01 -6.37988687e-01
-2.25687176e-01 2.07149014e-01 1.19932853e-01 -3.37613463e-01
1.49785161e+00 1.14564627e-01 -2.45008677e-01 5.68200827e-01
1.46584308e+00 1.12372369e-01 -1.35331810e+00 -2.12291211e-01
-4.53042805e-01 -4.55302507e-01 1.61620937e-02 -9.94929671e-01
-1.49122107e+00 7.55298436e-01 7.38752306e-01 7.72215500e-02
1.12591755e+00 1.88043654e-01 3.10626566e-01 4.85500693e-01
8.00311565e-02 -6.50566995e-01 4.82399166e-01 3.30109954e-01
9.62023675e-01 -1.11228299e+00 1.04878746e-01 -1.64380059e-01
-6.83282316e-01 1.09905577e+00 3.16523880e-01 -4.97346252e-01
6.87552989e-01 5.43819785e-01 3.49801779e-02 -4.91053522e-01
-8.32963824e-01 -1.63012385e-01 6.91487610e-01 7.19362319e-01
8.65222365e-02 3.97643626e-01 5.31563222e-01 5.85103214e-01
-6.50603712e-01 -3.66139680e-01 5.43434680e-01 6.01364732e-01
1.39511868e-01 -1.09487355e+00 -1.62826717e-01 1.34209946e-01
-2.83027917e-01 -4.25221734e-02 -9.18764830e-01 7.99663067e-01
2.21719190e-01 8.21807384e-01 2.01811776e-01 -6.40353322e-01
-1.31505793e-02 2.69148916e-01 9.07127857e-01 -6.99650526e-01
-7.12326095e-02 -1.21687949e-01 -1.85329318e-01 -6.40947700e-01
-4.72288728e-01 -5.67571640e-01 -8.43549609e-01 -2.32185408e-01
-1.12613596e-01 -2.04852015e-01 6.61294162e-01 1.05225623e+00
2.12297663e-01 7.12043166e-01 7.72821248e-01 -8.11869442e-01
-3.62965703e-01 -8.37396026e-01 -8.31697047e-01 4.89393383e-01
4.74604785e-01 -6.50184751e-01 -6.75156653e-01 3.84501964e-01] | [11.633016586303711, -0.03266732394695282] |
87539c1e-3006-4083-9be0-74ea63e75104 | voicefixer-a-unified-framework-for-high | 2204.05841 | null | https://arxiv.org/abs/2204.05841v2 | https://arxiv.org/pdf/2204.05841v2.pdf | VoiceFixer: A Unified Framework for High-Fidelity Speech Restoration | Speech restoration aims to remove distortions in speech signals. Prior methods mainly focus on a single type of distortion, such as speech denoising or dereverberation. However, speech signals can be degraded by several different distortions simultaneously in the real world. It is thus important to extend speech restoration models to deal with multiple distortions. In this paper, we introduce VoiceFixer, a unified framework for high-fidelity speech restoration. VoiceFixer restores speech from multiple distortions (e.g., noise, reverberation, and clipping) and can expand degraded speech (e.g., noisy speech) with a low bandwidth to 44.1 kHz full-bandwidth high-fidelity speech. We design VoiceFixer based on (1) an analysis stage that predicts intermediate-level features from the degraded speech, and (2) a synthesis stage that generates waveform using a neural vocoder. Both objective and subjective evaluations show that VoiceFixer is effective on severely degraded speech, such as real-world historical speech recordings. Samples of VoiceFixer are available at https://haoheliu.github.io/voicefixer. | ['Yuxuan Wang', 'Chuanzeng Huang', 'DeLiang Wang', 'Yan Zhao', 'Qiao Tian', 'Qiuqiang Kong', 'Xubo Liu', 'Haohe Liu'] | 2022-04-12 | null | null | null | null | ['speech-denoising'] | ['speech'] | [ 9.93374363e-02 -1.75351128e-01 4.10922080e-01 -1.49679676e-01
-9.40207958e-01 -4.22633320e-01 2.80096382e-01 -2.78937608e-01
1.08932316e-01 5.88035464e-01 8.26208830e-01 -5.82140684e-01
2.55905867e-01 -4.00092363e-01 -6.01231515e-01 -6.41019344e-01
1.30514219e-01 -3.95035774e-01 -1.74907614e-02 -4.35133636e-01
-1.64164498e-01 3.71580988e-01 -1.70043671e+00 3.17885458e-01
1.12490022e+00 7.07416296e-01 7.43853927e-01 1.02313697e+00
2.27250054e-01 3.51197690e-01 -1.21398604e+00 4.79192659e-02
6.62600771e-02 -5.79781711e-01 -2.69064754e-01 1.84009537e-01
6.38960227e-02 -6.83001518e-01 -7.14325845e-01 1.23122048e+00
8.73914003e-01 4.64834422e-01 2.86484152e-01 -7.42484987e-01
-8.60609531e-01 6.09672010e-01 8.28445926e-02 4.42104578e-01
5.82070172e-01 2.26726115e-01 4.84511077e-01 -1.28297508e+00
1.40075639e-01 1.54173064e+00 4.87034470e-01 6.81888342e-01
-1.03627551e+00 -4.54724342e-01 -1.44293815e-01 4.11162496e-01
-1.02083683e+00 -1.17012846e+00 7.85979927e-01 -1.64501607e-01
1.06674993e+00 6.42570674e-01 2.15632841e-01 1.19508719e+00
2.28926793e-01 5.79593539e-01 7.41418540e-01 -4.82627571e-01
2.66593814e-01 -2.27971986e-01 -1.41454432e-02 -1.86057195e-01
-2.66021371e-01 5.92644691e-01 -4.24858153e-01 -1.61949713e-02
3.88140142e-01 -2.51781225e-01 -9.93696928e-01 5.98400414e-01
-8.61130953e-01 3.21059912e-01 3.23952138e-01 2.95234233e-01
-4.30460095e-01 -1.22970983e-01 2.88964897e-01 5.35528541e-01
5.66765606e-01 1.42199755e-01 -1.67806581e-01 -7.36893937e-02
-8.62427175e-01 2.93349326e-02 5.59700906e-01 8.94759834e-01
2.29708448e-01 8.82920980e-01 -9.50943977e-02 1.45638251e+00
3.65049660e-01 7.62836754e-01 6.96139872e-01 -8.06334674e-01
4.27100480e-01 -6.23672903e-01 1.39161393e-01 -5.60314536e-01
-1.37740389e-01 -6.03791475e-01 -1.02461982e+00 2.98063934e-01
-7.01930523e-02 -2.13328928e-01 -1.21681952e+00 1.60760307e+00
1.76586688e-01 2.02385083e-01 3.39102805e-01 1.16236448e+00
9.55077529e-01 1.24697506e+00 -3.70390087e-01 -6.35424912e-01
9.87726331e-01 -1.17830539e+00 -1.50406134e+00 -3.57099235e-01
-1.67276308e-01 -1.31841993e+00 1.24940133e+00 6.08777583e-01
-1.36072004e+00 -8.35497499e-01 -1.06572855e+00 -1.02508657e-01
7.51307467e-03 -8.96576419e-02 -3.08867306e-01 6.96527183e-01
-1.31062829e+00 5.85461736e-01 -6.36470556e-01 5.76137230e-02
-9.21739563e-02 -4.90860380e-02 -1.68909445e-01 -1.80285230e-01
-1.42152488e+00 8.53317738e-01 -5.08269779e-02 2.66293019e-01
-1.08997691e+00 -7.43204355e-01 -8.92085791e-01 2.15949431e-01
2.27744997e-01 -4.71196234e-01 1.63204205e+00 -7.73506045e-01
-1.82403588e+00 -1.59295704e-02 -5.18512189e-01 -2.92988449e-01
2.54267544e-01 -3.55883658e-01 -1.20680463e+00 3.27427536e-02
-2.08913654e-01 -9.23849270e-02 1.28852403e+00 -1.36498761e+00
-4.02053118e-01 -8.46413374e-02 -4.31348324e-01 2.49323010e-01
-2.26365477e-01 1.95691124e-01 -2.28136346e-01 -1.09774184e+00
2.77326316e-01 -5.29358864e-01 6.43445030e-02 -3.02156776e-01
-5.53372622e-01 2.71469891e-01 9.40068483e-01 -1.37647617e+00
1.36629105e+00 -2.54605222e+00 2.67133415e-02 -2.71830529e-01
-2.22257599e-01 8.24478269e-01 -3.12882006e-01 5.61820805e-01
-2.90957481e-01 1.90853834e-01 -2.81982541e-01 -6.45434141e-01
-1.14869606e-02 1.22398578e-01 -6.49100304e-01 4.19509917e-01
-7.21039250e-02 3.48804921e-01 -7.62500346e-01 1.73047960e-01
4.41033036e-01 9.13553834e-01 -4.90272790e-01 2.99178809e-01
2.06054047e-01 4.69436079e-01 2.49046713e-01 6.81375623e-01
9.53949869e-01 5.31074345e-01 -2.39330158e-01 -1.78883076e-01
-3.18748444e-01 8.12328696e-01 -1.02426767e+00 1.42435145e+00
-6.13367140e-01 8.27260613e-01 7.77101040e-01 -4.68926460e-01
9.42869246e-01 8.29182923e-01 -1.38859451e-01 -6.48201287e-01
-7.72058517e-02 6.27458334e-01 7.31919259e-02 -5.45700133e-01
5.20564616e-01 -2.57194638e-01 4.30116236e-01 -9.07735340e-03
4.76422068e-03 -4.85442281e-01 9.10892189e-02 -1.04837008e-01
1.13548267e+00 -4.16749299e-01 1.82783082e-01 -9.83244330e-02
5.78520179e-01 -8.03222477e-01 6.59920931e-01 3.76614332e-01
-3.56605589e-01 9.16257143e-01 -1.73271149e-01 3.40640545e-01
-1.07772470e+00 -1.38454032e+00 -2.46836618e-01 8.46389592e-01
-6.54246751e-03 -2.01267123e-01 -9.35534477e-01 1.77958205e-01
-6.86229169e-02 1.02382600e+00 1.35063738e-01 -4.61542428e-01
-6.10000372e-01 -2.80749440e-01 5.28709114e-01 1.28622413e-01
2.42416933e-01 -1.03014672e+00 1.45879760e-01 4.61861312e-01
-4.45150822e-01 -8.87010217e-01 -1.12139022e+00 1.42035857e-01
-6.35918498e-01 -4.72783864e-01 -8.36743653e-01 -9.20313835e-01
3.49623859e-01 6.61312580e-01 7.59335399e-01 3.85240577e-02
4.10968922e-02 3.68119068e-02 -5.16183019e-01 -3.17712456e-01
-9.59368050e-01 -4.42853689e-01 5.62431335e-01 1.40150249e-01
-2.82663763e-01 -8.46896350e-01 -5.01417279e-01 4.24643219e-01
-1.05543327e+00 -3.39692175e-01 2.93543637e-01 8.63596618e-01
3.72977376e-01 5.14630139e-01 1.07401049e+00 -8.98301527e-02
1.13749182e+00 -4.59195167e-01 -4.52582687e-01 -1.76821277e-01
-2.71397829e-01 -6.14385605e-01 1.04379976e+00 -6.68998837e-01
-1.34891891e+00 -4.62656468e-01 -7.97877133e-01 -6.62608325e-01
-2.95399457e-01 6.95747674e-01 -6.14863455e-01 4.06863213e-01
8.84682953e-01 3.95164758e-01 -7.84012973e-02 -9.84018505e-01
1.82714969e-01 1.43833709e+00 9.97290194e-01 -9.03976411e-02
9.13079262e-01 -5.20403534e-02 -6.09561026e-01 -1.38214374e+00
-2.19367757e-01 -2.75756896e-01 -9.01544020e-02 -2.54874945e-01
2.29018673e-01 -9.80873704e-01 -3.37854803e-01 8.72426093e-01
-1.27748716e+00 -4.90679026e-01 -2.84045756e-01 7.53948629e-01
-3.77444118e-01 4.36160088e-01 -9.25128698e-01 -9.85733747e-01
-4.04983282e-01 -1.18692064e+00 6.38300180e-01 3.21238726e-01
-7.14646950e-02 -6.81862056e-01 -1.98731184e-01 3.48204404e-01
7.62933075e-01 -2.36953765e-01 5.92291534e-01 -1.88376442e-01
-2.96160042e-01 6.53255358e-02 4.50037897e-01 9.76446927e-01
7.44458973e-01 -6.19319221e-03 -1.32258344e+00 -4.99962777e-01
4.93829906e-01 1.56551853e-01 6.65274024e-01 5.61379850e-01
8.63568664e-01 -5.62019050e-01 2.04282135e-01 6.68396771e-01
8.59204948e-01 8.66918862e-01 1.00220287e+00 -1.42257631e-01
2.26664379e-01 4.41215515e-01 3.97849590e-01 2.80416101e-01
-7.52618238e-02 6.46400154e-01 3.37748796e-01 -8.64335522e-02
-9.13290143e-01 -3.24280113e-01 7.34585106e-01 1.59447539e+00
4.70124006e-01 -6.73798501e-01 -7.05282450e-01 8.26344788e-01
-1.20646036e+00 -8.65318775e-01 -1.24832042e-01 2.14203286e+00
1.04256237e+00 1.74569301e-02 -1.76195383e-01 6.50788903e-01
1.15205300e+00 2.31196806e-01 -7.54034042e-01 -5.72493851e-01
-2.97512621e-01 1.52409509e-01 -1.01037305e-02 1.04239416e+00
-5.86845100e-01 6.68014944e-01 5.96626282e+00 8.68197739e-01
-1.28872406e+00 2.35770479e-01 2.77192205e-01 -2.86153495e-01
-4.65975255e-01 -3.19948792e-01 -3.23449701e-01 7.10850596e-01
1.15489936e+00 -4.74159628e-01 1.06126595e+00 4.76964772e-01
9.17457342e-01 2.53609121e-01 -7.03498840e-01 9.15283859e-01
7.46222436e-02 -8.63082647e-01 -1.74425662e-01 -1.31523430e-01
5.38999379e-01 1.33453477e-02 1.69360489e-01 2.28326768e-01
-2.16664195e-01 -9.59407985e-01 1.01651609e+00 4.75111157e-01
9.38133299e-01 -6.84716761e-01 4.19022769e-01 4.92202103e-01
-1.04492497e+00 -6.52101487e-02 -2.06561774e-01 1.16907367e-02
6.44344628e-01 9.80558097e-01 -6.78602755e-01 5.12897789e-01
5.29946029e-01 5.65412462e-01 1.77270383e-01 1.28130305e+00
-6.50016189e-01 9.33102548e-01 -1.64286375e-01 5.15619159e-01
-3.39789867e-01 3.94474752e-02 1.03940296e+00 1.19226193e+00
7.14517295e-01 3.45188230e-01 -1.68339312e-01 6.22661829e-01
-7.20123649e-02 -1.76608622e-01 -2.11542666e-01 8.54714308e-04
9.04814184e-01 7.73956835e-01 1.51548371e-01 -1.15530193e-01
-1.92437232e-01 9.52761829e-01 -4.46899027e-01 8.98072660e-01
-6.53015375e-01 -7.54834831e-01 1.17219734e+00 7.86667392e-02
2.29226261e-01 -2.13132709e-01 -9.61260498e-02 -1.05254972e+00
1.84422910e-01 -1.19306314e+00 -3.02482903e-01 -1.14845705e+00
-1.11002290e+00 9.41933632e-01 -4.30266500e-01 -1.22770798e+00
-3.27907175e-01 -3.32649887e-01 -7.36305416e-01 1.42863011e+00
-1.64290833e+00 -4.82160896e-01 -1.44710988e-01 5.78993380e-01
1.08478999e+00 -3.71853113e-02 7.48444676e-01 6.86130643e-01
-7.15440154e-01 4.41722125e-01 3.35208952e-01 -3.46824706e-01
7.01906204e-01 -1.04390764e+00 8.56400251e-01 1.34496927e+00
-2.86352813e-01 5.42078078e-01 1.09724462e+00 -6.56064332e-01
-1.33727300e+00 -1.09349859e+00 9.76154685e-01 3.02765369e-01
3.64011705e-01 -3.70713413e-01 -1.38060033e+00 4.01957721e-01
2.68379569e-01 7.14132711e-02 4.20519143e-01 -4.81034487e-01
-1.48889139e-01 -2.52777576e-01 -1.21864188e+00 6.85029566e-01
9.46725965e-01 -8.53638470e-01 -6.79829657e-01 1.09242506e-01
1.08858073e+00 -7.44197369e-01 -6.80616200e-01 1.94457531e-01
8.90951306e-02 -8.93525720e-01 9.31017399e-01 -7.66679784e-03
1.81202307e-01 -6.02118492e-01 -5.40851235e-01 -2.11943936e+00
-1.90512955e-01 -1.28121638e+00 -3.12075049e-01 1.41501319e+00
3.06324691e-01 -9.29289877e-01 -1.06182978e-01 1.90414295e-01
-9.99927938e-01 -2.52626330e-01 -1.05091119e+00 -1.15394402e+00
5.58375241e-03 -4.59318310e-01 7.58799255e-01 5.43285728e-01
1.42316192e-01 2.12722257e-01 -6.15713000e-01 6.48199081e-01
4.72964168e-01 -4.58839417e-01 5.37287235e-01 -6.62645459e-01
-2.94870675e-01 -2.80092359e-01 1.39043853e-01 -1.39809227e+00
-5.04632741e-02 -5.52071095e-01 6.06542289e-01 -1.77488565e+00
-6.64246380e-01 -3.15661132e-02 -6.77631199e-02 2.00674549e-01
-2.07773447e-01 -2.07566619e-01 3.38333756e-01 1.64990723e-01
4.76289332e-01 1.02925718e+00 1.37238193e+00 -1.04450092e-01
-4.08334315e-01 2.16124207e-01 -6.08341038e-01 5.91984749e-01
1.05720377e+00 -3.63755107e-01 -5.35622776e-01 -6.66807652e-01
-5.89789033e-01 5.22376537e-01 1.17306046e-01 -9.99991179e-01
2.32021436e-01 -8.74665231e-02 -1.29137829e-01 -4.09547597e-01
7.19109416e-01 -5.80591559e-01 2.67812967e-01 3.57945770e-01
-2.12601557e-01 -3.45305353e-01 4.72939342e-01 4.22773123e-01
-5.70102036e-01 -1.69362679e-01 1.00106430e+00 1.94607437e-01
-2.50594854e-01 -1.61967367e-01 -7.28767812e-01 -2.32804313e-01
3.30079585e-01 -4.97889938e-03 -7.74560988e-01 -7.71808326e-01
-9.71402824e-01 -1.73683822e-01 1.19829468e-01 6.65452003e-01
9.50948715e-01 -1.26350045e+00 -9.76253927e-01 5.14224052e-01
-5.10136187e-01 -2.67055482e-01 5.66664040e-01 4.82525140e-01
-1.76599830e-01 3.74002337e-01 1.86327919e-01 -2.84976363e-01
-1.36144757e+00 5.24002969e-01 5.94249249e-01 5.23422897e-01
-6.43298924e-01 6.99121356e-01 1.54779464e-01 -2.44775727e-01
3.62221062e-01 -5.43741703e-01 3.99228260e-02 -1.42737076e-01
9.95502114e-01 5.60966969e-01 5.16025245e-01 -7.65588343e-01
-2.30818629e-01 7.48673156e-02 1.60015553e-01 -3.65156204e-01
1.05490792e+00 -5.31431854e-01 3.36034223e-02 4.37173814e-01
1.12102163e+00 2.54555821e-01 -1.22066021e+00 -2.00166985e-01
-4.21501487e-01 -6.82648242e-01 4.81301874e-01 -1.07352304e+00
-8.92446518e-01 7.65605450e-01 5.27483702e-01 5.19043863e-01
1.60020804e+00 -3.89643878e-01 1.10440779e+00 -1.42146274e-01
1.92949712e-01 -9.35779750e-01 1.14523225e-01 6.18208528e-01
1.64858079e+00 -6.93355799e-01 -4.52359706e-01 -4.09887224e-01
-2.18756601e-01 9.64102566e-01 1.75953016e-01 2.43910730e-01
7.09987402e-01 5.09178042e-01 4.34065044e-01 5.57551146e-01
-7.00212240e-01 -1.65550828e-01 1.92376360e-01 7.45780647e-01
2.85836220e-01 1.59098700e-01 -9.92280617e-02 6.74487770e-01
-6.11288309e-01 -3.10750484e-01 8.32734525e-01 7.22474456e-01
-7.04690158e-01 -9.68337119e-01 -1.07443643e+00 3.98704745e-02
-4.15002733e-01 -3.47547859e-01 -3.98157425e-02 8.10505301e-02
-1.08581096e-01 1.88372982e+00 -1.04531348e-01 -4.77167100e-01
8.47001433e-01 -5.67640290e-02 -2.58653648e-02 -5.49254060e-01
-5.17413437e-01 6.62550509e-01 1.23267196e-01 -2.40027994e-01
6.54611960e-02 -6.56820953e-01 -1.24249184e+00 -4.32292849e-01
-3.88155818e-01 2.54757144e-02 1.02287757e+00 6.65448308e-01
3.06928575e-01 1.24794888e+00 9.05225098e-01 -9.79366481e-01
-7.16843784e-01 -1.28514838e+00 -7.23217070e-01 1.17495902e-01
1.28797197e+00 -1.70805305e-01 -1.03701699e+00 2.04537585e-01] | [14.935456275939941, 6.042664527893066] |
8d4ce18e-01e7-45ea-9b4f-91d6b0f81292 | what-does-a-car-ssette-tape-tell | 1905.13448 | null | https://arxiv.org/abs/1905.13448v2 | https://arxiv.org/pdf/1905.13448v2.pdf | Audio Caption in a Car Setting with a Sentence-Level Loss | Captioning has attracted much attention in image and video understanding while a small amount of work examines audio captioning. This paper contributes a Mandarin-annotated dataset for audio captioning within a car scene. A sentence-level loss is proposed to be used in tandem with a GRU encoder-decoder model to generate captions with higher semantic similarity to human annotations. We evaluate the model on the newly-proposed Car dataset, a previously published Mandarin Hospital dataset and the Joint dataset, indicating its generalization capability across different scenes. An improvement in all metrics can be observed, including classical natural language generation (NLG) metrics, sentence richness and human evaluation ratings. However, though detailed audio captions can now be automatically generated, human annotations still outperform model captions on many aspects. | ['Mengyue Wu', 'Xuenan Xu', 'Kai Yu', 'Heinrich Dinkel'] | 2019-05-31 | null | null | null | null | ['audio-captioning'] | ['audio'] | [ 7.91555882e-01 6.27083898e-01 -5.65944724e-02 -4.81099308e-01
-1.53454876e+00 -3.27116907e-01 7.99807489e-01 1.20765734e-02
-2.46550635e-01 9.28053558e-01 7.04259872e-01 8.29980597e-02
4.61593539e-01 -3.05248231e-01 -1.02804470e+00 -3.06309789e-01
9.52585191e-02 4.21850383e-01 8.75906944e-02 -1.06972933e-01
8.68510921e-03 -7.92569071e-02 -1.66584623e+00 7.24143028e-01
6.30760849e-01 1.11288989e+00 3.99405241e-01 8.92575085e-01
2.30663136e-01 1.03660476e+00 -6.34553909e-01 -6.14136994e-01
-1.02565251e-01 -7.05287755e-01 -9.66282368e-01 2.16219142e-01
9.03499126e-01 -2.41454989e-01 -2.79187024e-01 7.01837003e-01
6.85070872e-01 -1.50764734e-01 6.47171080e-01 -1.22896171e+00
-8.97868752e-01 8.93328786e-01 -1.40352994e-01 -1.42528147e-01
8.14995587e-01 1.63078666e-01 1.15549338e+00 -8.87231469e-01
7.01699615e-01 1.24331975e+00 3.97816032e-01 8.05972755e-01
-1.10138035e+00 -4.74952638e-01 -1.34743616e-01 4.89070803e-01
-1.35005331e+00 -3.92371118e-01 6.79212093e-01 -4.77258980e-01
6.43530726e-01 1.42471507e-01 2.59360969e-01 1.54125035e+00
-2.53941625e-01 9.55816150e-01 6.58402145e-01 -4.34781700e-01
1.68548435e-01 3.47008646e-01 -2.59721518e-01 -6.79342216e-03
-1.13356665e-01 1.93700835e-01 -7.39421844e-01 1.97937548e-01
4.18350607e-01 -7.89981246e-01 -4.09621239e-01 -8.38863254e-02
-1.49416113e+00 9.44070160e-01 4.40404981e-01 1.03100799e-01
-5.14635503e-01 5.12792945e-01 6.60132289e-01 8.90150219e-02
3.26555550e-01 5.65750241e-01 -4.49806079e-02 -2.90888399e-01
-9.63396192e-01 3.44765782e-01 5.25217175e-01 1.32114196e+00
5.30689597e-01 2.05989987e-01 -8.18925500e-01 9.09707367e-01
6.55404255e-02 7.35370815e-01 3.66203487e-01 -1.06569195e+00
6.40673339e-01 -1.28221586e-01 2.54956067e-01 -7.91979969e-01
8.03577378e-02 -4.21667635e-01 -6.49916649e-01 -2.96093792e-01
5.50766289e-02 -1.70673102e-01 -7.25732327e-01 1.69177890e+00
-2.27360368e-01 4.69295204e-01 3.13041389e-01 1.15342784e+00
8.97286057e-01 9.29172635e-01 2.77144760e-01 -2.69621313e-02
1.40620685e+00 -1.30400586e+00 -9.60306168e-01 -3.44194680e-01
3.76964539e-01 -9.38035190e-01 1.03628194e+00 4.13806528e-01
-1.08723128e+00 -9.30103898e-01 -1.02047265e+00 4.61409502e-02
2.19776090e-02 2.99026430e-01 3.52225244e-01 6.41191602e-01
-1.24143922e+00 6.89684749e-02 -2.73704767e-01 -2.98615217e-01
3.66286457e-01 -7.43236095e-02 -3.22441757e-01 -3.29531252e-01
-1.33162785e+00 8.43390822e-01 6.43433750e-01 2.97679417e-02
-1.51323724e+00 -5.34666181e-01 -1.10098529e+00 -1.37116715e-01
1.98928431e-01 -7.48402417e-01 1.68100429e+00 -1.14865434e+00
-1.50969291e+00 7.05011010e-01 -1.08323261e-01 -1.15571475e+00
5.76798975e-01 -4.03221369e-01 -4.77787733e-01 6.30860269e-01
1.17239319e-01 1.81911397e+00 8.95363033e-01 -1.47810054e+00
-6.04508281e-01 3.07579845e-01 2.98039913e-02 2.01038748e-01
-2.34868914e-01 1.27963293e-02 -3.55645180e-01 -9.16368306e-01
-5.22956014e-01 -9.79576468e-01 -1.42544568e-01 -3.31754208e-01
-3.11012506e-01 -1.17183588e-01 7.52689242e-01 -8.87684464e-01
9.80748177e-01 -1.99821436e+00 1.18416529e-02 -4.17283356e-01
-4.61479276e-01 3.08073670e-01 -5.77840388e-01 3.79099429e-01
-8.89951959e-02 1.28814891e-01 -4.76870358e-01 -4.36316252e-01
-1.39124736e-01 1.27707301e-02 -6.18468881e-01 -1.67548239e-01
7.29963958e-01 1.04799736e+00 -9.16493416e-01 -5.26733398e-01
2.25709170e-01 7.41128922e-01 -7.28551149e-01 3.16474199e-01
-4.33109939e-01 4.55653846e-01 -9.37913135e-02 3.84791821e-01
4.76624489e-01 6.84059560e-02 -3.54296625e-01 -2.30925884e-02
3.20442587e-01 1.17786452e-01 -6.80105329e-01 1.92423785e+00
-7.16876507e-01 9.77299094e-01 -2.10026458e-01 -7.87219107e-01
1.11463094e+00 8.85626435e-01 1.59072310e-01 -6.73509598e-01
3.09156422e-02 1.18200108e-01 -3.11883330e-01 -5.56895375e-01
7.87197888e-01 -1.32142499e-01 -3.56305182e-01 5.66456243e-02
2.63072431e-01 -4.50748444e-01 2.97588736e-01 2.56909192e-01
8.48506629e-01 1.38288826e-01 -5.32781184e-02 1.02910139e-01
7.13766694e-01 1.57703638e-01 -1.55364722e-01 7.92047143e-01
-1.54889032e-01 1.35110617e+00 2.01602310e-01 2.95199901e-02
-1.28761590e+00 -1.01447606e+00 -7.13081881e-02 8.98613572e-01
-5.54163940e-02 -3.31211448e-01 -1.08940244e+00 -4.87458706e-01
-5.17839968e-01 1.06939352e+00 -5.00483572e-01 -1.82111219e-01
-4.47087377e-01 -1.46963686e-01 7.49051690e-01 6.02667272e-01
4.28073734e-01 -1.45102465e+00 -3.45956713e-01 2.94150889e-01
-7.60347962e-01 -1.79973030e+00 -5.53566813e-01 -3.18858713e-01
-5.83407104e-01 -5.88365257e-01 -1.07189059e+00 -1.07000697e+00
2.68182129e-01 1.10678889e-01 1.17828393e+00 -4.80464160e-01
-1.46236628e-01 5.66418409e-01 -8.32729518e-01 -7.00373530e-01
-9.04269993e-01 1.36682376e-01 -2.15206474e-01 2.40219310e-01
1.11797273e-01 -1.89798713e-01 -2.85830081e-01 4.35882181e-01
-1.03138626e+00 2.74996102e-01 7.23466516e-01 8.14634919e-01
4.89809334e-01 -6.01598620e-01 1.12194800e+00 -3.41136605e-01
5.82796693e-01 -2.78245240e-01 -2.95531720e-01 -1.53598875e-01
-1.62487298e-01 -1.18875638e-01 5.02477884e-01 -4.44701284e-01
-1.03132463e+00 3.18960339e-01 -2.18328029e-01 -6.25016093e-01
-5.04518211e-01 4.51577008e-01 -2.54528850e-01 3.91607225e-01
7.20201790e-01 2.53763914e-01 5.76010160e-02 -2.35113218e-01
6.11012340e-01 1.02934682e+00 1.08683872e+00 -4.74452257e-01
6.45450711e-01 3.03145707e-01 -3.98978263e-01 -8.89710546e-01
-1.04236054e+00 -4.05953497e-01 -3.07901889e-01 -4.90706921e-01
1.27646267e+00 -1.43556106e+00 -3.09552342e-01 6.80838376e-02
-1.67757344e+00 -1.31778326e-02 -3.64467323e-01 6.70370400e-01
-1.02189267e+00 2.85833836e-01 -3.96069765e-01 -8.94901454e-01
-3.04754972e-01 -1.18926585e+00 1.68757665e+00 -1.28443718e-01
-4.05299097e-01 -5.18454373e-01 6.30587200e-03 9.55468178e-01
2.89950460e-01 3.00610453e-01 2.97772616e-01 -5.69021344e-01
-7.29302764e-01 -4.74356085e-01 -2.87716508e-01 8.19339573e-01
-3.74728739e-01 -5.03588438e-01 -1.31722796e+00 -1.12637833e-01
-2.99593449e-01 -7.39858449e-01 8.93409610e-01 3.09346050e-01
9.89022255e-01 -3.72284979e-01 2.89496724e-02 -1.02102622e-01
1.17907453e+00 2.22188085e-01 1.10447156e+00 6.35460988e-02
4.26777840e-01 7.99588025e-01 7.94604421e-01 1.43036574e-01
1.75970629e-01 8.63516629e-01 4.98046100e-01 4.92328405e-02
-4.77033436e-01 -6.52729511e-01 6.73493683e-01 7.13527679e-01
2.18999058e-01 -5.75532615e-01 -7.87589431e-01 8.68265152e-01
-1.77065396e+00 -1.11737072e+00 -2.11246490e-01 1.94913840e+00
6.56987906e-01 1.80418968e-01 1.92529395e-01 2.22418070e-01
1.01560569e+00 -4.56901081e-02 -7.47262985e-02 -5.34384429e-01
-3.43205124e-01 -3.55588384e-02 2.25755170e-01 5.56094050e-01
-1.11972713e+00 8.86823058e-01 6.90999746e+00 9.61132646e-01
-9.02645111e-01 1.96126252e-01 7.59704113e-01 1.03569850e-01
-2.63817966e-01 -7.01301917e-02 -5.74633479e-01 3.29578608e-01
1.33356011e+00 -1.57291323e-01 7.81273469e-02 9.61949706e-01
5.48538327e-01 4.97944914e-02 -1.26638412e+00 1.13658547e+00
7.15927660e-01 -1.41657341e+00 5.53342342e-01 -4.80639078e-02
8.09996247e-01 -3.85953858e-02 4.53241318e-02 3.50580215e-01
-2.15151191e-01 -1.17767107e+00 1.05785346e+00 5.36002278e-01
9.28076744e-01 -7.60071456e-01 1.15713573e+00 3.19891945e-02
-9.36694324e-01 7.55456090e-02 -1.88019410e-01 -1.34255201e-01
7.82658339e-01 3.60620588e-01 -1.34438264e+00 7.12738335e-01
3.96756411e-01 6.43044174e-01 -7.30794847e-01 1.19872022e+00
-2.61308521e-01 8.24878097e-01 5.93593530e-02 -6.62052035e-02
3.95412862e-01 2.18153119e-01 6.08960450e-01 1.38690162e+00
3.77361059e-01 -4.06576157e-01 1.65306300e-01 7.41536975e-01
-1.04149774e-01 2.94821680e-01 -7.17569709e-01 -1.02765396e-01
2.13128045e-01 9.89675820e-01 -2.50269860e-01 -4.12864387e-01
-2.46409371e-01 1.01711130e+00 -1.90070003e-01 2.53591478e-01
-1.22934711e+00 -3.48498166e-01 2.30116174e-01 1.15249634e-01
4.59647983e-01 7.31655806e-02 -9.68140066e-02 -7.63318717e-01
1.16300799e-01 -7.79856861e-01 1.09211050e-01 -1.37137520e+00
-1.06381416e+00 9.09514546e-01 1.85902119e-01 -1.62041259e+00
-6.78676426e-01 -3.66643995e-01 -2.80313790e-01 4.33161646e-01
-1.51757884e+00 -1.20999920e+00 -2.85145402e-01 3.16495717e-01
1.06154358e+00 -2.49546811e-01 6.98269725e-01 4.81273830e-01
-1.16490886e-01 7.28736758e-01 -3.79781246e-01 3.15240435e-02
7.48502254e-01 -8.84042561e-01 1.44899517e-01 8.01278770e-01
4.31637228e-01 -1.82634115e-01 1.15569377e+00 -3.69462639e-01
-7.40928292e-01 -1.58437777e+00 1.08323407e+00 -4.94624555e-01
4.02036577e-01 -4.54090208e-01 -6.97604835e-01 5.18479228e-01
7.38339722e-01 -3.76801193e-01 4.28348929e-01 -4.68103707e-01
-3.39224041e-01 -2.69766897e-03 -8.25538993e-01 4.76131678e-01
9.34802175e-01 -5.05211234e-01 -6.36406541e-01 3.95041406e-01
1.26890254e+00 -1.19739100e-01 -6.53255463e-01 4.12487566e-01
2.67045110e-01 -6.97457731e-01 8.39872539e-01 -3.52444798e-01
9.17656541e-01 -4.71367270e-01 -2.29091123e-01 -1.05886459e+00
1.13347150e-01 -8.28030109e-01 2.79600322e-01 1.39231551e+00
7.99552798e-01 1.17232874e-01 7.21188009e-01 9.18750688e-02
-4.88563061e-01 -1.82294711e-01 -1.01020932e+00 -9.62561131e-01
-1.44630238e-01 -9.94578183e-01 3.35736036e-01 4.86245483e-01
4.90176529e-02 1.05888569e+00 -9.02453244e-01 3.50428885e-03
3.91236365e-01 -2.36402988e-01 9.42215562e-01 -9.19451714e-01
-1.26706436e-01 -2.20126912e-01 -6.83340013e-01 -9.43787694e-01
2.55831331e-01 -1.03689361e+00 4.19253260e-01 -1.60141182e+00
1.41515255e-01 3.08641762e-01 2.38362521e-01 2.85317063e-01
5.05834296e-02 7.12130427e-01 4.09642398e-01 -4.69273739e-02
-8.37847352e-01 8.42254043e-01 1.28034759e+00 -4.29894060e-01
1.46064684e-01 -2.73395717e-01 -6.01409912e-01 3.17051053e-01
5.88515759e-01 -3.99828732e-01 -4.12892997e-01 -4.18895334e-01
-5.75442389e-02 3.29986691e-01 6.46958768e-01 -1.48120534e+00
-7.42358575e-03 2.95890719e-01 -6.36341125e-02 -5.34333587e-01
5.24328351e-01 -6.68084025e-01 1.87443376e-01 1.85019389e-01
-8.30069721e-01 -1.24910116e-01 2.72988319e-01 8.10811222e-01
-8.56823742e-01 -2.38734484e-01 7.17981517e-01 1.71581909e-01
-5.22097886e-01 2.36947183e-02 -5.48043132e-01 9.91677716e-02
9.79054809e-01 -2.69694239e-01 -6.88534230e-02 -1.09611475e+00
-7.82387912e-01 1.03255868e-01 6.42129108e-02 9.53817546e-01
7.32691944e-01 -1.68022823e+00 -1.37552154e+00 -3.37499678e-01
6.65424705e-01 -1.20613001e-01 3.44696432e-01 5.34901261e-01
-3.77751499e-01 8.87693405e-01 -1.24040745e-01 -9.75853145e-01
-9.93977904e-01 6.25753582e-01 -4.16328125e-02 -8.80728941e-03
-2.72296637e-01 6.85258210e-01 2.14220718e-01 3.63094434e-02
3.77800614e-01 -3.40275258e-01 -3.11831564e-01 8.01994354e-02
7.37802863e-01 -8.88107251e-03 6.72095194e-02 -9.11670566e-01
2.61963662e-02 2.93786883e-01 1.36720061e-01 -5.18925250e-01
9.66618836e-01 -1.89900875e-01 4.25514281e-01 3.42771381e-01
1.28057706e+00 -3.56453359e-01 -1.12203312e+00 8.12189728e-02
-1.79172996e-02 -1.96069047e-01 -1.73724964e-01 -8.26605439e-01
-6.94431424e-01 9.98005092e-01 6.49337649e-01 7.63810277e-02
1.07696986e+00 3.44903380e-01 9.17776108e-01 3.25414628e-01
3.36551487e-01 -9.85221088e-01 5.79370737e-01 4.66724396e-01
1.36676395e+00 -1.33185065e+00 -7.38000989e-01 -4.79239464e-01
-1.25665689e+00 7.20696568e-01 6.20116532e-01 1.30086914e-01
7.87494928e-02 1.39433565e-02 9.64542478e-02 2.55156696e-01
-7.93929279e-01 -3.25803518e-01 2.47833923e-01 9.58286941e-01
4.62654263e-01 -4.41160835e-02 -2.04985783e-01 5.08579969e-01
-5.83503425e-01 1.34730518e-01 8.18971455e-01 2.79206872e-01
-3.39244515e-01 -8.51762295e-01 -3.77455711e-01 -2.50682272e-02
-3.19128692e-01 -2.32354358e-01 -4.58721310e-01 5.41143537e-01
-1.21745311e-01 1.30648685e+00 2.34409004e-01 -3.61081392e-01
2.98103064e-01 9.70034227e-02 1.50858790e-01 -6.76755786e-01
-4.29133087e-01 7.32303485e-02 5.16674340e-01 -3.05176020e-01
-6.60679221e-01 -5.94394147e-01 -9.08407986e-01 4.88237321e-01
-3.33170801e-01 4.80674267e-01 7.17421174e-01 6.32891834e-01
4.48015541e-01 5.35986245e-01 5.06614685e-01 -8.82930577e-01
-2.01563954e-01 -1.30953121e+00 -5.83312027e-02 6.11689210e-01
1.88841984e-01 -1.96742699e-01 -4.02168512e-01 7.06959248e-01] | [15.255708694458008, 4.8449015617370605] |
ff77a3fd-046b-4c46-b0c7-b2452aaf1e17 | contrastive-feature-learning-for-fault | 2208.13288 | null | https://arxiv.org/abs/2208.13288v1 | https://arxiv.org/pdf/2208.13288v1.pdf | Contrastive Feature Learning for Fault Detection and Diagnostics in Railway Applications | A railway is a complex system comprising multiple infrastructure and rolling stock assets. To operate the system safely, reliably, and efficiently, the condition many components needs to be monitored. To automate this process, data-driven fault detection and diagnostics models can be employed. In practice, however, the performance of data-driven models can be compromised if the training dataset is not representative of all possible future conditions. We propose to approach this problem by learning a feature representation that is, on the one hand, invariant to operating or environmental factors but, on the other hand, sensitive to changes in the asset's health condition. We evaluate how contrastive learning can be employed on supervised and unsupervised fault detection and diagnostics tasks given real condition monitoring datasets within a railway system - one image dataset from infrastructure assets and one time-series dataset from rolling stock assets. First, we evaluate the performance of supervised contrastive feature learning on a railway sleeper defect classification task given a labeled image dataset. Second, we evaluate the performance of unsupervised contrastive feature learning without access to faulty samples on an anomaly detection task given a railway wheel dataset. Here, we test the hypothesis of whether a feature encoder's sensitivity to degradation is also sensitive to novel fault patterns in the data. Our results demonstrate that contrastive feature learning improves the performance on the supervised classification task regarding sleepers compared to a state-of-the-art method. Moreover, on the anomaly detection task concerning the railway wheels, the detection of shelling defects is improved compared to state-of-the-art methods. | ['Olga Fink', 'Stefan Koller', 'Wilfried Bürzle', 'Lucian-Stefan Ancu', 'Kajan Ratnasabapathy', 'Gabriel Michau', 'Katharina Rombach'] | 2022-08-28 | null | null | null | null | ['fault-detection'] | ['miscellaneous'] | [ 4.28833216e-01 -1.50446996e-01 2.94241875e-01 -3.56603742e-01
-4.78483945e-01 1.04540564e-01 3.07729691e-01 2.37204120e-01
-1.51318416e-01 5.47506034e-01 -3.52013886e-01 3.32157835e-02
-6.99045062e-01 -9.41095233e-01 -6.75939143e-01 -1.02332127e+00
-1.82646945e-01 4.76661146e-01 4.47986573e-01 -5.50680935e-01
2.85744876e-01 7.92510986e-01 -2.32666564e+00 2.84076720e-01
5.72155714e-01 1.29203749e+00 1.50807515e-01 5.79894423e-01
2.28322998e-01 7.39489079e-01 -7.95747817e-01 2.89821178e-01
1.45663276e-01 -3.29604864e-01 -8.39816332e-01 5.55560708e-01
2.85643293e-03 1.27325002e-02 -2.62079626e-01 8.62034917e-01
1.42549053e-01 -1.87183484e-01 6.84737504e-01 -1.38412857e+00
5.88221401e-02 2.49798503e-02 7.16550788e-03 4.09346282e-01
3.00059497e-01 2.96011001e-01 6.46447718e-01 -6.05573595e-01
3.37375283e-01 7.68693209e-01 5.96161604e-01 1.68869913e-01
-1.19710171e+00 -2.02907786e-01 -5.03181107e-03 5.58283091e-01
-1.23008978e+00 -3.67937744e-01 9.42190111e-01 -5.92257798e-01
6.72106981e-01 3.22197884e-01 7.54735947e-01 7.12834179e-01
4.94048476e-01 4.39907789e-01 1.10056925e+00 -5.07124484e-01
4.05618310e-01 -5.18113896e-02 5.37594706e-02 7.38677561e-01
1.63746119e-01 3.69377673e-01 -6.48934901e-01 1.90145165e-01
1.32368475e-01 1.71936713e-02 -3.17788005e-01 -1.62408158e-01
-8.12725723e-01 4.77180332e-01 1.01949707e-01 7.00187266e-01
-6.28308833e-01 -4.54052836e-02 5.55105984e-01 9.66878235e-01
6.16577566e-01 4.78246897e-01 -6.55116260e-01 -1.56888425e-01
-1.06382203e+00 -4.83145118e-02 6.50481105e-01 3.47405463e-01
9.42334771e-01 2.81603813e-01 7.78556541e-02 7.08539426e-01
1.00299334e-02 5.71496010e-01 4.67693359e-01 -3.88750792e-01
2.82103028e-02 8.04810286e-01 -9.46270749e-02 -7.03468561e-01
-8.92422438e-01 -4.89753425e-01 -7.46251941e-01 4.13451463e-01
2.75139689e-01 1.22629933e-01 -9.06736851e-01 1.19203484e+00
1.30607113e-01 -1.06451876e-01 6.61507025e-02 5.46029866e-01
1.82616651e-01 3.94093156e-01 -5.94566584e-01 -4.04857278e-01
1.12115729e+00 -2.23601177e-01 -7.54379630e-01 -4.73311283e-02
7.43116856e-01 -5.03463149e-01 9.87003207e-01 8.16801786e-01
-7.26417840e-01 -7.25175619e-01 -1.28196800e+00 7.25191712e-01
-4.06072736e-01 1.73395991e-01 6.07744679e-02 6.88787103e-01
-7.65363872e-01 8.57097626e-01 -1.00449073e+00 -3.87995511e-01
1.89954370e-01 5.03507495e-01 -6.18807912e-01 -3.15020084e-01
-1.12206721e+00 1.07649410e+00 2.54964173e-01 2.89361060e-01
-1.05634093e+00 -4.84759778e-01 -7.04946697e-01 1.00649193e-01
5.17833650e-01 4.11918834e-02 8.87775600e-01 -8.06555986e-01
-1.20832968e+00 6.00410461e-01 1.84688613e-01 -4.89097446e-01
5.20338416e-01 -1.28906682e-01 -8.75897229e-01 3.54095995e-01
-9.29923449e-03 -3.07115108e-01 9.95008349e-01 -1.23315585e+00
-8.09201062e-01 -4.25014764e-01 -2.03432411e-01 -3.80546659e-01
-3.21136147e-01 -2.41388842e-01 2.99361888e-02 -3.75820875e-01
4.28803653e-01 -1.00856364e+00 9.95571315e-02 3.10881976e-02
-3.99499714e-01 9.80655923e-02 1.14959955e+00 -5.75050175e-01
9.03267801e-01 -2.40513968e+00 3.71567197e-02 5.16909778e-01
-3.59822154e-01 3.75053175e-02 9.78015363e-02 6.05352223e-01
-3.84375691e-01 -2.82196879e-01 -5.69016695e-01 1.46590889e-01
-2.24832043e-01 5.02304733e-01 2.42541373e-01 9.07585382e-01
7.40616202e-01 2.25558549e-01 -6.70984328e-01 -2.27013066e-01
2.28925422e-01 1.49871007e-01 -1.61267981e-01 3.87658715e-01
4.74492088e-03 5.44512212e-01 -2.80239165e-01 7.67957151e-01
2.77250767e-01 1.99909240e-01 -9.59459320e-02 -3.46169919e-01
-1.58352330e-01 -4.27349284e-02 -1.21106088e+00 1.32547247e+00
-7.55321801e-01 6.56745374e-01 -3.81705314e-01 -1.59988153e+00
1.36139703e+00 6.18090808e-01 8.67839456e-01 -1.12307835e+00
-1.02699474e-01 4.44409013e-01 1.11436762e-01 -1.06835127e+00
2.56174088e-01 -4.52750325e-01 -1.49099454e-01 3.66302043e-01
2.56641179e-01 -1.62912935e-01 2.05508232e-01 -3.33940923e-01
1.61503780e+00 -5.79856448e-02 -1.12216465e-01 -3.27413857e-01
7.58688271e-01 -1.04977965e-01 4.81230140e-01 2.83968985e-01
1.47972226e-01 4.10319537e-01 7.05695987e-01 -6.15620077e-01
-8.15044940e-01 -8.82030606e-01 -2.92462379e-01 6.50535226e-01
-1.10616840e-01 7.59661123e-02 -4.72019613e-01 -8.36426377e-01
1.41988173e-01 6.27096593e-01 -8.31530333e-01 -8.18596244e-01
-3.66900027e-01 -9.87746000e-01 3.62334877e-01 3.88284296e-01
3.47800821e-01 -1.00207293e+00 -1.02401078e+00 3.21682394e-01
1.02519408e-01 -8.20452094e-01 3.36795360e-01 7.05227435e-01
-8.38035166e-01 -1.46425128e+00 -1.81078166e-01 -3.96569848e-01
7.58131981e-01 -1.55503720e-01 1.04380989e+00 4.47592169e-01
-6.56471252e-01 5.74015558e-01 -5.74131966e-01 -3.02266657e-01
-8.44060540e-01 -1.39810503e-01 1.79115161e-01 4.71970975e-01
-3.12780514e-02 -5.01150846e-01 -3.07009280e-01 6.21375144e-01
-1.22359729e+00 -6.10680819e-01 6.21844649e-01 9.89981771e-01
4.14808154e-01 8.79725754e-01 8.33905756e-01 -8.36417615e-01
5.80487400e-02 -4.13694590e-01 -4.91881549e-01 2.55105674e-01
-8.92004788e-01 2.79287279e-01 4.76228952e-01 -4.85191010e-02
-8.69951785e-01 1.67218864e-01 -3.81735831e-01 -2.34195367e-01
-3.33813637e-01 6.32745683e-01 -3.76413524e-01 -2.86660623e-02
6.67008519e-01 3.73691082e-01 2.13620380e-01 -5.56483746e-01
-2.38018349e-01 7.62358427e-01 6.26362979e-01 -5.47208726e-01
7.94827282e-01 4.38959450e-01 4.64907795e-01 -1.11521089e+00
-6.13263249e-01 -5.98064542e-01 -9.35960412e-01 -7.12762892e-01
4.81590420e-01 -5.63852012e-01 -2.59610742e-01 8.68474662e-01
-5.95597863e-01 -2.84952700e-01 -7.22866356e-01 2.29145810e-01
-7.41806507e-01 8.60183835e-02 -3.23130757e-01 -9.44716334e-01
-2.31970698e-02 -1.03652453e+00 9.13055241e-01 -2.24508494e-01
1.47139490e-01 -8.82939637e-01 -7.61348605e-02 5.81853241e-02
3.27252120e-01 5.40228426e-01 9.80179489e-01 -7.98526466e-01
-7.29786456e-02 -7.29994953e-01 3.32389116e-01 8.55327845e-01
6.28406763e-01 1.24115549e-01 -9.10783708e-01 -4.38046217e-01
3.61257166e-01 -1.32000178e-01 7.46804595e-01 -3.22592258e-02
9.70041633e-01 1.88999474e-02 -1.26148816e-02 5.26318746e-03
1.49542296e+00 1.82519093e-01 7.34718442e-01 5.96890569e-01
2.77615577e-01 9.77857232e-01 1.18870497e+00 4.70406502e-01
-1.29447952e-02 7.72385359e-01 8.17179322e-01 -2.20214147e-02
1.21650077e-01 4.11243618e-01 2.84151763e-01 5.12483895e-01
-1.11205764e-01 4.08361182e-02 -9.83484328e-01 6.75435662e-01
-1.50910962e+00 -9.16497886e-01 -2.99649954e-01 2.29041815e+00
3.55800956e-01 6.56681716e-01 2.05449238e-01 1.27741098e+00
6.49851441e-01 -2.29338706e-01 -4.35086101e-01 -2.05899373e-01
-8.99953097e-02 1.98876232e-01 1.97470069e-01 -2.14786798e-01
-9.03826177e-01 7.83611760e-02 5.09990072e+00 2.57670105e-01
-1.33165145e+00 5.54706566e-02 3.27100277e-01 2.22924314e-02
1.53778881e-01 -2.41638675e-01 -2.05699682e-01 6.51933491e-01
1.28865719e+00 2.64334559e-01 -4.75455821e-02 5.85563421e-01
2.58042216e-01 -3.47367167e-01 -1.23432720e+00 4.92949814e-01
1.40457973e-01 -9.11470652e-01 -5.93450427e-01 2.28958484e-02
3.97054791e-01 -1.97548926e-01 -1.62038982e-01 2.00536877e-01
-5.75197399e-01 -7.64948189e-01 8.45469594e-01 8.95473778e-01
5.49467206e-01 -8.79042327e-01 1.21435201e+00 2.98562407e-01
-9.90976036e-01 -4.97964740e-01 2.06607848e-01 6.22258522e-02
1.97984561e-01 8.37372005e-01 -5.45664907e-01 9.35860574e-01
8.15686464e-01 6.28837764e-01 -7.99761534e-01 1.16004658e+00
-9.58589464e-02 6.28257871e-01 -1.27592012e-01 4.89614367e-01
-5.78217395e-03 1.55809194e-01 3.13179463e-01 9.66539741e-01
4.56607401e-01 -3.43166798e-01 8.24177042e-02 2.70160615e-01
4.79578644e-01 -2.15574563e-01 -6.72427356e-01 3.40433538e-01
1.44780949e-01 9.16854382e-01 -8.35379422e-01 1.14091951e-03
-3.92482102e-01 8.56282532e-01 -1.77140325e-01 -1.11745456e-02
-4.89631444e-01 -2.72222936e-01 3.91935855e-01 4.27108914e-01
4.53698725e-01 5.96261956e-02 1.13296717e-01 -6.24238372e-01
2.08329141e-01 -7.48859286e-01 3.38885278e-01 -5.82327366e-01
-1.26003575e+00 5.55748105e-01 1.80326446e-04 -1.61322486e+00
-2.37447962e-01 -6.49010837e-01 -6.25628829e-01 3.83631676e-01
-1.53814304e+00 -1.06000149e+00 -2.58004189e-01 6.54351711e-01
5.66686213e-01 -2.60704964e-01 6.71214581e-01 3.82828057e-01
-5.95757842e-01 2.25982040e-01 2.36569159e-02 -5.88435940e-02
4.59930331e-01 -1.29432583e+00 4.35786173e-02 9.25713599e-01
4.96856309e-02 -1.79901570e-01 8.94592106e-01 -4.76261586e-01
-1.45834589e+00 -1.14579856e+00 4.82685983e-01 5.28133735e-02
7.10902333e-01 -2.71499772e-02 -1.29453874e+00 2.98919588e-01
-3.33838165e-01 4.89537925e-01 5.25338113e-01 -4.60560527e-03
4.99998517e-02 -4.13713157e-01 -1.44033837e+00 -1.54044241e-01
4.92509782e-01 -6.03817403e-01 -5.76078892e-01 3.21530133e-01
4.52068038e-02 -1.32153317e-01 -1.30271435e+00 6.01191163e-01
4.34955388e-01 -1.02093172e+00 5.20142138e-01 -3.67251635e-01
3.22771549e-01 -4.00941491e-01 -2.27634579e-01 -1.44883263e+00
-3.01589863e-03 -1.14262663e-01 -1.80462807e-01 1.22455859e+00
2.70027071e-01 -5.09444714e-01 6.22697353e-01 2.27224171e-01
-4.61115837e-01 -8.39436352e-01 -1.30143654e+00 -1.16947079e+00
-2.04306439e-01 -6.07104778e-01 5.29702008e-01 6.49783373e-01
-2.76244879e-01 -2.66390145e-02 -1.83067784e-01 5.72287261e-01
1.93480983e-01 -1.12002358e-01 4.04818445e-01 -1.66754615e+00
-1.37803629e-01 -3.48430909e-02 -1.11314809e+00 1.29271612e-01
7.73626268e-02 -4.52143878e-01 4.83220249e-01 -1.31766498e+00
-3.79639864e-01 -5.07882178e-01 -5.57573378e-01 4.70083207e-01
1.25713974e-01 2.73527801e-01 -4.51635495e-02 6.10597879e-02
-2.35987231e-01 4.58933353e-01 7.31168985e-01 -3.26354504e-01
1.20469138e-01 4.49358553e-01 -2.08029728e-02 4.65175956e-01
6.98321819e-01 -6.75349593e-01 -2.42842466e-01 -7.80713931e-03
3.13679874e-01 1.43345341e-01 5.72405457e-01 -1.55720365e+00
5.59317917e-02 1.43129289e-01 1.08067870e-01 -5.03986061e-01
2.08973438e-01 -1.20908689e+00 2.44406298e-01 9.77841735e-01
9.57495943e-02 1.45466849e-01 1.85739189e-01 6.28658056e-01
-4.84508067e-01 -4.42999184e-01 7.54782319e-01 3.92480828e-02
-8.10747564e-01 -5.20439558e-02 -7.27189481e-01 -3.66602868e-01
9.88059640e-01 -2.38878652e-01 -1.06189534e-01 9.56775993e-02
-6.87646151e-01 -2.43021306e-02 4.70068634e-01 3.87777448e-01
6.84682071e-01 -1.05181170e+00 -7.44290113e-01 5.21827757e-01
5.26108205e-01 -2.40757484e-02 3.67484748e-01 1.05797601e+00
-4.01106596e-01 -7.05885515e-03 -4.40421462e-01 -9.32722330e-01
-1.21863103e+00 5.43677866e-01 5.21190882e-01 -1.28002405e-01
-5.72432995e-01 4.73548442e-01 -5.51727772e-01 -8.85445923e-02
-8.45136568e-02 -3.46879840e-01 -2.11619526e-01 2.73317069e-01
3.18377078e-01 5.31190634e-01 9.59225416e-01 -6.95613027e-01
-3.31492901e-01 5.14734149e-01 1.00583859e-01 1.93543866e-01
1.51021659e+00 -2.61357296e-02 3.95586807e-03 9.31864381e-01
1.06018853e+00 -5.05975604e-01 -1.24303961e+00 -4.06941734e-02
5.71370602e-01 -3.11603189e-01 2.32328475e-01 -7.33704329e-01
-1.52592921e+00 6.20145440e-01 1.22071540e+00 8.42647374e-01
1.53450096e+00 -9.75530818e-02 2.78948754e-01 5.12700975e-01
4.64646161e-01 -1.27195084e+00 2.20364228e-01 1.43662915e-01
8.77898991e-01 -1.25252163e+00 1.06071718e-01 -1.36086017e-01
-4.73264813e-01 1.44829488e+00 3.04582208e-01 -6.71398342e-02
9.45122302e-01 4.33987379e-01 -2.36332640e-02 -7.04741001e-01
-7.95284390e-01 -3.88195515e-01 1.91373274e-01 6.45980716e-01
7.93046504e-02 -1.97081745e-01 1.74986199e-02 2.47022226e-01
8.84119887e-04 -2.29158700e-02 5.80807865e-01 1.38142002e+00
-5.04216790e-01 -1.04813564e+00 -4.80383307e-01 7.89920330e-01
-3.56212288e-01 5.03653049e-01 1.98147409e-02 1.07837188e+00
3.85003418e-01 1.21292615e+00 2.77860165e-01 -4.96358484e-01
1.00703299e+00 3.15408170e-01 3.29664767e-01 -6.49925053e-01
-5.59226334e-01 -3.61696601e-01 1.87643811e-01 -5.78559995e-01
-5.96279442e-01 -1.08941019e+00 -1.10278714e+00 2.34663501e-01
-5.97530186e-01 7.39887655e-02 9.18486953e-01 1.53818429e+00
-1.16826072e-01 1.10682285e+00 1.18222618e+00 -7.05828965e-01
-4.58458483e-01 -9.35066342e-01 -1.06829286e+00 4.69635874e-01
5.54724872e-01 -1.03178883e+00 -3.50910842e-01 1.71992451e-01] | [6.797791004180908, 2.371485471725464] |
ad931477-ae6b-43af-9e1c-bb0964eb5b43 | quota-the-quantile-option-architecture-for | 1811.02073 | null | http://arxiv.org/abs/1811.02073v2 | http://arxiv.org/pdf/1811.02073v2.pdf | QUOTA: The Quantile Option Architecture for Reinforcement Learning | In this paper, we propose the Quantile Option Architecture (QUOTA) for
exploration based on recent advances in distributional reinforcement learning
(RL). In QUOTA, decision making is based on quantiles of a value distribution,
not only the mean. QUOTA provides a new dimension for exploration via making
use of both optimism and pessimism of a value distribution. We demonstrate the
performance advantage of QUOTA in both challenging video games and physical
robot simulators. | ['Linglong Kong', 'Shangtong Zhang', 'Borislav Mavrin', 'Hengshuai Yao', 'Bo Liu'] | 2018-11-05 | null | null | null | null | ['distributional-reinforcement-learning'] | ['methodology'] | [-6.05582178e-01 1.91725776e-01 -6.25658274e-01 -2.32222110e-01
-5.25670588e-01 -4.19267297e-01 3.98672938e-01 8.77812505e-02
-6.24429762e-01 1.30467713e+00 2.57899940e-01 -4.53196853e-01
-3.71681243e-01 -1.03858221e+00 -5.46075284e-01 -5.75323224e-01
-4.63599533e-01 4.13380235e-01 -1.18879460e-01 -6.28800035e-01
5.62653065e-01 4.17317927e-01 -1.43320000e+00 -1.31636053e-01
7.94242442e-01 1.13556445e+00 1.13790445e-02 4.56215143e-01
-1.36816069e-01 7.96182752e-01 -8.54131877e-01 5.70304319e-02
4.25606132e-01 -3.70269030e-01 -4.49124396e-01 -5.11915147e-01
-5.49040198e-01 -4.76916939e-01 -4.03537452e-02 1.12635398e+00
4.13468897e-01 5.01623213e-01 8.71030986e-01 -1.47986364e+00
-4.32864040e-01 8.53668213e-01 -4.57943976e-01 1.63207129e-01
3.37218285e-01 4.13148612e-01 1.11915290e+00 -4.79866594e-01
4.24244344e-01 1.34821928e+00 2.30079561e-01 4.89184827e-01
-1.08780193e+00 -6.64302528e-01 -4.43452410e-03 2.25236416e-01
-1.10032654e+00 2.96514165e-02 5.52739561e-01 -3.19651425e-01
9.96549189e-01 -1.60757214e-01 1.20498478e+00 8.60309541e-01
9.09066617e-01 8.80077064e-01 1.41311383e+00 -3.16458702e-01
1.19750905e+00 -1.78836882e-01 -6.06596053e-01 1.09589927e-01
8.43271539e-02 1.07924986e+00 -4.39582914e-01 -3.09011489e-01
1.04191017e+00 -3.21535647e-01 6.73857927e-02 -7.68353045e-01
-6.13521755e-01 1.34149909e+00 3.42920184e-01 -3.53620708e-01
-6.93808794e-01 5.20408452e-01 1.90888360e-01 5.07089674e-01
1.32217526e-01 8.12492311e-01 -3.86370510e-01 -8.27244580e-01
-6.74971044e-01 5.51681995e-01 1.01316738e+00 5.60884774e-01
7.30800569e-01 6.04974806e-01 -3.28536212e-01 3.45222771e-01
5.47323406e-01 6.70026481e-01 6.35213077e-01 -1.14662135e+00
1.72004715e-01 1.59551501e-01 5.59808433e-01 -4.79901135e-01
-4.39429700e-01 -1.88903958e-01 -5.53711317e-02 1.13760710e+00
4.64324772e-01 -7.25047052e-01 -6.16781473e-01 1.87497902e+00
2.78074145e-01 -1.49351835e-01 3.01162094e-01 8.32461298e-01
-9.28872228e-02 5.01301527e-01 2.27466896e-01 -1.57909900e-01
7.09748983e-01 -3.71577680e-01 -6.99955523e-01 9.10613984e-02
2.63756007e-01 -9.44746882e-02 1.14660180e+00 6.41338825e-01
-9.32163835e-01 -2.06960857e-01 -1.26591587e+00 6.94621265e-01
-2.47642264e-01 -6.41178966e-01 7.17953086e-01 9.62771833e-01
-1.00436628e+00 9.08829391e-01 -7.89994299e-01 2.06302390e-01
4.35210496e-01 2.96007454e-01 2.19978064e-01 4.60014582e-01
-1.34505713e+00 1.00308752e+00 6.36030436e-01 -5.51932871e-01
-1.04398549e+00 -5.59024930e-01 -8.04729640e-01 1.49882481e-01
7.49007046e-01 -1.51956633e-01 1.52853346e+00 -6.03363395e-01
-2.55700493e+00 5.16308397e-02 8.45292449e-01 -8.46529067e-01
4.97984499e-01 -4.44725215e-01 -2.25191593e-01 1.47366241e-01
-1.62424371e-01 6.35103464e-01 7.83763051e-01 -6.58408105e-01
-5.00506759e-01 -1.19960681e-01 1.39248386e-01 4.05208796e-01
2.78366745e-01 -3.20630640e-01 7.39353299e-01 -4.20194715e-01
-4.12734687e-01 -7.01069474e-01 -6.22094572e-01 -1.85791716e-01
-9.95784253e-02 -3.20656508e-01 1.62604406e-01 -9.92346108e-02
1.07354665e+00 -1.83709705e+00 -1.75226480e-01 3.61653656e-01
-6.46406785e-02 -1.46504313e-01 1.72923103e-01 6.66893721e-01
1.01432100e-01 3.60252298e-02 6.85182661e-02 4.66163725e-01
5.81208825e-01 3.81588638e-01 -7.59966612e-01 3.94027531e-01
-1.21013567e-01 9.34666157e-01 -9.98029649e-01 -2.37339765e-01
1.91059828e-01 -3.53846490e-01 -8.26678038e-01 2.92643815e-01
-6.79483116e-01 1.74219295e-01 -5.87014914e-01 5.98756850e-01
5.80624640e-01 2.66839832e-01 2.89633870e-01 8.23597908e-01
-3.52060407e-01 3.08952838e-01 -1.16989863e+00 1.59175122e+00
-2.71307111e-01 1.48791626e-01 -1.67091310e-01 -8.81229103e-01
1.15403426e+00 -1.49853257e-02 5.26309371e-01 -9.87244666e-01
3.77185196e-01 2.76556045e-01 1.06827237e-01 -2.12497056e-01
7.04408109e-01 -6.55444980e-01 -4.59538609e-01 9.12392914e-01
9.34665650e-02 -6.72658265e-01 9.20278281e-02 -1.67607754e-01
7.01473296e-01 7.37236738e-01 8.42366874e-01 -8.27908337e-01
-1.44689351e-01 -8.73586088e-02 5.33542037e-01 9.14259970e-01
-4.79837000e-01 6.25693277e-02 1.13980603e+00 -2.07592517e-01
-1.01937258e+00 -1.66473794e+00 -3.81650664e-02 1.28496814e+00
8.29376280e-02 -1.01593643e-01 -5.09366691e-01 -6.27948582e-01
4.92374122e-01 1.43726945e+00 -7.42889166e-01 -4.25436437e-01
-3.68033260e-01 -4.47573841e-01 4.07568574e-01 7.17185318e-01
1.86928779e-01 -1.29973280e+00 -1.62118518e+00 2.25927368e-01
4.54374880e-01 -2.33181253e-01 -3.67379785e-01 4.14143801e-01
-7.92480826e-01 -8.84239793e-01 -6.17763638e-01 2.21163183e-01
-2.53936052e-01 -3.95157754e-01 1.17867374e+00 -4.98167217e-01
-1.18497364e-01 6.08420849e-01 -2.17584059e-01 -7.54331827e-01
-2.93614775e-01 -3.77765238e-01 2.72016644e-01 -4.27704096e-01
1.72275960e-01 -7.93975830e-01 -6.52063429e-01 2.05984652e-01
-5.62678814e-01 -4.45815712e-01 4.54714060e-01 9.11308229e-01
5.70672512e-01 6.33724267e-04 1.13872361e+00 -3.10812145e-01
1.10093629e+00 -9.10587430e-01 -1.10777009e+00 -7.79877603e-02
-8.57650936e-01 3.35254163e-01 4.41196144e-01 -4.56050843e-01
-1.10527575e+00 -4.88953978e-01 -5.86619750e-02 -3.26426685e-01
1.02447860e-01 7.02747643e-01 -6.95767105e-02 2.58379489e-01
7.49487698e-01 4.04205732e-02 4.77676630e-01 -2.83851177e-01
5.62838137e-01 3.74165237e-01 3.00645709e-01 -1.04676437e+00
9.86556560e-02 2.05363199e-01 1.86222732e-01 -6.40556931e-01
-4.42525297e-01 1.00205250e-01 -2.45044872e-01 -4.02553529e-01
5.83780050e-01 -5.16488373e-01 -1.33261180e+00 1.21808308e-03
-4.36730832e-01 -6.11389220e-01 -1.11246455e+00 7.19955087e-01
-1.37977350e+00 -6.21416643e-02 -1.46410540e-01 -1.32712066e+00
-7.96624180e-03 -9.11372364e-01 3.96013737e-01 7.62964487e-01
-1.43785819e-01 -9.20925975e-01 3.83300036e-01 -5.78192413e-01
7.25058854e-01 6.62405074e-01 7.59250283e-01 -6.84161544e-01
-3.33425701e-01 1.26856953e-01 4.96519729e-02 -4.49468382e-02
-2.13567793e-01 -2.07922131e-01 -4.59488004e-01 -3.21639091e-01
9.13769081e-02 -8.88422489e-01 5.98177969e-01 5.48792124e-01
8.65619957e-01 -3.03490430e-01 8.82497877e-02 3.07406873e-01
1.42274868e+00 5.32995045e-01 5.38601041e-01 6.47061646e-01
-1.60078853e-01 4.55847681e-01 9.92601156e-01 1.43345869e+00
4.13609207e-01 4.96382505e-01 9.30874348e-01 6.25120401e-01
5.14286757e-01 -6.21989727e-01 3.49233896e-01 -1.71788037e-01
8.39801803e-02 3.01154256e-02 -9.55538869e-01 4.85857904e-01
-1.83464324e+00 -9.16006684e-01 6.44830108e-01 2.47708631e+00
1.03297007e+00 4.43753272e-01 6.16188526e-01 -3.50131959e-01
3.73619169e-01 -8.36110860e-02 -1.05591166e+00 -9.39924240e-01
1.24146678e-01 4.28531110e-01 6.22374654e-01 3.95939618e-01
-7.13582695e-01 7.33632803e-01 7.54744339e+00 1.10988414e+00
-9.58561540e-01 -4.15256947e-01 4.72236663e-01 -3.34742278e-01
-4.26947653e-01 1.04123382e-02 -4.64156866e-01 5.08419812e-01
1.04725134e+00 -7.12536037e-01 5.45407116e-01 1.22074592e+00
-9.30293724e-02 -7.39317715e-01 -9.37570572e-01 6.56742215e-01
-5.72860301e-01 -1.20742130e+00 -2.52269953e-01 3.41386288e-01
5.75177491e-01 -5.19413911e-02 4.70137656e-01 6.60768092e-01
1.10796666e+00 -1.57706559e+00 1.12238669e+00 6.57617867e-01
6.65642440e-01 -1.47859573e+00 7.07467377e-01 4.02262807e-01
-7.25540996e-01 -4.87747520e-01 -6.63068891e-01 -5.09094298e-01
7.90668800e-02 2.09545627e-01 -8.07958126e-01 2.85541177e-01
5.01612365e-01 1.76069409e-01 -1.31782610e-02 9.99893606e-01
-4.69316721e-01 6.60085857e-01 -4.70018148e-01 -4.76685554e-01
6.05880976e-01 -4.12680954e-01 5.32360911e-01 7.81728625e-01
2.53410429e-01 9.76592451e-02 3.56900215e-01 1.14902747e+00
3.51233363e-01 5.84841557e-02 -5.37975490e-01 -1.86722413e-01
6.85010791e-01 7.35458195e-01 -5.99380732e-01 1.37802782e-02
1.38843581e-01 5.57034612e-02 3.24192375e-01 2.20797420e-01
-6.88974023e-01 -5.14604330e-01 9.35413182e-01 -3.35383922e-01
4.76969808e-01 -3.93829852e-01 -5.82603931e-01 -6.79707646e-01
-5.01395226e-01 -6.37055099e-01 6.27535760e-01 -3.73803794e-01
-1.00220335e+00 1.94956496e-01 3.85776699e-01 -1.13995135e+00
-9.37576294e-01 -6.20861411e-01 -6.93157554e-01 8.49599779e-01
-1.41714311e+00 -4.13756043e-01 4.41922098e-01 3.83354843e-01
1.82727560e-01 -3.80939305e-01 6.73230886e-01 -6.48171186e-01
-1.02597065e-02 4.79504257e-01 4.18355584e-01 -3.55196953e-01
3.13473046e-01 -1.86494744e+00 3.05327654e-01 1.19342597e-03
-4.93942499e-02 2.30717778e-01 1.02856612e+00 -7.54532516e-01
-1.28569365e+00 -2.18689561e-01 -1.16346747e-01 -1.01721905e-01
9.45836365e-01 -1.15288235e-01 -5.93200982e-01 3.19688052e-01
3.55282128e-01 -3.61252189e-01 7.68565118e-01 2.24924400e-01
-1.11489430e-01 2.07957715e-01 -1.42265284e+00 6.70780241e-01
4.03575480e-01 -6.06827885e-02 -6.02826416e-01 -2.31216609e-01
7.27033377e-01 -5.35294175e-01 -6.82183087e-01 1.12251326e-01
7.47791767e-01 -1.16703844e+00 8.96142840e-01 -7.26283014e-01
3.53570998e-01 2.50085145e-01 -4.79724526e-01 -1.78240633e+00
-2.82489974e-02 -9.04892325e-01 -2.44514301e-01 5.96139967e-01
7.97547922e-02 -7.42416382e-01 8.33205819e-01 6.36633396e-01
4.20155600e-02 -1.13880634e+00 -1.42907846e+00 -1.13035464e+00
7.66818941e-01 -3.82949740e-01 1.07070315e+00 4.54642028e-01
5.87083280e-01 -2.10536838e-01 -3.71815890e-01 -3.90591025e-01
9.08435464e-01 2.67745256e-01 3.62110019e-01 -8.14670444e-01
-6.48272157e-01 -6.78127408e-01 -1.83109164e-01 -7.67794728e-01
5.74763343e-02 -5.75393260e-01 1.97564036e-01 -9.78829443e-01
-3.35314840e-01 -2.18020022e-01 -6.36476159e-01 1.84082747e-01
2.68891662e-01 -2.42862448e-01 3.97685468e-01 -8.07510018e-02
-6.80017471e-01 1.23560560e+00 1.16646612e+00 2.75022417e-01
-4.46781665e-01 2.20315874e-01 -5.94326556e-01 7.87178457e-01
1.27501571e+00 -3.71964663e-01 -4.71002817e-01 2.50635833e-01
5.57422638e-01 9.02022481e-01 -4.18401025e-02 -8.46801102e-01
-7.10819736e-02 -9.58998263e-01 3.89814079e-01 -4.93533999e-01
3.33070576e-01 -3.55847687e-01 -3.15364033e-01 7.63795197e-01
-3.67210895e-01 2.68087566e-01 3.65067840e-01 8.03510129e-01
8.80926624e-02 -2.82085866e-01 9.25922215e-01 -1.56297997e-01
-7.36245275e-01 -1.04395837e-01 -8.90454531e-01 4.03188139e-01
1.05752599e+00 -2.11751014e-01 -7.45370053e-03 -7.30761945e-01
-5.78351498e-01 6.05668783e-01 4.37851608e-01 1.88447967e-01
8.42912734e-01 -1.31987977e+00 -5.66015601e-01 2.98528075e-02
-2.14414537e-01 -5.69475710e-01 7.00045824e-02 3.82766157e-01
-3.59886587e-01 2.82209337e-01 -1.00804698e+00 -2.34421685e-01
-2.96812236e-01 5.80442607e-01 3.84593338e-01 -4.73888606e-01
-4.39109445e-01 7.20027447e-01 -2.94600949e-02 -4.84556377e-01
6.96705468e-03 -1.34669155e-01 -4.24724221e-01 9.09052640e-02
5.05128860e-01 7.65102327e-01 -5.60938954e-01 1.40518665e-01
-1.73842505e-01 1.15009407e-02 2.88439631e-01 -8.96218181e-01
1.23138762e+00 -3.62073034e-02 2.36763984e-01 7.19693899e-01
2.84260035e-01 -1.18661389e-01 -1.83899355e+00 1.52226821e-01
4.03043568e-01 -5.76783717e-01 -7.39159808e-02 -1.18451130e+00
-4.11308855e-01 7.45936096e-01 4.02199656e-01 2.26163313e-01
7.67875433e-01 -3.37886125e-01 3.25897902e-01 3.06727469e-01
9.46518362e-01 -1.84158123e+00 4.19437945e-01 6.80274367e-01
8.93334329e-01 -8.70562673e-01 2.50935499e-02 5.74113488e-01
-1.14980865e+00 1.10895681e+00 6.41157568e-01 -5.80776513e-01
9.85638857e-01 2.30036378e-01 7.37828389e-02 4.04031090e-02
-1.00540614e+00 -1.92447022e-01 -1.17261447e-01 9.05765712e-01
-1.12801030e-01 5.75033426e-01 -4.09324646e-01 8.65590990e-01
-5.57744265e-01 -4.27176384e-03 9.39647198e-01 1.17012668e+00
-9.80972707e-01 -9.62002158e-01 -4.63791072e-01 3.43876541e-01
-3.23508471e-01 9.05080289e-02 6.05240650e-02 8.61743867e-01
-3.33781570e-01 9.20504332e-01 1.04129918e-01 -2.40366399e-01
8.77545029e-02 -3.15169655e-02 5.79137146e-01 -4.48237389e-01
-2.99315512e-01 -2.15587839e-01 -1.07830457e-01 -9.49919999e-01
4.18224663e-01 -7.59401441e-01 -1.55369401e+00 -2.13926479e-01
-9.69685242e-02 5.07673919e-01 7.96558440e-01 7.93660164e-01
1.56339005e-01 3.62477750e-01 6.96839929e-01 -7.39869177e-01
-1.72329366e+00 -7.94111192e-01 -1.31515157e+00 -3.23359877e-01
2.18699768e-01 -1.21202648e+00 -1.71866864e-01 -9.46229398e-01] | [4.157121658325195, 2.4862983226776123] |
e6da164b-b5ce-4b93-98dc-678aeef4db4a | adversarial-unsupervised-domain-adaptation | 2101.00701 | null | https://arxiv.org/abs/2101.00701v1 | https://arxiv.org/pdf/2101.00701v1.pdf | Adversarial Unsupervised Domain Adaptation for Harmonic-Percussive Source Separation | This paper addresses the problem of domain adaptation for the task of music source separation. Using datasets from two different domains, we compare the performance of a deep learning-based harmonic-percussive source separation model under different training scenarios, including supervised joint training using data from both domains and pre-training in one domain with fine-tuning in another. We propose an adversarial unsupervised domain adaptation approach suitable for the case where no labelled data (ground-truth source signals) from a target domain is available. By leveraging unlabelled data (only mixtures) from this domain, experiments show that our framework can improve separation performance on the new domain without losing any considerable performance on the original domain. The paper also introduces the Tap & Fiddle dataset, a dataset containing recordings of Scandinavian fiddle tunes along with isolated tracks for 'foot-tapping' and 'violin'. | ['Patrik Ohlsson', 'Sven Ahlbäck', 'Simon Dixon', 'Emmanouil Benetos', 'Carlos Lordelo'] | 2021-01-03 | null | null | null | null | ['music-source-separation'] | ['music'] | [ 4.55566227e-01 -4.22967486e-02 -8.05456191e-03 -9.76665616e-02
-1.46727693e+00 -9.86641169e-01 4.09660131e-01 -2.82423615e-01
-2.89093316e-01 8.25584710e-01 3.54259640e-01 1.61187395e-01
-4.74944651e-01 -2.36867771e-01 -5.67672014e-01 -9.25626516e-01
-1.71053000e-02 7.57498622e-01 1.73002258e-01 -2.74562359e-01
-5.11523411e-02 2.48416588e-01 -1.18791986e+00 2.78383553e-01
7.10006475e-01 7.49959409e-01 -2.92949043e-02 8.80063713e-01
2.38842323e-01 5.91002166e-01 -1.11045873e+00 -2.50535190e-01
6.76504314e-01 -9.01890516e-01 -7.78792202e-01 -3.78550496e-03
4.76371795e-01 2.22491145e-01 -2.82259345e-01 8.79335344e-01
1.01037228e+00 2.50360101e-01 7.50979125e-01 -1.16028416e+00
-3.05194080e-01 9.50621367e-01 -5.05051196e-01 2.29625717e-01
1.37999237e-01 -1.01345167e-01 9.07392800e-01 -4.26785856e-01
3.79259795e-01 9.17145431e-01 1.01987851e+00 5.27590096e-01
-1.57392454e+00 -9.74898160e-01 -3.01176935e-01 1.89678185e-02
-1.05541170e+00 -8.79049480e-01 1.15633035e+00 -5.75953901e-01
4.36179221e-01 -2.98994724e-02 1.48471877e-01 1.59017301e+00
-7.65550315e-01 7.14840949e-01 1.03622746e+00 -5.77736259e-01
4.27045166e-01 1.07866742e-01 -1.61761165e-01 -2.32445911e-01
-5.70140332e-02 3.87522399e-01 -8.90114188e-01 -3.46138507e-01
5.34058034e-01 -8.04183006e-01 -3.23746175e-01 -5.13334632e-01
-1.31783581e+00 6.33481681e-01 -3.76634882e-03 5.33501565e-01
-2.06401214e-01 -1.70909137e-01 6.64554477e-01 6.37796342e-01
3.59879524e-01 7.53120661e-01 -6.79066181e-01 -3.76835823e-01
-1.35261345e+00 4.76313502e-01 1.06998479e+00 7.22410679e-01
1.95068225e-01 5.16453624e-01 -1.06653824e-01 1.14981091e+00
-8.19732174e-02 6.76190615e-01 8.65321577e-01 -1.06707227e+00
5.63656509e-01 -2.21882746e-01 2.10482016e-01 -5.17866015e-01
-2.21273452e-01 -7.84977138e-01 -4.45599020e-01 2.43902475e-01
8.95604372e-01 -5.59530139e-01 -8.18182945e-01 2.03677511e+00
2.23937839e-01 5.93926966e-01 4.34852123e-01 1.00807559e+00
8.12055469e-01 3.89852434e-01 -3.37791473e-01 -1.62861854e-01
8.58823657e-01 -9.48187947e-01 -5.53238690e-01 -3.00688505e-01
5.37647791e-02 -9.05688226e-01 9.19187486e-01 9.62159932e-01
-9.90379095e-01 -9.64174986e-01 -1.18562543e+00 2.07565546e-01
-5.95274009e-02 1.10149190e-01 -1.89613197e-02 8.15257847e-01
-4.69764143e-01 7.39568472e-01 -3.91979456e-01 -5.32905497e-02
2.91307032e-01 4.07940626e-01 -3.35851729e-01 2.45917812e-01
-1.29296494e+00 4.49805647e-01 4.46808130e-01 -4.88158494e-01
-1.23944926e+00 -8.02539051e-01 -6.06223762e-01 -1.21840857e-01
2.27658108e-01 -2.24230170e-01 1.43017733e+00 -1.38762796e+00
-1.80907798e+00 8.87964010e-01 4.02493894e-01 -8.15955460e-01
4.73912358e-01 -5.77371836e-01 -8.73688281e-01 1.63336217e-01
2.95995355e-01 8.59933496e-02 1.49353814e+00 -1.28006649e+00
-5.10198057e-01 -2.32288256e-01 -3.46936196e-01 1.74368516e-01
-1.13403745e-01 1.54125571e-01 4.23668884e-02 -1.10606134e+00
-2.80017108e-01 -9.42613125e-01 1.97952464e-01 -7.35303998e-01
-5.25297880e-01 2.46456429e-01 7.71678030e-01 -9.74115014e-01
8.26981723e-01 -2.56331205e+00 4.63012606e-01 2.52303511e-01
-1.36844441e-01 4.79395777e-01 -3.67533028e-01 2.72361249e-01
-4.06190366e-01 -4.14019972e-01 -5.73708355e-01 -3.44331205e-01
1.78490758e-01 2.49674454e-01 -6.52806640e-01 3.85007590e-01
8.84925053e-02 4.35384512e-01 -9.14892375e-01 -1.07442357e-01
-1.61049560e-01 2.98321456e-01 -4.91571099e-01 2.72616982e-01
-6.94170818e-02 1.09756792e+00 1.63771346e-01 2.76012450e-01
5.45460224e-01 4.23996985e-01 2.52305150e-01 -6.13215263e-04
2.45359451e-01 5.86417198e-01 -1.59323823e+00 2.20882368e+00
-2.74163216e-01 7.76792824e-01 5.07023156e-01 -1.24096537e+00
1.08604884e+00 6.24604106e-01 5.20793915e-01 -3.78245085e-01
9.40425619e-02 5.55985332e-01 3.83164823e-01 -3.02489638e-01
5.69708943e-02 -7.40594089e-01 -4.81285214e-01 4.03430402e-01
8.20190430e-01 -3.06906223e-01 3.45554128e-02 -1.80107951e-01
1.10694790e+00 3.83577347e-01 7.54929036e-02 -1.20527521e-01
3.70572358e-01 6.05420284e-02 7.18971968e-01 7.26896822e-01
-2.67530739e-01 9.02092099e-01 2.26129070e-01 1.68486506e-01
-9.05094445e-01 -1.27775705e+00 -2.29873601e-02 1.33881366e+00
-1.72140300e-01 -2.23784119e-01 -6.28882051e-01 -6.51883602e-01
6.92932755e-02 7.28517473e-01 -4.87368375e-01 -3.41719955e-01
-6.99085593e-01 -5.59761584e-01 1.28826523e+00 4.19702977e-01
4.29308981e-01 -8.44033837e-01 -3.39240670e-01 2.94689834e-01
-4.46501493e-01 -1.08212245e+00 -3.47433627e-01 7.20987141e-01
-7.88015366e-01 -9.38665986e-01 -1.01578736e+00 -8.04817080e-01
-2.68602729e-01 -2.40617886e-01 1.20405233e+00 -9.71256793e-01
5.38275540e-02 4.61715788e-01 -3.99195164e-01 -8.74803424e-01
-8.08820665e-01 1.88053071e-01 4.25716549e-01 2.14667469e-01
4.05353904e-01 -1.03635299e+00 -1.18148774e-01 2.99595922e-01
-6.38078213e-01 -5.76106310e-01 3.19004685e-01 8.76962841e-01
3.34735394e-01 3.77009511e-01 1.09987044e+00 -7.87519634e-01
6.77967012e-01 -5.08642375e-01 -1.33691475e-01 -2.27723494e-01
-6.26440942e-02 -7.42213279e-02 6.55453920e-01 -8.15310836e-01
-1.02510965e+00 9.87049937e-02 5.35119027e-02 -7.72541225e-01
-6.28108203e-01 1.83576271e-01 -4.57277179e-01 1.76885232e-01
1.35000086e+00 9.47010443e-02 -6.85331970e-02 -1.02587903e+00
2.36231536e-01 7.37953901e-01 1.38530803e+00 -7.62157679e-01
1.13692057e+00 3.49680692e-01 -2.65635520e-01 -7.20219970e-01
-8.19813132e-01 -7.42606401e-01 -9.83383596e-01 1.97264552e-01
5.33894897e-01 -1.07441401e+00 -1.02364570e-01 6.13773346e-01
-8.27015996e-01 -4.95774299e-01 -9.05836403e-01 7.85829186e-01
-8.28790605e-01 2.30631843e-01 -2.00036049e-01 -7.21400440e-01
-7.81983230e-03 -5.84818125e-01 1.02909184e+00 1.97878815e-02
-4.16153282e-01 -1.03182495e+00 8.26117635e-01 2.52057970e-01
9.02978927e-02 3.63819152e-01 5.04081666e-01 -1.38587499e+00
2.73361832e-01 -9.51117650e-02 5.01832962e-01 9.76144135e-01
2.02873096e-01 -5.01799703e-01 -1.62467575e+00 -1.93672672e-01
4.24470961e-01 -4.64237154e-01 8.25525403e-01 2.60480314e-01
3.94619316e-01 -1.20663308e-01 3.53364944e-02 5.97188354e-01
1.02054119e+00 1.19736567e-01 4.65097427e-01 3.26355010e-01
5.72717607e-01 5.76698542e-01 4.55550373e-01 3.07873785e-01
-2.36947984e-01 7.90883839e-01 1.73299675e-04 -3.74457799e-02
-3.63156378e-01 -2.23592699e-01 6.26342773e-01 5.86735010e-01
-1.60262823e-01 -5.53543754e-02 -8.35964978e-01 9.77033079e-01
-1.65135050e+00 -1.08174062e+00 1.27181932e-01 2.33968186e+00
1.08117068e+00 1.05760671e-01 9.28923786e-01 9.48761582e-01
8.24593306e-01 -3.34635191e-02 -5.01217782e-01 -2.52074033e-01
-2.40997270e-01 8.29026997e-01 3.02744925e-01 3.41914654e-01
-1.55177009e+00 5.65465569e-01 6.22720003e+00 9.06791031e-01
-1.08436990e+00 2.50846744e-01 -3.74893606e-01 -3.48091245e-01
2.10795775e-01 -1.61439583e-01 -2.68456191e-01 6.02097690e-01
1.23968196e+00 -2.02133413e-02 5.33471048e-01 5.77439189e-01
-1.95892379e-01 2.03479871e-01 -1.22447503e+00 1.08555615e+00
1.01229951e-01 -7.07529247e-01 -5.69088638e-01 -1.22577049e-01
7.15127945e-01 4.20485698e-02 1.41193599e-01 5.97820282e-01
4.64910448e-01 -8.45193386e-01 8.37135851e-01 1.42028630e-01
7.21397877e-01 -8.33375871e-01 4.03024226e-01 7.08441257e-01
-5.81818283e-01 -1.33991182e-01 -5.17755635e-02 5.47338799e-02
-4.75801295e-03 6.14383221e-01 -9.54144418e-01 7.94792891e-01
6.36951268e-01 6.96871936e-01 -3.24461281e-01 1.03176320e+00
-1.74549654e-01 1.23069477e+00 -4.44696099e-01 9.15584803e-01
-1.97434351e-01 5.20193391e-02 1.23309433e+00 1.40306902e+00
2.95065373e-01 -3.11306208e-01 1.28951659e-02 6.22681677e-01
5.14867576e-03 -1.24226533e-01 -5.99831581e-01 -5.22792339e-02
3.32933426e-01 9.64472055e-01 -4.11767542e-01 -1.96292892e-01
-1.11062534e-01 1.09562147e+00 -2.17643738e-01 3.55980426e-01
-8.53537798e-01 -7.14855969e-01 6.37064755e-01 -5.48128458e-03
6.35378361e-01 -3.74401510e-02 -3.09422672e-01 -1.09785569e+00
-4.09691870e-01 -1.18142259e+00 4.86411929e-01 -5.12213945e-01
-1.60665298e+00 5.33612549e-01 1.32369861e-01 -1.57375419e+00
-6.60811841e-01 -3.80474538e-01 -4.96956974e-01 1.12581885e+00
-1.23772764e+00 -9.47650313e-01 1.78969055e-01 9.43482399e-01
3.08361501e-01 -7.99506068e-01 9.64838386e-01 5.08731902e-01
-1.08760633e-01 6.59120023e-01 5.36900520e-01 4.15628523e-01
1.41044486e+00 -1.53872323e+00 2.87654936e-01 7.62563169e-01
8.84358823e-01 1.47533849e-01 9.86629546e-01 -3.47806782e-01
-6.07007861e-01 -9.50170875e-01 4.41200942e-01 -6.24436498e-01
5.28389931e-01 -3.79132569e-01 -9.46909785e-01 4.32685971e-01
2.83155739e-01 -2.92327166e-01 1.22125638e+00 1.89675286e-01
-5.53437173e-01 -6.54172301e-02 -1.19222462e+00 6.29572645e-02
7.71189451e-01 -7.82866597e-01 -1.21004927e+00 1.52067430e-02
3.25633138e-01 -4.70345944e-01 -7.25857437e-01 4.77318704e-01
4.90701526e-01 -1.05534256e+00 1.14495087e+00 -7.86759973e-01
2.10005820e-01 -2.76054561e-01 -2.29425058e-01 -1.77082396e+00
-1.98428765e-01 -8.76282096e-01 -2.96199650e-01 1.56301391e+00
1.29601419e-01 -2.63989896e-01 4.47320700e-01 -8.04074332e-02
-1.46189243e-01 3.11911762e-01 -1.00421154e+00 -1.15499878e+00
3.40864003e-01 -5.48181355e-01 4.07368332e-01 1.19843256e+00
-1.93278566e-01 7.18627751e-01 -5.25401354e-01 1.31965235e-01
6.34352803e-01 3.01979989e-01 1.12416017e+00 -1.55493236e+00
-1.12685204e+00 -1.97309658e-01 -3.24790329e-01 -6.18258059e-01
3.25034916e-01 -1.10496211e+00 1.65184677e-01 -9.38002110e-01
-4.30268705e-01 -1.60196215e-01 -5.73223293e-01 3.79466981e-01
1.12793885e-01 5.69889605e-01 3.68393719e-01 3.59843343e-01
-1.81340486e-01 1.91706866e-01 8.65629733e-01 -2.28260100e-01
-4.88556117e-01 3.97595972e-01 -9.00234401e-01 9.07732904e-01
8.30183685e-01 -7.51053095e-01 -5.20176470e-01 -2.73130655e-01
-3.25017840e-01 8.78541544e-02 5.39183617e-01 -1.49627209e+00
-1.49134532e-01 2.89969593e-01 3.53142321e-01 -1.35075495e-01
5.42761028e-01 -7.66598761e-01 1.65821061e-01 -1.96757708e-02
-5.84073186e-01 -8.61726224e-01 5.12852669e-01 7.07767069e-01
-4.44773674e-01 -3.79502416e-01 9.74503160e-01 1.56096578e-01
-4.44646835e-01 -3.73901784e-01 -1.64361566e-01 6.48044765e-01
8.06443989e-01 -2.31066704e-01 8.73578936e-02 -4.63303298e-01
-1.30679154e+00 -3.30135614e-01 2.69464493e-01 3.90232563e-01
1.21906372e-02 -1.49501729e+00 -1.06161427e+00 4.59463805e-01
-8.52351040e-02 -9.39591900e-02 1.06245212e-01 7.08912969e-01
1.37694076e-01 4.21928428e-02 -3.92206281e-01 -4.74078476e-01
-1.13344276e+00 6.03177547e-01 3.44426632e-01 -2.36650005e-01
-3.69087189e-01 1.04855514e+00 2.28839621e-01 -7.25667417e-01
2.45047390e-01 2.34257542e-02 -1.13864124e-01 3.27379763e-01
3.28878909e-01 5.65265179e-01 1.48640752e-01 -7.66666770e-01
-3.36467177e-01 3.60787421e-01 3.86821508e-01 -7.58143902e-01
1.22021651e+00 1.11128077e-01 3.95720571e-01 8.92503858e-01
1.05990946e+00 8.37246835e-01 -1.16352308e+00 -5.09764791e-01
5.35497665e-02 -3.51330042e-01 -2.06390187e-01 -1.16288173e+00
-8.42803121e-01 1.00476909e+00 4.88274932e-01 1.68823704e-01
1.41337729e+00 -4.65349071e-02 6.88398540e-01 1.42462984e-01
1.49373040e-01 -1.22858346e+00 9.31926295e-02 5.35046279e-01
9.30082858e-01 -8.53329599e-01 -3.96788865e-01 -3.77205573e-03
-9.36111391e-01 9.49009895e-01 8.48221928e-02 -5.99000573e-01
6.80591881e-01 4.21840489e-01 2.72898555e-01 1.92028984e-01
-2.38638014e-01 -3.78527373e-01 5.84805489e-01 1.23036182e+00
1.74104616e-01 -1.10004293e-02 2.96027660e-01 1.33115482e+00
-4.93791044e-01 2.21412703e-02 2.66779631e-01 6.87109530e-01
-4.61914018e-02 -1.38946295e+00 -9.30714011e-01 1.57752559e-01
-8.54916692e-01 1.00680161e-02 -7.31710613e-01 6.97622418e-01
4.72017586e-01 1.01947880e+00 -1.43434197e-01 -4.21172529e-01
4.22002405e-01 5.80405235e-01 6.74873590e-01 -8.10431480e-01
-7.42743611e-01 6.89548910e-01 2.45393440e-01 -4.31287140e-02
-8.65173161e-01 -7.59726346e-01 -9.79601324e-01 2.54987419e-01
-1.08204141e-01 5.10888815e-01 2.57169187e-01 7.56672084e-01
1.10177994e-01 7.87019193e-01 5.25167644e-01 -1.10173845e+00
-7.13398874e-01 -1.14621890e+00 -1.12318015e+00 6.07645154e-01
6.81387246e-01 -6.85600340e-01 -4.25475448e-01 5.24043024e-01] | [15.549175262451172, 5.474595069885254] |
982bd7b1-3ccd-4ba0-a170-6d9d367b65d0 | consistency-and-accuracy-of-celeba-attribute | 2210.07356 | null | https://arxiv.org/abs/2210.07356v2 | https://arxiv.org/pdf/2210.07356v2.pdf | Consistency and Accuracy of CelebA Attribute Values | We report the first systematic analysis of the experimental foundations of facial attribute classification. Two annotators independently assigning attribute values shows that only 12 of 40 common attributes are assigned values with >= 95% consistency, and three (high cheekbones, pointed nose, oval face) have essentially random consistency. Of 5,068 duplicate face appearances in CelebA, attributes have contradicting values on from 10 to 860 of the 5,068 duplicates. Manual audit of a subset of CelebA estimates error rates as high as 40% for (no beard=false), even though the labeling consistency experiment indicates that no beard could be assigned with >= 95% consistency. Selecting the mouth slightly open (MSO) for deeper analysis, we estimate the error rate for (MSO=true) at about 20% and (MSO=false) at about 2%. A corrected version of the MSO attribute values enables learning a model that achieves higher accuracy than previously reported for MSO. Corrected values for CelebA MSO are available at https://github.com/HaiyuWu/CelebAMSO. | ['Kevin W. Bowyer', 'Michael C. King', 'Terrance Boult', 'Manuel Günther', 'Grace Bezold', 'Haiyu Wu'] | 2022-10-13 | null | null | null | null | ['facial-attribute-classification'] | ['computer-vision'] | [-1.73258021e-01 5.20124793e-01 -1.84183165e-01 -8.05955827e-01
-5.79265654e-01 -5.23842037e-01 3.39214176e-01 2.65186816e-01
-2.41899252e-01 6.06520832e-01 -6.20888658e-02 -6.20192029e-02
-4.45409082e-02 -4.54447538e-01 -4.69954252e-01 -5.38216293e-01
-1.04221746e-01 6.81448936e-01 4.62268181e-02 6.77615218e-03
-5.19169755e-02 5.45059025e-01 -1.97077632e+00 3.76039416e-01
3.13151300e-01 1.21996832e+00 -3.83099943e-01 3.97531718e-01
-5.86317629e-02 3.69742006e-01 -7.59226024e-01 -1.12364018e+00
3.80476087e-01 -2.08904833e-01 -9.50516045e-01 -2.23262180e-02
1.30883026e+00 -4.30976391e-01 2.05657572e-01 9.22778726e-01
5.03184736e-01 -3.79190683e-01 8.47210586e-01 -1.70089769e+00
-8.27073336e-01 4.71653312e-01 -6.78341091e-01 -7.93959424e-02
5.61949790e-01 -1.34584755e-02 1.23970342e+00 -9.40362811e-01
7.06192970e-01 1.26460063e+00 9.13096964e-01 7.39077568e-01
-1.36812568e+00 -1.15664351e+00 -1.70745000e-01 8.34894851e-02
-1.97232556e+00 -9.08892274e-01 1.69654548e-01 -3.85055155e-01
4.15032297e-01 4.54734772e-01 7.40076065e-01 8.78890157e-01
-1.25296399e-01 2.66621977e-01 1.34506607e+00 -3.77805322e-01
9.75909606e-02 5.24548411e-01 -7.37767518e-02 8.06104660e-01
5.02310276e-01 -1.63907662e-01 -6.72031999e-01 -5.94818592e-01
4.71441925e-01 -5.80053449e-01 1.35288671e-01 1.89925492e-01
-7.59538949e-01 5.56189716e-01 2.18349218e-01 -4.82809022e-02
-5.43557666e-02 -5.74970851e-03 1.71578199e-01 3.97423536e-01
3.23547632e-01 1.82137594e-01 -4.11479294e-01 5.81596047e-02
-6.73361897e-01 1.96445763e-01 8.24226558e-01 1.36620033e+00
7.70138979e-01 -6.59523681e-02 2.75691897e-01 1.27762663e+00
2.35393539e-01 4.94386286e-01 1.66658461e-01 -1.28693330e+00
-5.93651412e-03 3.73719096e-01 3.17989916e-01 -9.29904878e-01
-4.61879700e-01 2.77427793e-01 -3.02420020e-01 4.73559588e-01
9.30984974e-01 2.57166959e-02 -7.82607198e-01 1.77336526e+00
4.44326252e-01 -3.49223286e-01 -3.87995273e-01 6.94913149e-01
1.15462935e+00 -1.59754194e-02 4.51322526e-01 -2.06593890e-02
1.90790570e+00 -1.05315924e-01 -7.03046978e-01 -1.27226681e-01
3.92465800e-01 -1.12337434e+00 1.08647108e+00 3.64825964e-01
-9.79118764e-01 -2.74465799e-01 -8.64376783e-01 5.60101867e-02
-3.67445201e-01 2.77439505e-01 6.36014879e-01 1.06256306e+00
-1.07834768e+00 3.22312653e-01 -2.73761958e-01 -7.13025630e-01
6.58330142e-01 5.32248795e-01 -9.90743697e-01 1.81997940e-01
-1.00985098e+00 1.02544594e+00 3.59868743e-02 1.03969522e-01
-5.43460250e-01 -4.35951322e-01 -6.95186377e-01 -3.07789147e-01
3.12996268e-01 -2.04082832e-01 1.17611110e+00 -1.01613700e+00
-1.15417159e+00 1.52870882e+00 -4.64360803e-01 5.83667159e-02
3.70545268e-01 -4.89432998e-02 -7.60922849e-01 2.38666102e-01
1.22524507e-01 9.36057746e-01 6.55885994e-01 -1.51386917e+00
-6.82995260e-01 -3.75568330e-01 -8.75884108e-03 -5.91224767e-02
-2.50704855e-01 3.73650551e-01 8.42943266e-02 -4.70180124e-01
3.60189050e-01 -1.03224945e+00 5.15092134e-01 7.18144596e-01
-1.55892491e-01 -5.08317947e-01 2.21813530e-01 -8.10420752e-01
1.01861680e+00 -2.02834749e+00 -6.78548813e-01 2.00594112e-01
4.06301677e-01 -1.66240502e-02 -2.40543541e-02 2.67321855e-01
-2.78149903e-01 3.67058516e-01 -3.57166417e-02 -2.87401766e-01
4.28526364e-02 1.89668238e-01 2.63424009e-01 6.64268076e-01
2.89110154e-01 3.34059626e-01 -7.23758519e-01 -6.57634318e-01
-9.96797606e-02 1.94033489e-01 -2.37604350e-01 -1.05850823e-01
3.69288474e-01 -7.27230161e-02 4.84467559e-02 1.32075143e+00
9.75123644e-01 1.24850541e-01 2.83905178e-01 -4.43796366e-01
7.74262822e-04 -9.74684432e-02 -1.17467666e+00 7.96352327e-01
-1.68018132e-01 6.30419672e-01 3.29789937e-01 -1.87190115e-01
1.16511548e+00 4.02360618e-01 3.72479975e-01 -4.01778042e-01
1.85973510e-01 4.38257039e-01 9.83932689e-02 -4.89967823e-01
3.51710886e-01 -4.81981903e-01 9.59204361e-02 3.02767485e-01
7.17932880e-02 -1.00838959e-01 -1.69843823e-01 5.13165034e-02
5.92830420e-01 1.66886151e-02 4.32759762e-01 -4.46820378e-01
2.73242742e-01 -2.22528622e-01 7.81194985e-01 4.00570929e-01
-5.23619592e-01 9.10418034e-01 6.83997273e-01 -5.43090761e-01
-1.08201408e+00 -1.06014383e+00 -8.37282121e-01 1.13993859e+00
-1.29110515e-01 -6.16987646e-01 -9.52265501e-01 -3.82076055e-01
1.92185506e-01 5.43645501e-01 -9.26172197e-01 -6.82834089e-02
1.79156169e-01 -8.34039330e-01 8.47560406e-01 4.12125558e-01
2.95379430e-01 -9.02727365e-01 -3.22323889e-01 -2.63003737e-01
-2.39756674e-01 -1.07990038e+00 -2.11452335e-01 -2.43469141e-02
-3.17542076e-01 -1.20620298e+00 -4.75826919e-01 -5.31208634e-01
7.02984989e-01 -2.17580751e-01 1.14450812e+00 3.97209436e-01
-4.95847553e-01 3.05993259e-01 -2.85570771e-01 -5.86662412e-01
-4.75831389e-01 -4.23214585e-01 4.14021432e-01 6.79649878e-03
8.22964013e-01 -3.29091698e-01 -2.97615439e-01 6.56552196e-01
-3.92701626e-01 -4.66928631e-01 2.68819690e-01 6.75454140e-01
3.87849063e-01 -3.69462222e-01 5.57242930e-01 -9.79051828e-01
1.04556635e-01 -5.29286087e-01 -3.56034249e-01 1.60148323e-01
-6.84170842e-01 -4.29574847e-01 2.40349248e-02 -5.48475683e-01
-9.22771454e-01 7.78971389e-02 -2.15739116e-01 -2.62181878e-01
-6.08551085e-01 -9.52061564e-02 -2.86987927e-02 3.92664894e-02
7.68319964e-01 -5.01084924e-01 3.28979433e-01 -5.48264921e-01
-2.36668848e-02 8.99708450e-01 5.12036622e-01 -7.35448003e-01
4.19959664e-01 5.98696768e-01 -6.37382418e-02 -9.50401902e-01
-7.15626180e-01 -1.13687694e-01 -7.65705109e-01 -2.53582954e-01
7.31505692e-01 -8.86541724e-01 -1.24764144e+00 4.94881511e-01
-6.96657717e-01 -1.96783260e-01 1.18532673e-01 3.78805935e-01
-4.45881784e-01 2.08508164e-01 -4.56572562e-01 -9.21372652e-01
-7.96421468e-02 -9.15215671e-01 1.05953741e+00 9.58485082e-02
-9.97002602e-01 -5.81175268e-01 -4.14296240e-01 6.40207231e-01
3.37850749e-02 2.56332129e-01 8.30177546e-01 -7.64427066e-01
3.14916074e-01 -4.19516355e-01 -2.44146928e-01 5.61654121e-02
5.42348862e-01 4.86072809e-01 -1.37913644e+00 -9.58351642e-02
-4.64541852e-01 -4.11351055e-01 2.77650118e-01 4.69415784e-02
1.02660811e+00 -3.29782754e-01 -2.11021915e-01 3.49955082e-01
1.04455221e+00 1.33592337e-01 5.82311869e-01 3.41355085e-01
4.11825567e-01 1.09221482e+00 7.35260427e-01 5.54798663e-01
4.60703552e-01 9.17939126e-01 4.55566525e-01 8.88973102e-02
-3.54896456e-01 1.90978888e-02 2.07704231e-01 -3.59384790e-02
-4.14556921e-01 1.10835597e-01 -9.38714921e-01 4.61594611e-01
-1.11628294e+00 -9.39206004e-01 -4.12996590e-01 2.59880900e+00
8.98877978e-01 3.77601944e-02 4.41141665e-01 -3.59159969e-02
1.00506973e+00 -3.27562243e-01 -1.77719116e-01 -6.21507466e-01
-8.21680855e-03 8.31277072e-02 4.33747143e-01 5.51820219e-01
-8.47255230e-01 8.04374039e-01 7.48697948e+00 6.05414987e-01
-8.30749393e-01 -9.11408365e-02 7.91008234e-01 -2.82548934e-01
-1.56727895e-01 -2.85578016e-02 -1.09262693e+00 6.58729076e-01
8.18596959e-01 1.05964221e-01 2.11652905e-01 6.97978675e-01
-8.58853459e-02 -2.38698259e-01 -9.87712860e-01 1.04838300e+00
1.80154905e-01 -6.88785136e-01 -5.31745255e-02 2.02541858e-01
3.48479986e-01 -3.17614138e-01 5.20660952e-02 -1.78112283e-01
2.37557754e-01 -1.19150591e+00 1.26289308e+00 4.12347198e-01
1.27503729e+00 -6.48469150e-01 8.50931108e-01 -2.95471221e-01
-1.03280270e+00 -3.21617462e-02 -6.37224793e-01 -5.06701432e-02
-2.87577033e-01 4.28368688e-01 -1.00255394e+00 1.38957873e-01
1.23820984e+00 3.34792495e-01 -9.24336910e-01 6.54816687e-01
-2.11760327e-01 4.72130865e-01 -7.11431623e-01 8.32016394e-02
-2.58837134e-01 -8.41372982e-02 3.50101292e-01 9.46529806e-01
2.38341436e-01 2.45843813e-01 -3.87602925e-01 6.41010582e-01
-1.74476489e-01 9.37161744e-02 -3.45511615e-01 1.34518415e-01
9.45732713e-01 1.41244137e+00 -7.29145586e-01 -1.82952687e-01
-5.00920475e-01 8.20882738e-01 2.64216512e-01 1.02806516e-01
-7.14637518e-01 -1.19046867e-01 9.40270841e-01 4.62016761e-01
6.10657483e-02 3.66160154e-01 -2.07740620e-01 -6.64606214e-01
8.93034562e-02 -9.29848850e-01 6.78519785e-01 -8.85200620e-01
-1.45989001e+00 7.09922969e-01 2.22911790e-01 -1.25216448e+00
6.32413998e-02 -8.40485036e-01 -3.00738424e-01 8.77283216e-01
-8.16781223e-01 -1.36807299e+00 -6.12709939e-01 2.23722249e-01
7.06679001e-02 -1.66465729e-01 1.12848711e+00 2.64420509e-01
-4.31154996e-01 1.15663421e+00 -2.52360433e-01 9.89563689e-02
1.24669147e+00 -1.19117463e+00 2.08513349e-01 -1.61543190e-01
-6.30479530e-02 8.49180460e-01 7.09060192e-01 -5.36796927e-01
-6.43034697e-01 -5.71043074e-01 1.21152258e+00 -9.04335558e-01
5.54191470e-01 -2.67067254e-01 -8.59420896e-01 8.31443310e-01
-3.70417349e-02 2.14465678e-01 1.04171348e+00 3.61287683e-01
-7.59495437e-01 -1.80185243e-01 -1.59836519e+00 5.32447934e-01
1.03692257e+00 -3.97189796e-01 -3.48411918e-01 2.65056849e-01
-7.53077194e-02 -1.87824562e-01 -1.37978804e+00 1.42770499e-01
1.22910190e+00 -1.09780061e+00 8.42873156e-01 -5.42667389e-01
2.25644469e-01 -3.27530175e-01 -2.20858082e-01 -8.22557151e-01
-3.38301450e-01 -3.38246524e-01 3.33337277e-01 1.71485519e+00
4.72844303e-01 -8.58989835e-01 7.58217812e-01 1.00699294e+00
1.96032494e-01 -7.23444939e-01 -1.14670694e+00 -7.44783819e-01
7.30420873e-02 -1.92926168e-01 7.71876276e-01 1.20725560e+00
-5.94819561e-02 -1.27793908e-01 -2.37197548e-01 1.31037369e-01
6.08967304e-01 -2.90327400e-01 4.97423589e-01 -1.40288520e+00
1.43585756e-01 -4.12769616e-01 -6.91813231e-01 -1.31602272e-01
4.23212945e-02 -8.30819905e-01 -2.46232316e-01 -1.09052014e+00
3.54808569e-01 -8.47086310e-01 7.38931894e-02 1.29188788e+00
-1.15341730e-02 9.70709860e-01 2.59944028e-03 2.32310623e-01
6.43357635e-02 3.32976400e-04 7.81485260e-01 1.34428233e-01
3.64442468e-01 -2.16036275e-01 -8.97778869e-01 9.91629064e-01
7.28926361e-01 -5.55991292e-01 2.09091348e-03 -2.06002533e-01
3.08420777e-01 -3.44277263e-01 4.14739043e-01 -7.67171741e-01
-1.97380081e-01 -1.31045491e-01 7.82900333e-01 -2.57577207e-02
5.86421907e-01 -8.74372780e-01 4.12049204e-01 2.53678739e-01
-1.19917184e-01 -8.81999284e-02 4.72596258e-01 7.89093077e-02
1.39682382e-01 -5.69074869e-01 8.23009610e-01 -2.42101341e-01
-6.61460698e-01 8.53699520e-02 -4.79709953e-01 -2.02805728e-01
9.57493544e-01 -4.74522173e-01 -4.01650339e-01 -4.96426404e-01
-1.09929192e+00 1.15002934e-02 7.01183975e-01 3.47943306e-01
3.33707511e-01 -1.40721357e+00 -7.39741147e-01 1.75140277e-01
5.26112914e-01 -5.42273700e-01 -1.02506064e-01 6.99341059e-01
-4.71488148e-01 -2.23060608e-01 -5.24145842e-01 -2.79618353e-01
-1.84904945e+00 2.14296609e-01 2.50759989e-01 8.68897140e-01
-2.56705254e-01 1.00531602e+00 -2.85450481e-02 -3.72288048e-01
-6.49825707e-02 2.65723765e-01 -7.49538988e-02 5.81479609e-01
4.93969411e-01 5.81672490e-01 1.28564090e-01 -1.08249652e+00
-7.21927404e-01 7.09975243e-01 -1.14183694e-01 -9.23641305e-03
1.05497062e+00 -1.31962433e-01 -3.77756327e-01 4.93302226e-01
1.01587284e+00 3.30077410e-01 -9.07583833e-01 3.54271352e-01
-1.94347143e-01 -7.23485172e-01 -4.60002065e-01 -9.04796124e-01
-7.24257648e-01 4.86889035e-01 7.21666157e-01 5.67179769e-02
9.45494831e-01 4.79078114e-01 2.44253963e-01 1.91688210e-01
5.26331723e-01 -9.32955742e-01 -3.53262663e-01 -8.64630491e-02
8.72014582e-01 -1.33610404e+00 4.21716452e-01 -8.55134726e-01
-8.75544369e-01 1.10959935e+00 7.43349612e-01 2.09106550e-01
4.26944196e-01 3.33547652e-01 4.09610331e-01 -3.47178608e-01
-6.56315625e-01 -3.75195369e-02 1.90736294e-01 6.60340667e-01
8.30615342e-01 1.81152016e-01 -5.09497225e-01 4.93457675e-01
-6.85293317e-01 -2.57469326e-01 5.45572460e-01 7.57751882e-01
-2.02114090e-01 -1.08244419e+00 -5.93381047e-01 6.73145711e-01
-5.44565856e-01 9.36621800e-02 -6.72766387e-01 9.66510653e-01
3.60412747e-01 9.17082131e-01 5.00149429e-01 -3.69756877e-01
1.56985804e-01 3.38753700e-01 6.99134231e-01 -3.28581512e-01
-3.99770111e-01 -8.85001197e-02 6.19659841e-01 -3.85387033e-01
-4.48118240e-01 -8.74307573e-01 -1.17731452e+00 -6.17712915e-01
-4.30525184e-01 -2.67457273e-02 3.80503029e-01 6.02991879e-01
-2.62721088e-02 -3.21283013e-01 3.27349097e-01 -3.53952318e-01
-2.78204888e-01 -9.74167705e-01 -7.17783928e-01 6.97687149e-01
1.65466860e-01 -9.85664546e-01 -6.56375170e-01 3.14405411e-01] | [13.048161506652832, 1.193516492843628] |
4ddb04f2-77f5-4fe3-b182-25b8fade67a6 | df-vo-what-should-be-learnt-for-visual | 2103.00933 | null | https://arxiv.org/abs/2103.00933v1 | https://arxiv.org/pdf/2103.00933v1.pdf | DF-VO: What Should Be Learnt for Visual Odometry? | Multi-view geometry-based methods dominate the last few decades in monocular Visual Odometry for their superior performance, while they have been vulnerable to dynamic and low-texture scenes. More importantly, monocular methods suffer from scale-drift issue, i.e., errors accumulate over time. Recent studies show that deep neural networks can learn scene depths and relative camera in a self-supervised manner without acquiring ground truth labels. More surprisingly, they show that the well-trained networks enable scale-consistent predictions over long videos, while the accuracy is still inferior to traditional methods because of ignoring geometric information. Building on top of recent progress in computer vision, we design a simple yet robust VO system by integrating multi-view geometry and deep learning on Depth and optical Flow, namely DF-VO. In this work, a) we propose a method to carefully sample high-quality correspondences from deep flows and recover accurate camera poses with a geometric module; b) we address the scale-drift issue by aligning geometrically triangulated depths to the scale-consistent deep depths, where the dynamic scenes are taken into account. Comprehensive ablation studies show the effectiveness of the proposed method, and extensive evaluation results show the state-of-the-art performance of our system, e.g., Ours (1.652%) v.s. ORB-SLAM (3.247%}) in terms of translation error in KITTI Odometry benchmark. Source code is publicly available at: \href{https://github.com/Huangying-Zhan/DF-VO}{DF-VO}. | ['Ian Reid', 'Ravi Garg', 'Jia-Wang Bian', 'Chamara Saroj Weerasekera', 'Huangying Zhan'] | 2021-03-01 | null | null | null | null | ['monocular-visual-odometry'] | ['robots'] | [-3.18717062e-01 -4.25721884e-01 -2.09229261e-01 -3.26161444e-01
-5.87122381e-01 -6.70224428e-01 4.57918763e-01 -7.32867837e-01
-1.83270127e-01 5.48353016e-01 1.79785043e-01 3.48362140e-02
1.82844162e-01 -6.13238931e-01 -1.02108979e+00 -6.37131691e-01
2.87654698e-01 4.28208500e-01 1.07589580e-01 -1.22643605e-01
3.97371411e-01 4.45654511e-01 -1.36868680e+00 -2.58605778e-01
8.47707629e-01 1.01082647e+00 2.34033003e-01 6.99479043e-01
2.68869549e-01 8.25386107e-01 -1.28851756e-01 -2.01169759e-01
6.59742117e-01 -1.67945415e-01 -7.68148899e-01 -1.80592295e-03
1.21217155e+00 -9.64241743e-01 -8.45839202e-01 1.18446672e+00
5.63209414e-01 -5.65643422e-02 2.66657203e-01 -1.27084291e+00
-6.34500980e-01 -1.45325974e-01 -6.93081677e-01 1.05573170e-01
4.64872420e-01 4.22815949e-01 9.45283949e-01 -1.22446299e+00
9.27430868e-01 1.19676137e+00 8.35000575e-01 3.64879072e-01
-9.83989060e-01 -7.87757874e-01 1.33719072e-01 2.15866596e-01
-1.43120360e+00 -6.05631709e-01 7.86174297e-01 -6.84374750e-01
1.03787422e+00 -2.50676036e-01 6.33501112e-01 9.98268962e-01
5.48425257e-01 5.65571964e-01 7.92693973e-01 6.25208858e-03
-2.23870367e-01 -3.48655134e-01 -3.05871844e-01 9.70892727e-01
5.56841552e-01 4.87848163e-01 -7.55304515e-01 3.10882926e-01
1.15011764e+00 1.81508601e-01 -3.79455656e-01 -9.09385979e-01
-1.44943154e+00 6.70094132e-01 8.16966772e-01 -1.00562908e-01
-1.21753655e-01 4.97908473e-01 2.98003763e-01 2.58675903e-01
4.82586235e-01 2.82441288e-01 -4.31795359e-01 -1.48484275e-01
-6.39827371e-01 3.06558162e-01 4.26745296e-01 1.33810163e+00
1.20582366e+00 2.98172772e-01 3.94906461e-01 4.91706640e-01
3.76632184e-01 8.37940872e-01 3.38778138e-01 -1.41387761e+00
7.67775953e-01 4.50606287e-01 2.14496046e-01 -1.18935502e+00
-5.38430750e-01 -4.30140287e-01 -8.61918151e-01 2.35612288e-01
4.84037369e-01 -1.96150199e-01 -7.93232560e-01 1.67859650e+00
3.92140448e-01 3.00525218e-01 7.37150107e-03 1.26976776e+00
8.10456634e-01 3.34883541e-01 -5.13080060e-01 2.04498842e-01
9.13326561e-01 -1.21176004e+00 -4.56250876e-01 -5.97880781e-01
6.21770322e-01 -8.92420888e-01 8.36584568e-01 3.82428586e-01
-8.69089782e-01 -6.70109808e-01 -1.20766866e+00 -4.03678268e-01
2.55761445e-02 1.69555634e-01 5.73711097e-01 2.42018536e-01
-1.26369667e+00 5.77132583e-01 -9.34794247e-01 -4.82015878e-01
3.55383366e-01 2.64653951e-01 -5.11466026e-01 -3.05698693e-01
-9.71094370e-01 8.04530203e-01 8.76991600e-02 2.63088793e-01
-1.13491690e+00 -6.85933232e-01 -1.16441119e+00 -3.71880770e-01
3.20169121e-01 -1.14950955e+00 1.24330401e+00 -9.70542073e-01
-1.65041506e+00 8.84464622e-01 -4.49826926e-01 -1.89822614e-01
8.15987349e-01 -7.20239401e-01 -6.45820936e-03 2.30328456e-01
3.29963058e-01 7.81933844e-01 4.82863873e-01 -1.29367554e+00
-6.13977849e-01 -4.70273644e-01 2.33373925e-01 5.19605935e-01
7.31423423e-02 -4.83372360e-01 -5.15378356e-01 -2.26599485e-01
7.05637157e-01 -1.19995427e+00 -5.88066131e-02 3.15666586e-01
-4.29273158e-01 2.22405523e-01 5.85092425e-01 -6.04285538e-01
7.81843126e-01 -1.75299644e+00 1.92445442e-01 -4.17329073e-01
3.28082263e-01 6.54094759e-03 1.15542449e-01 1.80265918e-01
1.39036551e-01 -2.09642291e-01 1.19464457e-01 -5.65532327e-01
-1.43198267e-01 1.99626014e-01 -2.51175493e-01 1.02046955e+00
-3.82226594e-02 9.89920497e-01 -1.04615915e+00 -2.31392771e-01
6.85518205e-01 5.25236547e-01 -7.00199783e-01 1.43351436e-01
3.68205942e-02 8.22295487e-01 -3.36424887e-01 8.17542255e-01
8.58141959e-01 -4.01426286e-01 -7.65675977e-02 -3.80867869e-01
-3.41855645e-01 4.33933526e-01 -1.36031377e+00 2.36217880e+00
-4.94423777e-01 8.17711651e-01 -1.03037231e-01 -5.71956098e-01
8.15457880e-01 3.10448445e-02 5.95705807e-01 -7.65593946e-01
2.07682237e-01 4.50459480e-01 -3.18672866e-01 -3.61457914e-01
6.68893635e-01 -5.39771765e-02 1.12285160e-01 -1.22519307e-01
2.89841518e-02 -3.62548620e-01 -1.77386582e-01 4.93152104e-02
6.85162961e-01 8.25748622e-01 3.16129148e-01 -1.60452679e-01
6.15401983e-01 1.46475554e-01 9.59777296e-01 3.84368539e-01
-4.64390814e-01 9.31642175e-01 1.12244420e-01 -6.63029313e-01
-1.20388782e+00 -1.12186837e+00 -5.83461784e-02 3.55227590e-01
8.45028043e-01 -1.32533923e-01 -3.19748819e-01 -1.46006480e-01
9.72352624e-02 6.15423247e-02 -3.08264405e-01 -8.01545456e-02
-7.98543990e-01 -4.00764078e-01 3.90254766e-01 5.15088201e-01
9.09402430e-01 -5.82593501e-01 -5.49078822e-01 1.34077385e-01
-3.61058742e-01 -1.55567217e+00 -5.61442733e-01 -4.40591037e-01
-1.08295870e+00 -1.07157731e+00 -6.08748794e-01 -6.60245597e-01
4.51037735e-01 8.12934875e-01 1.08256924e+00 -1.74880520e-01
-2.48248558e-02 1.73610032e-01 -6.43957928e-02 -3.91340069e-02
9.93514806e-02 1.37858525e-01 4.02590215e-01 -1.56714424e-01
2.09412366e-01 -6.31569803e-01 -1.04103899e+00 6.10681534e-01
-3.93821150e-01 2.32177973e-01 4.16283339e-01 7.98101366e-01
6.55010283e-01 -6.54145837e-01 1.02329060e-01 -4.27417874e-01
-3.24948817e-01 -2.97725737e-01 -1.01273072e+00 -2.48717189e-01
-6.70068562e-01 -1.54330641e-01 5.81063032e-01 3.03724073e-02
-8.86998117e-01 2.86196917e-01 4.36498933e-02 -8.22334826e-01
-1.15740485e-02 3.98724452e-02 -1.07845061e-01 -4.65083897e-01
5.64094722e-01 2.74258852e-01 8.92582685e-02 -2.30303645e-01
2.64513671e-01 4.25168753e-01 6.63866401e-01 -3.82080138e-01
1.04482806e+00 1.10799813e+00 1.74206540e-01 -5.82704961e-01
-1.09497011e+00 -6.66108727e-01 -9.38323617e-01 -2.71344334e-01
8.81754339e-01 -1.66944969e+00 -6.68448865e-01 7.84842551e-01
-1.23583925e+00 -2.94788897e-01 2.86877006e-01 8.28974605e-01
-8.80719423e-01 6.21617258e-01 -6.04685843e-01 -3.53829920e-01
-3.30815762e-01 -1.32068193e+00 1.30120015e+00 2.98040271e-01
1.26283139e-01 -9.69623625e-01 2.87912220e-01 5.17334521e-01
1.49161115e-01 4.31749046e-01 1.11092888e-01 1.81829929e-01
-1.25253248e+00 7.92851523e-02 -3.58315915e-01 2.60505110e-01
2.05086246e-01 -3.12790163e-02 -1.19477117e+00 -6.92907989e-01
-2.24753574e-01 -2.12584570e-01 6.82710528e-01 5.27017891e-01
6.57690704e-01 1.30071640e-01 -2.29646489e-01 1.44740701e+00
1.82142770e+00 3.97537760e-02 5.84486485e-01 7.22621679e-01
1.31907046e+00 2.66748458e-01 6.64987743e-01 3.30988139e-01
8.38976264e-01 8.89185369e-01 7.59083211e-01 4.46342640e-02
-1.93861902e-01 -5.03867805e-01 4.75253820e-01 9.37635541e-01
-1.19166017e-01 3.66076306e-02 -8.48664641e-01 6.20401680e-01
-1.82418776e+00 -7.88976252e-01 -2.56728142e-01 2.14388657e+00
3.74685466e-01 1.13094300e-02 -1.42313391e-01 -2.67493248e-01
4.42997813e-01 4.20828670e-01 -8.18086386e-01 -2.82404590e-02
-1.86968893e-01 -3.76935303e-01 8.95718575e-01 7.74907053e-01
-1.11214507e+00 1.23094559e+00 4.80661106e+00 1.10979900e-02
-1.48917520e+00 1.77111089e-01 2.13709146e-01 -2.13466510e-01
-5.30825667e-02 2.08704993e-01 -1.05048275e+00 2.51365125e-01
5.42983055e-01 -6.25476055e-03 4.15612042e-01 8.47056866e-01
2.50028670e-01 -1.27769753e-01 -1.06566882e+00 1.38286090e+00
2.44475558e-01 -1.51353621e+00 -1.04151674e-01 1.83001816e-01
1.13552928e+00 6.50946140e-01 -1.12120815e-01 1.84473284e-02
3.05874079e-01 -7.76592195e-01 8.83891940e-01 4.29453790e-01
9.82272148e-01 -4.39146340e-01 7.01747417e-01 2.48286009e-01
-1.48858035e+00 9.93666574e-02 -5.27635932e-01 -3.31351817e-01
3.55599314e-01 5.01636326e-01 -4.60167050e-01 1.00501871e+00
8.98611724e-01 1.45200861e+00 -3.48506719e-01 1.04348004e+00
-3.22650492e-01 -5.22352718e-02 -2.36049712e-01 3.16211581e-01
3.08898360e-01 -2.51728594e-01 6.22428477e-01 6.03645742e-01
4.96443123e-01 -1.45337611e-01 1.99253991e-01 8.58207703e-01
-4.36156131e-02 -2.61841178e-01 -8.82812142e-01 5.76828301e-01
4.69090670e-01 1.11816597e+00 -2.14976817e-01 -1.67124629e-01
-6.11422300e-01 9.52005208e-01 3.89453173e-01 3.08770627e-01
-8.07271659e-01 -1.24071553e-01 1.12417150e+00 4.03198265e-02
2.67024904e-01 -6.30558729e-01 -4.01194900e-01 -1.70296764e+00
3.40456158e-01 -5.63105226e-01 5.54899536e-02 -1.01138151e+00
-1.05219233e+00 5.57256997e-01 -2.91687846e-01 -1.70570743e+00
-2.54162043e-01 -6.86180234e-01 -2.23289937e-01 8.40731382e-01
-2.00491500e+00 -9.84788775e-01 -9.63120639e-01 5.44139564e-01
6.13091171e-01 2.24105064e-02 3.23819697e-01 4.97617453e-01
-4.73386496e-01 3.42422962e-01 1.87528610e-01 2.92738289e-01
1.00299609e+00 -1.05274796e+00 7.56053150e-01 1.09722626e+00
9.57186520e-02 5.40948629e-01 5.15279591e-01 -5.10673106e-01
-1.76675522e+00 -1.10954833e+00 8.04431677e-01 -7.74612725e-01
5.80260932e-01 -2.93226749e-01 -6.35327458e-01 9.71328795e-01
-4.40909714e-02 3.51828516e-01 -1.31330594e-01 -2.87783623e-01
-3.77143890e-01 -3.89949441e-01 -8.67943883e-01 5.03111541e-01
1.45780027e+00 -6.62361383e-01 -2.96719134e-01 2.04609051e-01
7.79149175e-01 -1.03172469e+00 -7.26736844e-01 3.63424569e-01
6.69268370e-01 -1.41015160e+00 1.13188958e+00 -1.04523003e-01
4.26121444e-01 -6.97756946e-01 -3.04944247e-01 -1.07119524e+00
-2.31263220e-01 -7.85277843e-01 -6.71707734e-04 7.91141391e-01
-9.40027162e-02 -8.57932210e-01 8.54410648e-01 1.22186802e-01
-2.99064755e-01 -5.41823089e-01 -1.04991198e+00 -8.23106766e-01
1.24607898e-01 -2.54303902e-01 4.72992778e-01 9.17518854e-01
-5.49300849e-01 2.42974132e-01 -6.94074810e-01 5.57059050e-01
8.48217726e-01 2.67802387e-01 1.30027866e+00 -1.08610666e+00
-1.22020185e-01 -9.22239125e-02 -7.03427732e-01 -1.70275247e+00
9.92087051e-02 -6.69176280e-01 1.34465039e-01 -1.51386428e+00
-3.75949815e-02 -3.46512973e-01 1.40360460e-01 1.26954839e-02
4.01207292e-03 3.42956543e-01 1.95980519e-01 5.58546603e-01
-4.62375581e-01 6.75789237e-01 1.53325570e+00 1.99393153e-01
-3.91389281e-02 -1.93023086e-01 -3.66682827e-01 8.18898499e-01
6.22640789e-01 -3.29312831e-01 -3.82986993e-01 -9.94079590e-01
2.96326697e-01 2.06067696e-01 7.47035742e-01 -1.13309109e+00
2.79084682e-01 -2.26925835e-01 3.86402577e-01 -6.79813445e-01
4.57127154e-01 -5.90327740e-01 1.94227412e-01 5.64037859e-01
2.53898084e-01 3.76435459e-01 -9.82123241e-03 6.20050490e-01
-2.26456270e-01 2.48469904e-01 8.46806943e-01 -2.15757847e-01
-1.16487551e+00 7.50432372e-01 2.96904325e-01 2.62454927e-01
7.16231465e-01 -4.30021703e-01 -6.28506303e-01 -4.71375853e-01
-2.05297634e-01 2.07680002e-01 9.28139150e-01 6.30423009e-01
5.33623993e-01 -1.40722144e+00 -4.73786801e-01 2.17457891e-01
3.37375194e-01 6.72082782e-01 3.20809394e-01 1.05673611e+00
-1.15853906e+00 5.39826572e-01 -3.30260575e-01 -1.15009212e+00
-8.34785461e-01 1.69176191e-01 7.23501980e-01 1.29441932e-01
-8.76611948e-01 7.96294987e-01 3.93914968e-01 -6.57351017e-01
5.42839840e-02 -5.16392648e-01 2.81536251e-01 -3.11289221e-01
2.84543037e-01 4.14198756e-01 -9.57069825e-03 -9.13953781e-01
-5.22261858e-01 1.32973886e+00 1.93151623e-01 1.60504013e-01
1.26101267e+00 -5.32166898e-01 4.26428095e-02 3.91403675e-01
1.47357953e+00 -2.27017716e-01 -1.89604020e+00 -3.38096678e-01
-3.15201819e-01 -8.46591115e-01 7.69961178e-02 -2.22137779e-01
-1.20392835e+00 1.03160846e+00 5.77878475e-01 -5.74081838e-01
8.29651415e-01 -3.25816154e-01 9.30785179e-01 3.04944336e-01
7.06811070e-01 -7.75504887e-01 1.78598776e-01 8.46642196e-01
6.38816535e-01 -1.61934698e+00 2.86642194e-01 -4.00082111e-01
-4.68451887e-01 1.17910016e+00 9.24893916e-01 -5.00424862e-01
4.47057754e-01 -1.60279408e-01 3.67126524e-01 -1.51705682e-01
-5.06922007e-01 8.40011984e-03 1.45922363e-01 4.58990961e-01
2.49393269e-01 -2.15763032e-01 2.55674124e-01 -1.90420851e-01
-3.12875479e-01 1.03119686e-01 6.89021468e-01 7.50482976e-01
-2.75104135e-01 -6.72289193e-01 -2.29980558e-01 -1.42987207e-01
-2.04133894e-02 9.95424856e-03 -1.19408816e-01 9.30863500e-01
2.79393625e-02 6.55173182e-01 8.41297656e-02 -5.22090852e-01
3.62463117e-01 -3.48681718e-01 5.68611860e-01 -3.22698474e-01
-2.41101280e-01 -9.05788988e-02 -1.46506624e-02 -1.04124320e+00
-5.74225783e-01 -7.98346698e-01 -1.15189183e+00 -6.21090114e-01
-6.44936338e-02 -5.11060238e-01 5.42678893e-01 8.86965692e-01
5.36041737e-01 2.07575470e-01 7.36383319e-01 -1.21129811e+00
-5.34453988e-01 -6.72451198e-01 -2.47024745e-01 3.15599740e-01
7.69871235e-01 -8.99928689e-01 -5.69458008e-01 -1.33463442e-01] | [8.228723526000977, -2.21358323097229] |
b85c9661-c7cc-4ec2-b27f-d674d38a737d | improving-automatic-quotation-attribution-in | 2307.03734 | null | https://arxiv.org/abs/2307.03734v1 | https://arxiv.org/pdf/2307.03734v1.pdf | Improving Automatic Quotation Attribution in Literary Novels | Current models for quotation attribution in literary novels assume varying levels of available information in their training and test data, which poses a challenge for in-the-wild inference. Here, we approach quotation attribution as a set of four interconnected sub-tasks: character identification, coreference resolution, quotation identification, and speaker attribution. We benchmark state-of-the-art models on each of these sub-tasks independently, using a large dataset of annotated coreferences and quotations in literary novels (the Project Dialogism Novel Corpus). We also train and evaluate models for the speaker attribution task in particular, showing that a simple sequential prediction model achieves accuracy scores on par with state-of-the-art models. | ['Adam Hammond', 'Graeme Hirst', 'Frank Rudzicz', 'Krishnapriya Vishnubhotla'] | 2023-07-07 | null | null | null | null | ['coreference-resolution'] | ['natural-language-processing'] | [ 2.78307438e-01 4.40857649e-01 -3.22877914e-01 -3.58556360e-01
-1.22507918e+00 -9.12524819e-01 1.18413794e+00 2.50766426e-01
-4.99095380e-01 7.62138724e-01 8.65504205e-01 -5.38538024e-03
2.71345377e-01 -7.74749964e-02 -4.13989216e-01 -1.09465398e-01
6.03853047e-01 1.15566576e+00 1.15136802e-01 -6.00814819e-01
4.96990085e-01 1.12606101e-01 -1.18331397e+00 8.02926183e-01
3.61374527e-01 6.88039660e-01 -2.07576588e-01 1.00635242e+00
-2.49548897e-01 1.45564377e+00 -1.08917582e+00 -1.32665753e+00
-3.56455743e-01 -7.79242218e-01 -1.50806499e+00 -3.12689543e-01
4.06760663e-01 2.37644017e-01 -1.54273167e-01 9.26393628e-01
9.04660404e-01 1.75830275e-01 8.80357146e-01 -1.22112131e+00
-8.71379137e-01 1.48171163e+00 -3.75599824e-02 4.75804567e-01
6.74636960e-01 -3.25519115e-01 1.54762077e+00 -8.71672571e-01
9.76901770e-01 1.35743487e+00 8.39442372e-01 8.12449813e-01
-1.66060293e+00 -5.92430472e-01 -4.09688622e-01 8.41988742e-01
-9.83932376e-01 -8.71575475e-01 9.17116702e-01 -6.59688056e-01
1.02034128e+00 6.08448803e-01 1.34296298e-01 1.69476426e+00
-2.09022492e-01 9.50212598e-01 1.06626761e+00 -4.56155241e-01
-5.44866323e-02 1.68889269e-01 3.05929720e-01 1.44954354e-01
-8.94734204e-01 -1.34924710e-01 -1.03443503e+00 -3.42968076e-01
-3.53559665e-02 -7.66424119e-01 -1.10820338e-01 2.90605694e-01
-1.24024153e+00 9.28468823e-01 1.20746329e-01 1.86369210e-01
-4.38571684e-02 9.56999660e-02 9.11888778e-01 5.47603309e-01
3.77746612e-01 9.08036530e-01 -3.71783853e-01 -7.98732400e-01
-7.33575702e-01 5.70431232e-01 1.46507883e+00 8.04308474e-01
2.77999848e-01 -3.54293972e-01 -2.59419650e-01 1.33284044e+00
-1.76061749e-01 2.18045846e-01 7.07416534e-01 -9.68647182e-01
4.49187726e-01 2.79471219e-01 -9.59989708e-03 -7.44995713e-01
-5.68014264e-01 -2.48586610e-01 -2.61946976e-01 -4.20139760e-01
5.51337600e-01 -2.77813315e-01 1.34645507e-01 1.89879096e+00
2.05403686e-01 4.44593169e-02 3.23557138e-01 7.91752934e-01
1.40775144e+00 4.04863536e-01 7.75394496e-03 -3.88111025e-01
1.64107585e+00 -1.22060633e+00 -8.20527256e-01 -2.08556995e-01
5.67362666e-01 -1.27739704e+00 1.09925747e+00 -9.99878421e-02
-1.14044487e+00 -4.58139062e-01 -1.01270163e+00 -4.66796666e-01
-4.57087979e-02 3.90903885e-03 5.16032502e-02 2.51596272e-01
-2.71501511e-01 6.15065277e-01 -3.94118071e-01 -2.20935225e-01
1.02028623e-01 -1.83987617e-01 2.91581191e-02 7.40865707e-01
-1.36568785e+00 1.49371171e+00 1.34934336e-01 -3.89080346e-01
-7.69538760e-01 -1.09379935e+00 -5.08951247e-01 8.40786397e-02
2.77874261e-01 -2.94414788e-01 2.01465249e+00 -1.05029356e+00
-1.75708210e+00 1.49735701e+00 2.24928018e-02 -5.28747261e-01
8.39626729e-01 -2.04245791e-01 -6.20032489e-01 -1.31000221e-01
2.20019862e-01 4.25725758e-01 5.26339531e-01 -1.02010608e+00
-3.21221977e-01 -1.50438368e-01 1.57054618e-01 3.42668563e-01
-1.33115813e-01 9.13051248e-01 2.27931842e-01 -4.25595552e-01
-3.73665661e-01 -8.53608787e-01 4.76292610e-01 -4.58107859e-01
-7.74648547e-01 -7.05701470e-01 6.74821973e-01 -5.82795024e-01
1.05872810e+00 -1.89924490e+00 5.36972225e-01 -6.14034593e-01
1.65974230e-01 -2.23061427e-01 1.09058179e-01 6.46827936e-01
-3.42049897e-02 -1.52051508e-01 -3.06186844e-02 -9.29738998e-01
5.76942205e-01 2.79147416e-01 -5.33691525e-01 3.76196742e-01
2.61320192e-02 1.00329852e+00 -8.97164345e-01 -7.85469234e-01
-2.79749721e-01 7.81457275e-02 -1.79161385e-01 3.79053652e-01
-4.94523019e-01 3.59854937e-01 -8.99174511e-02 2.65978456e-01
-1.37602970e-01 -3.33779156e-01 3.41587514e-01 2.79568061e-02
-8.74617696e-02 1.14089918e+00 -6.46673083e-01 1.77968132e+00
-3.93334150e-01 1.18206739e+00 -2.16222823e-01 -7.53489435e-01
9.54358101e-01 4.65306431e-01 -1.15691938e-01 -6.55547917e-01
4.81095552e-01 2.80765623e-01 2.57093042e-01 -6.25039220e-01
8.30609441e-01 -6.09906316e-01 -7.22034812e-01 9.29072440e-01
2.83876806e-01 -5.10598011e-02 1.28709584e-01 2.96306938e-01
9.00015831e-01 1.15692336e-02 5.83788276e-01 -4.20015931e-01
6.93830371e-01 3.20514083e-01 4.13418144e-01 6.92972839e-01
-8.67421851e-02 5.71598411e-01 9.17369664e-01 -1.95796043e-01
-1.14518774e+00 -1.03997970e+00 -5.63456893e-01 1.73214126e+00
-3.59411873e-02 -5.94543755e-01 -8.12079370e-01 -8.83270383e-01
-4.10614051e-02 1.25387013e+00 -1.06096423e+00 2.94152021e-01
-9.19978321e-01 -6.11780643e-01 1.22992599e+00 5.15375793e-01
-3.17410985e-03 -1.36105967e+00 -8.20626736e-01 3.04866403e-01
-6.00415766e-01 -1.25557101e+00 -5.18871963e-01 1.63242385e-01
1.28079742e-01 -1.27082574e+00 -7.52196461e-02 -5.51484048e-01
-1.97037950e-01 -5.72025359e-01 1.65179074e+00 -9.20236707e-02
4.00794894e-02 2.67624948e-02 -3.45506132e-01 -7.05059052e-01
-1.27928376e+00 4.11170155e-01 -2.64736652e-01 -2.92833418e-01
3.81633818e-01 -6.01035893e-01 1.68019943e-02 4.45634305e-01
-1.39350325e-01 1.05516814e-01 -8.10015117e-05 1.24702311e+00
1.08322531e-01 -1.23791218e+00 9.08859789e-01 -1.12910140e+00
7.18318760e-01 -4.47597265e-01 -1.83148935e-01 1.69697739e-02
-2.68501550e-01 -2.47906059e-01 8.18642914e-01 -6.38757288e-01
-1.01526415e+00 -5.83916485e-01 -4.34034824e-01 -8.17359537e-02
-6.11837246e-02 5.89391291e-01 -7.75241926e-02 4.50835317e-01
1.15841568e+00 -1.26965921e-02 2.32587874e-01 -6.44987464e-01
5.78903019e-01 1.00164509e+00 1.20422769e+00 -7.51108885e-01
5.04318535e-01 8.97098035e-02 -1.85077906e-01 -7.16480136e-01
-1.76335239e+00 -5.21343946e-01 -8.58514965e-01 -2.41809204e-01
8.43875170e-01 -6.69674277e-01 -1.17037487e+00 -5.67295142e-02
-1.49771917e+00 -2.71441042e-01 -4.59647477e-01 -3.49898003e-02
-9.33264077e-01 2.06025392e-01 -7.92792320e-01 -5.48361659e-01
-5.00514388e-01 -8.63013566e-01 6.43435657e-01 3.42460237e-02
-1.10363662e+00 -1.20544779e+00 4.61863428e-01 8.15666556e-01
2.39579201e-01 2.98846811e-01 1.09898782e+00 -1.48061752e+00
2.14089751e-01 2.58215219e-02 -1.32575676e-01 1.54020175e-01
-3.99731874e-01 -3.49498063e-01 -1.05960119e+00 2.83537745e-01
-1.83541715e-01 -8.82936835e-01 6.00487649e-01 -5.08329570e-01
5.93802154e-01 -5.03973305e-01 4.43590526e-03 1.87477529e-01
7.38902092e-01 -4.87270594e-01 2.85982996e-01 5.77758491e-01
4.95422095e-01 5.08331180e-01 4.33285356e-01 3.97927701e-01
7.48594344e-01 1.00666630e+00 1.02815375e-01 4.19743717e-01
-4.60921228e-01 -2.64376998e-01 2.82736391e-01 9.38891113e-01
3.39456171e-01 -1.51471078e-01 -9.94767368e-01 7.72123575e-01
-2.38793516e+00 -1.37761366e+00 -3.27240616e-01 1.42316961e+00
1.43169928e+00 5.43097220e-02 2.98765004e-01 -7.15114102e-02
6.70498312e-01 3.51355493e-01 -3.21189910e-01 -8.55648756e-01
-4.64692175e-01 1.07588224e-01 -2.70456582e-01 4.74401444e-01
-1.03409874e+00 9.14019704e-01 6.98759890e+00 7.30537474e-01
-9.07854855e-01 6.36163354e-01 2.27426767e-01 -1.05447322e-02
-2.19778761e-01 1.36045158e-01 -9.00272667e-01 4.15641606e-01
1.29209793e+00 -5.64381361e-01 1.92312703e-01 8.79224539e-01
-4.05212671e-01 3.32144111e-01 -1.60301614e+00 7.18491197e-01
8.03795815e-01 -1.89195120e+00 -1.04939952e-01 -3.40435714e-01
5.69702566e-01 3.43510985e-01 -3.18068713e-01 4.63235229e-01
4.32286769e-01 -1.08409369e+00 1.26080501e+00 2.25451320e-01
5.86421669e-01 -5.44446468e-01 8.63083005e-01 5.91870070e-01
-2.09991500e-01 -1.30995885e-01 -2.49968231e-01 -4.61999714e-01
5.15974998e-01 -3.67153019e-01 -1.09491789e+00 1.41936123e-01
3.54155034e-01 6.77816272e-01 -6.01420939e-01 4.06363606e-01
-6.47505879e-01 7.16216862e-01 -4.91636023e-02 -5.87252617e-01
1.73988134e-01 2.48533845e-01 9.11938965e-01 1.55447817e+00
-3.90811384e-01 1.73494115e-01 1.61186293e-01 1.22418547e+00
-6.85617030e-01 3.70638311e-01 4.00185585e-01 1.08183406e-01
8.24849069e-01 1.36599624e+00 -7.17379451e-02 -6.05967402e-01
-2.29680538e-01 7.12393939e-01 8.59106481e-01 -2.54811347e-01
-8.45332921e-01 6.49936050e-02 6.47155225e-01 -9.83330887e-03
1.43179342e-01 3.01524103e-01 -5.15196502e-01 -9.21640515e-01
-4.33241665e-01 -9.69860733e-01 7.95160413e-01 -7.79091716e-01
-1.72075355e+00 4.13784713e-01 -6.80786967e-02 -6.31493866e-01
-6.79794729e-01 -2.67056137e-01 -8.57851803e-01 7.98111141e-01
-1.32921529e+00 -1.47774959e+00 -1.23956412e-01 2.99192488e-01
7.09040761e-01 -2.98138857e-01 1.36927330e+00 -5.55482693e-02
-4.12327766e-01 8.85991931e-01 -4.70821783e-02 7.92491019e-01
1.21234643e+00 -1.51371849e+00 5.13150573e-01 5.35322547e-01
5.23039162e-01 3.41055274e-01 1.19203413e+00 -1.54760316e-01
-9.51993823e-01 -5.37521601e-01 1.24439621e+00 -1.23362064e+00
1.37775779e+00 -4.84875321e-01 -8.62875283e-01 9.64236081e-01
6.60314560e-01 -5.37660778e-01 1.29355204e+00 7.92053401e-01
-9.01073456e-01 5.24234533e-01 -7.64004290e-01 5.44417024e-01
9.66569126e-01 -8.53651345e-01 -1.46560967e+00 6.90872788e-01
6.04777932e-01 -8.76737475e-01 -1.11059797e+00 -7.66502768e-02
6.49978459e-01 -7.28425384e-01 7.53125966e-01 -1.18309999e+00
1.14817917e+00 8.70379508e-02 -4.57856566e-01 -1.22729957e+00
6.28393143e-02 -8.75351250e-01 -1.62942261e-01 1.58486581e+00
8.28415453e-01 -1.54114924e-02 4.74118590e-01 3.19485724e-01
-4.77215439e-01 -2.34298021e-01 -1.23057759e+00 -4.92531270e-01
4.52777743e-01 -3.19993973e-01 4.45811987e-01 1.32776892e+00
8.57656002e-01 1.62413526e+00 -1.84618339e-01 -3.75053495e-01
6.69442296e-01 5.51340103e-01 1.02380848e+00 -1.37641859e+00
-5.82092762e-01 -7.77726710e-01 -4.35882866e-01 -8.73120189e-01
1.11660075e+00 -1.26254237e+00 4.00099158e-03 -1.10352480e+00
5.42434335e-01 -1.52455226e-01 1.17812432e-01 3.35319579e-01
-1.87670320e-01 5.40508747e-01 3.50752443e-01 4.64055300e-01
-7.61575580e-01 4.99661863e-01 7.89834738e-01 -1.01157807e-01
7.16199577e-02 -3.40124846e-01 -7.23941207e-01 8.52562308e-01
4.41781461e-01 -6.66529715e-01 2.79126465e-01 -3.41059268e-01
2.76800781e-01 4.46763158e-01 4.42034990e-01 -4.21815336e-01
5.57252705e-01 -7.08373189e-02 8.86393934e-02 -5.74421108e-01
7.18979120e-01 1.01576395e-01 -1.48657039e-01 1.20470293e-01
-1.03503728e+00 2.45475724e-01 7.16267303e-02 2.73844421e-01
-9.78469178e-02 -4.58703876e-01 9.21015203e-01 -6.57389089e-02
-5.11779249e-01 -5.27084470e-01 -2.49288216e-01 8.72617364e-01
6.28417432e-01 8.38512182e-02 -8.10612857e-01 -2.98803717e-01
-9.60164309e-01 9.54620391e-02 3.73654157e-01 7.72772014e-01
7.98992664e-02 -1.04655826e+00 -1.40060222e+00 -5.22986233e-01
4.69670445e-01 -5.68957448e-01 5.73970936e-02 9.24615562e-01
-6.89369515e-02 -4.05403525e-02 -1.58082485e-01 -7.14546144e-01
-1.64818537e+00 3.53647113e-01 3.12532902e-01 -5.39661907e-02
-6.08746171e-01 1.04974139e+00 -4.04272199e-01 -8.11837852e-01
-1.26402467e-01 6.33089900e-01 -3.40160280e-01 5.04256308e-01
6.54388964e-01 3.32705408e-01 -1.98984165e-02 -1.11708140e+00
-3.65047932e-01 -6.28120750e-02 -2.70649552e-01 -3.11334699e-01
1.12435055e+00 -2.74403095e-01 -3.60306919e-01 1.08205831e+00
1.02447546e+00 1.60854012e-01 -7.15992689e-01 -5.31486750e-01
3.67058814e-01 -1.44014940e-01 -3.47967118e-01 -1.24356365e+00
1.16089970e-01 7.08083391e-01 -3.78591716e-01 2.14939654e-01
1.81920454e-01 5.50393581e-01 6.75622702e-01 4.97541994e-01
-3.06932747e-01 -1.31329775e+00 2.41008937e-01 8.14917982e-01
1.09292448e+00 -8.85324597e-01 -8.30851793e-02 -2.33768597e-01
-8.95605624e-01 1.08743155e+00 4.16564524e-01 -5.41987456e-02
3.57037067e-01 5.49961567e-01 5.33663750e-01 -2.92915463e-01
-1.28650618e+00 9.45391655e-02 2.48688251e-01 2.71622837e-01
8.24262142e-01 2.36536905e-01 -4.31628019e-01 1.09345067e+00
-1.15519786e+00 -8.11461687e-01 1.03448606e+00 1.81831077e-01
-1.28025472e-01 -9.29915071e-01 -1.06990792e-01 6.78184330e-02
-7.25964725e-01 -2.15185538e-01 -1.10317791e+00 7.54598618e-01
-3.29457790e-01 9.23878312e-01 2.09249541e-01 -1.27347112e-01
5.14489293e-01 4.55340296e-01 4.22339767e-01 -5.25557876e-01
-1.36522090e+00 -2.44266719e-01 1.09449446e+00 -2.79887527e-01
-5.76050043e-01 -1.14530778e+00 -1.22304976e+00 -7.20382810e-01
-4.12993580e-01 5.88304996e-01 5.76091349e-01 1.45095444e+00
4.92746606e-02 5.92805207e-01 2.58648843e-01 -5.57899296e-01
-1.26078665e+00 -1.34632933e+00 -1.40446633e-01 9.19304848e-01
1.80261880e-02 -5.63369989e-01 -3.78463089e-01 6.67665899e-02] | [11.092257499694824, 8.924667358398438] |
1ca659b6-d37a-4a5c-b9c5-b02a43213370 | a-large-scale-study-of-a-sleep-tracking-and | 2211.02592 | null | https://arxiv.org/abs/2211.02592v1 | https://arxiv.org/pdf/2211.02592v1.pdf | A Large-Scale Study of a Sleep Tracking and Improving Device with Closed-loop and Personalized Real-time Acoustic Stimulation | Various intervention therapies ranging from pharmaceutical to hi-tech tailored solutions have been available to treat difficulty in falling asleep commonly caused by insomnia in modern life. However, current techniques largely remain ill-suited, ineffective, and unreliable due to their lack of precise real-time sleep tracking, in-time feedback on the therapies, an ability to keep people asleep during the night, and a large-scale effectiveness evaluation. Here, we introduce a novel sleep aid system, called Earable, that can continuously sense multiple head-based physiological signals and simultaneously enable closed-loop auditory stimulation to entrain brain activities in time for effective sleep promotion. We develop the system in a lightweight, comfortable, and user-friendly headband with a comprehensive set of algorithms and dedicated own-designed audio stimuli. We conducted multiple protocols from 883 sleep studies on 377 subjects (241 women, 119 men) wearing either a gold-standard device (PSG), Earable, or both concurrently. We demonstrate that our system achieves (1) a strong correlation (0.89 +/- 0.03) between the physiological signals acquired by Earable and those from the gold-standard PSG, (2) an 87.8 +/- 5.3% agreement on sleep scoring using our automatic real-time sleep staging algorithm with the consensus scored by three sleep technicians, and (3) a successful non-pharmacological stimulation alternative to effectively shorten the duration of sleep falling by 24.1 +/- 0.1 minutes. These results show that the efficacy of Earable exceeds existing techniques in intentions to promote fast falling asleep, track sleep state accurately, and achieve high social acceptance for real-time closed-loop personalized neuromodulation-based home sleep care. | ['Tam Vu', 'Sangtae Ha', 'Sy Duong-Quy', 'Hoang Huu Nguyen', 'Linh Nguyen', 'Nhat Pham', 'Hoang Truong', 'Nam Bui', 'Ban Xuan Dong', 'Galen Pogoncheff', 'Anh Nguyen'] | 2022-11-04 | null | null | null | null | ['sleep-staging'] | ['medical'] | [-1.71088874e-01 -1.58867374e-01 1.74933583e-01 -3.16971719e-01
-8.08154702e-01 -3.56017947e-01 -4.67115134e-01 -6.72637671e-02
-5.98470688e-01 8.30585539e-01 4.88463759e-01 -2.11606935e-01
-1.04250811e-01 -2.29864806e-01 -1.62839025e-01 -5.15137076e-01
-1.33231580e-01 2.96381593e-01 1.14627808e-01 -1.04923718e-01
-2.27634192e-01 1.21207058e-01 -1.79248524e+00 -2.04586312e-01
9.82107222e-01 1.15527046e+00 4.14658993e-01 3.88934642e-01
4.39295501e-01 -1.11774273e-01 -8.21601689e-01 3.24603647e-01
-7.25901872e-02 -5.94416142e-01 -1.70594946e-01 -4.60418880e-01
6.00511655e-02 -2.70550996e-01 -1.51362672e-01 4.62828130e-01
1.32999551e+00 -2.50961189e-03 -8.69920254e-02 -1.20521593e+00
-1.21320114e-01 2.48524487e-01 -9.56218541e-02 4.11899865e-01
8.44371915e-01 3.89926940e-01 3.79091531e-01 -4.01108354e-01
9.70835239e-03 5.06849468e-01 8.74518871e-01 9.50897455e-01
-1.48189235e+00 -1.15058708e+00 -6.28071845e-01 3.68516266e-01
-1.58745360e+00 -1.14868116e+00 5.82738638e-01 -1.17336817e-01
1.13025355e+00 6.07224047e-01 1.46585214e+00 8.59165251e-01
5.70881367e-01 -1.06229924e-01 1.23050928e+00 -9.37033072e-02
6.27635777e-01 2.07488775e-01 1.10778593e-01 3.53724211e-01
2.57457882e-01 -2.69337799e-02 -1.07498312e+00 -2.98755348e-01
6.44921437e-02 1.10930286e-01 -7.74539888e-01 4.01452631e-01
-7.85634816e-01 3.73314857e-01 1.58833921e-01 5.15452683e-01
-5.43548405e-01 -6.18263595e-02 1.76197916e-01 2.37908270e-02
3.29557329e-01 3.89275402e-01 -2.36137375e-01 -7.57605135e-01
-1.20643532e+00 -1.86276332e-01 8.29628825e-01 4.15447921e-01
9.55584198e-02 -6.87799556e-03 -4.81030166e-01 7.75073469e-01
3.39779973e-01 1.14023662e+00 9.59114134e-01 -9.57866073e-01
-2.20119506e-01 2.81108201e-01 3.78710240e-01 -6.99636221e-01
-1.17272151e+00 -4.38519478e-01 -4.11855221e-01 -1.08013317e-01
-2.43435502e-02 4.85821860e-03 -3.19514036e-01 1.83784854e+00
-3.39650642e-03 4.35062684e-02 -3.89867872e-01 7.64687955e-01
9.18138325e-01 2.85782963e-01 1.37745351e-01 -9.08172667e-01
1.75582111e+00 -3.04322004e-01 -1.01496649e+00 -5.33118904e-01
1.72145635e-01 -4.41638082e-01 1.65468526e+00 5.39509237e-01
-1.20847356e+00 -4.16864812e-01 -1.10472715e+00 1.95777953e-01
1.50177404e-01 3.04845329e-02 -4.44076434e-02 1.22561204e+00
-1.48335779e+00 3.63411099e-01 -1.17481208e+00 -6.89896584e-01
2.06507131e-01 9.26906049e-01 -2.38555923e-01 3.48080933e-01
-9.69920933e-01 7.78051734e-01 -2.75707006e-01 2.83785108e-02
-4.18103635e-01 -7.96539545e-01 -5.16875386e-01 2.02935651e-01
-1.25880405e-01 -1.03713822e+00 1.19873726e+00 -2.79692411e-01
-1.81570661e+00 9.74225879e-01 -5.20939171e-01 -3.46257538e-01
-1.69830903e-01 -4.35951561e-01 -7.80297399e-01 2.91025579e-01
4.14806813e-01 2.24596635e-01 4.99710888e-01 -4.70822960e-01
-1.40357003e-01 -7.07513690e-01 -6.15761817e-01 1.30684689e-01
-5.99103928e-01 2.95084238e-01 1.31058097e-01 -8.91739223e-03
1.13404304e-01 -9.29209411e-01 1.53007403e-01 -2.96349134e-02
-1.16831131e-01 2.15025377e-02 5.74012995e-01 -7.14099884e-01
1.56183207e+00 -2.38971090e+00 -3.60616356e-01 1.40926614e-03
4.79622036e-01 2.41844341e-01 3.65482241e-01 1.78771839e-01
1.73133835e-01 -1.01936787e-01 -9.48699638e-02 -4.89111692e-01
1.08819194e-02 1.03153311e-01 2.13141218e-01 7.19043732e-01
-7.50511527e-01 7.68631458e-01 -8.12096000e-01 -1.86378047e-01
4.32235152e-01 5.32635272e-01 -4.06352401e-01 4.67880934e-01
7.93446243e-01 4.56663966e-01 7.00415745e-02 6.80652499e-01
2.12651014e-01 -8.57543871e-02 -1.25225127e-01 -1.68229148e-01
-3.36744368e-01 6.53798401e-01 -6.75171196e-01 1.71479619e+00
-4.36536103e-01 4.82115597e-01 4.09616351e-01 -2.79660493e-01
9.20976937e-01 5.17345428e-01 6.39257073e-01 -1.06485677e+00
3.25156301e-01 3.61540467e-01 -2.55338531e-02 -9.17686641e-01
6.95877001e-02 -8.88064742e-01 5.22194877e-02 5.15881479e-01
-1.92286596e-01 -2.88539141e-01 -2.66095310e-01 -1.19635351e-01
1.47849989e+00 -6.00720346e-01 4.16197121e-01 -6.16388977e-01
1.89228192e-01 -6.95595562e-01 4.89844412e-01 6.36038661e-01
-6.39082015e-01 5.63337803e-01 -2.71810412e-01 1.27840623e-01
-2.18546808e-01 -1.21881187e+00 -1.25869006e-01 9.34908032e-01
-8.75586644e-03 -8.28711331e-01 -6.98552072e-01 1.24575093e-01
-1.14244372e-01 9.20120716e-01 -3.07119131e-01 -7.15784371e-01
7.84831494e-02 -4.81884867e-01 6.67125225e-01 5.03120422e-01
4.48334455e-01 -1.11119223e+00 -9.84815657e-01 5.22982717e-01
-7.00143576e-01 -8.01795065e-01 -6.19116664e-01 2.79337466e-01
-5.53740442e-01 -6.19554937e-01 -3.87882680e-01 -2.51310587e-01
1.17678344e-01 4.05891240e-01 8.67182732e-01 -1.00162603e-01
-4.00331050e-01 7.47856438e-01 -3.24734710e-02 -6.12519681e-01
1.49902835e-01 -2.31852040e-01 8.56146991e-01 -1.99359432e-01
5.56585908e-01 -1.21113801e+00 -1.15487111e+00 4.88568872e-01
-1.51013404e-01 -4.09265518e-01 2.75005400e-01 2.94881165e-01
3.28326017e-01 -2.96306640e-01 7.83304691e-01 4.17010747e-02
9.88841534e-01 -2.58906186e-01 1.59506816e-02 -2.44994864e-01
-9.20447767e-01 -4.70561236e-01 4.10107881e-01 -2.93818414e-01
-5.99876404e-01 -1.60084128e-01 -4.53228831e-01 -6.96129054e-02
-2.10953146e-01 9.72714275e-02 -2.15029672e-01 1.31842494e-01
9.92440164e-01 2.47223973e-01 3.22025836e-01 -2.01439008e-01
-1.10655859e-01 1.27627981e+00 1.01063204e+00 -2.11702749e-01
3.85974288e-01 4.94174689e-01 -4.00223345e-01 -1.09937084e+00
-6.79070473e-01 -7.38125145e-01 -2.07763314e-01 -3.51313651e-01
1.00909650e+00 -9.63346064e-01 -1.09984708e+00 3.61641586e-01
-4.32362318e-01 -5.55530131e-01 -2.38642335e-01 7.93045282e-01
-5.77922344e-01 3.34601775e-02 -2.88114399e-01 -9.81823146e-01
-1.30773520e+00 -7.35617757e-01 8.98920476e-01 5.38203180e-01
-1.01183200e+00 -3.63815427e-01 1.82984427e-01 5.20745277e-01
8.55340838e-01 -1.27238696e-02 2.63142228e-01 -1.40316516e-01
1.67894244e-01 -2.84079343e-01 2.91595072e-01 1.22165032e-01
5.89650214e-01 -6.18367851e-01 -1.31471193e+00 -3.20424825e-01
6.32396221e-01 -1.76678002e-01 6.64157495e-02 8.01812947e-01
8.14379871e-01 -6.81925938e-02 -4.56597328e-01 5.16343534e-01
7.82008946e-01 4.23771024e-01 8.74027610e-01 2.14527190e-01
-3.34831141e-02 2.21522316e-01 2.84062773e-01 5.23340762e-01
3.46316487e-01 7.22565114e-01 4.73832190e-02 5.08167408e-02
6.47561029e-02 -1.49032380e-02 7.21276522e-01 8.29995394e-01
-2.04505417e-02 1.71380024e-02 -5.16640007e-01 2.82259166e-01
-1.26868939e+00 -9.41334188e-01 -2.16765031e-01 2.66241312e+00
7.28008509e-01 2.35720083e-01 6.21678472e-01 2.56295472e-01
4.15045232e-01 -4.79129046e-01 -5.98154366e-01 -4.00504917e-01
2.40016699e-01 7.06482887e-01 2.94306666e-01 1.59330770e-01
-4.05482024e-01 2.34360978e-01 6.69058418e+00 1.70876026e-01
-1.29179835e+00 4.36438560e-01 2.28429791e-02 -8.89593959e-01
1.67561304e-02 -4.59562749e-01 -5.04147172e-01 9.82712090e-01
1.77935004e+00 -4.89241481e-01 5.75601637e-01 5.83636522e-01
1.16181946e+00 -2.65002221e-01 -8.41357172e-01 1.43787682e+00
1.97421625e-01 -7.67904699e-01 -1.06768012e+00 8.31232369e-02
-2.65082181e-01 2.98474655e-02 -2.31471300e-01 9.39730182e-02
-7.43224740e-01 -8.22420835e-01 7.60751605e-01 7.31639385e-01
1.21845222e+00 -4.06020910e-01 4.24771875e-01 2.29232430e-01
-1.15788805e+00 -3.29822302e-01 8.22384879e-02 -4.82697397e-01
4.69369680e-01 6.40711367e-01 -6.07820690e-01 -1.99748315e-02
1.17165172e+00 4.55303967e-01 -3.91831249e-01 1.32697749e+00
-3.02380294e-01 1.13843381e+00 -8.18409920e-01 -2.62840509e-01
-5.89117706e-01 3.11847702e-02 5.18053234e-01 5.92225075e-01
5.01360357e-01 5.18476963e-01 -2.89394438e-01 9.54515457e-01
2.99644262e-01 -1.15051270e-01 -4.34981823e-01 2.64647126e-01
5.82345963e-01 1.23016465e+00 -7.22269535e-01 5.46767302e-02
-3.39676231e-01 7.06621885e-01 -4.68239844e-01 3.77073251e-02
-7.93965459e-01 -5.91662645e-01 6.74239278e-01 5.74563682e-01
-4.16402787e-01 -1.17749490e-01 -4.83854502e-01 -7.40036249e-01
2.06288308e-01 -5.46145558e-01 1.94710657e-01 -1.26087999e+00
-9.25229728e-01 5.64306736e-01 -2.82926559e-01 -1.26092625e+00
4.95374277e-02 2.74336990e-02 -7.98028111e-01 7.50041723e-01
-6.76070094e-01 -5.46865404e-01 -6.71015322e-01 6.75126970e-01
2.09193811e-01 4.08220232e-01 1.07698154e+00 7.11006165e-01
-5.58702052e-01 6.17372692e-01 -1.92693502e-01 -8.77940476e-01
1.01278436e+00 -8.80158901e-01 -3.72913271e-01 4.01189178e-01
-4.39186573e-01 9.34982657e-01 9.16787565e-01 -3.80650192e-01
-1.29471743e+00 -5.05278945e-01 8.56584847e-01 -4.54633683e-01
2.61034667e-01 -4.94239628e-01 -5.97115755e-01 3.21645021e-01
-2.69283839e-02 -4.78832781e-01 1.14555323e+00 1.55348673e-01
4.34064776e-01 -7.40758121e-01 -1.31643379e+00 2.60466456e-01
1.03681755e+00 -6.87572598e-01 -7.54662275e-01 1.82136819e-01
4.37081218e-01 -2.29298308e-01 -6.46122515e-01 1.25855327e-01
1.10850668e+00 -1.01226676e+00 7.40871489e-01 1.30116865e-01
-3.02008927e-01 -1.54235676e-01 1.34194300e-01 -1.03899992e+00
-3.59817713e-01 -1.15393198e+00 1.47802085e-01 1.22092199e+00
1.17887363e-01 -1.05894649e+00 3.81974965e-01 1.24256325e+00
-7.03568339e-01 -5.59470594e-01 -1.19085574e+00 -6.70164585e-01
-6.79212749e-01 -5.27300894e-01 3.00998151e-01 2.06930891e-01
7.06804693e-01 4.82098311e-01 -3.25989097e-01 8.02459717e-02
1.53253615e-01 -2.23257974e-01 4.12260264e-01 -1.32319534e+00
-1.02200739e-01 -1.06797695e-01 -6.11719966e-01 -6.49218321e-01
-5.89511991e-02 -5.83489656e-01 2.93419033e-01 -1.69122708e+00
1.73000067e-01 -8.52953717e-02 -4.96859759e-01 7.68005013e-01
2.51587834e-02 3.14237267e-01 -1.98917165e-01 -8.88577253e-02
-4.04920220e-01 7.29248583e-01 5.20317614e-01 1.74280867e-01
-8.35404277e-01 3.72384340e-01 -1.10854375e+00 5.16686082e-01
7.69354820e-01 -6.50546312e-01 -6.38793409e-01 1.49963215e-01
-1.26801869e-02 1.04015157e-01 3.43866825e-01 -1.43551409e+00
3.09555352e-01 3.60653281e-01 1.05950408e-01 -2.41734564e-01
6.60404027e-01 -7.13480830e-01 4.21482027e-01 7.08819985e-01
3.33823711e-01 -8.31909254e-02 5.20292282e-01 1.74753726e-01
5.72155237e-01 2.17285156e-01 7.20649242e-01 1.21173270e-01
-7.29522705e-02 -1.67695358e-01 -8.22091937e-01 5.59124015e-02
5.93853533e-01 -5.37941754e-01 -3.39775681e-01 -6.94720566e-01
-1.05470634e+00 1.39395982e-01 3.97904515e-01 2.63027787e-01
5.24033070e-01 -9.97088432e-01 -9.97955874e-02 4.43657875e-01
-2.05815043e-02 -4.88588214e-01 4.24045324e-01 1.46700609e+00
-6.71539232e-02 4.51825082e-01 -3.83656174e-01 -7.34547138e-01
-1.45730722e+00 8.90417546e-02 6.11360967e-01 4.12050635e-01
-9.35042441e-01 6.58918083e-01 -1.94704160e-01 2.29853958e-01
1.69733658e-01 -3.28845888e-01 1.41364664e-01 -1.95259288e-01
7.77767062e-01 4.81046408e-01 4.88873124e-01 -2.56058604e-01
-7.38438964e-01 5.08250713e-01 5.86766064e-01 -2.56656200e-01
1.13351297e+00 -6.59000278e-01 -1.50826275e-01 7.68293083e-01
6.62096024e-01 2.41807058e-01 -5.79096973e-01 5.31236827e-01
-5.15463531e-01 -2.88476020e-01 1.49657205e-01 -9.81337965e-01
-7.07593262e-01 4.78911847e-01 1.43201053e+00 2.50197709e-01
1.64219618e+00 1.95907757e-01 1.22078454e+00 1.50759056e-01
6.43323004e-01 -8.28567207e-01 -1.76649615e-01 -3.68678331e-01
7.26636291e-01 -3.86816412e-01 -1.43879801e-01 2.50100885e-02
-2.33118311e-01 6.12908006e-01 3.57791007e-01 3.88651878e-01
8.56022537e-01 3.02412897e-01 1.94579050e-01 -3.94682735e-01
-5.13476551e-01 -1.45336226e-01 3.80407423e-02 7.65144765e-01
3.50111902e-01 2.94240832e-01 -7.48952806e-01 1.28917193e+00
-8.38741779e-01 5.44277608e-01 3.83841246e-01 8.56589437e-01
-6.45235598e-01 -6.64938629e-01 -4.38959897e-01 9.22758043e-01
-2.92683452e-01 1.40483797e-01 -8.90765339e-02 3.53693873e-01
8.14907029e-02 1.66968763e+00 -4.05686311e-02 -5.36417842e-01
7.40760624e-01 4.19266552e-01 3.87737870e-01 -6.71148062e-01
-6.05232239e-01 1.78066611e-01 1.30524188e-01 -8.91164422e-01
-3.25449497e-01 -6.87391758e-01 -1.37848926e+00 -1.86623558e-01
-1.95488334e-01 3.92083347e-01 5.74556231e-01 1.04604888e+00
9.33143437e-01 4.78167415e-01 4.09783810e-01 -8.26372266e-01
-6.48795515e-02 -1.05144548e+00 -1.18490350e+00 5.77643998e-02
2.77038872e-01 -8.46416414e-01 -5.78062236e-01 -1.70168757e-01] | [13.541888236999512, 3.3997299671173096] |
7c71d14c-dd4c-458d-bafb-42eb4df6a7e8 | deepscaffold-a-comprehensive-tool-for | 1908.07209 | null | https://arxiv.org/abs/1908.07209v4 | https://arxiv.org/pdf/1908.07209v4.pdf | DeepScaffold: a comprehensive tool for scaffold-based de novo drug discovery using deep learning | The ultimate goal of drug design is to find novel compounds with desirable pharmacological properties. Designing molecules retaining particular scaffolds as the core structures of the molecules is one of the efficient ways to obtain potential drug candidates with desirable properties. We proposed a scaffold-based molecular generative model for scaffold-based drug discovery, which performs molecule generation based on a wide spectrum of scaffold definitions, including BM-scaffolds, cyclic skeletons, as well as scaffolds with specifications on side-chain properties. The model can generalize the learned chemical rules of adding atoms and bonds to a given scaffold. Furthermore, the generated compounds were evaluated by molecular docking in DRD2 targets and the results demonstrated that this approach can be effectively applied to solve several drug design problems, including the generation of compounds containing a given scaffold and de novo drug design of potential drug candidates with specific docking scores. Finally, a command line interface is created. | ['Zhenming Liu', 'Jianxing Hu', 'Yibo Li', 'Jielong Zhou', 'Yanxing Wang', 'Liangren Zhang'] | 2019-08-20 | null | null | null | null | ['molecular-docking'] | ['medical'] | [ 4.15899694e-01 4.29747067e-02 -6.49303913e-01 -1.22427158e-01
-3.21204633e-01 -7.91643977e-01 2.87351936e-01 3.94130200e-01
1.14078417e-01 1.50714314e+00 4.82266657e-02 -5.37017226e-01
-5.30921035e-02 -9.53214824e-01 -6.70716286e-01 -7.93852568e-01
-1.40020192e-01 3.87313128e-01 2.71395177e-01 -1.58494830e-01
5.15936375e-01 9.81226206e-01 -1.15144336e+00 -9.13217515e-02
1.32887471e+00 5.23317337e-01 4.41796839e-01 -8.40505287e-02
-1.84143275e-01 1.86045870e-01 -5.67664683e-01 -1.57500148e-01
-9.62116346e-02 -6.20462239e-01 -5.16835272e-01 -3.18569392e-02
-3.87283191e-02 7.21948296e-02 1.21718526e-01 7.37585664e-01
6.19458258e-01 3.02080721e-01 9.79098916e-01 -4.53830540e-01
-7.72933960e-01 2.25247115e-01 4.58405055e-02 -2.82290310e-01
6.92243516e-01 -1.82425175e-02 7.95128703e-01 -1.28231096e+00
5.83428741e-01 8.73306274e-01 1.35404468e-01 7.38247275e-01
-1.34185970e+00 -8.25632751e-01 -1.41683789e-02 1.25032917e-01
-1.45242512e+00 -1.63017765e-01 5.18583059e-01 -5.35669506e-01
1.01674688e+00 3.99953604e-01 6.66583896e-01 8.06220770e-01
5.33233881e-01 2.01494232e-01 6.51430547e-01 2.15995554e-02
7.07483947e-01 -9.50495079e-02 -2.08153367e-01 7.16373503e-01
5.46523631e-01 -2.02888418e-02 -7.40554035e-01 -6.88429773e-01
6.62400484e-01 1.78866133e-01 -2.53068328e-01 -3.91365409e-01
-9.39063430e-01 1.03222191e+00 5.99815965e-01 1.83662906e-01
-5.60113311e-01 -2.66169518e-01 -1.96033530e-02 -4.83575702e-01
-4.20179181e-02 1.15237975e+00 -6.29492700e-01 5.58895469e-01
-5.46437442e-01 2.06038341e-01 6.21235430e-01 8.26510549e-01
7.62952805e-01 1.59059584e-01 -1.87309697e-01 3.66128743e-01
5.00654399e-01 1.49821341e-01 1.92827672e-01 -4.27437037e-01
-2.08145380e-01 5.82475483e-01 1.85971081e-01 -6.19890988e-01
-6.02887034e-01 -6.52376890e-01 -6.49192810e-01 -7.38053247e-02
-1.12559825e-01 -1.68828174e-01 -1.11302042e+00 1.71488523e+00
6.07670903e-01 3.34379286e-01 2.04826221e-01 5.22765875e-01
1.07742143e+00 7.75205314e-01 5.57783723e-01 -8.43053758e-01
1.09019172e+00 -5.73004484e-01 -5.13996184e-01 4.88355756e-01
4.94784832e-01 -7.06471801e-01 5.90068579e-01 4.72073704e-01
-8.42531085e-01 -2.77684420e-01 -1.09505892e+00 4.33919668e-01
-2.99387395e-01 4.24025580e-03 1.20650876e+00 6.64145708e-01
-5.08633077e-01 1.06108284e+00 -5.64085007e-01 1.12972111e-01
4.73369807e-01 7.88843751e-01 -3.62006634e-01 1.41142294e-01
-1.13800800e+00 8.12142551e-01 8.66164148e-01 -4.51883720e-03
-1.33155775e+00 -9.23874915e-01 -6.34243250e-01 -2.88801454e-02
3.52413148e-01 -1.17587721e+00 6.30057573e-01 -4.45691139e-01
-1.86202860e+00 2.47594133e-01 -2.68117726e-01 -1.73157752e-01
-1.75471589e-01 1.82824895e-01 -3.17533135e-01 -7.67863914e-02
1.61987841e-01 4.82100576e-01 4.47615117e-01 -8.66711795e-01
-2.76537597e-01 -2.66639769e-01 -2.03089088e-01 3.02871585e-01
2.26370692e-01 -4.67306077e-02 1.49504066e-01 -5.67000926e-01
1.51086628e-01 -7.91363418e-01 -7.63295472e-01 -2.90729553e-01
-8.65670025e-01 -3.02730829e-01 1.83512226e-01 -2.26466000e-01
1.17291605e+00 -1.57795048e+00 6.09279811e-01 7.36887157e-01
5.54272421e-02 3.99848670e-01 -1.14958256e-01 8.54369581e-01
-4.87457961e-01 4.43704426e-01 9.28306356e-02 6.60027981e-01
-6.19285226e-01 -2.70561010e-01 -2.16682091e-01 2.22707942e-01
1.73410565e-01 6.92100167e-01 -9.25673246e-01 -3.20403911e-02
-1.10760830e-01 3.83754760e-01 -7.56527185e-01 2.73527116e-01
-8.16116750e-01 1.05712199e+00 -1.24500775e+00 8.62964392e-01
5.06348014e-01 -2.37738088e-01 4.07856345e-01 -1.49715751e-01
-9.90589522e-03 4.09433246e-01 -8.72345150e-01 1.50915599e+00
1.69380128e-01 -4.78167832e-01 -8.65219831e-01 -4.00886357e-01
1.13151085e+00 4.11372155e-01 5.47394633e-01 -3.77804071e-01
-5.18946201e-02 3.93992424e-01 8.78407583e-02 -9.22977775e-02
-6.24053404e-02 -5.39762378e-01 3.78630251e-01 -1.64211527e-01
-7.44895861e-02 8.07551071e-02 3.37453812e-01 -1.13096930e-01
8.62874210e-01 2.30828941e-01 4.39828426e-01 -1.98783100e-01
7.92492270e-01 1.02886647e-01 7.20072210e-01 1.96004167e-01
7.32688665e-01 2.66469210e-01 3.09501380e-01 -5.19424021e-01
-9.43518281e-01 -9.63304758e-01 -4.48239088e-01 7.71158636e-01
-4.55199517e-02 -6.45451427e-01 -5.08146286e-01 -3.84705782e-01
-2.57088572e-01 6.66810632e-01 -5.76125860e-01 -3.45497012e-01
-2.32748479e-01 -8.10340285e-01 2.70200491e-01 -8.15008134e-02
3.85264829e-02 -9.00306702e-01 1.24696217e-01 7.73181081e-01
4.04436529e-01 -2.59955645e-01 -4.55773354e-01 4.72466081e-01
-9.37405348e-01 -1.34602988e+00 -8.16580296e-01 -1.10095441e+00
9.10700262e-01 1.95060134e-01 4.13539290e-01 6.38068542e-02
-1.78230643e-01 -6.19915366e-01 -3.41964632e-01 -6.89503729e-01
-4.36814815e-01 -1.85064506e-02 2.18354538e-01 -2.45096549e-01
-1.41523793e-01 -7.04794109e-01 -7.27572739e-01 4.92510945e-01
-8.69074106e-01 5.76401465e-02 5.40137708e-01 8.86784732e-01
1.34367883e+00 2.40847562e-02 1.07038629e+00 -8.63062978e-01
8.33069026e-01 -5.96767545e-01 -5.49971700e-01 3.92239511e-01
-4.81434852e-01 4.18153256e-01 7.30767310e-01 -6.38938308e-01
-7.76426136e-01 5.29444754e-01 -4.04001594e-01 2.79515237e-01
4.98868227e-02 1.00826979e+00 -8.60113263e-01 -3.16280484e-01
7.59745419e-01 4.85562623e-01 9.45161954e-02 -6.78355634e-01
2.59874612e-01 1.47452384e-01 4.29130420e-02 -9.26620901e-01
4.83266175e-01 4.03963774e-03 6.13091707e-01 -7.54736722e-01
-4.75998402e-01 -1.05943553e-01 -3.94691110e-01 3.42266113e-01
8.54487062e-01 -8.26135457e-01 -9.59401309e-01 -2.64452621e-02
-1.12237406e+00 1.79406524e-01 5.47274835e-02 7.67420828e-01
-5.06369531e-01 4.30390298e-01 -4.89537045e-02 -9.79815573e-02
-4.83473450e-01 -1.33527100e+00 4.44131136e-01 3.74972880e-01
-3.31122071e-01 -6.99237168e-01 3.32547277e-01 -9.52719972e-02
-8.60998258e-02 5.78357041e-01 1.63360798e+00 -9.16123807e-01
-1.00329471e+00 -1.83266420e-02 3.23166221e-01 5.64046390e-02
6.18650615e-01 1.19386815e-01 -2.23020613e-01 -2.28844732e-01
-5.10573447e-01 3.43623199e-03 5.08494675e-01 5.96028328e-01
1.10069191e+00 -4.38577920e-01 -6.76973999e-01 5.59210181e-01
1.23849797e+00 9.87685382e-01 1.03909349e+00 1.47097766e-01
5.72275639e-01 2.23685190e-01 7.27243483e-01 6.14382982e-01
-3.17044497e-01 8.34582031e-01 4.92787182e-01 -1.88153893e-01
1.38106018e-01 -5.52254260e-01 5.81834652e-02 -6.48710951e-02
-3.98095161e-01 -1.76169753e-01 -5.55441380e-01 6.06768839e-02
-1.48941541e+00 -1.01040494e+00 -3.43447179e-01 2.50735855e+00
1.37712562e+00 -1.77660808e-01 3.64563376e-01 -4.22914922e-01
6.21389568e-01 -6.96810722e-01 -1.01809585e+00 -1.53949484e-01
-1.23204276e-01 8.97346914e-01 3.03215683e-01 3.73162478e-01
-7.25343883e-01 1.18692243e+00 7.45619869e+00 9.51060951e-01
-1.09815490e+00 -6.98596656e-01 5.39807856e-01 2.38421962e-01
-5.06472230e-01 4.65736747e-01 -1.12054551e+00 3.59317631e-01
6.94238722e-01 -6.12068713e-01 1.57868266e-01 7.16081738e-01
5.40678620e-01 -6.68669818e-03 -1.13438416e+00 5.99466801e-01
-4.75707352e-01 -2.13949990e+00 7.25183010e-01 2.46884808e-01
8.67025852e-01 -6.42143786e-01 -1.33846954e-01 -3.25750023e-01
5.39720850e-03 -1.37212086e+00 3.87811929e-01 5.86791813e-01
7.94917226e-01 -9.31780159e-01 2.32952163e-01 2.00183913e-01
-7.80587673e-01 1.86710835e-01 -4.53277171e-01 3.02634686e-01
-3.02290432e-02 2.81222016e-01 -1.22959125e+00 5.56179762e-01
-1.05287969e-01 7.08300114e-01 -4.42899644e-01 1.60292768e+00
-5.32138169e-01 2.92685717e-01 -1.03072956e-01 -4.39687997e-01
-3.79741821e-03 -5.30679226e-01 2.78034776e-01 6.67548776e-01
3.36432010e-01 2.42402256e-01 3.67482275e-01 9.23337638e-01
-1.13565652e-02 5.68092763e-01 -5.46752095e-01 -1.75931141e-01
5.33017814e-01 7.36353993e-01 -5.91972768e-01 -2.35018447e-01
-6.72245324e-02 6.44445360e-01 -1.06089979e-01 6.49301767e-01
-7.91923761e-01 -5.28459191e-01 7.81309128e-01 3.63151133e-01
1.71258062e-01 -1.18531495e-01 1.41695797e-01 -6.50807321e-01
-5.14221191e-01 -1.10160112e+00 1.87846348e-01 -5.97216845e-01
-9.85792696e-01 4.97576177e-01 -2.33781524e-02 -1.18374491e+00
3.94185573e-01 -4.12881434e-01 -6.98952854e-01 1.20088041e+00
-1.14024651e+00 -8.73359144e-01 1.41570255e-01 5.17162085e-01
3.49484652e-01 -4.72719342e-01 1.19104707e+00 2.65084118e-01
-5.69567561e-01 1.73992440e-01 3.14068258e-01 -7.60156870e-01
6.30408704e-01 -7.43668556e-01 -8.67752358e-02 3.08903545e-01
-1.95939347e-01 1.26329267e+00 8.29616606e-01 -1.13471627e+00
-1.16331410e+00 -1.02205431e+00 4.72580820e-01 -5.68319187e-02
3.31064075e-01 -9.23043117e-03 -7.12576330e-01 2.10194260e-01
-3.55504274e-01 -6.42502367e-01 1.39751828e+00 -1.41217753e-01
1.81343313e-03 3.84301305e-01 -9.33299184e-01 6.83742464e-01
9.11155105e-01 3.24837118e-02 -6.41299188e-02 8.34907472e-01
8.81822646e-01 -5.62964201e-01 -1.03796554e+00 3.86077672e-01
3.98867846e-01 -3.32487077e-01 1.20794523e+00 -1.34397507e+00
1.56457663e-01 -9.88174975e-01 8.70334879e-02 -1.26393414e+00
-7.73050904e-01 -8.48282695e-01 -8.70857481e-03 6.28957093e-01
9.79504168e-01 -3.65397125e-01 8.93879235e-01 4.04961228e-01
-3.55819374e-01 -9.17011559e-01 -8.20124269e-01 -7.84664214e-01
5.93092516e-02 3.07104290e-01 9.22731400e-01 5.96875608e-01
1.63653776e-01 9.85948384e-01 -4.61894959e-01 8.85314867e-02
1.40183881e-01 4.62569594e-02 6.93477392e-01 -1.37794626e+00
-4.24510688e-01 -2.92316794e-01 -3.15887749e-01 -9.83675003e-01
3.29057649e-02 -1.11847460e+00 -6.72242165e-01 -1.75182176e+00
2.63992876e-01 -5.38574040e-01 6.77805990e-02 4.18858737e-01
-1.00987237e-02 -4.41198319e-01 -2.54177958e-01 1.98587954e-01
-1.20353915e-01 6.77206755e-01 1.62067270e+00 -1.81205198e-01
-8.59978259e-01 7.23827302e-01 -1.14791298e+00 2.73985952e-01
7.63296306e-01 -5.02935171e-01 -5.45476258e-01 5.25266349e-01
6.03425503e-01 1.58510640e-01 -1.09403342e-01 -5.28141558e-01
-1.84206396e-01 -8.65988553e-01 5.38520217e-01 -5.62884569e-01
1.24384165e-01 -6.97032273e-01 1.05012035e+00 6.99444473e-01
-1.25360429e-01 -6.08542860e-01 1.48872420e-01 7.24416673e-01
-4.14842293e-02 -3.40203047e-01 6.37427926e-01 -1.15073688e-01
-3.55152249e-01 6.34865999e-01 -5.73035479e-01 -6.62474692e-01
1.32019699e+00 -7.16828167e-01 -4.32885699e-02 1.39488233e-02
-1.14008498e+00 -1.02366820e-01 3.11343849e-01 1.83121219e-01
9.58421230e-01 -1.23820853e+00 -4.32263494e-01 -1.23493401e-02
3.47485304e-01 1.69329834e-03 -7.95608610e-02 4.34912175e-01
-8.27268302e-01 6.54449642e-01 -2.32445315e-01 -1.73039302e-01
-1.43815422e+00 9.09492075e-01 5.71004093e-01 1.65360384e-02
-4.07474004e-02 7.21722722e-01 5.01876831e-01 -2.93749794e-02
2.27640405e-01 -1.76649272e-01 -3.55839074e-01 -1.60930663e-01
6.14634931e-01 5.29815145e-02 1.26803145e-01 -3.33582193e-01
-5.53397298e-01 5.92473328e-01 -1.62993297e-01 7.41500854e-01
1.51366651e+00 4.41983670e-01 -4.75616604e-02 -3.73425692e-01
9.12464440e-01 7.13862032e-02 -8.74146879e-01 -5.81947006e-02
-5.70780300e-02 -2.67233223e-01 -2.66428411e-01 -1.00003493e+00
-4.04834032e-01 2.11692601e-01 4.78889823e-01 -6.57657087e-01
7.12531984e-01 1.66510403e-01 3.92261118e-01 5.01376867e-01
7.18992233e-01 -6.48871422e-01 3.35151702e-01 3.06098819e-01
1.09072185e+00 -6.91040814e-01 4.21034724e-01 -7.85443366e-01
-3.83991450e-01 1.21921146e+00 3.23106825e-01 2.80442029e-01
5.57320476e-01 -6.69396281e-01 -7.12364137e-01 -4.76908386e-01
-5.63353240e-01 -2.00648442e-01 5.01131296e-01 7.70147443e-01
7.54663527e-01 7.41339400e-02 -1.02360272e+00 5.03993630e-01
2.90086389e-01 2.54145041e-02 3.20318550e-01 8.90636325e-01
-7.25170255e-01 -1.95330250e+00 -2.48575151e-01 8.94042179e-02
-3.86305571e-01 -5.11349082e-01 -8.35720956e-01 3.13353807e-01
2.56460190e-01 8.14503014e-01 -6.93389535e-01 -2.42673203e-01
5.20995617e-01 -1.50113314e-01 5.97548008e-01 -1.23117113e+00
-4.97029632e-01 4.95473415e-01 1.22993708e-01 2.96542775e-02
-2.14121744e-01 -1.63624972e-01 -1.75436842e+00 2.51076780e-02
-8.33197117e-01 6.90702975e-01 6.12037301e-01 6.34678900e-01
8.77774358e-01 4.46194381e-01 9.82169867e-01 -5.03440320e-01
-1.08384481e-02 -3.91792506e-01 -8.01648080e-01 -8.02669600e-02
-1.22179151e-01 -6.32979333e-01 2.80805022e-01 2.98144519e-01] | [4.965819358825684, 5.726573467254639] |
df014c51-149c-40e8-84c4-9463c4a5b19a | reading-car-license-plates-using-deep | 1601.05610 | null | http://arxiv.org/abs/1601.05610v1 | http://arxiv.org/pdf/1601.05610v1.pdf | Reading Car License Plates Using Deep Convolutional Neural Networks and LSTMs | In this work, we tackle the problem of car license plate detection and
recognition in natural scene images. Inspired by the success of deep neural
networks (DNNs) in various vision applications, here we leverage DNNs to learn
high-level features in a cascade framework, which lead to improved performance
on both detection and recognition.
Firstly, we train a $37$-class convolutional neural network (CNN) to detect
all characters in an image, which results in a high recall, compared with
conventional approaches such as training a binary text/non-text classifier.
False positives are then eliminated by the second plate/non-plate CNN
classifier. Bounding box refinement is then carried out based on the edge
information of the license plates, in order to improve the
intersection-over-union (IoU) ratio. The proposed cascade framework extracts
license plates effectively with both high recall and precision. Last, we
propose to recognize the license characters as a {sequence labelling} problem.
A recurrent neural network (RNN) with long short-term memory (LSTM) is trained
to recognize the sequential features extracted from the whole license plate via
CNNs. The main advantage of this approach is that it is segmentation free. By
exploring context information and avoiding errors caused by segmentation, the
RNN method performs better than a baseline method of combining segmentation and
deep CNN classification; and achieves state-of-the-art recognition accuracy. | ['Chunhua Shen', 'Hui Li'] | 2016-01-21 | null | null | null | null | ['license-plate-detection'] | ['computer-vision'] | [ 4.62866366e-01 -3.36387873e-01 2.58287955e-02 -2.06001267e-01
-8.14600766e-01 -5.68643630e-01 5.60534179e-01 -3.58671844e-01
-7.04888880e-01 3.77555102e-01 -3.51536065e-01 -1.79988816e-01
4.56398517e-01 -8.77549529e-01 -8.30238819e-01 -8.66165161e-01
5.40887296e-01 2.40144819e-01 6.22249246e-01 -4.60307114e-02
6.35277629e-01 7.34867513e-01 -1.62438750e+00 3.93861115e-01
8.56537938e-01 1.23184872e+00 1.68846384e-01 4.80312824e-01
-3.50496233e-01 9.42911088e-01 -6.35505497e-01 -6.88928366e-01
4.76551294e-01 -1.51499823e-01 -6.37190342e-01 4.50312018e-01
2.28512824e-01 -1.94312274e-01 -3.11751992e-01 1.25991535e+00
1.35289669e-01 5.04930690e-02 7.34283686e-01 -7.07657874e-01
-3.70226175e-01 3.30334008e-02 -6.89127445e-01 1.18503958e-01
-4.15239036e-02 1.74858481e-01 9.33188915e-01 -1.23276949e+00
3.89690191e-01 6.60408914e-01 8.66249323e-01 4.19656217e-01
-8.16402495e-01 -6.40713513e-01 2.16395855e-02 2.24519789e-01
-1.66462958e+00 -4.64998603e-01 5.11794746e-01 -4.61073369e-01
9.53979075e-01 7.64157027e-02 6.14614725e-01 6.45241141e-01
1.01318561e-01 1.21250474e+00 7.60024011e-01 -5.87261319e-01
-9.01815016e-03 1.58036545e-01 1.27247199e-01 8.75329316e-01
5.40536232e-02 -1.33140698e-01 -4.89754081e-02 5.01892924e-01
9.70294654e-01 5.88621616e-01 -1.97868720e-02 1.39680624e-01
-7.47281015e-01 7.53331959e-01 5.39920092e-01 4.52839673e-01
-4.07394409e-01 -8.50057676e-02 3.46705824e-01 -2.16942966e-01
3.18791211e-01 2.83783257e-01 1.33426767e-02 1.13127299e-01
-1.26692331e+00 2.85360933e-04 6.42572284e-01 8.44135821e-01
8.84545207e-01 1.48387358e-01 -1.64477885e-01 1.16973436e+00
2.46616051e-01 3.58997643e-01 4.23627824e-01 -3.71376693e-01
4.73387271e-01 9.71232176e-01 -1.14121750e-01 -9.05101120e-01
-3.85783702e-01 -5.65729797e-01 -9.79521513e-01 4.14165199e-01
4.62405413e-01 -3.26512903e-02 -1.35087776e+00 9.01621521e-01
-2.10307434e-01 2.68022954e-01 7.99669325e-02 9.73331332e-01
5.71813047e-01 9.00771439e-01 -6.19623363e-02 1.15973264e-01
1.53613234e+00 -1.19528878e+00 -4.75035876e-01 -5.98962247e-01
6.56442046e-01 -8.43346834e-01 6.02364123e-01 2.50355363e-01
-7.05237210e-01 -6.11877263e-01 -1.06527781e+00 2.43063943e-04
-6.57303691e-01 6.61751568e-01 1.17649227e-01 6.12380385e-01
-8.51192594e-01 2.30896190e-01 -5.61366618e-01 -6.39837235e-02
6.71049178e-01 6.68569803e-01 -9.92320552e-02 -2.75795996e-01
-8.97419393e-01 8.26982319e-01 3.21779817e-01 6.60115898e-01
-5.15320897e-01 -1.28875524e-01 -8.23289394e-01 1.76969811e-01
5.31844735e-01 -1.68827549e-01 1.07797277e+00 -1.35466027e+00
-1.62121749e+00 9.19116020e-01 -1.64602995e-01 -5.50382614e-01
3.24676931e-01 -1.01510346e-01 -4.25010622e-01 9.38347131e-02
-1.01186201e-01 7.09440827e-01 8.86753201e-01 -9.75157619e-01
-1.09025824e+00 -3.36902350e-01 -2.31540531e-01 1.12108268e-01
-1.60382733e-01 1.87950492e-01 -9.33185458e-01 -6.03679776e-01
2.14331895e-01 -1.07431102e+00 -2.72977889e-01 -4.96659786e-01
-5.86901426e-01 -2.85710037e-01 8.94378364e-01 -8.48924875e-01
9.28782880e-01 -2.31064510e+00 -3.95436049e-01 3.97199661e-01
1.41294122e-01 9.23581183e-01 4.86738533e-02 -1.60771720e-02
1.98714584e-01 8.41013789e-02 -4.27090436e-01 -4.27614152e-01
-2.29941159e-01 -4.11908235e-03 -1.39298856e-01 4.80905056e-01
6.34935439e-01 1.03071439e+00 -2.41389781e-01 -4.62981045e-01
3.40240270e-01 3.84970874e-01 -1.29323035e-01 2.19940152e-02
3.59679125e-02 5.93583621e-02 -2.39451513e-01 9.41748500e-01
7.96656132e-01 -1.91352949e-01 -5.33328913e-02 4.91850749e-02
-3.95845294e-01 -6.03087880e-02 -9.85905170e-01 1.05200970e+00
-4.26241606e-01 9.02731299e-01 -7.33392164e-02 -1.19699717e+00
1.32419074e+00 2.15288386e-01 3.34285088e-02 -8.67618859e-01
2.44259834e-01 4.81228143e-01 -1.45247772e-01 -4.06730503e-01
3.92721593e-01 -3.12982388e-02 -1.18075721e-01 -1.44556597e-01
-1.55720999e-02 1.34764746e-01 1.67571545e-01 -2.13437021e-01
7.24107206e-01 -1.85668971e-02 -9.51813348e-03 -3.20291817e-02
9.10219669e-01 3.51617485e-02 6.40933752e-01 6.96576118e-01
-1.48317413e-02 6.87247455e-01 3.90017152e-01 -6.19499385e-01
-1.02962053e+00 -5.30329525e-01 8.32814500e-02 9.21157002e-01
1.69747293e-01 1.40879557e-01 -8.26855898e-01 -5.35328984e-01
-3.48209143e-01 3.90045017e-01 -3.77408803e-01 1.04648970e-01
-9.68284249e-01 -7.44007945e-01 7.97060072e-01 7.59782314e-01
9.40906227e-01 -1.10028958e+00 -4.43272918e-01 3.32308829e-01
-3.57987918e-02 -1.41860807e+00 -4.29634869e-01 2.23024085e-01
-7.60205925e-01 -9.26374972e-01 -9.63232994e-01 -1.31379187e+00
6.29837871e-01 5.54591045e-03 6.02770627e-01 1.25923291e-01
-2.06273511e-01 -2.64289826e-01 -1.96497813e-01 -1.92017376e-01
-2.78941989e-01 1.07204176e-01 -3.28777671e-01 5.34937024e-01
5.97231925e-01 -3.97664681e-03 -5.08775175e-01 3.92885149e-01
-8.55785191e-01 4.06404585e-02 1.23382306e+00 7.34835982e-01
6.06614232e-01 1.97797958e-02 3.69957775e-01 -6.88020349e-01
2.24552244e-01 -9.51761100e-03 -9.91242230e-01 2.57061571e-01
-4.17048186e-01 -1.71590343e-01 8.31777573e-01 -1.10041440e-01
-9.76738572e-01 5.52173615e-01 -6.35201156e-01 -4.67034936e-01
-3.77442807e-01 3.31530035e-01 -3.13538089e-02 -2.41255149e-01
1.29347160e-01 7.33942866e-01 5.63777424e-03 -3.92428935e-01
-9.87300128e-02 8.57342064e-01 7.00867653e-01 7.09028244e-02
5.11936188e-01 2.76566356e-01 -1.54135168e-01 -1.06946993e+00
-6.98188365e-01 -8.74133110e-01 -7.75865316e-01 -4.32155371e-01
1.07941985e+00 -7.82849371e-01 -9.32302952e-01 1.01882255e+00
-1.16088498e+00 -3.99626307e-02 8.17407295e-02 4.31157172e-01
-9.45413262e-02 2.30386853e-01 -8.38725626e-01 -1.03893650e+00
-2.72412032e-01 -1.41885448e+00 1.05104709e+00 5.91830313e-01
4.16051835e-01 -7.17896283e-01 -4.21894550e-01 4.75887567e-01
3.34952295e-01 7.02948943e-02 5.53604841e-01 -1.05006862e+00
-8.45665574e-01 -7.42131531e-01 -5.77177525e-01 6.83450997e-01
-2.92637467e-01 1.05240822e-01 -9.98486817e-01 1.84975028e-01
-1.17034845e-01 -1.45728767e-01 1.29464090e+00 6.01078928e-01
8.60055447e-01 -2.45034471e-01 -3.92864883e-01 5.64152002e-01
1.57364595e+00 4.15830791e-01 1.01145685e+00 5.11388421e-01
8.58217895e-01 4.59658027e-01 3.86146188e-01 1.00320056e-01
1.05938703e-01 5.39223850e-01 3.38495702e-01 -3.67518008e-01
3.44729424e-02 1.30029833e-02 4.25522566e-01 4.14461285e-01
-2.09144399e-01 -2.47934073e-01 -1.23840988e+00 4.26206499e-01
-1.61931860e+00 -8.73473644e-01 -2.43697390e-01 1.96127176e+00
3.21384668e-01 5.13937235e-01 2.66273201e-01 2.65086949e-01
1.12132561e+00 -3.68519351e-02 -2.89344221e-01 -4.64948565e-01
-2.46479526e-01 4.88609672e-02 8.86041224e-01 2.62364268e-01
-1.36412489e+00 1.28257918e+00 5.50751591e+00 1.10140598e+00
-1.39329231e+00 -2.03085795e-01 1.00458074e+00 3.41166288e-01
3.45353544e-01 -3.07272941e-01 -1.21326005e+00 3.47359419e-01
8.21087956e-01 5.83451569e-01 1.34476945e-01 7.37344682e-01
-7.48696849e-02 -8.21885392e-02 -6.74618363e-01 8.78108501e-01
2.25128815e-01 -1.59298050e+00 -2.14824174e-02 1.23220801e-01
5.70421100e-01 7.35765174e-02 1.07274517e-01 5.54816246e-01
-1.10057801e-01 -1.03701258e+00 7.02442229e-01 6.56986892e-01
8.42989862e-01 -1.07224536e+00 1.25707042e+00 5.87548673e-01
-1.25687087e+00 -3.97311151e-01 -2.66265154e-01 9.22370329e-02
1.14924191e-02 5.33737004e-01 -1.08839059e+00 2.45855436e-01
5.17656326e-01 6.64924622e-01 -4.92757142e-01 1.24866736e+00
-1.20348081e-01 6.04422212e-01 -1.65893376e-01 -2.64394164e-01
8.07464838e-01 -4.84079957e-01 2.89913118e-01 1.65556526e+00
1.27653956e-01 1.05486572e-01 1.78366259e-01 8.58648717e-01
-3.90459031e-01 5.23373373e-02 -3.73746246e-01 9.89725231e-04
1.42967314e-01 1.24755836e+00 -1.26748240e+00 -4.59325880e-01
-6.52832568e-01 1.00341046e+00 6.43837154e-02 3.22846204e-01
-7.65240550e-01 -6.69539690e-01 2.60031939e-01 -3.63541283e-02
8.57128978e-01 -4.74340133e-02 -3.70543718e-01 -9.26986516e-01
1.11210771e-01 -5.87050498e-01 1.89469725e-01 -3.25467438e-01
-9.53572631e-01 6.02026939e-01 -7.02312768e-01 -1.36253655e+00
-1.48573026e-01 -9.39507246e-01 -6.30510330e-01 7.95254767e-01
-1.64636528e+00 -1.24536359e+00 1.30476884e-03 4.36456889e-01
9.73802745e-01 -1.97040081e-01 4.80991930e-01 3.04928094e-01
-9.21478271e-01 5.58482707e-01 2.62351900e-01 9.35718536e-01
2.73749620e-01 -9.13828790e-01 4.53817308e-01 1.12814283e+00
2.36473382e-02 3.44864219e-01 9.07476023e-02 -5.80934048e-01
-1.01066482e+00 -1.35877192e+00 7.75915146e-01 1.88847054e-02
2.96773553e-01 -5.70505500e-01 -8.68214428e-01 5.46842992e-01
-1.12221844e-01 1.69827923e-01 4.76620734e-01 -3.01740676e-01
-2.36211449e-01 -1.33404419e-01 -9.55647707e-01 3.77175570e-01
5.08453906e-01 -4.52853858e-01 -5.57663023e-01 1.79227382e-01
3.49939466e-01 -2.12299749e-01 -3.69548798e-01 3.51569355e-01
6.32810116e-01 -9.51087117e-01 9.05842602e-01 -4.03687060e-02
4.37070400e-01 -3.37319672e-01 4.67169136e-02 -7.48214364e-01
-2.96913832e-01 -1.02542125e-01 3.04904997e-01 1.20125329e+00
7.00991154e-01 -5.98007023e-01 9.28969860e-01 4.72255439e-01
-4.45092350e-01 -7.95721114e-01 -9.69286144e-01 -7.24460661e-01
-1.28629863e-01 -2.31968507e-01 1.97218150e-01 4.16422516e-01
-5.21851540e-01 1.86949924e-01 -4.28546220e-01 2.48457804e-01
3.01203907e-01 1.58291429e-01 4.17308390e-01 -1.18159366e+00
6.63994104e-02 -7.73676157e-01 -7.44822681e-01 -1.25884640e+00
2.32871935e-01 -6.37820601e-01 4.63031948e-01 -1.39810538e+00
1.41548917e-01 -1.39084235e-01 -2.90195793e-01 4.93607491e-01
-5.27928956e-02 8.57081115e-01 2.79435188e-01 5.18745661e-01
-8.32202911e-01 1.72629908e-01 1.00683355e+00 -3.29631805e-01
-3.32795650e-01 4.82045919e-01 -3.36796850e-01 8.90868425e-01
6.46255732e-01 -4.12484318e-01 4.14844066e-01 -2.84296423e-01
-8.88273567e-02 1.10926680e-01 2.82802016e-01 -1.10879910e+00
6.77392185e-01 3.70839715e-01 7.74886012e-01 -7.76251495e-01
3.25086802e-01 -6.13597691e-01 -4.45628941e-01 5.44633925e-01
8.80363025e-03 -2.07609743e-01 3.46708000e-01 5.43528974e-01
-6.06211483e-01 -5.25083899e-01 8.84527922e-01 -2.12488607e-01
-1.16525209e+00 1.33217633e-01 -9.49104309e-01 -3.97275895e-01
9.83702958e-01 -6.74965560e-01 -2.17373613e-02 -1.72918309e-02
-4.74715382e-01 -6.93360046e-02 2.52636701e-01 2.25556985e-01
7.18096197e-01 -9.68038619e-01 -7.14115202e-01 3.41428548e-01
8.38831365e-02 9.26889107e-03 1.74621224e-01 9.07691777e-01
-7.77171195e-01 5.77221394e-01 -1.18011892e-01 -8.67225647e-01
-1.14478731e+00 2.48989463e-01 5.04297495e-01 -3.45991641e-01
-6.70212030e-01 9.92593169e-01 6.85942546e-02 -3.38074565e-01
5.31940237e-02 -1.57576725e-01 -6.02701128e-01 1.51735932e-01
5.50082922e-01 1.99528217e-01 3.66485536e-01 -8.09434116e-01
-4.77606446e-01 8.66659760e-01 -2.83128560e-01 5.73448092e-02
1.39742553e+00 1.36891067e-01 -1.00589074e-01 1.12397164e-01
1.24711454e+00 -1.24697939e-01 -1.29517376e+00 -2.08530769e-01
3.89199138e-01 -2.45132312e-01 5.71490861e-02 -6.80485904e-01
-1.24326408e+00 8.51041377e-01 5.32308817e-01 7.59495497e-02
1.10632062e+00 -3.30226392e-01 9.66674507e-01 5.42161405e-01
-9.83261690e-02 -1.26259458e+00 -4.71142866e-02 7.60767639e-01
4.75364774e-01 -1.30469489e+00 -3.75337809e-01 -2.67499775e-01
-7.77123332e-01 1.58302784e+00 5.38557947e-01 -3.99198890e-01
2.83313274e-01 4.09438491e-01 5.93473464e-02 -6.90628141e-02
-1.15411513e-01 -5.23657262e-01 3.69186938e-01 8.34136903e-02
1.67817667e-01 -1.42244771e-01 1.31924987e-01 6.12861216e-01
2.66671181e-01 -7.31000751e-02 4.79121655e-01 8.88884306e-01
-7.47010767e-01 -7.27636576e-01 -5.43616831e-01 5.74540615e-01
-5.52608848e-01 -1.39555618e-01 -4.13573623e-01 8.71435940e-01
2.77270019e-01 8.72326553e-01 3.65811884e-01 -3.77746373e-01
3.50889117e-01 2.90301412e-01 5.41946627e-02 -5.69687903e-01
-6.60094678e-01 4.27000612e-01 -1.95609674e-01 -2.33802333e-01
-3.07975322e-01 -6.05334401e-01 -1.29855645e+00 2.48390157e-02
-7.80257702e-01 -5.12442365e-02 8.00506175e-01 1.30593812e+00
1.00103736e-01 5.47820628e-01 6.89128399e-01 -9.15469229e-01
-2.26765931e-01 -8.07085991e-01 -5.85250020e-01 1.75709412e-01
4.16984916e-01 -3.36933970e-01 -2.45785117e-01 1.81551874e-01] | [9.844403266906738, -4.933106899261475] |
4fe69ba8-8d72-4d2a-8af8-379b7b56bc43 | a-robust-and-lightweight-deep-attention | 2206.00455 | null | https://arxiv.org/abs/2206.00455v1 | https://arxiv.org/pdf/2206.00455v1.pdf | A robust and lightweight deep attention multiple instance learning algorithm for predicting genetic alterations | Deep-learning models based on whole-slide digital pathology images (WSIs) become increasingly popular for predicting molecular biomarkers. Instance-based models has been the mainstream strategy for predicting genetic alterations using WSIs although bag-based models along with self-attention mechanism-based algorithms have been proposed for other digital pathology applications. In this paper, we proposed a novel Attention-based Multiple Instance Mutation Learning (AMIML) model for predicting gene mutations. AMIML was comprised of successive 1-D convolutional layers, a decoder, and a residual weight connection to facilitate further integration of a lightweight attention mechanism to detect the most predictive image patches. Using data for 24 clinically relevant genes from four cancer cohorts in The Cancer Genome Atlas (TCGA) studies (UCEC, BRCA, GBM and KIRC), we compared AMIML with one popular instance-based model and four recently published bag-based models (e.g., CHOWDER, HE2RNA, etc.). AMIML demonstrated excellent robustness, not only outperforming all the five baseline algorithms in the vast majority of the tested genes (17 out of 24), but also providing near-best-performance for the other seven genes. Conversely, the performance of the baseline published algorithms varied across different cancers/genes. In addition, compared to the published models for genetic alterations, AMIML provided a significant improvement for predicting a wide range of genes (e.g., KMT2C, TP53, and SETD2 for KIRC; ERBB2, BRCA1, and BRCA2 for BRCA; JAK1, POLE, and MTOR for UCEC) as well as produced outstanding predictive models for other clinically relevant gene mutations, which have not been reported in the current literature. Furthermore, with the flexible and interpretable attention-based MIL pooling mechanism, AMIML could further zero-in and detect predictive image patches. | ['Xu Steven Xu', 'Hong Zhang', 'Miaomiao Yang', 'Xingyu Li', 'Bangwei Guo'] | 2022-05-31 | null | null | null | null | ['deep-attention', 'deep-attention'] | ['computer-vision', 'natural-language-processing'] | [ 2.88073957e-01 2.47462526e-01 -4.35628086e-01 -6.59068227e-02
-1.31872571e+00 -1.44613340e-01 2.88668871e-01 6.17001235e-01
-1.40755326e-01 8.08410883e-01 8.89414027e-02 -5.13057053e-01
-3.00515056e-01 -6.25265598e-01 -7.99581349e-01 -1.21991360e+00
-1.44213885e-01 3.89808863e-01 1.46434858e-01 -1.46128535e-01
4.27543595e-02 3.47953886e-01 -1.20094454e+00 4.63382930e-01
9.92692232e-01 9.84046519e-01 1.18604086e-01 8.26579154e-01
-1.29649818e-01 7.35880673e-01 -1.73749223e-01 -5.83111584e-01
-4.36509520e-01 -3.15342814e-01 -6.91753864e-01 -3.17024320e-01
5.16611993e-01 2.90875956e-02 -3.49083066e-01 7.64403820e-01
7.90480852e-01 -5.42193234e-01 6.21958017e-01 -9.01987374e-01
-5.43305933e-01 3.94457102e-01 -8.01044285e-01 9.50943902e-02
1.54631183e-01 3.35558057e-01 1.27414382e+00 -7.93552458e-01
8.09723318e-01 9.05132174e-01 9.63746250e-01 6.27521753e-01
-1.02753723e+00 -5.58505893e-01 -2.10376039e-01 3.02124828e-01
-1.41584218e+00 -1.82351083e-01 2.96686083e-01 -3.71499836e-01
1.15037775e+00 6.10138655e-01 5.38883507e-01 1.03280878e+00
6.22068763e-01 9.84744549e-01 7.19905794e-01 -4.05034304e-01
-5.55237904e-02 -1.09272286e-01 3.27186197e-01 1.01395059e+00
2.00718731e-01 -3.40927154e-01 -2.96721667e-01 -4.23513860e-01
2.99392790e-01 2.30548739e-01 -4.56565410e-01 2.92513818e-01
-1.12994289e+00 6.68823659e-01 4.57868457e-01 4.94694233e-01
-7.24916682e-02 2.10980415e-01 5.68459988e-01 -9.40594748e-02
5.43708444e-01 1.92794919e-01 -5.18819928e-01 1.08013213e-01
-6.49644256e-01 -1.55277811e-02 2.70303428e-01 3.78758341e-01
4.13428575e-01 -4.66760784e-01 -6.12598300e-01 8.98139119e-01
3.10932189e-01 -1.19486615e-01 9.50251341e-01 4.28781062e-02
2.37557918e-01 1.14826274e+00 -2.50230968e-01 -7.60112703e-01
-1.03699315e+00 -6.10437036e-01 -1.00878942e+00 -2.12859184e-01
2.86761940e-01 2.80470163e-01 -1.15234768e+00 1.46640897e+00
6.04249053e-02 4.14935112e-01 -6.72733262e-02 7.53843129e-01
1.17828953e+00 2.84924507e-01 3.02881628e-01 3.26640382e-02
1.44398797e+00 -6.81727469e-01 -5.69087148e-01 1.19928248e-01
1.08684015e+00 -2.61654168e-01 8.86079669e-01 1.14290729e-01
-8.62641335e-01 -3.49561572e-01 -9.31651711e-01 -2.95282543e-01
-5.97280085e-01 3.05981487e-01 8.30013216e-01 5.14226437e-01
-1.38288367e+00 2.80082375e-01 -8.62043440e-01 -7.77951062e-01
8.79213989e-01 4.78892237e-01 -4.97119129e-01 -1.12093650e-01
-1.06781948e+00 7.47590184e-01 2.97646224e-01 -5.97893633e-02
-9.06178117e-01 -1.03625345e+00 -6.88362837e-01 2.32405424e-01
-2.14552626e-01 -6.66258812e-01 7.99642980e-01 -7.43484437e-01
-1.35338902e+00 1.25372648e+00 -2.44139895e-01 -4.87233400e-01
1.92815989e-01 2.63500154e-01 -4.40041512e-01 5.34836538e-02
-2.08423555e-01 6.95619345e-01 2.39421234e-01 -6.14185929e-01
-3.82663548e-01 -4.38826710e-01 -3.30415487e-01 2.57441518e-03
-3.41637075e-01 -2.38969892e-01 -5.31807482e-01 -4.10840750e-01
-1.38384253e-01 -6.86607718e-01 -1.56291902e-01 9.86849591e-02
-7.33828187e-01 -2.87254453e-01 6.26780570e-01 -8.48150611e-01
1.13119423e+00 -2.33084822e+00 1.48143336e-01 1.00948125e-01
2.16740608e-01 4.88314360e-01 -2.25228250e-01 1.44377708e-01
-3.47240537e-01 5.51132023e-01 -1.31023541e-01 -1.57655120e-01
-5.81438728e-02 -1.17063984e-01 1.48153692e-01 6.63680971e-01
6.03490412e-01 1.33245468e+00 -7.37240314e-01 -4.93740678e-01
-5.19288070e-02 5.00715673e-01 -5.22050381e-01 1.13007985e-02
-3.59538257e-01 7.32320994e-02 -2.88447857e-01 1.19972444e+00
5.67210019e-01 -5.55446923e-01 2.21490443e-01 -4.57693227e-02
3.55387092e-01 -1.74329445e-01 -3.31267804e-01 1.45812714e+00
-8.85704160e-02 6.47444129e-01 -6.50323406e-02 -8.95886302e-01
7.91506469e-01 4.04829800e-01 5.24616480e-01 -5.34777045e-01
4.64011542e-02 1.63777605e-01 3.73121440e-01 -6.81598663e-01
-5.81191070e-02 -5.43814227e-02 1.49318308e-01 -1.23794861e-01
1.43353388e-01 2.26637274e-01 4.25751060e-02 1.10157132e-01
1.73066449e+00 -2.60869741e-01 4.75971520e-01 -3.27795185e-02
6.71508670e-01 4.03850265e-02 7.69925237e-01 6.06395483e-01
-3.67542863e-01 7.57600009e-01 8.40167403e-01 -4.23677713e-01
-6.10139251e-01 -7.25995004e-01 -4.76550996e-01 8.02296162e-01
-1.13232709e-01 -4.04072195e-01 -2.96103418e-01 -7.33915627e-01
2.99586803e-01 4.31249470e-01 -7.93190062e-01 -1.83628500e-01
-5.12516946e-02 -1.52899587e+00 9.42675531e-01 4.33613777e-01
4.61804837e-01 -5.70953548e-01 1.72964279e-02 3.12877119e-01
-1.53583824e-04 -5.13718188e-01 -1.93874225e-01 3.40105653e-01
-6.77663326e-01 -1.53144228e+00 -5.55614829e-01 -6.23615026e-01
6.98249280e-01 -2.96485752e-01 8.98755491e-01 3.06778193e-01
-8.89780343e-01 -3.34102586e-02 -1.57811150e-01 -7.08165705e-01
-4.50392783e-01 -1.20859370e-01 -3.16913247e-01 1.14157662e-01
3.68064851e-01 -6.59762695e-02 -4.47384864e-01 1.14440411e-01
-8.11105072e-01 2.24524722e-01 1.08458471e+00 1.25691390e+00
7.97749758e-01 -1.48217425e-01 7.86673248e-01 -8.25927556e-01
2.93381035e-01 -7.04849780e-01 -3.01715732e-01 3.77102196e-01
-2.16672316e-01 -4.67730582e-01 5.56014180e-01 -5.12672924e-02
-8.74522388e-01 -1.28248818e-02 -4.56440806e-01 1.51275954e-05
-1.90443054e-01 7.33966172e-01 -2.15203598e-01 -2.64402986e-01
4.72558320e-01 1.96849763e-01 3.05151790e-01 -1.81073844e-01
-2.90309340e-01 5.97955763e-01 1.22315705e-01 6.07751720e-02
4.40062247e-02 3.48263621e-01 2.29874060e-01 -5.93775928e-01
-6.29734099e-01 -4.06259477e-01 -1.16334707e-01 7.68177807e-02
8.35872531e-01 -8.86080444e-01 -9.57217515e-01 6.84848428e-01
-7.45395243e-01 -4.03406411e-01 1.27266541e-01 2.67352700e-01
-2.77329892e-01 3.66005786e-02 -1.17823827e+00 -3.35854352e-01
-4.92079675e-01 -1.33269954e+00 1.35482013e+00 2.08280325e-01
-1.81445777e-01 -1.04263949e+00 1.22433759e-01 3.55324894e-01
5.20489514e-01 4.66016769e-01 1.47626066e+00 -9.56194758e-01
-4.61687595e-01 -6.05997741e-01 -4.52608466e-01 -1.59044445e-01
8.07937682e-02 4.42307234e-01 -1.05983555e+00 -2.99750507e-01
-8.07702601e-01 -3.17261517e-01 1.28029907e+00 5.17388403e-01
1.57122970e+00 -5.34736142e-02 -1.20497775e+00 7.69665360e-01
1.56633294e+00 2.92162776e-01 8.52330863e-01 2.80322164e-01
4.08379883e-01 8.76569934e-03 3.02254051e-01 -5.83283268e-02
1.45660266e-01 6.48029208e-01 6.28434300e-01 -5.09896159e-01
-1.23268411e-01 5.55571392e-02 2.16726229e-01 2.80194670e-01
1.25300325e-02 -5.87182164e-01 -1.13887691e+00 5.93714893e-01
-1.67520094e+00 -7.34919786e-01 -1.71690434e-01 1.84975564e+00
7.93748498e-01 -4.58995067e-02 -3.55960310e-01 1.03578519e-03
6.07197702e-01 -6.15891106e-02 -7.09747016e-01 -3.42930049e-01
-3.48741829e-01 2.47760251e-01 3.25903237e-01 1.59556959e-02
-1.14013147e+00 4.70500737e-01 6.25312614e+00 1.02534461e+00
-1.08642840e+00 1.09905265e-01 1.47855949e+00 -1.24125265e-01
-1.85408652e-01 -3.97817045e-01 -9.41101015e-01 5.38403928e-01
9.71099317e-01 1.30657479e-01 -3.85872334e-01 4.87533659e-01
1.21021271e-01 -4.77453955e-02 -1.06010699e+00 6.93674326e-01
1.68127883e-02 -1.81226766e+00 -1.76508382e-01 2.63814062e-01
6.52132392e-01 2.33368486e-01 6.47694692e-02 3.48567486e-01
5.46317138e-02 -1.25238454e+00 -3.12449992e-01 8.56457412e-01
1.03865588e+00 -6.20583594e-01 1.36593401e+00 4.97835614e-02
-4.05793697e-01 -2.29594216e-01 -4.22962040e-01 4.93531317e-01
-4.53402251e-01 9.23023701e-01 -1.21510339e+00 7.28542745e-01
6.58695102e-01 8.38705897e-01 -7.68783331e-01 1.10102654e+00
7.17174634e-02 1.10366762e+00 -1.27504572e-01 -1.69160485e-01
1.76142916e-01 4.69564617e-01 3.24122071e-01 1.47465801e+00
4.50976431e-01 1.05646700e-01 -2.42900670e-01 7.27154374e-01
-2.07146823e-01 2.54872024e-01 5.33641083e-03 -1.21430412e-01
9.17683989e-02 1.60481417e+00 -4.76059377e-01 -1.21897914e-01
-5.00666082e-01 5.67656577e-01 4.87774640e-01 2.60105342e-01
-9.12868500e-01 -4.29847062e-01 7.55417407e-01 1.95334151e-01
1.60680130e-01 5.10236323e-01 -8.38038400e-02 -7.45052755e-01
-4.53188568e-01 -7.85562575e-01 8.99393559e-01 -5.99434197e-01
-1.40394974e+00 2.68976957e-01 -6.83464408e-01 -1.06454289e+00
2.11458445e-01 -8.91084611e-01 -9.78347778e-01 8.57499003e-01
-1.64570475e+00 -1.19916999e+00 -3.80196780e-01 2.58029550e-01
2.23982245e-01 -2.98709333e-01 1.09276927e+00 1.69732094e-01
-1.20486140e+00 9.38134193e-01 2.17880711e-01 7.13521242e-02
9.20865893e-01 -1.30535281e+00 3.58557515e-02 2.76774019e-01
-6.12276435e-01 4.16186750e-01 8.84739310e-02 -5.50828040e-01
-1.52690840e+00 -1.48917234e+00 6.51368618e-01 -9.33224522e-03
6.02659941e-01 -1.45565227e-01 -9.76032495e-01 7.07113922e-01
8.34678039e-02 3.33695292e-01 1.17042017e+00 -6.57580718e-02
-7.62912035e-02 -1.99821964e-01 -1.33569002e+00 4.03859645e-01
6.51250958e-01 -1.20433033e-01 2.24698588e-01 6.65921807e-01
6.04785800e-01 -6.93922758e-01 -1.09341621e+00 4.58689451e-01
4.33188409e-01 -8.47259700e-01 8.80412042e-01 -6.15552962e-01
6.45594597e-01 -3.11422914e-01 -3.37317884e-02 -1.33263195e+00
-8.07726502e-01 -1.61828622e-01 -1.35083765e-01 1.24734914e+00
6.79037690e-01 -8.58318329e-01 8.13299358e-01 3.93494129e-01
-4.61467832e-01 -1.36623001e+00 -1.14559174e+00 -1.58056274e-01
-4.93523926e-02 -2.30172530e-01 5.98732173e-01 7.60872006e-01
4.02371854e-01 -1.67815834e-01 1.21831186e-01 2.36787990e-01
3.04036081e-01 -1.58181325e-01 4.06024992e-01 -8.69551897e-01
-3.48292202e-01 -7.33145714e-01 -1.13485038e+00 -1.92279309e-01
6.47697374e-02 -1.43593824e+00 -3.72246385e-01 -1.55000269e+00
6.75383568e-01 -4.85664010e-01 -7.73910344e-01 8.68765295e-01
-5.59515655e-01 2.69477010e-01 -4.64916945e-01 -1.23790517e-01
-4.34813976e-01 1.21761300e-01 9.60381508e-01 -5.89863062e-01
1.96515962e-01 -1.62719220e-01 -1.01489949e+00 4.87864017e-01
8.19556236e-01 3.26985270e-02 1.26550093e-01 -6.38401648e-03
1.07730605e-01 8.63262564e-02 5.56658626e-01 -9.16249454e-01
2.19006732e-01 3.15991342e-02 9.09221411e-01 -7.36161649e-01
1.54398844e-01 -4.01207089e-01 1.58475742e-01 6.37665808e-01
-3.37651581e-01 -4.63153124e-01 4.87932920e-01 6.21964037e-01
-3.14685106e-01 1.24338344e-01 7.56945491e-01 8.21492672e-02
-5.99795222e-01 5.02045810e-01 -4.12516922e-01 -4.98265624e-01
1.19291687e+00 -3.90671194e-01 -8.65074873e-01 -3.72895226e-02
-5.87915301e-01 3.95306230e-01 1.84829921e-01 1.92850202e-01
6.06641710e-01 -1.08756959e+00 -1.08254087e+00 2.39191920e-01
5.23295403e-01 6.68686628e-02 6.85606778e-01 1.25111449e+00
-4.10670519e-01 5.71522892e-01 1.09253295e-01 -7.33470798e-01
-1.17006814e+00 3.75953704e-01 6.56670809e-01 -7.00670540e-01
-2.81333297e-01 1.35440648e+00 2.76237786e-01 -4.10063237e-01
1.60049766e-01 -4.80017841e-01 -1.05030566e-01 -8.43556970e-02
6.11637950e-01 1.43241659e-01 5.34912050e-01 -2.73111075e-01
-5.88588238e-01 3.26714873e-01 -4.56675440e-01 8.75931025e-01
1.37991047e+00 3.57068002e-01 -4.25824523e-01 2.93948919e-01
1.31567597e+00 -3.00147355e-01 -9.23382580e-01 5.00175469e-02
-1.13094952e-02 -8.24004039e-02 1.00417726e-01 -1.23741889e+00
-1.12255692e+00 5.46860099e-01 8.35487187e-01 -2.32878059e-01
1.25119293e+00 8.44673887e-02 5.82392752e-01 -1.84333414e-01
1.15732022e-01 -6.02028608e-01 -2.62654155e-01 1.58273429e-01
7.23874509e-01 -1.30471218e+00 -4.31472668e-04 -2.68321276e-01
-1.98225841e-01 1.27658367e+00 7.35055566e-01 3.85010809e-01
4.18361038e-01 3.29225123e-01 -2.03862548e-01 -3.27079147e-01
-1.30095625e+00 7.91315064e-02 3.06778640e-01 3.20584893e-01
9.34109628e-01 2.12420955e-01 -3.48204911e-01 1.01619291e+00
2.65353382e-01 7.01018646e-02 3.82905960e-01 5.78814566e-01
-3.69931906e-01 -7.91501105e-01 -1.91927209e-01 1.12298572e+00
-7.62537062e-01 -1.20267108e-01 -5.57348251e-01 1.04191434e+00
1.88701227e-01 5.60236871e-01 2.03843698e-01 -3.15952748e-01
9.93740931e-02 2.76979417e-01 1.19253539e-01 -3.13818187e-01
-7.97627211e-01 1.33557856e-01 1.27537876e-01 -4.13047820e-01
-1.67307392e-01 -5.33327639e-01 -1.15788543e+00 4.33673104e-03
-4.49635655e-01 -2.40808040e-01 3.91923040e-01 7.35514998e-01
7.51731813e-01 8.70962501e-01 2.08523706e-01 -3.80667657e-01
-1.02929361e-01 -1.17254305e+00 -5.97479582e-01 2.38527045e-01
2.90212065e-01 -5.35447240e-01 -1.77923054e-01 -2.15391338e-01] | [15.207609176635742, -2.8497698307037354] |
c534b163-ab07-4af7-89e2-47c202f614d7 | how-to-train-bert-with-an-academic-budget | 2104.07705 | null | https://arxiv.org/abs/2104.07705v2 | https://arxiv.org/pdf/2104.07705v2.pdf | How to Train BERT with an Academic Budget | While large language models a la BERT are used ubiquitously in NLP, pretraining them is considered a luxury that only a few well-funded industry labs can afford. How can one train such models with a more modest budget? We present a recipe for pretraining a masked language model in 24 hours using a single low-end deep learning server. We demonstrate that through a combination of software optimizations, design choices, and hyperparameter tuning, it is possible to produce models that are competitive with BERT-base on GLUE tasks at a fraction of the original pretraining cost. | ['Omer Levy', 'Moshe Berchansky', 'Peter Izsak'] | 2021-04-15 | null | https://aclanthology.org/2021.emnlp-main.831 | https://aclanthology.org/2021.emnlp-main.831.pdf | emnlp-2021-11 | ['linguistic-acceptability'] | ['natural-language-processing'] | [-3.22150409e-01 3.10676903e-01 -1.89702734e-01 -6.69296682e-01
-1.28348780e+00 -7.38245666e-01 2.89374858e-01 -1.05233416e-01
-6.06463730e-01 6.61517620e-01 6.14991300e-02 -7.93038368e-01
4.04918164e-01 -5.38401663e-01 -9.29484546e-01 -1.46229282e-01
-9.05882716e-02 9.63480532e-01 1.14491813e-01 -2.78816968e-01
-2.86571264e-01 5.12923121e-01 -1.08336115e+00 5.21426678e-01
6.36925638e-01 5.63622236e-01 5.03962040e-02 7.02789247e-01
-3.91429007e-01 9.32864606e-01 -5.98602057e-01 -4.71358359e-01
6.43486023e-01 -5.46594970e-02 -8.58490944e-01 -2.09702596e-01
3.79644364e-01 -2.91740030e-01 1.27672460e-02 5.88668942e-01
5.68347394e-01 -1.21109679e-01 1.47797734e-01 -9.37064350e-01
-4.59600031e-01 1.05513895e+00 -2.99917489e-01 1.82992205e-01
5.12432307e-02 4.17295694e-01 9.97126579e-01 -7.65903533e-01
3.88448954e-01 1.16934514e+00 7.38292217e-01 3.23559970e-01
-1.49639738e+00 -8.18799555e-01 2.73947001e-01 -4.38729495e-01
-1.46219802e+00 -8.32759023e-01 4.28863078e-01 -3.06293458e-01
1.59408391e+00 -2.93099303e-02 2.85890996e-01 1.14521205e+00
2.07931712e-01 7.32282221e-01 1.02053821e+00 -4.63262528e-01
1.34861991e-01 3.71871740e-01 1.36466488e-01 8.96274626e-01
1.59025639e-01 -1.14855558e-01 -5.37901044e-01 -2.21443310e-01
5.39066911e-01 -2.73878247e-01 1.38806939e-01 -1.17252618e-01
-1.22607613e+00 7.92467177e-01 2.35973716e-01 4.32207733e-01
-4.55716997e-02 6.29881024e-01 2.56733924e-01 6.19667113e-01
5.58541656e-01 8.46665978e-01 -1.03701210e+00 -2.93516099e-01
-1.06117749e+00 3.02268207e-01 1.16013348e+00 1.11847520e+00
8.92397285e-01 2.12193474e-01 3.36147904e-01 5.33466876e-01
1.79438621e-01 2.27031678e-01 4.16027486e-01 -8.29500318e-01
4.60454315e-01 3.14239889e-01 1.52612478e-01 -4.35272634e-01
-3.57495964e-01 -5.68113983e-01 -3.09422195e-01 8.99390057e-02
4.57575828e-01 -7.04768240e-01 -9.05103564e-01 1.55655396e+00
9.89762917e-02 2.06448749e-01 -4.93696034e-02 7.07996666e-01
4.75073487e-01 6.57168150e-01 4.84659672e-02 1.64564520e-01
1.11587858e+00 -1.22992623e+00 -2.16321915e-01 -7.81133294e-01
9.53731477e-01 -8.65631104e-01 1.60041785e+00 6.31137192e-01
-1.35558820e+00 -2.76515275e-01 -1.12630403e+00 -5.28367579e-01
-3.41707379e-01 -2.19876729e-02 1.30879760e+00 8.04442942e-01
-1.20658433e+00 5.24640918e-01 -1.12943327e+00 -2.77501047e-01
4.01632898e-02 4.97078210e-01 -3.30910087e-01 2.08320692e-01
-9.21565056e-01 9.50900376e-01 3.57356310e-01 8.81659836e-02
-1.05568409e+00 -8.63547623e-01 -3.99536133e-01 1.99341953e-01
5.34123957e-01 -7.14964807e-01 1.52826965e+00 -7.20812976e-01
-1.74714947e+00 8.33943367e-01 -1.41145304e-01 -5.46935558e-01
4.67306942e-01 -3.20125312e-01 -2.80698091e-01 -1.52943760e-01
-3.18719059e-01 7.07133293e-01 7.77544320e-01 -9.19239581e-01
-4.65729475e-01 -4.68561687e-02 3.33624691e-01 3.67251858e-02
-4.23868865e-01 6.39621139e-01 -3.85991424e-01 -3.90794903e-01
-9.90556777e-02 -7.70687222e-01 -5.14476955e-01 -2.43379623e-01
-1.04749531e-01 -9.33784097e-02 3.39603961e-01 -6.33986235e-01
9.76432562e-01 -2.03055334e+00 -1.50711834e-01 1.05979376e-01
2.13433579e-01 7.15142041e-02 -2.86972880e-01 4.19118792e-01
2.19788536e-01 5.95389903e-01 6.55169860e-02 -5.64124405e-01
2.76351273e-01 6.79402128e-02 -5.83968401e-01 2.55878985e-01
7.84976631e-02 9.60088313e-01 -7.18191385e-01 -3.20312589e-01
-2.20927581e-01 2.22751111e-01 -8.91309321e-01 2.16316998e-01
-7.81861007e-01 1.38182804e-01 -2.45949000e-01 5.52550495e-01
3.40964377e-01 -4.58793759e-01 3.32934350e-01 3.57799530e-01
-5.44623025e-02 7.66404748e-01 -1.07127476e+00 1.88059509e+00
-7.14896441e-01 4.55362350e-01 5.46010911e-01 -5.02160728e-01
7.26623058e-01 4.79111433e-01 1.94833383e-01 -6.20266870e-02
-3.23611195e-03 4.14341241e-01 7.93545321e-02 -3.40688348e-01
2.85651654e-01 -3.49221051e-01 -7.52607435e-02 7.23974705e-01
1.77409798e-01 -4.55855608e-01 1.27210602e-01 8.55106413e-02
1.28156734e+00 9.20687765e-02 -7.98362046e-02 -2.08283290e-01
-1.35068133e-01 1.49166912e-01 6.12163603e-01 8.73095989e-01
-1.11591808e-01 3.57085168e-01 6.28664851e-01 -7.51834571e-01
-1.24555767e+00 -8.82629514e-01 1.38990462e-01 1.57932270e+00
-6.74679041e-01 -5.39121091e-01 -7.65271068e-01 -6.62060738e-01
-2.63689570e-02 6.28833175e-01 -3.93505096e-02 2.10448831e-01
-6.68007970e-01 -6.71821058e-01 8.53105962e-01 4.92427319e-01
2.87639678e-01 -8.62121046e-01 -2.16386810e-01 2.63411969e-01
2.52477020e-01 -1.18388057e+00 -5.10920584e-01 6.59373760e-01
-8.41696322e-01 -3.21251899e-01 -3.83877575e-01 -1.01270342e+00
3.59876126e-01 2.22955346e-01 1.46295106e+00 2.24312931e-01
-1.33434206e-01 -1.04306348e-01 1.08064972e-01 -4.59769398e-01
-4.06966895e-01 5.90371966e-01 6.53765583e-03 -4.93769944e-01
6.29634082e-01 -8.58009994e-01 -1.98011771e-01 -9.91903152e-03
-6.87087774e-01 1.74296409e-01 7.95980155e-01 7.01644838e-01
1.62322283e-01 -2.68477462e-02 3.87520343e-01 -1.17619777e+00
7.17447937e-01 -4.20626223e-01 -9.70231831e-01 3.46715003e-01
-8.14745486e-01 1.13224514e-01 6.94665194e-01 -4.18702453e-01
-8.75601709e-01 1.88978985e-01 -2.68891037e-01 -2.97798753e-01
-2.56528676e-01 6.30071759e-01 -1.09717451e-01 -3.51410180e-01
8.86772037e-01 -2.08257154e-01 -1.78792149e-01 -7.71878958e-01
5.90981722e-01 6.01667225e-01 1.68116212e-01 -9.99698579e-01
7.32503831e-01 5.34940660e-02 -5.13275623e-01 -5.09176493e-01
-8.28476071e-01 -2.64631122e-01 -4.58917767e-01 5.02621889e-01
6.93556309e-01 -1.32996690e+00 -6.84605539e-01 1.74945686e-02
-1.15619934e+00 -8.87823164e-01 -9.21954215e-02 4.30773824e-01
-2.72566974e-01 -2.34293542e-03 -7.62963831e-01 -4.33244556e-01
-3.04077744e-01 -1.20465291e+00 9.34744775e-01 -1.55921817e-01
-2.45948672e-01 -9.49830413e-01 1.52923577e-02 5.95823228e-01
6.80518448e-01 -2.71200985e-01 1.16832876e+00 -9.45193172e-01
-9.05403793e-01 -2.61155158e-01 -1.19857542e-01 3.99192035e-01
-3.93922001e-01 4.65052724e-02 -1.03744113e+00 -4.03160870e-01
-4.34035212e-02 -6.46669805e-01 5.05698860e-01 5.39923348e-02
1.06765711e+00 -4.99410659e-01 -1.74804047e-01 9.35379446e-01
1.33063829e+00 -1.29093423e-01 1.11758180e-01 1.95966661e-01
5.86524367e-01 1.78529128e-01 -6.66659102e-02 5.65570779e-04
2.92663217e-01 4.47505414e-01 -5.50762862e-02 -1.59764498e-01
1.57775924e-01 -3.58948946e-01 4.91689622e-01 8.89350891e-01
1.91942960e-01 -7.05413520e-02 -1.48372793e+00 5.34952998e-01
-1.59673798e+00 -6.21824443e-01 3.52408051e-01 1.93935764e+00
1.18649590e+00 5.64837873e-01 5.62076457e-02 -5.33710599e-01
2.18925416e-01 1.15952201e-01 -4.43719447e-01 -6.94430411e-01
9.63698700e-02 5.00189722e-01 6.25459492e-01 8.33211839e-01
-8.02900970e-01 1.52836120e+00 8.55287075e+00 7.81660736e-01
-1.32389569e+00 4.06924516e-01 7.00233996e-01 -4.92048949e-01
-6.36979461e-01 3.97203863e-01 -1.09703577e+00 1.79771289e-01
1.59856522e+00 -1.98993057e-01 9.28631008e-01 1.31328392e+00
2.86909699e-01 3.52548301e-01 -1.27850807e+00 7.24706769e-01
-1.51288435e-01 -1.46023118e+00 -2.50101745e-01 7.05536678e-02
7.36902535e-01 6.76309049e-01 -7.05536082e-02 7.81257629e-01
9.22418535e-01 -1.36305892e+00 6.69659495e-01 1.70220330e-01
3.06165546e-01 -6.24732971e-01 5.10798454e-01 6.69761717e-01
-6.08373702e-01 -1.70970766e-03 -5.42010665e-01 -3.09723675e-01
2.12076083e-01 6.08946562e-01 -1.16112983e+00 5.87292537e-02
4.92126018e-01 4.27574031e-02 -6.37035370e-01 8.81697714e-01
-3.67013425e-01 1.04045856e+00 -6.12529814e-01 -2.67107487e-02
3.94751251e-01 2.05649193e-02 1.75347120e-01 1.34179294e+00
4.37319800e-02 -1.07792594e-01 3.86376649e-01 8.86325777e-01
-4.36811179e-01 1.36111468e-01 -5.24008751e-01 -5.92099726e-01
4.96834129e-01 1.31930697e+00 -5.65715909e-01 -2.25092515e-01
-5.64002573e-01 7.59358823e-01 8.10242116e-01 4.53216434e-01
-8.80539179e-01 -3.18223923e-01 6.97100401e-01 8.18577558e-02
-2.79108025e-02 -6.69232011e-01 -4.99996394e-01 -1.41383231e+00
5.22002913e-02 -1.35575068e+00 4.12687808e-02 -8.33874583e-01
-1.19916725e+00 7.27398634e-01 -1.78549647e-01 -4.66968477e-01
-3.96743774e-01 -8.07650208e-01 -3.81082565e-01 1.08093154e+00
-1.49428546e+00 -1.27679706e+00 1.27448425e-01 2.18051806e-01
3.09658557e-01 -3.04145068e-01 1.12711346e+00 2.35907704e-01
-6.01135731e-01 6.79886281e-01 1.03798360e-02 -2.10672915e-02
7.77778685e-01 -1.24675322e+00 8.52533638e-01 7.88827240e-01
5.60830176e-01 1.16660655e+00 7.95770347e-01 -2.88773000e-01
-1.64405954e+00 -7.11019218e-01 1.26463902e+00 -5.95097959e-01
1.02507222e+00 -1.00081313e+00 -6.86931670e-01 1.30330288e+00
3.87236357e-01 7.83791915e-02 7.65513301e-01 5.40588975e-01
-5.07034004e-01 -2.71501690e-01 -8.67580891e-01 7.23066270e-01
9.34664190e-01 -7.90807545e-01 -4.00156617e-01 8.99791241e-01
1.28580523e+00 -5.99716246e-01 -9.34433639e-01 1.78020224e-01
3.75118166e-01 -5.87068081e-01 8.41837108e-01 -1.02023065e+00
1.14382453e-01 -8.46276805e-02 -3.31317484e-02 -1.00988984e+00
-2.25299343e-01 -1.38600135e+00 -6.36361400e-03 1.05394125e+00
9.56437767e-01 -5.44830799e-01 1.09066200e+00 1.07104385e+00
-1.63557068e-01 -8.09891939e-01 -6.05477035e-01 -8.17411900e-01
5.78690767e-01 -6.37997150e-01 8.10241818e-01 6.99925840e-01
8.54993910e-02 6.04359865e-01 -2.95585334e-01 1.32140338e-01
1.24277242e-01 -8.13503563e-02 1.05751073e+00 -8.34299564e-01
-8.68582308e-01 -3.18614990e-01 1.06241889e-01 -1.15920019e+00
3.48013431e-01 -1.09410512e+00 1.21720463e-01 -1.25245023e+00
8.60065594e-02 -6.15953267e-01 -2.98585027e-01 9.69496012e-01
1.83520436e-01 1.53850056e-02 2.47513235e-01 1.34783685e-01
-4.12461370e-01 6.15086108e-02 7.56080210e-01 4.15922813e-02
-2.51762480e-01 -2.72715807e-01 -1.12319279e+00 8.24890554e-01
7.52627373e-01 -4.65303123e-01 -4.01516259e-01 -1.20243394e+00
6.37438118e-01 -2.56493270e-01 -5.15114367e-02 -9.73259985e-01
2.34107003e-01 -3.42443496e-01 1.45514846e-01 -1.98004961e-01
3.50334376e-01 -7.99048007e-01 -4.39332984e-03 -4.29277420e-02
-5.87680638e-01 2.21257553e-01 4.50948417e-01 1.52554378e-01
-8.05757940e-03 -3.76407653e-01 5.54993749e-01 -4.27527338e-01
-4.06226218e-01 2.31391713e-01 -3.64618391e-01 1.55798674e-01
7.40719795e-01 3.28904629e-01 -3.45139444e-01 -4.77150559e-01
-5.63273072e-01 1.12196855e-01 4.95126456e-01 2.43608519e-01
3.63495462e-02 -7.60498643e-01 -5.34935057e-01 2.50889927e-01
-3.14476252e-01 1.42036173e-02 -2.94869125e-01 5.84358752e-01
-8.62013280e-01 6.14231825e-01 2.77188629e-01 -2.10488960e-01
-8.45457971e-01 6.46696448e-01 4.21048671e-01 -5.68732440e-01
-6.70359135e-01 1.26314175e+00 -6.06247336e-02 -7.90848017e-01
3.00228179e-01 -5.65887690e-01 6.98101759e-01 -4.21517015e-01
4.88636613e-01 -1.72582656e-01 3.45170885e-01 5.75219728e-02
-2.99454451e-01 5.03042005e-02 -3.27877700e-01 -4.64222997e-01
1.46187401e+00 2.03829691e-01 -1.06372833e-01 3.95563155e-01
1.00977170e+00 2.93706059e-01 -1.29751170e+00 -1.22495897e-01
2.29833081e-01 -1.32561952e-01 3.19465846e-01 -1.05885327e+00
-6.89131796e-01 1.02225411e+00 6.19035102e-02 -1.26843611e-02
7.93003380e-01 -2.37251520e-01 9.81423318e-01 9.69816148e-01
6.83626354e-01 -1.09608102e+00 -3.39115262e-02 6.88366532e-01
7.57459342e-01 -1.06382847e+00 -1.21030778e-01 -1.88405350e-01
-5.74465215e-01 8.20048511e-01 4.34967160e-01 -2.96822995e-01
7.69788682e-01 1.02051532e+00 2.86288679e-01 8.13956782e-02
-1.25148785e+00 5.45820929e-02 -1.71546236e-01 2.67706513e-01
8.77212286e-01 2.20181905e-02 -5.73204551e-03 8.10602486e-01
-5.95008373e-01 2.31919050e-01 2.38588467e-01 9.67270434e-01
-6.53989613e-01 -1.24776411e+00 -1.29610792e-01 1.74989015e-01
-6.91751063e-01 -5.26356280e-01 -3.33156437e-01 7.78438628e-01
-4.27878760e-02 8.01452935e-01 -8.72673914e-02 -2.16727689e-01
6.03624880e-02 5.68614483e-01 4.30867255e-01 -1.22107446e+00
-9.72446263e-01 2.38021463e-01 5.05788088e-01 -6.29116654e-01
2.89688051e-01 -2.80205488e-01 -1.22188222e+00 -6.84067547e-01
-1.19025461e-01 1.79353565e-01 8.28931749e-01 1.05373347e+00
6.66145086e-01 1.36215433e-01 2.47922868e-01 -7.31893241e-01
-9.30276394e-01 -9.56964135e-01 -4.58147556e-01 -1.56335190e-01
1.18895084e-01 -3.26962993e-02 -4.40218240e-01 -1.03357555e-02] | [10.583257675170898, 8.187274932861328] |
04f9421f-3651-4e5e-905b-8026de0c0c5d | neural-machine-translation-with-contrastive | 2212.03140 | null | https://arxiv.org/abs/2212.03140v1 | https://arxiv.org/pdf/2212.03140v1.pdf | Neural Machine Translation with Contrastive Translation Memories | Retrieval-augmented Neural Machine Translation models have been successful in many translation scenarios. Different from previous works that make use of mutually similar but redundant translation memories~(TMs), we propose a new retrieval-augmented NMT to model contrastively retrieved translation memories that are holistically similar to the source sentence while individually contrastive to each other providing maximal information gains in three phases. First, in TM retrieval phase, we adopt a contrastive retrieval algorithm to avoid redundancy and uninformativeness of similar translation pieces. Second, in memory encoding stage, given a set of TMs we propose a novel Hierarchical Group Attention module to gather both local context of each TM and global context of the whole TM set. Finally, in training phase, a Multi-TM contrastive learning objective is introduced to learn salient feature of each TM with respect to target sentence. Experimental results show that our framework obtains improvements over strong baselines on the benchmark datasets. | ['Rui Yan', 'Dongyan Zhao', 'Lemao Liu', 'Shen Gao', 'Xin Cheng'] | 2022-12-06 | null | null | null | null | ['nmt'] | ['computer-code'] | [ 4.71705943e-01 -1.76321015e-01 -2.68464327e-01 -2.35795796e-01
-1.31328452e+00 -1.76838428e-01 8.58671367e-01 5.49222566e-02
-3.65333974e-01 7.90791154e-01 4.67696816e-01 -3.24685961e-01
1.72840655e-01 -5.81049442e-01 -1.04492962e+00 -7.93706894e-01
3.54124039e-01 7.65291512e-01 -5.25072403e-02 -4.96191531e-01
5.55633128e-01 5.65920100e-02 -1.07489622e+00 7.47935474e-01
1.10956287e+00 5.87542355e-01 7.12439775e-01 1.15030855e-01
-2.24494606e-01 6.94500446e-01 -6.25748575e-01 -2.90932506e-01
8.65475982e-02 -8.45379591e-01 -1.01912129e+00 -1.72126405e-02
5.56799948e-01 -2.49399573e-01 -4.34509844e-01 9.98372555e-01
7.97594309e-01 5.18657506e-01 5.55772007e-01 -4.62152451e-01
-1.27129745e+00 8.41700375e-01 -7.86465883e-01 6.97846174e-01
2.28347689e-01 -3.38636525e-02 9.58698213e-01 -1.59734249e+00
7.81044066e-01 1.19791830e+00 -7.50312582e-02 5.08055925e-01
-1.14414608e+00 -7.21209705e-01 3.60739678e-01 3.48232597e-01
-1.20973289e+00 -6.93508685e-01 8.36099625e-01 2.51257092e-01
1.54091704e+00 4.12475556e-01 4.07503068e-01 1.12448347e+00
6.82982802e-01 9.77392435e-01 1.22455883e+00 -6.84463501e-01
-1.25692308e-01 9.30538699e-02 1.98490322e-01 3.36495847e-01
-8.64530206e-02 -1.34606272e-01 -8.84137332e-01 -7.90073816e-03
4.57474202e-01 8.04599971e-02 -3.93449187e-01 -1.25123560e-01
-1.46946013e+00 6.21763945e-01 6.65779293e-01 5.67080855e-01
-6.21125579e-01 1.18731081e-01 1.96627036e-01 8.36293519e-01
7.98226476e-01 4.45803821e-01 -2.27993712e-01 3.66299778e-01
-1.10255837e+00 1.26712531e-01 1.06442034e-01 1.16812277e+00
1.08376110e+00 -2.78625250e-01 -6.76959515e-01 9.13582087e-01
1.76419497e-01 6.70400739e-01 1.03827941e+00 -3.20572346e-01
1.14717579e+00 6.07619941e-01 -5.46162017e-02 -8.47068548e-01
1.60136670e-01 -8.59524608e-01 -9.63492095e-01 -7.40157068e-01
-4.96785760e-01 3.91230017e-01 -9.47490096e-01 1.89262736e+00
-3.89962085e-02 1.70980722e-01 1.83902368e-01 1.03284645e+00
7.89254129e-01 9.95180130e-01 -1.07657030e-01 -6.90155923e-01
1.12444162e+00 -1.45067263e+00 -7.55892515e-01 -5.87454259e-01
7.06330180e-01 -9.35587049e-01 1.15377939e+00 -2.97346413e-01
-1.66373670e+00 -5.48896313e-01 -1.11775541e+00 -4.36727375e-01
-2.87894785e-01 -7.23293945e-02 2.92245716e-01 -6.50549009e-02
-1.26905942e+00 6.08542919e-01 -6.35626733e-01 -3.19757223e-01
3.26729119e-01 5.94978690e-01 -2.47241974e-01 -4.29790795e-01
-1.39043462e+00 1.32721663e+00 2.34873742e-01 3.51316631e-01
-9.47486460e-01 -3.11337352e-01 -6.45985365e-01 1.49314180e-01
1.92298904e-01 -1.12546813e+00 1.26161683e+00 -1.24254405e+00
-1.26647496e+00 8.06564927e-01 -7.23161280e-01 -3.48825365e-01
-3.04745529e-02 -3.50743622e-01 -2.55559951e-01 2.32845530e-01
2.09259912e-01 7.81325161e-01 1.02051163e+00 -1.11094391e+00
-2.55699515e-01 -5.14298975e-01 -2.95955837e-01 9.29726720e-01
-4.50323910e-01 2.03135580e-01 -5.15368164e-01 -7.47257531e-01
5.43541372e-01 -7.21131206e-01 -1.97109818e-01 -6.76891267e-01
-1.45581558e-01 -3.87969576e-02 5.56463718e-01 -7.82428205e-01
1.26597965e+00 -1.89832449e+00 9.77945626e-01 -2.67388374e-01
5.76653630e-02 1.19721092e-01 -5.65868556e-01 5.63927412e-01
3.16020809e-02 2.25224178e-02 -2.54949212e-01 -6.51300013e-01
-1.10675544e-01 3.63752469e-02 -7.48196125e-01 1.90565005e-01
6.50778472e-01 1.50053191e+00 -9.29860771e-01 -5.85870862e-01
-2.69487768e-01 2.99992234e-01 -2.48649180e-01 1.97384864e-01
-1.85349613e-01 5.43508768e-01 -5.15357375e-01 6.34996057e-01
6.32369280e-01 -2.72298455e-01 1.91936120e-01 1.56746954e-01
2.49055207e-01 8.76804233e-01 -2.17516199e-01 2.13046551e+00
-4.02567059e-01 4.37246948e-01 -3.02848876e-01 -1.02348042e+00
9.12684560e-01 3.52198780e-01 3.86164226e-02 -1.37962258e+00
1.23457491e-01 3.78439784e-01 -7.38045648e-02 -1.71889737e-01
1.04145956e+00 -2.91164011e-01 5.05484156e-02 9.01264369e-01
2.46889159e-01 2.70484239e-01 -2.92363882e-01 3.50333184e-01
9.10605788e-01 2.57909566e-01 1.28635913e-01 -2.69348115e-01
5.32579303e-01 9.25156772e-02 4.56352562e-01 6.05040967e-01
-1.13183729e-01 5.35236657e-01 -2.26188973e-01 -2.70247817e-01
-8.52948070e-01 -1.02428412e+00 4.00761843e-01 1.41093791e+00
5.30310571e-01 -2.01730549e-01 -5.06649196e-01 -4.04840618e-01
-5.84687412e-01 6.04505897e-01 -5.59948683e-01 -7.40524709e-01
-1.23606098e+00 -9.70315576e-01 1.02412544e-01 5.64740121e-01
6.61350429e-01 -1.36506438e+00 -4.14517462e-01 2.11529553e-01
-8.55864882e-01 -7.46027112e-01 -8.95010531e-01 3.10910046e-01
-1.48837757e+00 -1.80208519e-01 -7.82441556e-01 -1.31456554e+00
8.39107037e-01 1.01483595e+00 1.47167706e+00 2.48841807e-01
3.12454909e-01 -9.25979838e-02 -4.45533216e-01 2.82414500e-02
-2.59951591e-01 3.59333783e-01 -1.80905491e-01 -1.91106483e-01
4.44889486e-01 -5.96420050e-01 -6.44193411e-01 1.26903132e-01
-7.80772686e-01 2.36742571e-01 1.21872914e+00 1.06809664e+00
8.76151681e-01 -4.96730894e-01 7.11824000e-01 -4.79188949e-01
8.40577960e-01 -6.97571456e-01 4.41203490e-02 5.81920147e-01
-5.89237094e-01 6.58282787e-02 3.64824921e-01 -6.25822783e-01
-1.00427532e+00 -4.85935837e-01 4.63413864e-01 -5.08217573e-01
4.38208848e-01 8.37871611e-01 -2.19996467e-01 2.81914949e-01
4.11813051e-01 1.01383114e+00 -6.37206733e-02 -3.72139484e-01
3.67460340e-01 4.81850445e-01 3.13938558e-01 -6.84399009e-01
5.56764364e-01 1.04388744e-02 -2.57771224e-01 -2.53957272e-01
-4.83379126e-01 -3.08528870e-01 -7.13113248e-01 4.74839844e-02
4.60747093e-01 -1.08506691e+00 1.60352543e-01 1.35267034e-01
-1.11632252e+00 -9.13571045e-02 -5.17330989e-02 3.86031121e-01
-6.77554011e-01 4.74572450e-01 -9.00956511e-01 -5.04091084e-01
-1.00868976e+00 -1.24776864e+00 1.24297786e+00 -2.53440645e-02
6.21740706e-03 -6.67882681e-01 1.06647275e-01 5.48774719e-01
5.85892439e-01 -4.01439905e-01 1.19719791e+00 -5.95651567e-01
-8.99887085e-01 -2.45099869e-02 -1.58217013e-01 5.63807935e-02
8.46323445e-02 -6.62253141e-01 -6.88482702e-01 -7.70303786e-01
2.82505155e-01 -3.86639267e-01 1.29042745e+00 5.42634726e-02
4.41912979e-01 -5.46170235e-01 -4.52178299e-01 1.26583636e-01
1.21413720e+00 9.79280993e-02 7.70891666e-01 2.80669302e-01
4.08239275e-01 5.00773668e-01 9.02485669e-01 -3.76488507e-01
2.95790285e-01 6.60821021e-01 2.81754106e-01 1.24014221e-01
-1.75385430e-01 -2.28925154e-01 8.63998592e-01 1.55593050e+00
-2.82398090e-02 -2.27871522e-01 -4.44987804e-01 7.47197390e-01
-2.03698897e+00 -9.95716393e-01 2.61121780e-01 2.39678502e+00
8.55756998e-01 1.60936676e-02 -2.30886206e-01 -4.60841119e-01
7.80895770e-01 1.26353055e-01 -4.45747495e-01 -5.23567259e-01
-4.78663296e-01 5.02805412e-01 -6.71918914e-02 4.09634501e-01
-7.40565002e-01 1.45031571e+00 5.80396223e+00 7.96812415e-01
-1.11584139e+00 4.90711838e-01 4.90746856e-01 -5.42323887e-01
-5.60116768e-01 1.57613531e-01 -4.28103536e-01 1.92352340e-01
9.53169286e-01 -3.64199638e-01 6.06695950e-01 1.92540243e-01
-3.37145105e-02 1.67764381e-01 -1.14349186e+00 5.35113573e-01
3.79527837e-01 -1.28719163e+00 7.04069674e-01 7.71874115e-02
7.66782224e-01 1.78809896e-01 5.72184384e-01 7.15734184e-01
-2.22431391e-01 -9.59241092e-01 5.35631716e-01 5.19935668e-01
4.17143822e-01 -8.52005005e-01 7.14222252e-01 5.35613835e-01
-9.15113807e-01 1.38880193e-01 -7.80214369e-01 -9.10126269e-02
-1.64269269e-01 3.59103344e-02 -6.13187790e-01 9.81994629e-01
3.63783896e-01 7.28344142e-01 -6.48758531e-01 5.73716283e-01
2.72344369e-02 2.18154743e-01 3.19873989e-01 -1.01045899e-01
4.36026245e-01 -2.02339455e-01 6.66053712e-01 1.04911196e+00
3.55000168e-01 -5.12670306e-03 -5.61058149e-02 9.11626339e-01
-4.13592279e-01 6.00427926e-01 -6.62084818e-01 4.70549334e-03
6.51693463e-01 1.03398275e+00 -3.29116225e-01 -6.00752771e-01
-2.82229483e-01 1.57235765e+00 7.54765272e-01 5.81979096e-01
-5.50255835e-01 6.01651371e-02 3.43207896e-01 -2.22574338e-01
3.91767085e-01 -7.50655010e-02 -1.44848704e-01 -1.39593935e+00
4.65081483e-01 -9.39215302e-01 1.65451020e-01 -7.82955706e-01
-1.03844571e+00 8.47971022e-01 -2.33787209e-01 -1.49047852e+00
-2.58496940e-01 -5.05704470e-02 -6.31691396e-01 1.35860980e+00
-1.65448380e+00 -1.57593524e+00 4.17776734e-01 6.05179012e-01
9.76061165e-01 -2.67751306e-01 9.93802965e-01 2.62047887e-01
-6.56150341e-01 7.37201631e-01 1.28691062e-01 -1.76038563e-01
9.15679455e-01 -8.68295968e-01 3.61304164e-01 1.03963053e+00
1.94687992e-01 1.36704230e+00 3.71096700e-01 -7.64215052e-01
-1.80050325e+00 -1.27745140e+00 1.48213995e+00 -4.33464408e-01
3.12904537e-01 -2.46105522e-01 -1.10819471e+00 9.60631013e-01
8.39768112e-01 -3.85387063e-01 6.63951635e-01 6.36558011e-02
-4.56090659e-01 8.80630016e-02 -7.91913569e-01 9.80234146e-01
1.01155150e+00 -7.07955122e-01 -1.16765034e+00 3.69942456e-01
1.04926586e+00 -3.38700086e-01 -6.56474590e-01 3.97890806e-01
2.83374906e-01 -4.06708330e-01 8.74207437e-01 -8.93278003e-01
7.13079453e-01 -2.31844991e-01 -4.72479582e-01 -1.42716801e+00
-6.23325646e-01 -4.91527796e-01 -3.77764165e-01 9.52971280e-01
5.10278523e-01 -5.25441349e-01 2.48610944e-01 6.43075928e-02
-6.11820161e-01 -9.29549694e-01 -1.13924158e+00 -7.10047841e-01
6.68009579e-01 2.81918257e-01 5.73275268e-01 1.08081925e+00
3.98210347e-01 1.08673000e+00 -5.68297982e-01 3.47420163e-02
3.33500683e-01 7.48577952e-01 4.47618097e-01 -7.01644301e-01
-4.20331389e-01 -6.57915056e-01 1.72497734e-01 -1.29815841e+00
2.91298091e-01 -1.37095964e+00 1.49457067e-01 -1.53367794e+00
9.42592144e-01 1.10658832e-01 -8.10941637e-01 2.55048126e-01
-7.01788485e-01 3.29791814e-01 1.15358904e-01 8.04209888e-01
-8.77747059e-01 9.63739872e-01 1.57062650e+00 -4.76113200e-01
-1.75387174e-01 -4.41641539e-01 -8.91035497e-01 -1.64694697e-01
6.69740915e-01 -4.62293565e-01 -5.07673264e-01 -1.11091578e+00
2.14196980e-01 1.95527583e-01 1.37251183e-01 -5.17715693e-01
3.71251673e-01 -5.16934171e-02 3.33800733e-01 -9.71973479e-01
4.03861344e-01 -5.31959593e-01 -8.67079645e-02 3.89860779e-01
-5.57494640e-01 5.46012223e-01 1.38232484e-01 5.25774062e-01
-3.34906429e-01 7.02601373e-02 3.50495934e-01 -4.05301481e-01
-3.91850263e-01 2.51306057e-01 -4.45279926e-01 -2.62936980e-01
5.05320609e-01 9.50225741e-02 -5.94878137e-01 -2.20977619e-01
-5.48805952e-01 3.29884976e-01 3.90072525e-01 7.56468654e-01
1.00936687e+00 -1.60043800e+00 -9.79445398e-01 -7.47209638e-02
4.24932331e-01 -1.34281605e-01 2.38292530e-01 1.07458925e+00
1.09464519e-01 5.94617963e-01 -3.12942266e-01 -4.33600217e-01
-1.05654502e+00 1.05886853e+00 2.07830533e-01 -5.08169591e-01
-5.22783697e-01 7.11306155e-01 2.74518222e-01 -3.67330909e-01
-1.28233239e-01 -2.05469459e-01 -5.45602404e-02 -1.22644484e-01
7.18412519e-01 2.35368172e-03 3.93728822e-01 -9.49053526e-01
-2.92898178e-01 4.38680232e-01 -8.04791987e-01 -4.09079075e-01
1.16035497e+00 -4.65009779e-01 -6.05891168e-01 5.36670685e-01
1.14424992e+00 -2.50237465e-01 -5.13374269e-01 -8.16590726e-01
7.97164142e-02 -3.06024939e-01 2.32564583e-02 -9.36392188e-01
-9.40598845e-01 1.01740468e+00 3.97826165e-01 -6.45376325e-01
1.57692909e+00 -8.41848254e-02 9.22196090e-01 6.32115543e-01
4.61528331e-01 -8.81550372e-01 3.20985228e-01 5.18827558e-01
1.05683947e+00 -1.16504288e+00 -1.45332068e-01 -1.17590897e-01
-4.48985398e-01 6.97554052e-01 6.11176372e-01 -1.38093889e-01
1.36025045e-02 -3.96121442e-01 -3.40773351e-03 -1.77266195e-01
-1.32605028e+00 -4.72145807e-03 3.82115752e-01 2.42721394e-01
6.04726970e-01 -3.65261398e-02 -7.25268483e-01 4.95414823e-01
1.99921891e-01 -3.48581016e-01 1.44097254e-01 1.31105578e+00
-7.10241199e-01 -1.09446549e+00 -1.70215011e-01 2.32381478e-01
-2.95726001e-01 -7.94186175e-01 -8.39438021e-01 4.12234724e-01
-3.75890344e-01 1.07774830e+00 6.76390603e-02 -5.98758698e-01
9.75502804e-02 4.87746835e-01 7.77109027e-01 -8.34351063e-01
-1.03272009e+00 4.89602298e-01 -2.24849746e-01 -2.88581073e-01
-3.46608967e-01 -6.13513827e-01 -1.07765293e+00 -2.23305821e-01
-5.34113705e-01 2.32572541e-01 3.73408914e-01 1.05722010e+00
6.74310863e-01 4.02167767e-01 7.77547061e-01 -6.97425783e-01
-6.30383551e-01 -1.41549754e+00 -8.57890248e-02 3.12584430e-01
2.68344194e-01 -4.10076588e-01 4.20526303e-02 -3.41712475e-01] | [11.702323913574219, 10.017871856689453] |
c7547934-dcf6-4081-9d8b-788db5a59650 | pay-more-attention-to-history-a-context | 2112.08735 | null | https://arxiv.org/abs/2112.08735v2 | https://arxiv.org/pdf/2112.08735v2.pdf | Pay More Attention to History: A Context Modelling Strategy for Conversational Text-to-SQL | Conversational text-to-SQL aims at converting multi-turn natural language queries into their corresponding SQL (Structured Query Language) representations. One of the most intractable problems of conversational text-to-SQL is modelling the semantics of multi-turn queries and gathering the proper information required for the current query. This paper shows that explicitly modelling the semantic changes by adding each turn and the summarization of the whole context can bring better performance on converting conversational queries into SQLs. In particular, we propose two conversational modelling tasks in both turn grain and conversation grain. These two tasks simply work as auxiliary training tasks to help with multi-turn conversational semantic parsing. We conducted empirical studies and achieved new state-of-the-art results on the large-scale open-domain conversational text-to-SQL dataset. The results demonstrate that the proposed mechanism significantly improves the performance of multi-turn semantic parsing. | ['Yan Zhang', 'Wei Wu', 'Sirui Wang', 'Yutian Li', 'Hanchu Zhang', 'Yuntao Li'] | 2021-12-16 | null | null | null | null | ['text-to-sql'] | ['computer-code'] | [ 1.83046162e-01 6.41342878e-01 -1.39976948e-01 -9.77741480e-01
-1.31784046e+00 -7.37334311e-01 7.22249687e-01 2.04486266e-01
-5.05654365e-02 5.31696379e-01 6.44166350e-01 -5.66638947e-01
2.68269330e-01 -1.15460002e+00 -7.38549888e-01 2.20444147e-02
3.15996230e-01 1.06810236e+00 4.85945404e-01 -6.37361944e-01
1.31351382e-01 -1.89019650e-01 -1.44889927e+00 1.15955281e+00
9.84637380e-01 7.59229183e-01 4.35517401e-01 9.50138867e-01
-1.33922088e+00 1.20748484e+00 -6.82113647e-01 -6.34295225e-01
-2.59691745e-01 -4.34331656e-01 -1.54037488e+00 -1.41757309e-01
6.39084578e-02 4.01470484e-03 3.92742544e-01 5.31016648e-01
1.82265386e-01 1.70266345e-01 1.44895539e-01 -1.23717964e+00
-2.51229733e-01 9.15603817e-01 8.12943205e-02 -1.91379875e-01
9.50866282e-01 -2.12451011e-01 1.31453419e+00 -6.28832042e-01
6.45143270e-01 1.77323854e+00 3.58693808e-01 4.73618060e-01
-8.50868702e-01 -2.00661615e-01 2.15777636e-01 3.41448709e-02
-6.84151649e-01 -3.67176682e-01 5.92532456e-01 -1.59141213e-01
1.67385900e+00 6.98276341e-01 2.85082787e-01 7.63684928e-01
-1.06810421e-01 8.58750820e-01 6.99123085e-01 -4.56242919e-01
-4.25858274e-02 3.46370339e-01 5.91717303e-01 6.70435250e-01
-2.79260427e-01 -5.57371736e-01 -4.40974891e-01 -4.48502123e-01
1.88951775e-01 -4.70963754e-02 2.79602170e-01 -5.38020208e-03
-9.47847247e-01 9.88940120e-01 1.43296540e-01 1.78656518e-01
-2.01189488e-01 8.23071748e-02 7.57802010e-01 5.23188949e-01
6.15627170e-01 5.77379823e-01 -7.95250833e-01 -6.10770702e-01
-2.69915283e-01 5.11440516e-01 1.53604734e+00 1.38210034e+00
9.24476504e-01 -6.51230931e-01 -1.07001781e-01 1.09838128e+00
1.38787583e-01 5.70348859e-01 1.91938311e-01 -1.22029829e+00
1.06747782e+00 1.14384234e+00 1.80560946e-01 -7.49058545e-01
-2.26785004e-01 1.22937314e-01 -2.05037057e-01 -9.06926155e-01
2.39955515e-01 -1.57900631e-01 -3.14280480e-01 1.51230299e+00
4.72415090e-01 -5.35729885e-01 6.88588917e-01 2.74637550e-01
1.23274696e+00 9.79001105e-01 1.94095090e-01 -2.00618804e-01
1.93493128e+00 -1.29230237e+00 -7.17756450e-01 -5.86134911e-01
1.06289482e+00 -9.42402780e-01 1.49036300e+00 -1.90315396e-01
-1.08875000e+00 -4.67520237e-01 -4.76026297e-01 -6.16234541e-01
-5.37954271e-01 1.24129839e-02 8.99575710e-01 4.49377984e-01
-8.47369492e-01 -5.53680211e-02 -8.36353064e-01 -6.73370123e-01
-2.41192490e-01 -8.15478116e-02 4.94711027e-02 -2.15678036e-01
-1.27545583e+00 6.73913717e-01 1.50936335e-01 -2.04218641e-01
-2.90960640e-01 -6.08934045e-01 -9.14665103e-01 1.71068743e-01
8.36040616e-01 -9.13903892e-01 1.90610802e+00 -4.54553485e-01
-1.45454657e+00 1.09519565e+00 -1.10623312e+00 -3.75642359e-01
2.51704782e-01 -4.73452359e-01 -7.51995072e-02 9.85124335e-03
5.28940260e-01 5.82394779e-01 2.72150815e-01 -1.10165536e+00
-7.53259778e-01 -5.50944746e-01 5.05488098e-01 3.98513883e-01
1.23237260e-01 2.90663511e-01 -6.61124825e-01 -2.85143286e-01
1.24941811e-01 -8.08485031e-01 -4.27974612e-02 -7.62095451e-01
-4.85018253e-01 -8.91159654e-01 7.42085218e-01 -7.56262481e-01
1.22465956e+00 -1.80159116e+00 5.36640920e-03 -3.34369928e-01
2.31768470e-02 -1.38916180e-01 -1.04518406e-01 1.08464253e+00
1.82856128e-01 2.43478805e-01 -1.83456630e-01 -4.83014613e-01
8.90254229e-02 5.18650949e-01 -8.09767663e-01 -7.14170277e-01
1.17946655e-01 1.35488236e+00 -7.13131070e-01 -6.37680531e-01
-2.74096876e-02 -1.51914075e-01 -7.74392188e-01 7.48414159e-01
-9.86018896e-01 2.53047973e-01 -8.05841804e-01 3.77294093e-01
4.27042544e-01 -2.54626125e-01 3.57062340e-01 -3.20490710e-02
2.34128341e-01 9.92366970e-01 -4.11016732e-01 1.74979484e+00
-9.48142469e-01 3.40282530e-01 3.72173567e-03 -9.22796190e-01
1.04277229e+00 1.80038571e-01 3.27240050e-01 -7.50031769e-01
-2.74140507e-01 2.00768933e-02 -5.81909120e-01 -9.53856409e-01
7.37638891e-01 -1.87238201e-01 -7.02774942e-01 7.87027717e-01
-3.75332326e-01 -7.58734107e-01 5.44465333e-02 4.72257614e-01
8.27150404e-01 -1.40797243e-01 3.37823629e-01 -2.41486169e-02
6.65176213e-01 5.00063062e-01 2.75513887e-01 7.88837016e-01
2.55603611e-01 4.20429744e-02 9.56583381e-01 -3.93343985e-01
-6.11558497e-01 -7.79086411e-01 5.06769240e-01 1.75606644e+00
8.40963498e-02 -5.50810218e-01 -1.04285550e+00 -8.33787262e-01
-1.63595051e-01 1.28777039e+00 -9.90487486e-02 7.00547323e-02
-7.08605766e-01 -4.81546104e-01 6.66290879e-01 3.45734447e-01
6.35241210e-01 -1.17230868e+00 -4.85985160e-01 3.17701012e-01
-1.08480275e+00 -1.63605332e+00 -3.81789625e-01 6.13677464e-02
-9.99907315e-01 -1.38168311e+00 3.23905721e-02 -1.08283651e+00
2.77619302e-01 3.54951650e-01 1.39934409e+00 -1.57569943e-03
2.58464497e-02 4.65319693e-01 -6.08439326e-01 -4.73204911e-01
-8.14768195e-01 5.08306623e-01 -9.88307714e-01 -2.95037478e-01
6.67564511e-01 -2.64452875e-01 -8.54557157e-02 5.32885194e-01
-8.33067358e-01 3.96173507e-01 7.58615658e-02 4.56494004e-01
2.14479849e-01 -2.12945044e-01 6.63893819e-01 -1.42357874e+00
1.20076907e+00 -3.50569248e-01 -1.63950801e-01 6.99678242e-01
-3.78114939e-01 3.69223803e-01 5.69797754e-01 1.22682378e-01
-1.67983115e+00 -2.97889262e-01 -2.13392064e-01 5.53293586e-01
-1.74934670e-01 5.69112599e-01 -2.21662119e-01 6.80654347e-01
5.90844631e-01 3.46297562e-01 -4.76482995e-02 -3.93800139e-01
7.23836362e-01 9.39922929e-01 4.07615572e-01 -1.07046247e+00
2.23766416e-01 4.54781651e-01 -4.90997136e-01 -8.15479994e-01
-1.08908868e+00 -7.70892084e-01 -2.60818034e-01 1.95486113e-01
1.05537212e+00 -8.48251104e-01 -7.83732295e-01 5.14988005e-01
-1.57341814e+00 -4.53587711e-01 -2.46944577e-01 -2.54595190e-01
-7.39440024e-01 4.70649749e-01 -6.79489136e-01 -7.72523761e-01
-4.71219689e-01 -1.00517690e+00 1.50127447e+00 -1.20837592e-01
-4.09254611e-01 -1.16294706e+00 8.97548124e-02 1.23766696e+00
4.56663847e-01 -2.19361767e-01 1.64797068e+00 -9.95063901e-01
-6.02455437e-01 -7.75073990e-02 -2.93628037e-01 1.95419475e-01
1.37098238e-01 -2.69750267e-01 -8.65257978e-01 2.07645938e-01
4.24210161e-01 -5.82918346e-01 5.23674309e-01 -7.82180279e-02
8.84791613e-01 -5.75504184e-01 -2.49413654e-01 1.19658351e-01
1.00669420e+00 4.90174174e-01 6.41877651e-01 -3.07375547e-02
4.31794941e-01 1.00073957e+00 8.23317885e-01 2.97533244e-01
1.20454466e+00 6.04304731e-01 1.73945993e-01 1.96971744e-01
1.43285498e-01 -6.33515358e-01 1.62517279e-01 1.03538454e+00
4.87774491e-01 -3.51832539e-01 -9.69993532e-01 4.90692019e-01
-2.07871628e+00 -7.53595769e-01 -2.03108132e-01 1.69409490e+00
9.63374794e-01 -4.70080078e-02 -9.87158790e-02 -5.27790725e-01
6.82433903e-01 3.31640124e-01 -3.70106965e-01 -8.37314188e-01
9.46554244e-02 9.14493650e-02 -1.72465801e-01 9.69795525e-01
-7.09964752e-01 1.54218733e+00 6.38592386e+00 5.28782547e-01
-7.37904072e-01 1.15291104e-01 4.05596912e-01 1.78202659e-01
-7.16263294e-01 4.92975503e-01 -9.96419609e-01 4.92258221e-02
8.69632840e-01 -2.31820777e-01 6.38630629e-01 8.09798479e-01
2.74362922e-01 -4.03464317e-01 -1.39841616e+00 8.03792000e-01
-6.68711364e-02 -1.24467826e+00 5.11308968e-01 -5.86574852e-01
2.95713723e-01 -1.16221540e-01 -7.05882609e-01 8.73014152e-01
5.01664698e-01 -7.32680023e-01 1.67730078e-01 5.53281069e-01
3.65526646e-01 -4.46355283e-01 8.17781627e-01 6.87280834e-01
-1.27657652e+00 3.67077142e-02 -1.02749787e-01 -1.84463441e-01
3.02067667e-01 4.07433003e-01 -1.23206365e+00 6.56307876e-01
5.56449413e-01 5.62711537e-01 -4.94616002e-01 2.34082583e-02
-8.94057825e-02 4.35648888e-01 -2.09535033e-01 -3.73514384e-01
1.83935881e-01 -2.49995127e-01 3.95241559e-01 1.18399143e+00
-9.25399065e-02 2.75371939e-01 3.64833593e-01 7.77596653e-01
-1.29955739e-01 1.22415736e-01 -5.27762234e-01 -3.40669096e-01
5.09393334e-01 7.48896897e-01 -2.43039593e-01 -8.26181591e-01
-5.30300319e-01 7.89888978e-01 2.85672486e-01 4.18878675e-01
-5.96520245e-01 -5.36957502e-01 7.38537431e-01 2.07622752e-01
2.04604063e-02 -1.11677274e-02 -2.87435502e-01 -1.07369614e+00
6.08204007e-01 -1.00363553e+00 6.63427055e-01 -1.13692415e+00
-1.24254477e+00 5.69993138e-01 2.82252640e-01 -3.48251253e-01
-8.22449565e-01 -2.20763939e-03 -6.69665515e-01 7.78920114e-01
-1.44109082e+00 -1.18845344e+00 -5.06223381e-01 5.56025147e-01
1.09411609e+00 3.67775969e-02 1.05963814e+00 -2.65936870e-02
-1.51466042e-01 2.02786550e-01 -3.71519089e-01 1.41017258e-01
5.69371879e-01 -1.08329868e+00 7.97661841e-01 3.47146034e-01
-2.03030348e-01 7.73271739e-01 5.47037601e-01 -6.47924542e-01
-1.48334658e+00 -1.02925205e+00 1.64205980e+00 -6.37842953e-01
5.36330283e-01 -5.97991049e-01 -9.78466451e-01 9.93074298e-01
2.89937705e-01 -7.70745337e-01 4.87166613e-01 3.39897990e-01
-2.89740622e-01 -8.10865089e-02 -8.91317070e-01 4.56451416e-01
1.00935996e+00 -1.07393765e+00 -1.32766640e+00 6.14221215e-01
1.68411899e+00 -4.55571324e-01 -7.60989070e-01 2.36954436e-01
8.60307664e-02 -1.05867243e+00 9.00446057e-01 -1.04060042e+00
4.55407768e-01 1.89813361e-01 -4.51144785e-01 -9.74696159e-01
6.21545434e-01 -8.10873568e-01 3.12741250e-01 1.19290781e+00
5.29404759e-01 -1.03300381e+00 8.27594340e-01 9.27671611e-01
-2.02348039e-01 -6.85831845e-01 -6.65578067e-01 -3.92062128e-01
2.65699849e-02 -6.69401824e-01 8.05400789e-01 7.37320125e-01
4.70943302e-01 8.57140779e-01 1.90839738e-01 -1.82802618e-01
1.84166268e-01 6.97477758e-01 1.21166563e+00 -1.06108832e+00
-2.21571088e-01 -1.04288667e-01 2.09208295e-01 -1.66796267e+00
4.52436715e-01 -8.52715969e-01 1.17026726e-02 -1.87609959e+00
2.07333043e-01 -3.76056701e-01 6.83469951e-01 5.27287364e-01
-2.09347263e-01 -9.89277840e-01 1.16981782e-01 5.06722890e-02
-7.41734445e-01 6.12672567e-01 1.23134184e+00 -1.88404307e-01
-4.89108443e-01 3.35778028e-01 -1.06902421e+00 3.29467684e-01
6.59334600e-01 -5.99424183e-01 -8.11976910e-01 -6.94244623e-01
3.96519065e-01 8.41912985e-01 1.47021070e-01 -4.16602790e-01
3.09338450e-01 -3.77561420e-01 -7.76255012e-01 -7.16132581e-01
4.21216905e-01 -5.00468135e-01 -1.40216410e-01 2.20757276e-01
-8.88136387e-01 1.86345443e-01 1.21491775e-01 5.26261389e-01
-7.62895823e-01 -1.25979587e-01 3.35522473e-01 -4.93852854e-01
-5.23542106e-01 -2.84718603e-01 -3.49118501e-01 8.74150634e-01
8.06372225e-01 8.36640224e-02 -6.15881681e-01 -8.23028743e-01
-8.21763813e-01 6.94260359e-01 4.06120196e-02 8.35256934e-01
3.46398473e-01 -6.48484111e-01 -5.00814795e-01 1.29074141e-01
3.18591237e-01 5.09949625e-01 -5.16591966e-02 3.70306551e-01
-5.20073116e-01 1.15371966e+00 3.13069701e-01 -7.74383485e-01
-1.41088831e+00 1.70012534e-01 3.40048999e-01 -7.18240440e-01
-1.97017506e-01 7.90755510e-01 2.71464914e-01 -1.23402262e+00
2.86726922e-01 -8.39506626e-01 -1.44437477e-02 3.68090421e-02
1.22090168e-01 2.03247562e-01 1.95965081e-01 -1.23078085e-01
-3.05277884e-01 4.98669267e-01 -8.04343522e-02 -1.11578010e-01
1.23930967e+00 -4.03292179e-01 -7.30461061e-01 4.22878623e-01
1.39326727e+00 -2.11693555e-01 -5.03390491e-01 -5.73101938e-01
4.53784347e-01 -1.49124354e-01 -5.87409914e-01 -8.20035875e-01
-4.60824579e-01 9.58386242e-01 -3.03438127e-01 4.64015871e-01
8.29302967e-01 4.73402619e-01 1.31285071e+00 1.26077008e+00
5.42225957e-01 -9.81265247e-01 1.57519445e-01 1.01632786e+00
9.52045679e-01 -1.30328310e+00 -3.63816172e-01 -1.03701425e+00
-8.08872044e-01 1.05574989e+00 6.13996744e-01 3.17769974e-01
4.19578195e-01 1.44867403e-02 3.41139376e-01 -5.44605255e-01
-1.23751283e+00 5.44943884e-02 -1.37465671e-01 3.95768106e-01
4.19909984e-01 1.98129475e-01 -2.52760909e-02 7.63735414e-01
-6.25804007e-01 -1.89208224e-01 2.22993270e-01 1.08833885e+00
-6.31964207e-01 -1.28105867e+00 -9.32334810e-02 3.90173912e-01
-1.87621161e-01 -2.23545596e-01 -9.32288826e-01 5.93193114e-01
-5.85761845e-01 1.56659722e+00 1.18561737e-01 -1.62838355e-01
6.64595246e-01 7.70035684e-01 7.49228299e-02 -1.05776715e+00
-9.94746029e-01 -2.84482569e-01 8.83936226e-01 -8.38925183e-01
-3.43918353e-01 -4.25092369e-01 -1.76927590e+00 -3.76343608e-01
3.80395167e-02 7.23470330e-01 7.42675543e-01 1.17644536e+00
7.07756102e-01 3.64348918e-01 5.72204590e-01 3.79321314e-02
-6.42912388e-01 -9.25198793e-01 3.73816714e-02 6.14128232e-01
9.77879912e-02 -7.48946331e-03 -1.29225895e-01 1.65556118e-01] | [10.036736488342285, 7.886224269866943] |
9c8a5085-040d-4c4f-b031-c625e8c85595 | mobiprox-supporting-dynamic-approximate | 2303.11291 | null | https://arxiv.org/abs/2303.11291v1 | https://arxiv.org/pdf/2303.11291v1.pdf | Mobiprox: Supporting Dynamic Approximate Computing on Mobiles | Runtime-tunable context-dependent network compression would make mobile deep learning adaptable to often varying resource availability, input "difficulty", or user needs. The existing compression techniques significantly reduce the memory, processing, and energy tax of deep learning, yet, the resulting models tend to be permanently impaired, sacrificing the inference power for reduced resource usage. The existing tunable compression approaches, on the other hand, require expensive re-training, seldom provide mobile-ready implementations, and do not support arbitrary strategies for adapting the compression. In this paper we present Mobiprox, a framework enabling flexible-accuracy on-device deep learning. Mobiprox implements tunable approximations of tensor operations and enables runtime adaptation of individual network layers. A profiler and a tuner included with Mobiprox identify the most promising neural network approximation configurations leading to the desired inference quality with the minimal use of resources. Furthermore, we develop control strategies that depending on contextual factors, such as the input data difficulty, dynamically adjust the approximation level of a model. We implement Mobiprox in Android OS and through experiments in diverse mobile domains, including human activity recognition and spoken keyword detection, demonstrate that it can save up to 15% system-wide energy with a minimal impact on the inference accuracy. | ['Veljko Pejović', 'Saša Misailović', 'Yifan Zhao', 'Hashim Sharif', 'Octavian Machidon', 'Matevž Fabjančič'] | 2023-03-16 | null | null | null | null | ['human-activity-recognition', 'human-activity-recognition'] | ['computer-vision', 'time-series'] | [ 4.42476459e-02 -2.19167352e-01 -6.54191077e-01 -4.52814460e-01
-5.54419339e-01 -4.14669275e-01 7.41246194e-02 4.95933287e-04
-6.79598093e-01 2.62349010e-01 -1.67014420e-01 -5.84844351e-01
-3.24469149e-01 -7.09643185e-01 -8.05755258e-01 -5.90331197e-01
1.16182953e-01 3.17099482e-01 1.87830091e-01 1.17390245e-01
5.57122976e-02 3.11703593e-01 -1.75297129e+00 3.42099279e-01
7.20622480e-01 1.40201712e+00 4.27408546e-01 8.70876670e-01
2.09623538e-02 5.72664142e-01 -6.86386883e-01 -3.73868078e-01
3.26692797e-02 5.80335036e-02 -5.11535048e-01 -4.80710447e-01
3.72478187e-01 -5.81762493e-01 -4.32155639e-01 7.85632610e-01
7.96217144e-01 6.04296438e-02 2.44656101e-01 -1.00241184e+00
-2.54989624e-01 1.02972007e+00 9.39732268e-02 5.75189769e-01
-2.21838549e-01 2.76283771e-01 6.09443009e-01 -7.59978533e-01
-1.43008726e-02 8.76052916e-01 8.72582555e-01 7.91019738e-01
-1.10294890e+00 -6.28067017e-01 2.67330438e-01 4.31640446e-01
-1.57541883e+00 -9.91445184e-01 5.18134713e-01 1.71870261e-01
1.39908326e+00 5.69211662e-01 7.45570958e-01 1.19973266e+00
8.46628249e-02 4.61409628e-01 5.06520092e-01 -3.09528410e-01
7.87232816e-01 8.82705748e-02 1.16360709e-01 8.06889117e-01
1.64235219e-01 -5.12252092e-01 -9.41391051e-01 -2.67917067e-01
2.48506844e-01 8.96270424e-02 -2.56069124e-01 8.90567154e-02
-7.59639561e-01 3.51027697e-01 5.01587465e-02 1.59924865e-01
-4.25554901e-01 6.43534362e-01 5.59535801e-01 2.44895458e-01
3.40064615e-01 2.88017392e-01 -8.44303429e-01 -8.61939549e-01
-1.05476546e+00 1.39625400e-01 7.96120405e-01 7.58193672e-01
5.72934628e-01 3.05771649e-01 -1.64184794e-01 1.05991697e+00
1.32889584e-01 5.53607345e-01 1.00454378e+00 -1.19220090e+00
6.06348336e-01 4.65870827e-01 -3.49531084e-01 -8.65596414e-01
-3.89208406e-01 -7.08791256e-01 -6.86775506e-01 -3.14897567e-01
4.30482328e-02 -3.67085189e-01 -7.32800484e-01 1.73218656e+00
3.13919336e-01 1.26415133e-01 -1.12135679e-01 6.59774244e-01
4.81640786e-01 4.47437078e-01 2.60004448e-03 -1.51785478e-01
1.26341200e+00 -8.01342249e-01 -6.80518448e-01 -3.50438029e-01
7.14662671e-01 -4.44205731e-01 1.74593174e+00 6.76408470e-01
-1.31407893e+00 -3.58332336e-01 -1.23891640e+00 -1.27876386e-01
-3.09307992e-01 2.84061223e-01 6.94914877e-01 1.04230797e+00
-1.01897037e+00 7.21072316e-01 -1.27185631e+00 3.94347683e-02
3.51372629e-01 8.36616457e-01 4.34763849e-01 2.38648608e-01
-9.93074059e-01 4.84083295e-01 3.99951786e-01 7.32269138e-02
-7.59834409e-01 -9.53580737e-01 -1.72502786e-01 5.85679829e-01
4.79560167e-01 -7.98817754e-01 1.53677034e+00 -8.98054302e-01
-2.06963062e+00 2.04700604e-01 -8.51256773e-02 -8.61060321e-01
2.31481656e-01 -1.72172338e-01 -3.86860609e-01 2.76429411e-02
-6.21052623e-01 6.79896414e-01 9.60304081e-01 -4.81782764e-01
-5.22895038e-01 -3.20422739e-01 2.02173293e-01 3.30668807e-01
-1.27057803e+00 -3.94069642e-01 -8.31206620e-01 -4.32494760e-01
-2.41005123e-01 -1.00687969e+00 -3.57408635e-02 -1.20953068e-01
-3.68573554e-02 -6.51227832e-02 1.08317971e+00 -4.26294327e-01
1.66628587e+00 -2.22269654e+00 -4.55833040e-02 1.91137165e-01
8.41170475e-02 4.12860245e-01 5.52387303e-03 -2.74425656e-01
5.63231826e-01 2.20955074e-01 8.05827007e-02 -5.33002973e-01
5.65577187e-02 5.99284887e-01 -4.58491072e-02 8.67502615e-02
-3.42387617e-01 5.98393381e-01 -4.35782164e-01 -3.47673297e-01
2.70870375e-03 6.15244269e-01 -1.01449227e+00 -5.51650897e-02
-4.02361780e-01 -3.63061339e-01 -1.92839682e-01 6.80194020e-01
2.55268008e-01 -3.35997850e-01 4.24237609e-01 -3.68625194e-01
2.59568036e-01 7.46868610e-01 -1.12404728e+00 1.60372007e+00
-1.03864968e+00 5.68350434e-01 1.57334968e-01 -8.62139165e-01
5.79895198e-01 1.94111988e-01 2.55865902e-01 -8.10598254e-01
2.26952478e-01 3.06770593e-01 -2.35310663e-03 -4.01456445e-01
7.90473878e-01 7.01940358e-01 1.05615176e-01 3.72388870e-01
5.75695746e-02 3.90975416e-01 -6.89428523e-02 -1.45309821e-01
1.10394835e+00 -2.95668900e-01 -2.30712220e-02 -1.60539985e-01
1.47169963e-01 -5.01149893e-01 3.86964589e-01 7.88053393e-01
1.22313183e-02 -2.79080076e-03 2.62641042e-01 -4.12456185e-01
-8.17206383e-01 -5.55725336e-01 -1.86318874e-01 1.63100874e+00
-1.72216848e-01 -6.72960460e-01 -1.36989629e+00 -3.92560929e-01
-2.77539164e-01 5.52412450e-01 -1.95823967e-01 -4.30656075e-01
-6.06352568e-01 -7.94936717e-01 9.93146300e-01 4.89989340e-01
6.35151505e-01 -6.43062830e-01 -1.12533915e+00 1.24073341e-01
-1.88748628e-01 -1.26774001e+00 -5.70055425e-01 4.40037131e-01
-1.17358422e+00 -5.37693441e-01 -1.06792010e-01 -3.61956656e-01
2.99077928e-01 1.80149734e-01 1.04878318e+00 2.93994606e-01
3.86759713e-02 3.74144346e-01 3.21403965e-02 -3.67617637e-01
-3.44040811e-01 8.87403846e-01 3.65940452e-01 -2.28715912e-01
3.14812958e-01 -8.62718105e-01 -8.52158606e-01 1.67596593e-01
-7.60030448e-01 -4.64741811e-02 4.47134018e-01 5.98707557e-01
8.05524111e-01 1.05820253e-01 4.05903012e-01 -6.50992274e-01
8.81565869e-01 -4.56257343e-01 -5.10426760e-01 2.11830884e-01
-1.19328582e+00 1.96014792e-01 8.43734443e-01 -1.05298293e+00
-6.69114113e-01 -1.39380917e-01 -2.39514619e-01 -6.89775348e-01
2.58758664e-01 5.91900945e-01 -3.30956370e-01 -5.21980338e-02
9.47843552e-01 2.95072526e-01 -3.30621243e-01 -5.05814373e-01
2.06410602e-01 9.30991173e-01 5.70310354e-01 -5.76374948e-01
-8.16684309e-03 5.51467501e-02 -2.36791879e-01 -7.66322911e-01
-4.13488448e-01 1.51490673e-01 -1.22900762e-01 -7.19400942e-02
3.02600414e-01 -9.07645464e-01 -9.98919129e-01 2.67847836e-01
-8.04239333e-01 -8.25330853e-01 -1.55479938e-01 2.48366132e-01
-2.64549762e-01 -8.94919932e-02 -4.80933756e-01 -6.61787868e-01
-9.88443136e-01 -1.28417778e+00 8.22888076e-01 1.90479890e-01
-2.81654537e-01 -7.46753812e-01 -5.17479122e-01 4.35029030e-01
7.15259790e-01 -4.25985932e-01 1.00166965e+00 -4.63514388e-01
-5.76953113e-01 -8.28027725e-03 2.13147253e-01 3.49769294e-01
-1.20895699e-01 -2.91383304e-02 -1.11798966e+00 -1.83195710e-01
-8.15256462e-02 -1.69627398e-01 4.54269558e-01 3.86331260e-01
2.12502909e+00 -1.04368520e+00 -1.12150073e-01 9.68333006e-01
1.09447932e+00 1.51407555e-01 4.27756816e-01 1.36190817e-01
7.55674541e-01 -9.49421804e-03 2.48888329e-01 5.84763348e-01
3.02958727e-01 1.05073106e+00 3.75015169e-01 4.52193320e-01
-4.65228548e-03 -2.25647137e-01 5.74494123e-01 1.07829118e+00
-9.22995210e-02 -3.09220076e-01 -7.98035502e-01 3.11774313e-01
-1.55748880e+00 -6.55933917e-01 4.82291490e-01 2.25612974e+00
1.15714526e+00 3.98661941e-01 2.18227550e-01 5.28770089e-01
3.54860444e-03 9.45451632e-02 -1.04350221e+00 -6.98389173e-01
2.96524882e-01 4.25639570e-01 9.13197517e-01 4.14650440e-01
-6.20005906e-01 6.77199781e-01 5.66114902e+00 1.20747852e+00
-1.62443960e+00 5.16156733e-01 7.75207758e-01 -9.34379995e-01
-3.01954120e-01 -3.91702533e-01 -9.26377475e-01 8.80672276e-01
1.85324717e+00 4.40733619e-02 1.00944471e+00 1.27122450e+00
3.03086936e-01 2.12839037e-01 -1.09824657e+00 1.21314466e+00
-2.40727782e-01 -1.49836457e+00 -7.04075769e-02 1.24624304e-01
1.66437536e-01 2.94826776e-01 3.79041642e-01 3.90767664e-01
-2.58775681e-01 -8.78693700e-01 8.13751936e-01 4.75499392e-01
9.64278817e-01 -1.02046955e+00 4.63875860e-01 5.00411272e-01
-9.12104070e-01 -4.60765570e-01 -3.01091969e-01 2.58853361e-02
-3.02174777e-01 6.10479057e-01 -9.27433908e-01 -3.27169091e-01
1.13264596e+00 -3.39177959e-02 -6.06371880e-01 2.85221219e-01
2.90532917e-01 1.02593338e+00 -5.77723205e-01 -4.96507198e-01
-1.00680813e-01 3.35564345e-01 2.68126696e-01 1.27775538e+00
3.40574652e-01 1.22213528e-01 -1.43274054e-01 3.98738384e-01
-4.88690227e-01 1.03929639e-02 -8.40393528e-02 -2.84079332e-02
1.09024858e+00 1.05771983e+00 -4.15282905e-01 -2.82738477e-01
9.39078480e-02 8.79931927e-01 2.91453004e-01 1.68964401e-01
-9.71242309e-01 -1.05251856e-01 9.52408493e-01 3.69096041e-01
2.47313276e-01 -3.52086693e-01 -5.21200895e-01 -8.79264534e-01
3.66588861e-01 -1.22208428e+00 7.77098164e-02 -3.41088384e-01
-3.31046969e-01 4.47238982e-01 -3.27403247e-02 -6.12418175e-01
-4.10241306e-01 -4.16848004e-01 -1.83182031e-01 3.69730741e-01
-9.93327975e-01 -5.55117667e-01 -4.49610531e-01 5.35379469e-01
7.88534522e-01 -2.24885732e-01 9.52678800e-01 7.45865285e-01
-8.58964860e-01 1.47702754e+00 1.17251977e-01 -4.25106734e-01
2.09833175e-01 -8.83788466e-01 3.54625732e-01 5.69151878e-01
2.04947174e-01 8.28009427e-01 5.74893773e-01 -1.82882339e-01
-1.83145678e+00 -1.05257893e+00 5.12608886e-01 -1.12357877e-01
2.73863673e-01 -5.48987329e-01 -1.04692638e+00 2.55463541e-01
-1.72076270e-01 1.78160295e-01 7.88141787e-01 3.17415982e-01
-2.70892650e-01 -7.08232045e-01 -1.20018017e+00 8.89880240e-01
1.21916473e+00 -7.19821036e-01 3.61025900e-01 3.67060125e-01
9.59903836e-01 -7.48334825e-01 -1.03836906e+00 3.16816002e-01
7.74731755e-01 -8.65426064e-01 9.14643645e-01 -3.55382264e-01
7.31980726e-02 1.48687378e-01 -3.57798398e-01 -8.52174044e-01
-2.03027464e-02 -6.50001764e-01 -1.16694367e+00 1.02083266e+00
6.40587986e-01 -3.14545721e-01 1.04136729e+00 8.50992680e-01
-9.19596553e-02 -1.43097711e+00 -1.11448097e+00 -4.57973123e-01
-4.88046408e-01 -7.92738616e-01 9.66233671e-01 5.25626004e-01
-1.87807649e-01 2.56345302e-01 -3.00546110e-01 1.40065074e-01
1.73436612e-01 -5.98957241e-01 6.51746690e-01 -7.91738868e-01
-9.48244691e-01 -5.57988942e-01 -8.41326490e-02 -1.16808903e+00
-7.31972829e-02 -4.65272427e-01 -1.80271342e-01 -7.54960597e-01
-1.82838038e-01 -7.48861790e-01 -3.44078749e-01 6.82021618e-01
1.96046814e-01 1.35065898e-01 1.23944618e-02 2.55502403e-01
-6.38631821e-01 2.56452471e-01 4.23520952e-01 -1.78115338e-01
-5.91749310e-01 2.17035845e-01 -6.49887562e-01 7.59688795e-01
9.02678728e-01 -4.74463671e-01 -8.44821990e-01 -8.23776186e-01
4.42092210e-01 -1.30045444e-01 1.79354101e-01 -1.32253385e+00
4.04884607e-01 -9.95377526e-02 2.23439306e-01 -9.77567136e-02
4.74719286e-01 -7.41948068e-01 1.10985167e-01 4.90958661e-01
-5.44533074e-01 3.08458120e-01 3.71021897e-01 5.01282454e-01
3.85492951e-01 -2.55839825e-01 7.17904031e-01 2.75174528e-01
-4.01635140e-01 3.18740964e-01 -4.68219817e-01 -2.29656085e-01
4.86689568e-01 -2.70599425e-01 -4.71515357e-02 -2.39599019e-01
-3.83946538e-01 -1.58301622e-01 1.42520458e-01 4.00565714e-01
4.88077760e-01 -9.58437324e-01 -7.65888114e-03 1.30127892e-01
-2.23401368e-01 1.17203094e-01 3.09013039e-01 6.16442621e-01
-4.09804165e-01 3.73691410e-01 2.11496904e-01 -6.70317471e-01
-1.50949693e+00 3.01330805e-01 7.88374960e-01 -1.76493287e-01
-3.72300446e-01 7.00405836e-01 -8.03490162e-01 -1.95185930e-01
7.83183753e-01 -7.50031829e-01 1.84088305e-01 -2.11415619e-01
5.93668103e-01 5.37691593e-01 7.31750727e-01 7.42083192e-02
-3.30012679e-01 1.23059317e-01 -7.62176700e-03 4.46886383e-03
9.95489419e-01 -9.65702087e-02 2.71432579e-01 3.50481957e-01
1.10628676e+00 -1.27667755e-01 -1.20568264e+00 -2.62425125e-01
-2.02420563e-01 -1.77540019e-01 5.29214621e-01 -7.55137026e-01
-1.26150751e+00 7.17963338e-01 1.15317869e+00 4.89266887e-02
1.47789979e+00 -4.39399093e-01 1.09450305e+00 9.74961758e-01
3.86280000e-01 -1.36255944e+00 -3.48824710e-02 3.92388254e-01
2.81017721e-01 -8.27615023e-01 -2.58100796e-02 1.41130686e-01
-2.71305442e-01 9.63429928e-01 7.27879763e-01 4.96198356e-01
7.19088495e-01 7.10836828e-01 -1.99016288e-01 1.30676091e-01
-1.09979415e+00 5.14241576e-01 1.60209820e-01 3.87272269e-01
2.72020131e-01 1.28609121e-01 -1.55428618e-01 7.99850643e-01
-5.71722448e-01 1.70354918e-01 1.68125108e-01 5.55135548e-01
-4.70073134e-01 -8.88215780e-01 -1.64450899e-01 6.68526173e-01
-7.01745093e-01 -1.29067928e-01 7.30483010e-02 3.07117730e-01
3.65823925e-01 7.63269603e-01 3.40743214e-01 -7.90224969e-01
2.97507882e-01 4.23556603e-02 2.25340739e-01 -2.52869219e-01
-8.31796527e-01 -2.68218875e-01 1.33992493e-01 -6.17217541e-01
3.55288386e-02 -3.29788446e-01 -1.26755106e+00 -8.28709662e-01
-2.01449513e-01 -4.91439700e-02 1.09616339e+00 1.07548892e+00
9.51526821e-01 6.24860346e-01 2.92415649e-01 -7.37634242e-01
-7.90218711e-01 -9.15008366e-01 -1.34101615e-03 -3.91931623e-01
2.30322376e-01 -3.46558273e-01 -6.98711574e-02 -1.17853679e-01] | [8.587942123413086, 3.0493111610412598] |
2ebfb4cf-04fc-4eda-b2e4-bd1f601ff07b | inception-residual-block-based-neural-network | 1810.13169 | null | http://arxiv.org/abs/1810.13169v2 | http://arxiv.org/pdf/1810.13169v2.pdf | Inception-Residual Block based Neural Network for Thermal Image Denoising | Thermal cameras show noisy images due to their limited thermal resolution,
especially for the scenes of a low temperature difference. In order to deal
with a noise problem, this paper proposes a novel neural network architecture
with repeatable denoising inception-residual blocks(DnIRB) for noise learning.
Each DnIRB has two sub-blocks with difference receptive fields and one shortcut
connection to prevent a vanishing gradient problem. The proposed approach is
tested for thermal images. The experimental results indicate that the proposed
approach shows the best SQNR performance and reasonable processing time
compared with state-of-the-art denoising methods. | ['Jin-Young Kim', 'Seungyou Na', 'Nazeer Shahid', 'Huy Toan Nguyen', 'Seongmin Hwang', 'Doseong Sin', 'Gwanghyun Yu'] | 2018-10-31 | null | null | null | null | ['thermal-image-denoising'] | ['computer-vision'] | [ 3.22451204e-01 -6.39370203e-01 5.92063904e-01 -3.56302381e-01
-3.55099946e-01 2.03409512e-03 2.39323199e-01 -6.02848530e-01
-7.34822571e-01 5.91914117e-01 5.68199120e-02 1.68696959e-02
6.03432907e-03 -4.89746958e-01 -4.81445372e-01 -1.20025432e+00
2.43632779e-01 -6.64817214e-01 3.74364972e-01 -1.68767467e-01
5.52107394e-01 4.13897693e-01 -1.27401853e+00 4.77328271e-01
7.59293616e-01 1.27623034e+00 5.49277723e-01 5.53525865e-01
1.34639099e-01 7.83938646e-01 -7.01375186e-01 -4.62292023e-02
6.56682193e-01 -3.25278938e-01 -1.91456050e-01 -5.25229946e-02
5.74164510e-01 -7.10920393e-01 -8.28990102e-01 1.39740193e+00
7.17064917e-01 2.71207780e-01 4.47175115e-01 -7.28242815e-01
-8.48730505e-01 4.19247568e-01 -8.12181532e-01 6.15830064e-01
-3.19473773e-01 2.79702395e-01 3.32240283e-01 -7.36534357e-01
2.89162040e-01 1.48349583e+00 7.96443343e-01 6.65767550e-01
-1.23982608e+00 -5.37248909e-01 -4.10593003e-02 5.11442959e-01
-1.24467766e+00 -3.14454973e-01 7.72294343e-01 -2.03352259e-03
1.13713467e+00 5.02729826e-02 3.56597662e-01 1.13406849e+00
6.89167619e-01 3.32639366e-01 1.42402136e+00 -2.18309894e-01
2.61826098e-01 -3.24228048e-01 3.28774065e-01 1.25741929e-01
3.46914917e-01 1.86046511e-01 -2.67807871e-01 3.51870298e-01
1.01924312e+00 -1.31832575e-02 -5.46881735e-01 3.53451580e-01
-6.62562907e-01 4.04496819e-01 7.38288045e-01 4.86890733e-01
-3.04829985e-01 6.01798773e-01 5.45715272e-01 5.01655519e-01
4.01915163e-01 -5.69633506e-02 -1.59985662e-01 9.27745551e-03
-9.13962483e-01 -1.99047670e-01 4.65715706e-01 7.76999056e-01
6.15646839e-01 7.59706676e-01 -1.51702324e-02 1.06856441e+00
9.97758284e-02 5.33558011e-01 4.91691232e-01 -1.11420810e+00
4.05743718e-01 4.03217738e-03 -6.79534003e-02 -1.06126606e+00
-2.05874935e-01 -2.28074566e-01 -1.74956179e+00 7.95503378e-01
7.78113902e-02 -1.49827018e-01 -1.26834226e+00 1.12878287e+00
-1.76103741e-01 6.02075681e-02 2.30234951e-01 1.14674425e+00
8.66857767e-01 1.30453289e+00 -3.69206101e-01 -4.23549414e-01
9.31844056e-01 -9.20631886e-01 -1.00394952e+00 -3.51757467e-01
-2.29683384e-01 -8.21658671e-01 6.28345132e-01 1.14351261e+00
-9.23529685e-01 -1.13619184e+00 -1.20850575e+00 -1.56136662e-01
-1.29122883e-01 1.66433990e-01 1.58582374e-01 8.23756814e-01
-1.27383101e+00 9.67634261e-01 -6.40304565e-01 -2.42365956e-01
2.01908171e-01 2.83330262e-01 -2.09958687e-01 -2.58109182e-01
-1.03128326e+00 8.31164777e-01 5.42725861e-01 9.45680559e-01
-1.04311049e+00 9.37094316e-02 -5.95915496e-01 8.72751176e-02
2.16652453e-01 -3.89031291e-01 7.91866302e-01 -1.23902941e+00
-1.58434391e+00 3.70358229e-01 3.56853940e-02 -4.28878665e-01
4.61417943e-01 -4.35684294e-01 -4.63405102e-01 3.63243133e-01
-2.76711315e-01 4.39872026e-01 1.42620325e+00 -1.29679227e+00
-3.04943204e-01 -3.85920972e-01 -3.79448771e-01 1.09819457e-01
-2.98994362e-01 1.04588024e-01 -4.09977764e-01 -8.30704451e-01
4.53744888e-01 -3.90405655e-01 -4.71905887e-01 -1.45207554e-01
-2.23783299e-01 2.92881951e-02 1.08000088e+00 -8.14372659e-01
1.16591525e+00 -2.27255702e+00 3.61440261e-03 1.40515909e-01
-2.09725037e-01 4.00276333e-01 -2.28869125e-01 1.56748071e-01
-3.63919437e-01 1.52917460e-01 -1.72031730e-01 -5.33938482e-02
-5.92383504e-01 3.10975969e-01 1.17662279e-02 5.89436293e-01
8.30242708e-02 1.50908530e-01 -3.06109130e-01 -2.90781736e-01
4.85240817e-01 6.40702844e-01 3.00302524e-02 1.77379951e-01
2.56013781e-01 4.22616661e-01 -2.27138892e-01 3.04028869e-01
1.42036426e+00 4.98036623e-01 -1.66694134e-01 -7.13328481e-01
-4.00908887e-01 -3.24658394e-01 -1.59907603e+00 1.44459367e+00
-2.89654493e-01 8.34472895e-01 7.44220436e-01 -1.03725100e+00
1.29860687e+00 2.32777312e-01 2.07008079e-01 -9.72460568e-01
3.54016274e-01 2.65156597e-01 -1.08950198e-01 -7.64520288e-01
4.82893795e-01 -2.49955013e-01 4.78828311e-01 -1.34017274e-01
3.35116722e-02 -1.19546689e-01 1.06024005e-01 -7.19013810e-02
1.00015366e+00 1.27457932e-01 -8.00779164e-02 -4.56576288e-01
8.38139415e-01 -5.33755183e-01 9.22040343e-01 1.00171053e+00
-3.64667773e-01 9.55187142e-01 2.91573673e-01 -6.46236360e-01
-1.38438296e+00 -6.68677032e-01 -9.05586183e-02 4.98082936e-01
3.55375856e-01 1.11194022e-01 -9.43269789e-01 -5.15691265e-02
-5.21151662e-01 4.38551962e-01 -3.95338744e-01 1.06022106e-02
-7.46678531e-01 -7.37498522e-01 5.78271687e-01 6.11569285e-01
1.27866864e+00 -1.14966750e+00 -5.63449383e-01 1.81238249e-01
5.87421749e-03 -1.00322843e+00 -2.38387689e-01 5.16650915e-01
-1.12561929e+00 -6.66488409e-01 -7.99330354e-01 -1.01330841e+00
6.31836057e-01 5.37118852e-01 8.58762801e-01 -7.98554812e-03
-3.97785276e-01 5.16084842e-02 -3.16876113e-01 1.04666062e-01
-2.49532729e-01 -5.30927062e-01 -5.77774411e-03 -7.86345378e-02
2.36239955e-01 -6.20359719e-01 -8.63637269e-01 2.11679190e-01
-1.32730532e+00 -2.97664821e-01 6.67495251e-01 1.05628252e+00
5.08124650e-01 8.03406239e-01 2.22931996e-01 -4.83685911e-01
7.36033797e-01 5.18901385e-02 -9.56964970e-01 6.00964688e-02
-5.41560054e-01 -5.90548962e-02 9.45410371e-01 -2.55725831e-01
-1.47610354e+00 3.85890417e-02 -7.14718252e-02 -4.67744976e-01
-3.48266244e-01 1.67400464e-01 -1.42715409e-01 -9.65316221e-02
7.61985421e-01 4.01446104e-01 -1.92285866e-01 -6.72926784e-01
4.48637009e-02 8.82350564e-01 6.67466998e-01 -3.92230213e-01
7.74570942e-01 4.36057806e-01 7.94011280e-02 -1.13360739e+00
-2.75448501e-01 -4.30910438e-01 -5.81431508e-01 -3.31721842e-01
1.11578786e+00 -1.01118779e+00 -7.72670627e-01 1.11400485e+00
-1.27269149e+00 -8.68140310e-02 1.62862688e-01 4.42567080e-01
-2.23048806e-01 7.93169677e-01 -1.16515064e+00 -9.57562804e-01
-7.20663071e-01 -1.36623800e+00 4.42717165e-01 5.06977975e-01
6.75447524e-01 -5.37248015e-01 -3.61199200e-01 1.43981591e-01
5.96608043e-01 -7.40902200e-02 5.69299877e-01 4.84995358e-02
-6.67126060e-01 9.81124938e-02 -4.13393885e-01 1.41406178e+00
-3.24071757e-02 9.56378579e-02 -1.10993338e+00 -4.03204381e-01
7.11458504e-01 -3.22903603e-01 1.42194390e+00 7.74136484e-01
1.37485158e+00 -1.67504787e-01 2.00914755e-01 7.95632780e-01
2.13258648e+00 5.67667007e-01 1.22636044e+00 5.61832905e-01
5.84931135e-01 2.61876822e-01 3.27295899e-01 2.56765455e-01
-6.06194735e-01 5.09026796e-02 6.04459107e-01 -2.45614320e-01
-1.46154538e-01 3.63898546e-01 3.88430625e-01 7.46934354e-01
-1.50002733e-01 -4.22189832e-01 -5.92390418e-01 5.18282592e-01
-1.56900489e+00 -1.08058405e+00 -4.90597218e-01 1.81735849e+00
5.35227478e-01 3.51045787e-01 -5.27874649e-01 2.82097727e-01
9.29441988e-01 5.22707999e-01 -5.56168914e-01 -9.05404210e-01
-3.86052936e-01 3.17381799e-01 8.91022503e-01 3.13737422e-01
-1.08977926e+00 5.08491457e-01 6.69868612e+00 1.09704971e+00
-1.08181131e+00 2.76591200e-02 9.57516253e-01 2.59513855e-01
1.52499884e-01 -1.80827469e-01 -5.31727552e-01 3.96409690e-01
6.81535423e-01 5.63254714e-01 7.44779170e-01 6.41039491e-01
6.18355811e-01 -5.03135324e-01 -6.28041148e-01 1.26988447e+00
5.90777285e-02 -7.82277524e-01 3.84858926e-03 -3.72475952e-01
7.55615354e-01 8.95992368e-02 1.36633739e-01 -2.07291782e-01
-1.60537779e-01 -1.04744720e+00 3.97877872e-01 7.69326508e-01
6.17510498e-01 -9.63990867e-01 1.07525909e+00 2.05063850e-01
-9.25475836e-01 -3.35659087e-01 -1.16362417e+00 -1.31121650e-01
-1.09898150e-01 9.05713379e-01 -8.61961469e-02 5.38215518e-01
1.30777419e+00 7.65647471e-01 -3.38105202e-01 1.01948786e+00
-2.00455785e-01 7.64332175e-01 -3.19420636e-01 4.98158708e-02
3.58228922e-01 -8.11061263e-01 3.51488411e-01 1.24275815e+00
5.69189727e-01 2.91560799e-01 -2.79224902e-01 8.42766941e-01
-2.88653411e-02 -1.91858754e-01 -5.54961026e-01 3.64813715e-01
1.57060221e-01 1.28269362e+00 -7.80021667e-01 -3.31221044e-01
-5.00601530e-01 1.29912448e+00 -2.83909708e-01 6.67472720e-01
-4.47930485e-01 -6.11423254e-01 3.63405377e-01 -2.54112184e-01
5.70318282e-01 -3.40001225e-01 -5.86873591e-01 -9.46244419e-01
1.10023037e-01 -9.80425179e-01 -4.83524501e-02 -1.18809068e+00
-1.48359621e+00 7.61732757e-01 -3.37165982e-01 -1.36648822e+00
3.89238477e-01 -1.00721765e+00 -6.77642107e-01 1.14524984e+00
-1.23623776e+00 -8.80363107e-01 -5.13835490e-01 5.88985205e-01
7.70631135e-01 -9.22140926e-02 3.44911307e-01 2.24844813e-01
-7.07359791e-01 1.23617508e-01 7.97678888e-01 2.72964090e-02
6.93891585e-01 -1.04101801e+00 3.14456046e-01 1.34550691e+00
-5.59965730e-01 5.78117192e-01 7.45573044e-01 -6.53844535e-01
-1.25428927e+00 -8.39572787e-01 6.90821037e-02 4.34218854e-01
3.12012702e-01 -1.67873010e-01 -1.09178853e+00 3.23088288e-01
8.67418766e-01 -6.98748161e-04 -2.90491655e-02 -2.46699914e-01
-3.41996074e-01 -9.40563679e-01 -1.34942746e+00 3.24251771e-01
6.02718353e-01 -2.46702656e-01 -4.80618566e-01 -1.01262785e-01
3.63383025e-01 -2.47664869e-01 -6.30138099e-01 3.34530264e-01
2.72194356e-01 -1.31401920e+00 7.29882896e-01 3.25748146e-01
4.63260770e-01 -6.35563374e-01 -1.53399155e-01 -1.23898757e+00
-6.58979416e-01 -4.84824568e-01 4.90809441e-01 1.16044784e+00
1.74035635e-02 -2.76113272e-01 5.63084185e-01 3.85167331e-01
-2.05889925e-01 -2.79992521e-01 -1.11683512e+00 -6.40220940e-01
-1.57772630e-01 -3.45135897e-01 -1.29422009e-01 4.30614412e-01
-5.82907081e-01 4.40296173e-01 -8.82787943e-01 2.31439099e-01
7.85320640e-01 -4.77967232e-01 2.16180190e-01 -8.40189397e-01
-8.46041590e-02 -1.67360246e-01 -5.17370462e-01 -1.07231140e+00
-3.05757016e-01 -5.23582026e-02 4.37352568e-01 -1.61564434e+00
2.62671560e-01 4.51129451e-02 -5.32725215e-01 1.89821705e-01
-1.31236330e-01 4.35308844e-01 2.94263894e-03 5.06323762e-03
-2.79406577e-01 5.65644801e-01 1.21166289e+00 -4.34181571e-01
-1.30917765e-02 -1.77184105e-01 -2.10943818e-01 6.42505050e-01
8.55449796e-01 -4.30419564e-01 -3.94099057e-01 -8.93340826e-01
-5.70699126e-02 4.75373492e-02 2.61709124e-01 -1.54923344e+00
3.82068872e-01 2.80994065e-02 8.06674719e-01 -8.88174117e-01
3.34191978e-01 -1.15979493e+00 8.79909918e-02 5.40604293e-01
-2.39603162e-01 1.20659724e-01 1.95173681e-01 7.00034738e-01
-4.68696386e-01 -7.08449900e-01 1.08164513e+00 -4.12126243e-01
-9.28104758e-01 -1.10991105e-01 -8.16795707e-01 -4.49921519e-01
7.16938615e-01 -5.74873745e-01 -4.44816053e-01 -3.60498518e-01
-5.56173384e-01 -1.04333118e-01 2.95210660e-01 5.05178198e-02
1.12802660e+00 -9.65189695e-01 -6.97269917e-01 2.77821004e-01
-4.88330424e-01 6.07952997e-02 6.40949309e-01 3.62120628e-01
-8.89803052e-01 -1.22834392e-01 -6.26542628e-01 -5.50340712e-01
-1.28431273e+00 6.04623914e-01 7.85817981e-01 -5.11117280e-02
-6.93361878e-01 8.88486445e-01 3.19800004e-02 4.07654420e-02
4.07871336e-01 -4.56272274e-01 -2.29565695e-01 -3.88879746e-01
5.15366137e-01 6.82245612e-01 2.24133953e-01 -4.78159368e-01
-2.12889314e-02 7.35974491e-01 -5.39893731e-02 -5.28973453e-02
1.51689792e+00 -2.50634313e-01 -5.71689844e-01 2.21505925e-01
1.11366308e+00 -4.11129355e-01 -1.47269833e+00 3.16707208e-03
-3.30432832e-01 -5.83644629e-01 4.52654392e-01 -8.79634261e-01
-1.49422991e+00 9.99447584e-01 1.38565385e+00 1.49327725e-01
1.95125329e+00 -9.68918324e-01 7.42448390e-01 7.32530415e-01
6.46171421e-02 -1.58433998e+00 2.46618345e-01 6.06907368e-01
8.30754161e-01 -1.11558306e+00 1.45362914e-01 -2.02249587e-01
-4.04908270e-01 1.61441207e+00 8.81453276e-01 -4.64942902e-01
6.26107454e-01 3.64543498e-01 3.52986127e-01 1.48373723e-01
-5.41650236e-01 5.83457127e-02 -2.28961289e-01 7.53845334e-01
3.38182390e-01 -4.74473834e-01 -6.14618540e-01 1.43768251e-01
4.17089611e-01 -4.61639985e-02 7.98700392e-01 7.51795530e-01
-7.52636611e-01 -8.10129285e-01 -7.80374289e-01 4.05868679e-01
-8.15590441e-01 -2.70347714e-01 4.18610983e-02 3.94922584e-01
3.94271433e-01 1.41462505e+00 2.38166992e-02 -4.85453606e-01
3.79643828e-01 -3.98172647e-01 2.90039480e-01 4.93565910e-02
-7.91402042e-01 5.94076455e-01 -2.25409344e-01 -6.98348343e-01
-6.63447917e-01 -1.29120937e-02 -9.77066278e-01 -1.98212489e-01
-3.31559271e-01 -4.88308184e-02 7.64929950e-01 4.86728728e-01
-1.44362494e-01 6.82747245e-01 8.70431066e-01 -7.99072623e-01
-5.78915298e-01 -1.35494626e+00 -9.07469809e-01 4.43319499e-01
3.93174231e-01 8.77869278e-02 -4.86656457e-01 1.25202715e-01] | [11.390233039855957, -2.346128225326538] |
5756ab42-c735-4661-8719-b081feea9465 | metaadvdet-towards-robust-detection-of | 1908.02199 | null | https://arxiv.org/abs/1908.02199v1 | https://arxiv.org/pdf/1908.02199v1.pdf | MetaAdvDet: Towards Robust Detection of Evolving Adversarial Attacks | Deep neural networks (DNNs) are vulnerable to adversarial attack which is maliciously implemented by adding human-imperceptible perturbation to images and thus leads to incorrect prediction. Existing studies have proposed various methods to detect the new adversarial attacks. However, new attack methods keep evolving constantly and yield new adversarial examples to bypass the existing detectors. It needs to collect tens of thousands samples to train detectors, while the new attacks evolve much more frequently than the high-cost data collection. Thus, this situation leads the newly evolved attack samples to remain in small scales. To solve such few-shot problem with the evolving attack, we propose a meta-learning based robust detection method to detect new adversarial attacks with limited examples. Specifically, the learning consists of a double-network framework: a task-dedicated network and a master network which alternatively learn the detection capability for either seen attack or a new attack. To validate the effectiveness of our approach, we construct the benchmarks with few-shot-fashion protocols based on three conventional datasets, i.e. CIFAR-10, MNIST and Fashion-MNIST. Comprehensive experiments are conducted on them to verify the superiority of our approach with respect to the traditional adversarial attack detection methods. | ['Hailin Shi', 'Chenxu Zhao', 'Li Chen', 'Junhai Yong', 'Dan Zeng', 'Chen Ma'] | 2019-08-06 | null | null | null | null | ['adversarial-attack-detection', 'adversarial-attack-detection'] | ['computer-vision', 'knowledge-base'] | [ 2.18730286e-01 -3.78474444e-01 2.31160671e-01 -9.86473337e-02
-4.87290353e-01 -6.80920660e-01 7.26217687e-01 -2.96807736e-01
-5.78663707e-01 5.14035940e-01 -2.79981256e-01 -7.03029633e-02
3.62626463e-01 -1.02497804e+00 -7.88258195e-01 -8.08780670e-01
-9.63323638e-02 2.37308607e-01 8.08719516e-01 -4.93413299e-01
-6.62313774e-02 3.76298398e-01 -1.23250461e+00 2.79424489e-01
5.94273806e-01 8.90449882e-01 -2.35286742e-01 7.89984405e-01
2.14575946e-01 8.99634957e-01 -1.18708897e+00 -8.42676222e-01
7.38399327e-01 -4.19886500e-01 -2.82091826e-01 -1.96235284e-01
3.01003158e-01 -7.59758592e-01 -1.11553252e+00 1.67531955e+00
7.76364982e-01 2.10556895e-01 2.63591409e-01 -1.64137137e+00
-8.61356437e-01 7.34541774e-01 -5.06389439e-01 5.97579718e-01
5.68586104e-02 8.21464419e-01 3.40266466e-01 -4.37413424e-01
4.73709375e-01 1.51977515e+00 5.85648358e-01 1.13948262e+00
-7.34645128e-01 -1.24317336e+00 2.82467186e-01 4.39109296e-01
-1.05859089e+00 -4.91170198e-01 9.91842330e-01 -1.17757574e-01
4.68817800e-01 2.01969698e-01 3.99404913e-01 1.92674017e+00
2.70487249e-01 7.34837592e-01 9.03635085e-01 1.13983288e-01
4.34947789e-01 2.18375325e-01 1.18258297e-02 4.31108326e-01
3.69601965e-01 6.28854573e-01 1.13039553e-01 -2.59974927e-01
4.12037134e-01 4.69246566e-01 -1.40238136e-01 1.44820467e-01
-7.14726985e-01 8.20435286e-01 5.99485934e-01 4.37595621e-02
-1.34497583e-01 1.46079853e-01 1.05737936e+00 8.81771684e-01
2.94041753e-01 3.08658928e-01 -5.39570212e-01 1.63181052e-01
-4.06180531e-01 1.44404665e-01 7.35149026e-01 8.82346988e-01
5.05589366e-01 6.44046485e-01 -7.60305598e-02 4.75144923e-01
1.32608535e-02 5.69686413e-01 6.55794084e-01 -4.02503997e-01
4.01476413e-01 4.65725124e-01 -1.99103475e-01 -1.25699449e+00
-4.87707667e-02 -4.65186089e-01 -1.01583815e+00 3.60277563e-01
2.05722600e-01 -6.39060736e-01 -1.10222721e+00 1.84420109e+00
4.30874914e-01 8.37293923e-01 3.81185532e-01 7.37656116e-01
7.30828464e-01 6.89178109e-01 1.01356417e-01 -2.55968660e-01
1.24255359e+00 -9.85588074e-01 -5.74417591e-01 -2.50291675e-01
1.95938557e-01 -4.99616981e-01 9.51347351e-01 4.13485080e-01
-6.36478066e-01 -5.83276808e-01 -1.24114323e+00 7.98987687e-01
-6.29097581e-01 -5.85820913e-01 1.99849203e-01 9.54182148e-01
-3.03948432e-01 4.68420327e-01 -4.64752048e-01 8.31904169e-03
7.41054773e-01 1.66810706e-01 -1.91426829e-01 -9.80562121e-02
-1.83492744e+00 6.68157339e-01 7.05067992e-01 2.52401736e-03
-1.65780723e+00 -5.51208496e-01 -4.84909385e-01 -5.83986603e-02
5.86378634e-01 -2.50712812e-01 1.12390256e+00 -1.37161839e+00
-1.31485510e+00 6.02957964e-01 7.89182186e-01 -5.34577847e-01
9.80230510e-01 -6.84889704e-02 -9.08989489e-01 1.78205639e-01
-3.13931316e-01 -5.59051745e-02 1.37023437e+00 -1.10758674e+00
-4.97790426e-01 -3.02605689e-01 4.38079208e-01 -6.72430992e-02
-8.40894878e-01 4.25206602e-01 4.79420982e-02 -9.83070493e-01
-5.11689425e-01 -8.72492611e-01 -2.63465822e-01 -1.63400173e-02
-4.19861853e-01 7.08554387e-02 1.36818993e+00 -2.29070172e-01
1.05578125e+00 -2.20261145e+00 -1.31281763e-01 -7.82420561e-02
5.14518857e-01 9.64313328e-01 -3.34510863e-01 2.58128911e-01
-1.65947184e-01 9.94779915e-02 -1.61298618e-01 1.47517636e-01
-1.65193513e-01 2.91854650e-01 -6.48273110e-01 4.23157603e-01
9.25291628e-02 7.98549950e-01 -1.14048898e+00 -2.06302151e-01
-2.07210816e-02 1.22060344e-01 -2.74806052e-01 4.65464264e-01
-3.00838232e-01 3.42957824e-01 -6.37221158e-01 7.58067012e-01
9.53025341e-01 2.01157495e-01 -2.78314143e-01 -9.42763984e-02
3.85711402e-01 -5.73947012e-01 -1.22558200e+00 1.06205487e+00
-5.49265835e-03 1.42448574e-01 -4.86599244e-02 -1.02032411e+00
8.93072665e-01 3.65221649e-01 5.49905822e-02 -4.29718316e-01
6.38824582e-01 1.13351785e-01 3.06009054e-01 -6.45786464e-01
-2.07796454e-01 -1.57559454e-01 -2.81107128e-01 3.77686679e-01
2.71542102e-01 4.19499576e-01 -3.96842621e-02 3.13083470e-01
1.71599042e+00 -5.36891937e-01 1.50072441e-01 2.68047214e-01
7.87487149e-01 -1.68504938e-01 9.35905218e-01 1.02829707e+00
-7.26280510e-01 -9.29522514e-02 4.18930352e-01 -8.38919759e-01
-9.94234264e-01 -1.19645298e+00 2.10160315e-01 1.19594371e+00
3.37172091e-01 -1.11895643e-01 -9.00530040e-01 -1.20117569e+00
-6.18591197e-02 4.30614591e-01 -6.63436592e-01 -1.11358440e+00
-5.94235063e-01 -9.58793223e-01 1.08448172e+00 4.99187469e-01
1.01603591e+00 -1.39492095e+00 -4.18529004e-01 2.53233254e-01
3.43380719e-01 -1.22449446e+00 -3.88769388e-01 7.18599930e-02
-4.14929479e-01 -1.19972599e+00 -4.46275532e-01 -7.95221686e-01
5.60332537e-01 1.56800732e-01 6.28940701e-01 1.75939277e-01
-5.94246924e-01 1.27577201e-01 -4.74698275e-01 -6.87308252e-01
-8.25382829e-01 -2.98016459e-01 5.02500117e-01 2.06358492e-01
3.94572139e-01 -6.62517548e-01 -5.05014718e-01 4.26627219e-01
-1.12456870e+00 -7.31991827e-01 6.85954452e-01 9.40611541e-01
1.09346323e-01 4.41711038e-01 7.98822284e-01 -9.73347783e-01
8.38648736e-01 -7.56999612e-01 -6.42398477e-01 2.59713084e-01
-2.99316287e-01 -2.52595067e-01 1.46109593e+00 -1.34619927e+00
-8.77701879e-01 -1.22191720e-01 -6.85437843e-02 -1.15331531e+00
-3.25121462e-01 -3.72485407e-02 -4.41111892e-01 -4.65895653e-01
1.00417626e+00 3.64460975e-01 -1.85987473e-01 -1.41675219e-01
3.33509684e-01 4.97306198e-01 6.95653141e-01 -4.38417673e-01
1.63216496e+00 2.17397392e-01 -2.54594713e-01 -4.56596285e-01
-6.10349238e-01 1.43967703e-01 -3.40936363e-01 -3.67872387e-01
5.70490241e-01 -7.07225680e-01 -8.11571598e-01 1.19486988e+00
-1.07131267e+00 -1.00366689e-01 -1.91502079e-01 2.48100027e-01
-1.45838231e-01 3.84950876e-01 -1.02651834e+00 -6.78495347e-01
-5.29918194e-01 -1.19186151e+00 2.67687082e-01 3.34026098e-01
4.16443497e-01 -7.89768338e-01 2.49988601e-01 1.19318524e-02
6.27690375e-01 6.13942802e-01 7.79792547e-01 -1.19435489e+00
-3.62130255e-01 -7.28869140e-01 -5.80746196e-02 6.21257544e-01
4.52300049e-02 1.97546959e-01 -1.13789105e+00 -5.80410242e-01
4.26936507e-01 -6.53355896e-01 8.34135473e-01 -4.11861330e-01
1.22997284e+00 -7.84051955e-01 -1.26015157e-01 6.91013575e-01
1.38907850e+00 6.21239066e-01 5.67080081e-01 3.67554128e-01
8.37780833e-01 9.88446623e-02 3.36215556e-01 4.31012541e-01
-2.41534844e-01 2.28963062e-01 8.60626698e-01 2.14220047e-01
2.59152412e-01 -1.71157852e-01 5.61324716e-01 7.46152580e-01
8.81669596e-02 -4.57347989e-01 -7.39936292e-01 1.43888788e-02
-1.68988204e+00 -1.34211028e+00 3.76633972e-01 1.88780308e+00
8.21219742e-01 7.47328401e-01 6.61569461e-02 2.58029312e-01
9.10584092e-01 4.12366658e-01 -1.15453935e+00 -3.45813155e-01
-6.71036914e-02 1.30840257e-01 4.87370253e-01 -3.69542278e-02
-1.40444922e+00 1.07565594e+00 5.71211481e+00 1.00818861e+00
-1.22379279e+00 3.20714027e-01 5.78557312e-01 -2.39699572e-01
1.37713373e-01 -2.65066057e-01 -6.66148424e-01 7.74797857e-01
9.42272544e-01 -2.92307258e-01 5.44556797e-01 1.18518841e+00
-1.58587143e-01 7.47524321e-01 -7.99254894e-01 8.04923594e-01
1.91762388e-01 -1.00570202e+00 2.94931233e-01 -3.38391572e-01
6.19889200e-01 5.39844483e-02 3.27367544e-01 8.04745972e-01
7.93893039e-01 -6.64807379e-01 4.90110964e-01 3.14103335e-01
6.40354514e-01 -1.00351572e+00 8.29972982e-01 6.14985943e-01
-1.16334295e+00 -6.72881961e-01 -7.75548100e-01 1.08445011e-01
-2.20581487e-01 1.05523922e-01 -4.05301481e-01 2.59213746e-01
6.70900941e-01 4.50669825e-01 -6.54735684e-01 7.79700875e-01
-2.37859860e-01 6.63546503e-01 -2.27622632e-02 -4.13120613e-02
5.01711249e-01 2.18061805e-01 9.17516947e-01 7.81951845e-01
-7.06551373e-02 6.41787127e-02 4.80099648e-01 6.37597978e-01
-5.27677178e-01 -1.21741697e-01 -7.97117412e-01 -5.69022708e-02
7.44070053e-01 1.41922247e+00 -4.70162958e-01 -4.04552490e-01
-4.57328916e-01 1.12178850e+00 3.58184308e-01 2.62496382e-01
-1.20450056e+00 -6.86915874e-01 7.04436481e-01 -2.59681910e-01
1.09045841e-01 2.59787291e-01 4.70545679e-01 -1.26968479e+00
-6.65589646e-02 -1.37319589e+00 5.51866710e-01 -2.89699256e-01
-1.90672755e+00 8.51875007e-01 -2.66498029e-01 -1.39802277e+00
1.38436854e-01 -6.07585132e-01 -1.37903428e+00 3.64192545e-01
-1.10339177e+00 -1.03450334e+00 -3.57713282e-01 1.04375064e+00
6.62889659e-01 -8.36211741e-01 7.39086986e-01 5.36548495e-01
-1.12984490e+00 1.17963386e+00 5.79973347e-02 7.23110139e-01
8.52095902e-01 -8.51185501e-01 6.49669170e-01 1.11512387e+00
-2.33939752e-01 4.21640038e-01 6.66051090e-01 -8.30951691e-01
-1.36400616e+00 -1.45343435e+00 -3.24224591e-01 -3.50718230e-01
1.15034175e+00 -5.42463481e-01 -1.10989034e+00 6.73787951e-01
-8.11173208e-03 5.79024911e-01 4.67532665e-01 -6.75094247e-01
-7.93300092e-01 -3.42038363e-01 -1.49191737e+00 8.14473331e-01
1.00604546e+00 -3.79393309e-01 -6.52284682e-01 3.48432571e-01
1.27702487e+00 -2.26762563e-01 -5.58095515e-01 4.37219113e-01
3.06453407e-01 -7.73370683e-01 9.91980910e-01 -1.11497211e+00
2.26871774e-01 -5.05147427e-02 -1.42419353e-01 -1.33051288e+00
-4.42943275e-01 -7.72037208e-01 -5.40885031e-01 1.04952145e+00
7.77946552e-03 -8.98323536e-01 7.07937658e-01 2.54653841e-01
-2.38813199e-02 -6.39203370e-01 -9.56410170e-01 -1.01983476e+00
1.15876071e-01 -7.74178430e-02 8.34833026e-01 1.27877092e+00
-5.35224259e-01 2.02800915e-01 -7.33224571e-01 2.98589647e-01
8.91493380e-01 -5.52588284e-01 8.51792395e-01 -1.13983893e+00
-3.45481813e-01 -3.50857049e-01 -9.03319597e-01 -2.93575287e-01
1.20231457e-01 -5.97094893e-01 -9.48406309e-02 -5.48374414e-01
9.04402211e-02 -1.41890004e-01 -8.34821403e-01 4.57601130e-01
-5.01175225e-01 1.57398820e-01 6.43981323e-02 2.47669473e-01
-6.84719443e-01 4.66727406e-01 1.06184399e+00 -5.24036348e-01
8.64615440e-02 1.01758055e-01 -6.17489398e-01 1.02997637e+00
8.66397262e-01 -7.82251656e-01 -6.17352128e-01 -2.61758178e-01
-2.23106183e-02 -1.66806698e-01 3.95355642e-01 -1.24360120e+00
3.02000940e-01 -2.08080724e-01 4.57364410e-01 -2.58476019e-01
4.75769751e-02 -9.48317885e-01 -1.20318078e-01 9.44022715e-01
-2.63205886e-01 1.49449378e-01 8.32745060e-02 1.01462924e+00
1.53968185e-01 -3.06764364e-01 1.20970666e+00 -4.77271169e-01
-9.03167009e-01 9.32019770e-01 -2.21615508e-01 3.68843704e-01
1.66338909e+00 5.53839654e-02 -7.04956532e-01 -1.47703260e-01
-6.33795977e-01 1.87970594e-01 1.96688294e-01 6.30384743e-01
8.55769336e-01 -1.36448824e+00 -7.51666129e-01 3.02522302e-01
1.47223487e-01 -3.12952667e-01 4.39308465e-01 3.02840173e-01
-4.76342469e-01 -5.55440009e-01 -4.58542615e-01 -1.02716848e-01
-1.31525540e+00 1.38275564e+00 5.07342219e-01 -1.69828147e-01
-6.84617698e-01 9.76230145e-01 3.96085456e-02 -4.33178842e-01
4.26221520e-01 4.36324328e-01 -1.22141428e-01 -1.31913155e-01
8.02777827e-01 4.89483297e-01 -3.76862198e-01 -3.99671316e-01
-2.60275334e-01 8.53101257e-03 -6.02869570e-01 4.43596810e-01
1.13222086e+00 3.25777978e-01 -1.59952492e-02 3.40364456e-01
1.22883296e+00 -2.78339744e-01 -1.23365021e+00 -5.44016063e-01
-2.57667422e-01 -4.93453681e-01 -3.57880771e-01 -6.51254117e-01
-1.41379333e+00 8.32952142e-01 8.41088057e-01 2.74431348e-01
1.07615101e+00 -3.74908358e-01 1.20193505e+00 7.40701139e-01
2.73082376e-01 -9.38491762e-01 6.26993954e-01 3.32634717e-01
5.26357472e-01 -1.32389557e+00 -1.61641672e-01 9.01944004e-03
-5.39151430e-01 1.01202977e+00 1.28820717e+00 -5.78327000e-01
7.29367614e-01 4.27633703e-01 1.01647228e-01 -4.91265580e-02
-8.75960290e-01 3.35157037e-01 -2.22859025e-01 5.68279386e-01
-6.33021057e-01 -7.48344213e-02 2.00235620e-01 6.74083054e-01
2.34210026e-02 -4.37195808e-01 4.72961366e-01 9.08441007e-01
-5.66047668e-01 -7.96304286e-01 -3.72146666e-01 4.92761374e-01
-5.77119648e-01 4.22822535e-02 -5.05783439e-01 6.05587125e-01
3.60263646e-01 8.45978022e-01 -2.16252118e-01 -9.48101878e-01
4.13794518e-01 -2.16384113e-01 1.30661845e-01 -4.16658103e-01
-8.09185028e-01 -3.37450534e-01 -4.36812162e-01 -4.27329332e-01
1.84214592e-01 -1.20201364e-01 -8.29101145e-01 -5.27641356e-01
-3.51134092e-01 -5.29343784e-02 1.30145937e-01 7.64594555e-01
-1.28793679e-02 8.06778789e-01 1.30796254e+00 -4.66392457e-01
-1.36503065e+00 -8.24940622e-01 -4.74072009e-01 6.40485287e-01
2.00351447e-01 -6.36450350e-01 -8.58109236e-01 -3.22713196e-01] | [5.544356346130371, 7.92865514755249] |
a2a7e16c-4136-4719-a7c0-bd77fec7007a | bias-reduced-hindsight-experience-replay-with | 1905.05498 | null | https://arxiv.org/abs/1905.05498v5 | https://arxiv.org/pdf/1905.05498v5.pdf | Bias-Reduced Hindsight Experience Replay with Virtual Goal Prioritization | Hindsight Experience Replay (HER) is a multi-goal reinforcement learning algorithm for sparse reward functions. The algorithm treats every failure as a success for an alternative (virtual) goal that has been achieved in the episode. Virtual goals are randomly selected, irrespective of which are most instructive for the agent. In this paper, we present two improvements over the existing HER algorithm. First, we prioritize virtual goals from which the agent will learn more valuable information. We call this property the instructiveness of the virtual goal and define it by a heuristic measure, which expresses how well the agent will be able to generalize from that virtual goal to actual goals. Secondly, we reduce existing bias in HER by the removal of misleading samples. To test our algorithms, we built two challenging environments with sparse reward functions. Our empirical results in both environments show vast improvement in the final success rate and sample efficiency when compared to the original HER algorithm. A video showing experimental results is available at https://youtu.be/3cZwfK8Nfps . | ['Armin Biess', 'Binyamin Manela'] | 2019-05-14 | null | null | null | null | ['multi-goal-reinforcement-learning'] | ['methodology'] | [-1.26435772e-01 2.15881422e-01 -1.20093450e-01 -2.26299211e-01
-7.17285573e-01 -5.18779337e-01 4.21562761e-01 5.05571961e-02
-6.64714336e-01 1.45014167e+00 3.40573788e-01 -1.72393993e-02
-3.89711350e-01 -7.03668833e-01 -6.05020285e-01 -8.64695549e-01
-4.05081987e-01 5.30614555e-01 -1.44874565e-02 -5.15772760e-01
5.38703501e-01 1.56991512e-01 -1.51697004e+00 1.90308094e-02
7.21663952e-01 6.00082040e-01 5.12472391e-01 7.37496018e-01
3.51474732e-01 1.26286149e+00 -6.68274343e-01 -2.88228601e-01
4.38172668e-01 -7.74318933e-01 -1.01139963e+00 -3.57164107e-02
-1.94162905e-01 -6.69417202e-01 -3.51353705e-01 1.05284929e+00
4.76642847e-01 5.04960477e-01 4.71254021e-01 -1.52120256e+00
-2.82225370e-01 8.74466360e-01 -4.02388155e-01 2.41223961e-01
5.70335150e-01 2.96692878e-01 1.08668673e+00 -3.67982119e-01
7.05715477e-01 1.16126609e+00 1.34349242e-01 7.23872125e-01
-1.07308972e+00 -6.29193187e-01 2.41441414e-01 3.78543913e-01
-7.60712385e-01 -4.68081921e-01 7.03195155e-01 -1.08930215e-01
8.09259534e-01 1.21737912e-01 7.76556075e-01 1.12284338e+00
1.35914460e-01 1.14690375e+00 1.30396545e+00 -2.94246584e-01
8.71391296e-01 -3.35282795e-02 -1.10497415e-01 6.01997554e-01
1.09959826e-01 6.70107186e-01 -6.17366791e-01 -1.07199833e-01
7.36251116e-01 1.86651692e-01 -2.43865460e-01 -5.42870045e-01
-1.08456588e+00 1.06865704e+00 4.46704566e-01 3.03363986e-02
-7.90848792e-01 3.62241000e-01 2.44514212e-01 7.92895854e-01
-8.54690671e-02 6.74446940e-01 -2.95344114e-01 -4.45040375e-01
-6.03552341e-01 6.69521689e-01 7.18043029e-01 7.85932004e-01
8.69283617e-01 2.28089422e-01 -2.17187643e-01 5.12629926e-01
9.23010260e-02 3.75021815e-01 6.89629912e-01 -1.44162190e+00
2.70152330e-01 3.19574356e-01 5.14853179e-01 -5.23163795e-01
-3.33295673e-01 -4.43588108e-01 -1.75075069e-01 7.14699805e-01
4.60285246e-01 -4.21223193e-01 -5.88174045e-01 1.83467603e+00
1.89133108e-01 5.14156632e-02 4.24465865e-01 9.34044719e-01
6.32286370e-01 4.25979048e-01 -1.16648644e-01 -4.26266402e-01
5.71117103e-01 -1.07514310e+00 -5.44970334e-01 -3.25558990e-01
5.88831365e-01 -3.25827986e-01 1.12738502e+00 7.18897700e-01
-1.13501811e+00 -2.65010864e-01 -9.76473153e-01 5.93531907e-01
7.10062459e-02 -3.74714613e-01 9.93160188e-01 4.73591924e-01
-9.65161622e-01 9.73881125e-01 -5.82130492e-01 -2.42979631e-01
4.45870459e-01 2.60282636e-01 -1.89627275e-01 -4.37264293e-02
-1.08702004e+00 7.85641968e-01 3.40710074e-01 -4.63866353e-01
-1.55584836e+00 -1.36159390e-01 -7.60407329e-01 9.23067983e-03
7.75435746e-01 -3.98979932e-01 1.70682847e+00 -9.99578595e-01
-1.66354167e+00 5.67973673e-01 -5.98936453e-02 -6.97883487e-01
5.04807532e-01 -9.62242782e-02 -2.23536137e-02 2.22488597e-01
3.75364840e-01 5.23740709e-01 8.18247318e-01 -1.33556867e+00
-9.88127351e-01 -2.36014992e-01 4.24801767e-01 6.08943105e-01
1.26155570e-01 -1.62554190e-01 1.62991777e-01 -3.37625325e-01
-1.50824428e-01 -7.30938375e-01 -3.99770528e-01 -4.33947116e-01
-1.04589984e-02 -3.04533869e-01 2.17547297e-01 -8.72902498e-02
9.51837242e-01 -1.95940888e+00 1.61640421e-01 3.84769365e-02
2.34845415e-01 -8.97723883e-02 -2.75817484e-01 6.63406134e-01
2.11364537e-01 -1.99256048e-01 8.94799680e-02 -1.52740836e-01
1.38507336e-02 3.63221884e-01 -3.66124809e-01 4.90979671e-01
-3.24011385e-01 7.83615530e-01 -1.42512071e+00 -2.19962209e-01
9.20013636e-02 -2.45379969e-01 -6.45213127e-01 3.96906286e-01
-2.23051161e-01 4.52874273e-01 -6.38021648e-01 5.22524655e-01
2.46089935e-01 -1.10438243e-01 7.05398321e-02 6.94943786e-01
9.95412841e-03 2.55733460e-01 -1.19455886e+00 1.50205588e+00
-2.79978782e-01 1.62721112e-01 -1.62588745e-01 -1.06529796e+00
8.99715781e-01 2.58881807e-01 3.68583649e-01 -8.70388269e-01
2.66083330e-01 2.03882575e-01 6.05130009e-02 -4.21138763e-01
6.85202837e-01 -3.13152820e-01 -1.48811579e-01 6.51509106e-01
9.72910449e-02 -1.94826990e-01 4.71675903e-01 3.71136010e-01
1.14716804e+00 4.73489881e-01 6.21144712e-01 -1.16316490e-01
1.40234455e-01 7.48027414e-02 7.25001574e-01 1.28854430e+00
-6.69997215e-01 2.17558801e-01 6.96148157e-01 -3.45480293e-01
-9.04073060e-01 -1.16201508e+00 3.29933375e-01 1.23079324e+00
3.25824171e-01 -1.75510436e-01 -2.88245678e-01 -8.00221741e-01
1.88311487e-02 9.52361286e-01 -7.90416300e-01 -9.26593915e-02
-3.14857423e-01 -3.54241431e-01 2.20846146e-01 3.07093412e-01
3.61842543e-01 -1.74504900e+00 -1.19684207e+00 3.09959352e-01
-1.12981163e-01 -5.17911136e-01 -1.22993626e-01 5.06432533e-01
-8.48961651e-01 -1.06704748e+00 -6.23178720e-01 -6.54281616e-01
2.66789675e-01 3.48764449e-01 1.13209260e+00 1.24194898e-01
8.11598636e-03 5.00126123e-01 -8.76559258e-01 -4.72393870e-01
-2.80003965e-01 -3.72791708e-01 4.62688237e-01 -4.37968999e-01
3.89843792e-01 -6.84384346e-01 -5.04025519e-01 7.08787441e-02
-5.92864394e-01 -1.59390479e-01 4.70147252e-01 1.09964752e+00
4.49943483e-01 3.99666056e-02 1.05892301e+00 -7.06690490e-01
8.46746385e-01 -6.92010701e-01 -6.39359295e-01 -9.84004512e-02
-6.65601611e-01 3.32059979e-01 8.24279189e-01 -2.86381155e-01
-8.66023362e-01 -5.56160100e-02 -4.57532704e-02 -1.46398172e-01
-1.67088345e-01 3.38458836e-01 2.65058041e-01 2.07553431e-01
8.00841868e-01 4.19008136e-01 1.92206539e-02 -7.87598938e-02
2.74594724e-01 3.90331924e-01 1.81906149e-01 -6.87668443e-01
4.40602422e-01 1.73607990e-01 -6.72515854e-02 -4.67805326e-01
-7.83465981e-01 -2.89580196e-01 5.27588502e-02 -4.62884068e-01
2.98329920e-01 -7.15626597e-01 -1.14908302e+00 1.68254644e-01
-4.71742302e-01 -7.64381647e-01 -6.69551730e-01 6.84370935e-01
-1.18093872e+00 2.55799234e-01 -2.62143254e-01 -1.15792489e+00
-1.32418483e-01 -1.12678933e+00 5.53472221e-01 5.50624132e-01
-1.84041783e-01 -6.51061296e-01 2.12504879e-01 -5.71475439e-02
2.31473207e-01 2.30188340e-01 3.10680568e-01 -7.25888610e-01
-3.79243553e-01 1.94327593e-01 2.66208470e-01 9.87429172e-02
-6.99550882e-02 -5.40575266e-01 -7.28551447e-01 -6.03078544e-01
2.82956332e-01 -9.83456314e-01 8.09857547e-01 6.00380778e-01
6.71914458e-01 -3.96024883e-01 -5.58016561e-02 4.37327683e-01
1.50795388e+00 5.09536445e-01 6.24178767e-01 8.61950696e-01
2.74115596e-02 3.85744154e-01 1.12497246e+00 1.01771271e+00
2.69211084e-01 4.11777079e-01 8.98256838e-01 4.03444946e-01
3.06729853e-01 -3.71877134e-01 6.33787930e-01 1.86819747e-01
-1.37846306e-01 -1.48306906e-01 -3.91042531e-01 6.34072185e-01
-2.06701946e+00 -1.33788967e+00 2.84815639e-01 2.34404159e+00
7.60625303e-01 2.12221295e-01 5.41148484e-01 1.31790593e-01
3.35062325e-01 1.74484834e-01 -7.97462761e-01 -5.77304006e-01
7.86347315e-02 1.96837187e-01 3.54612738e-01 6.86884582e-01
-6.16092324e-01 1.05521548e+00 5.84156752e+00 6.19615793e-01
-6.06705487e-01 -4.54630926e-02 4.25890505e-01 -6.25607371e-01
-1.50442392e-01 7.30521604e-02 -6.83431387e-01 2.16199771e-01
6.19085908e-01 -5.40054977e-01 9.42409039e-01 1.06638634e+00
1.88471377e-01 -4.57062423e-01 -1.15425909e+00 7.79582322e-01
-2.33264387e-01 -9.45188582e-01 -2.87713647e-01 -3.19659114e-02
5.54287732e-01 4.89567146e-02 2.58414969e-02 8.22558582e-01
1.06206727e+00 -9.95023787e-01 8.76256704e-01 4.78536189e-01
4.65729326e-01 -1.04735410e+00 6.27357960e-01 6.56093001e-01
-7.08372891e-01 -3.76482069e-01 -5.58789968e-01 -6.20391965e-01
-7.56660551e-02 -2.79501379e-02 -9.64900613e-01 3.74666095e-01
7.12754011e-01 5.00010014e-01 -1.51071757e-01 1.03970385e+00
-5.61474800e-01 3.91626239e-01 -1.51963472e-01 -6.53716028e-01
5.59962451e-01 -1.88745931e-01 7.25397885e-01 6.11743212e-01
3.53943437e-01 1.22009508e-01 4.65046793e-01 7.37774134e-01
1.95123181e-01 3.68186384e-02 -7.45361626e-01 1.66475326e-01
5.57523787e-01 1.16152430e+00 -4.55471694e-01 -2.17646345e-01
-3.42115364e-03 9.91558433e-01 6.02440536e-01 3.72370511e-01
-6.01743639e-01 -3.29296976e-01 5.35463572e-01 -2.69554615e-01
2.72023737e-01 9.90746319e-02 1.27927974e-01 -9.30553079e-01
-1.65152952e-01 -9.56814945e-01 5.01020312e-01 -7.96089768e-01
-9.84035969e-01 6.71964407e-01 -2.15247214e-01 -1.31285763e+00
-6.25202656e-01 -2.69188613e-01 -5.25402546e-01 7.35373735e-01
-1.45137489e+00 -5.92004180e-01 -4.84840907e-02 6.97768033e-01
7.05414116e-01 -5.25084496e-01 8.72438252e-01 -3.46813023e-01
-6.80799335e-02 4.66007441e-01 2.67108768e-01 -1.99709147e-01
5.44504702e-01 -1.51933372e+00 -8.84926468e-02 7.35520661e-01
-3.98059115e-02 4.00265247e-01 1.02164137e+00 -6.14424288e-01
-1.13670886e+00 -7.18002617e-01 5.30393660e-01 -1.60957456e-01
5.82053185e-01 1.17296658e-01 -3.85589778e-01 7.88056850e-01
1.54986963e-01 -3.22295994e-01 4.93777841e-01 1.92599460e-01
1.78373810e-02 1.73006222e-01 -1.39414787e+00 9.22218442e-01
9.27587330e-01 7.60900453e-02 -8.13686132e-01 2.43263155e-01
5.91579795e-01 -3.25107336e-01 -4.65120137e-01 2.13035420e-02
5.12220383e-01 -1.49474359e+00 7.02270567e-01 -8.76134694e-01
6.61421359e-01 -2.75703460e-01 -3.11351925e-01 -1.76002407e+00
-5.70164979e-01 -8.60113621e-01 -1.63968831e-01 6.64972901e-01
1.87075809e-01 -5.91756344e-01 8.50593984e-01 1.96038231e-01
-7.65522495e-02 -9.91912901e-01 -8.39986384e-01 -9.87877250e-01
1.62954748e-01 -2.27334395e-01 6.15637600e-01 6.96964383e-01
5.39782882e-01 3.59173626e-01 -7.06782639e-01 -1.71361744e-01
7.66722918e-01 3.75817567e-01 7.44873285e-01 -1.00234389e+00
-6.84522867e-01 -4.70242411e-01 -1.01127610e-01 -1.11173451e+00
-3.15560028e-02 -8.68677080e-01 9.80964527e-02 -1.53881919e+00
3.13261509e-01 -3.80907029e-01 -2.73019314e-01 4.68870014e-01
-2.28240639e-01 -1.41624391e-01 3.28062594e-01 1.95247188e-01
-9.93558109e-01 7.00138748e-01 1.50634789e+00 3.32526743e-01
-6.17042661e-01 2.43559316e-01 -9.81713951e-01 6.73138440e-01
1.36736071e+00 -6.61286473e-01 -4.81424153e-01 1.59872957e-02
3.22328806e-01 6.78535402e-01 2.08534688e-01 -9.95068967e-01
-2.23328080e-02 -4.36320752e-01 3.23005676e-01 -2.76890457e-01
2.53296226e-01 -5.68838894e-01 -2.07024813e-01 8.62345695e-01
-5.97321630e-01 1.20016776e-01 -1.74875423e-01 6.63367629e-01
9.25500616e-02 -7.59355068e-01 6.61610425e-01 -4.91309047e-01
-8.71104777e-01 2.03945950e-01 -5.06162167e-01 1.27529413e-01
1.17337608e+00 -1.93685994e-01 -2.37220764e-01 -9.48151648e-01
-8.28802764e-01 6.50967419e-01 4.71952111e-01 1.12104453e-01
8.86538684e-01 -1.34966218e+00 -8.73927414e-01 -3.26427035e-02
1.66157708e-01 -3.11641753e-01 6.24170229e-02 5.77459097e-01
-1.33908838e-01 1.20072223e-01 -5.38527608e-01 -1.07379951e-01
-1.03758848e+00 7.53011107e-01 2.05770746e-01 -5.28220773e-01
-7.32948601e-01 8.43482137e-01 2.01760232e-02 -2.24734038e-01
3.87939334e-01 -5.61546162e-03 -4.84878123e-01 -1.36332586e-01
6.66645527e-01 5.49635231e-01 -4.14338171e-01 -2.89354205e-01
-1.49639130e-01 -2.76469458e-02 -2.34857589e-01 -4.09636945e-01
1.64488852e+00 -2.10371450e-01 4.13821965e-01 5.26138008e-01
6.66426480e-01 -6.14836924e-02 -1.58761883e+00 -2.14125216e-01
-2.37136140e-01 -6.91423118e-01 -1.50788933e-01 -9.80513513e-01
-7.19223619e-01 3.44444126e-01 4.24397886e-01 4.73798126e-01
1.07785738e+00 -1.54566482e-01 5.17572880e-01 6.37532234e-01
9.25399899e-01 -1.17556596e+00 2.83283383e-01 6.19953752e-01
9.98464882e-01 -1.26389098e+00 -3.04387156e-02 3.86439443e-01
-1.31974232e+00 8.99788380e-01 5.21576226e-01 -4.93673861e-01
9.87803712e-02 -1.04578309e-01 -7.36612231e-02 -2.55150825e-01
-7.86239624e-01 -5.31848311e-01 -4.63866472e-01 7.46649265e-01
-1.52955549e-02 1.63318098e-01 -3.83821815e-01 5.86262345e-01
-5.05354106e-01 2.21009716e-01 9.60788250e-01 1.03053260e+00
-9.44624901e-01 -9.27686870e-01 -2.66198605e-01 5.88785112e-01
-4.25372422e-01 -3.70719528e-04 -8.34471136e-02 5.84593832e-01
-3.33322704e-01 1.22140276e+00 -2.64716625e-01 -3.66093397e-01
2.11982369e-01 -2.01619998e-01 8.37908328e-01 -6.45565510e-01
-4.82271582e-01 -5.67515530e-02 1.05639674e-01 -8.06865096e-01
-3.42681885e-01 -8.81100118e-01 -1.55885780e+00 -3.72371465e-01
-3.25361565e-02 3.93178076e-01 2.03678221e-01 8.43696535e-01
-4.35693637e-02 5.85913658e-01 9.28333759e-01 -8.49515259e-01
-1.04572427e+00 -9.08671081e-01 -9.76043701e-01 3.52987260e-01
5.29902160e-01 -9.44086730e-01 -6.00670338e-01 -3.46703827e-01] | [4.110002517700195, 1.6647841930389404] |
9e4cc64e-f227-46eb-b399-0cebfef21cf0 | knowledge-graph-contrastive-learning-for | 2205.00976 | null | https://arxiv.org/abs/2205.00976v2 | https://arxiv.org/pdf/2205.00976v2.pdf | Knowledge Graph Contrastive Learning for Recommendation | Knowledge Graphs (KGs) have been utilized as useful side information to improve recommendation quality. In those recommender systems, knowledge graph information often contains fruitful facts and inherent semantic relatedness among items. However, the success of such methods relies on the high quality knowledge graphs, and may not learn quality representations with two challenges: i) The long-tail distribution of entities results in sparse supervision signals for KG-enhanced item representation; ii) Real-world knowledge graphs are often noisy and contain topic-irrelevant connections between items and entities. Such KG sparsity and noise make the item-entity dependent relations deviate from reflecting their true characteristics, which significantly amplifies the noise effect and hinders the accurate representation of user's preference. To fill this research gap, we design a general Knowledge Graph Contrastive Learning framework (KGCL) that alleviates the information noise for knowledge graph-enhanced recommender systems. Specifically, we propose a knowledge graph augmentation schema to suppress KG noise in information aggregation, and derive more robust knowledge-aware representations for items. In addition, we exploit additional supervision signals from the KG augmentation process to guide a cross-view contrastive learning paradigm, giving a greater role to unbiased user-item interactions in gradient descent and further suppressing the noise. Extensive experiments on three public datasets demonstrate the consistent superiority of our KGCL over state-of-the-art techniques. KGCL also achieves strong performance in recommendation scenarios with sparse user-item interactions, long-tail and noisy KG entities. Our implementation codes are available at https://github.com/yuh-yang/KGCL-SIGIR22 | ['Chenliang Li', 'Lianghao Xia', 'Chao Huang', 'Yuhao Yang'] | 2022-05-02 | null | null | null | null | ['general-knowledge'] | ['miscellaneous'] | [-2.68741012e-01 2.86244471e-02 -7.67362773e-01 -3.08298409e-01
-2.16405228e-01 -3.53308260e-01 1.12127893e-01 1.32239312e-01
7.64468759e-02 7.10696816e-01 7.19865024e-01 2.01274958e-02
-6.37723148e-01 -1.14016497e+00 -7.47246504e-01 -5.29871881e-01
-2.25521848e-02 2.32586756e-01 -1.18769728e-01 -5.49775362e-01
-8.83572549e-02 -2.76092608e-02 -1.17434597e+00 3.01005840e-01
1.33933306e+00 8.81819725e-01 6.39306679e-02 -2.16320977e-02
-3.72135341e-01 5.50685167e-01 -3.11309546e-01 -7.96576798e-01
3.05261672e-01 -2.54067540e-01 -4.02580023e-01 -6.78967759e-02
3.21862459e-01 -5.23851886e-02 -6.06269360e-01 1.14261830e+00
5.28949499e-01 3.30064148e-01 6.20972037e-01 -1.26512396e+00
-1.50326669e+00 1.14586496e+00 -7.93965816e-01 1.96928397e-01
3.10419261e-01 -1.74706012e-01 1.53468978e+00 -1.05117321e+00
3.72116983e-01 1.19109917e+00 7.23030508e-01 2.53219455e-01
-1.08266282e+00 -8.96003425e-01 8.10299695e-01 2.78512090e-01
-1.56173325e+00 6.94363797e-03 8.92112434e-01 -2.06861123e-01
5.77442348e-01 2.49220252e-01 8.16184640e-01 1.23228705e+00
-3.37522209e-01 1.01869237e+00 6.27500713e-01 -8.59240294e-02
-8.32536146e-02 3.68458450e-01 5.53530872e-01 6.35171235e-01
9.31849182e-01 -9.40061882e-02 -7.01787710e-01 -2.60742366e-01
8.14697087e-01 4.51357991e-01 -6.21446371e-01 -3.93381834e-01
-9.77193296e-01 8.02053750e-01 9.29620802e-01 2.64106333e-01
-5.15148222e-01 -1.56779870e-01 2.07389221e-01 4.51970786e-01
5.61033070e-01 4.34362143e-01 -6.00049675e-01 4.01343644e-01
-2.66909540e-01 9.14670378e-02 7.22764194e-01 1.09212291e+00
8.04788828e-01 2.79124945e-01 -2.39825606e-01 9.64652359e-01
4.71666962e-01 5.83408356e-01 4.86484110e-01 -1.86059237e-01
6.13872647e-01 9.36972916e-01 -1.42091528e-01 -1.50415599e+00
-3.32229465e-01 -1.16703022e+00 -1.01135504e+00 -5.96015453e-01
1.68036148e-01 -2.57821351e-01 -6.34209931e-01 1.76980400e+00
3.52142334e-01 4.28507566e-01 7.11863954e-03 1.11839736e+00
1.29563069e+00 4.04543281e-01 8.37240890e-02 -2.46068954e-01
1.18158484e+00 -9.21473742e-01 -8.31996977e-01 -6.19429648e-02
6.59280598e-01 -3.07149857e-01 1.21442199e+00 3.18520904e-01
-6.33343101e-01 -4.80660200e-01 -8.98304999e-01 2.05755234e-01
-4.12434995e-01 2.68869013e-01 1.10664213e+00 5.26674151e-01
-6.09187365e-01 4.00661558e-01 -2.98424661e-01 3.59457172e-02
5.73107600e-01 3.36067200e-01 -2.50911325e-01 -3.60575676e-01
-1.69698644e+00 4.30092126e-01 3.53560776e-01 2.14852810e-01
-2.68549353e-01 -9.35509026e-01 -7.70731211e-01 2.65260011e-01
8.16595554e-01 -9.89271462e-01 5.92204630e-01 -9.58690524e-01
-1.12718403e+00 1.11456811e-01 1.14834212e-01 -1.05956584e-01
6.29619658e-02 -3.96929950e-01 -9.71289933e-01 -3.95182490e-01
-3.51408002e-04 -6.43572658e-02 8.41081798e-01 -1.43215668e+00
-4.27097738e-01 -5.88496208e-01 2.57384300e-01 5.18014252e-01
-6.38391376e-01 -6.18502975e-01 -6.62698269e-01 -1.02290785e+00
9.69272405e-02 -6.85973823e-01 -1.99280426e-01 -5.69720566e-01
-4.08902764e-01 -2.52190500e-01 4.66365159e-01 -4.28121507e-01
1.58580959e+00 -2.03528428e+00 -1.03749020e-03 5.42037010e-01
4.77089256e-01 3.71091574e-01 -3.24282855e-01 4.30530399e-01
2.23217532e-02 8.70572105e-02 2.52886921e-01 5.08907475e-02
-9.59368944e-02 3.34871680e-01 -3.19727063e-01 8.24195147e-02
-7.93415383e-02 1.24622452e+00 -1.21764576e+00 -2.59553716e-02
-1.25983983e-01 5.06917059e-01 -5.72800219e-01 1.71766039e-02
-1.44511357e-01 2.09526911e-01 -9.13198411e-01 6.76169395e-01
6.57608509e-01 -7.28907526e-01 3.47243994e-01 -7.22769916e-01
4.26206380e-01 3.48378628e-01 -1.54403079e+00 1.54037797e+00
-2.82040775e-01 -2.52847373e-01 -7.93958735e-03 -1.02148819e+00
9.29365277e-01 -1.27952555e-02 3.50155532e-01 -6.70421720e-01
4.10393402e-02 3.11177373e-02 -4.09474634e-02 -2.77863115e-01
6.08716607e-01 1.44482881e-01 1.56607047e-01 3.12795669e-01
2.94916838e-01 5.64115286e-01 1.38308227e-01 7.46578097e-01
7.40714610e-01 -1.37433693e-01 2.56075084e-01 -1.81682676e-01
3.82181913e-01 -3.41795474e-01 8.64148736e-01 7.35899508e-01
2.80738205e-01 2.92209804e-01 1.60228640e-01 -2.15471759e-01
-2.63825446e-01 -8.77387702e-01 1.61335722e-01 1.26234114e+00
4.68343943e-01 -7.82780230e-01 -1.03977814e-01 -1.12750781e+00
4.80662256e-01 5.48650861e-01 -7.00301290e-01 -5.63287973e-01
-2.23755211e-01 -1.01284504e+00 1.03828728e-01 6.64638817e-01
2.10869253e-01 -7.41415977e-01 7.73176134e-01 1.70317248e-01
-1.97582722e-01 -7.27556288e-01 -7.63246894e-01 -2.77760178e-02
-8.17459643e-01 -1.16151178e+00 -5.81820726e-01 -4.34342831e-01
8.25690925e-01 8.55362713e-01 1.41746581e+00 2.03039378e-01
2.62086123e-01 4.71347153e-01 -8.10872793e-01 -3.57389271e-01
2.84050941e-01 2.29246085e-04 4.03509080e-01 2.34795392e-01
5.57435513e-01 -7.59992898e-01 -8.21539581e-01 5.04952908e-01
-7.95592129e-01 -1.76827237e-01 7.51316905e-01 9.14824903e-01
6.68138087e-01 1.91289306e-01 1.15686107e+00 -1.47936833e+00
9.93361771e-01 -9.71014142e-01 -2.69157946e-01 3.92513007e-01
-1.15755939e+00 -6.30122703e-03 8.30037057e-01 -4.91537809e-01
-1.13104665e+00 -5.02264082e-01 1.47327274e-01 -4.95686859e-01
3.22331756e-01 1.02210975e+00 -2.89726973e-01 -6.97951764e-02
6.61905766e-01 3.99773456e-02 -4.23732013e-01 -7.02715397e-01
7.87265599e-01 4.15907085e-01 1.60394192e-01 -5.67566097e-01
8.66557062e-01 2.42348731e-01 -3.10666859e-01 -3.97326469e-01
-1.24631119e+00 -7.42345095e-01 -1.76580235e-01 9.63178203e-02
1.19923696e-01 -1.32191145e+00 -3.88403893e-01 5.15951701e-02
-4.45723206e-01 2.35523194e-01 -4.87517744e-01 7.70612359e-01
1.82705268e-01 3.75521004e-01 -4.97428745e-01 -5.72323263e-01
-6.09538376e-01 -6.15433991e-01 5.93613803e-01 4.55037624e-01
2.38285914e-01 -9.64189053e-01 2.02508215e-02 5.76136768e-01
4.38120484e-01 -3.15924317e-01 7.33743012e-01 -8.59066427e-01
-5.68657398e-01 -9.68160406e-02 -4.26025003e-01 3.05133909e-01
5.69173574e-01 -3.59839082e-01 -6.30090177e-01 -2.89241552e-01
-4.85369593e-01 -2.71247268e-01 1.00191224e+00 3.29458594e-01
1.01814425e+00 -5.76782405e-01 -3.84001285e-01 5.50192535e-01
1.25175345e+00 -2.49411121e-01 1.90450549e-01 -1.30403265e-01
1.37975287e+00 3.43284369e-01 7.16349959e-01 6.07530951e-01
7.84414887e-01 5.71021914e-01 1.93803698e-01 -1.43349558e-01
-1.83408096e-01 -7.38660276e-01 2.26635262e-01 1.36260986e+00
-2.75843352e-01 -3.96921217e-01 -3.71280670e-01 5.47013283e-01
-2.14663434e+00 -8.36727023e-01 -3.51645261e-01 2.22538257e+00
8.22407842e-01 -1.61761060e-01 1.95875153e-01 -1.91385075e-01
6.40689731e-01 1.22818232e-01 -6.57206953e-01 2.59359658e-01
-3.15822899e-01 -1.52460471e-01 4.98545796e-01 1.93479985e-01
-9.00727630e-01 8.37394238e-01 4.78347492e+00 1.00213170e+00
-7.50617802e-01 -3.18066739e-02 3.44423562e-01 -8.78389478e-02
-8.09802234e-01 -3.07369411e-01 -9.05467212e-01 5.81259072e-01
5.20083189e-01 -4.30749953e-01 4.35648352e-01 9.52229321e-01
8.07591602e-02 3.54529798e-01 -6.53874695e-01 1.16425920e+00
9.44520086e-02 -1.29665709e+00 4.41244841e-01 7.30702505e-02
1.10470831e+00 9.62882712e-02 6.06807657e-02 7.20960021e-01
8.11144710e-01 -7.43416190e-01 2.10434377e-01 5.34289956e-01
4.85399783e-01 -8.25739801e-01 8.39162171e-01 2.25972474e-01
-1.45889449e+00 -1.47826463e-01 -6.98909342e-01 1.03242464e-01
6.73968568e-02 1.06881154e+00 -6.06933475e-01 1.29594052e+00
7.19260633e-01 1.18438566e+00 -4.47505623e-01 8.94458652e-01
-4.49808419e-01 8.32366168e-01 -1.89784467e-01 1.00127652e-01
3.98287512e-02 -6.69109702e-01 4.27293152e-01 1.06317103e+00
2.41229162e-01 4.49076056e-01 3.50239486e-01 7.11375177e-01
-6.09800100e-01 5.90395212e-01 -5.83374679e-01 -1.84352785e-01
5.83007991e-01 1.44030786e+00 -3.44084054e-01 -2.88159698e-01
-8.18846166e-01 6.45430028e-01 5.92620730e-01 5.89547455e-01
-5.07636249e-01 -9.82107520e-02 7.09183514e-01 1.22370563e-01
4.34523821e-01 1.45125911e-01 -9.13074389e-02 -1.63283718e+00
1.17684409e-01 -9.66656804e-01 8.06368768e-01 -4.28833574e-01
-2.02170157e+00 2.73346454e-01 -2.96130240e-01 -1.20175862e+00
2.23313600e-01 -3.01601797e-01 -4.74214733e-01 7.62361288e-01
-1.63451505e+00 -1.26348388e+00 -4.23190236e-01 9.28279519e-01
5.42342737e-02 -2.55767465e-01 5.96110344e-01 6.81040525e-01
-6.12739682e-01 9.35402393e-01 2.33752668e-01 1.18749008e-01
7.44024098e-01 -1.31387591e+00 4.57369946e-02 5.65796435e-01
5.76535940e-01 1.07667315e+00 2.14758739e-01 -9.52864647e-01
-1.73822963e+00 -1.31105793e+00 2.76804447e-01 -4.36693162e-01
7.64307916e-01 -9.33832601e-02 -1.19989824e+00 6.92703068e-01
-2.30609283e-01 2.53360659e-01 1.12318516e+00 9.45077121e-01
-6.57184899e-01 -2.73887366e-01 -9.04794931e-01 5.10181308e-01
1.41294634e+00 -3.80599916e-01 -5.12674809e-01 3.00641835e-01
8.13455462e-01 -7.15137646e-02 -1.02889502e+00 5.39213300e-01
4.86096889e-01 -5.49045324e-01 1.04074097e+00 -8.67351949e-01
6.14675358e-02 -4.95282263e-01 -6.05036542e-02 -1.71578324e+00
-8.59889269e-01 -3.04581434e-01 -7.66140044e-01 1.34624958e+00
5.35060823e-01 -6.28008366e-01 8.62093985e-01 4.01972830e-01
-8.45783129e-02 -9.29117441e-01 -2.03084737e-01 -6.76118195e-01
-3.57190967e-01 -1.36323720e-01 7.93565631e-01 1.46917105e+00
1.78275391e-01 7.33920693e-01 -6.99104011e-01 2.98337221e-01
5.59360564e-01 4.09545958e-01 7.05877066e-01 -1.40253723e+00
-4.94273067e-01 -2.74281919e-01 -2.28513360e-01 -1.17661953e+00
-6.17942810e-02 -1.15589511e+00 -5.80939770e-01 -1.61089933e+00
3.42492163e-01 -7.27044404e-01 -9.48930800e-01 4.85505372e-01
-7.24122703e-01 5.05660325e-02 -6.54007718e-02 3.40453088e-01
-9.03586090e-01 8.43573213e-01 1.43523693e+00 -1.44632250e-01
-3.58887702e-01 2.01400712e-01 -1.49540043e+00 5.76405764e-01
6.19401455e-01 -3.27847630e-01 -1.02240634e+00 -3.68979335e-01
7.49616444e-01 -2.97540724e-01 -5.97963221e-02 -3.89635861e-01
1.83128491e-01 -6.64880201e-02 4.54258770e-01 -2.93772012e-01
5.50791733e-02 -8.41204226e-01 2.79617488e-01 -9.06388089e-02
-2.15561047e-01 -2.16961712e-01 -2.04418793e-01 1.17498291e+00
-3.24048609e-01 1.11003526e-01 1.89413249e-01 -2.18353629e-01
-7.37088501e-01 7.45411694e-01 3.55230004e-01 2.10151792e-01
4.26194191e-01 7.15736905e-03 -4.41955626e-01 -4.51203585e-01
-6.96170032e-01 5.04937530e-01 2.33812809e-01 6.61768079e-01
6.27117276e-01 -1.59742308e+00 -7.25802362e-01 1.13648258e-01
4.08749700e-01 -1.52775750e-01 5.48477530e-01 8.67698312e-01
3.27222705e-01 2.08029523e-01 1.65597230e-01 -5.09331301e-02
-1.03608406e+00 6.23342752e-01 -2.25154907e-02 -4.04715419e-01
-5.57494342e-01 1.13020957e+00 4.60199177e-01 -5.47474802e-01
1.39941886e-01 -5.19349612e-02 -6.34524465e-01 2.30555311e-01
6.17035151e-01 3.32711965e-01 1.64401650e-01 -4.78672445e-01
-2.57495314e-01 2.98760980e-01 -5.11747241e-01 6.99801683e-01
1.40832400e+00 -3.00564587e-01 1.67666391e-01 2.50956655e-01
7.71079302e-01 4.28184509e-01 -7.31021821e-01 -9.15571630e-01
-1.94473863e-01 -7.35328615e-01 2.39574671e-01 -9.22115564e-01
-1.60542727e+00 3.17653596e-01 2.20000282e-01 3.39928657e-01
1.04371750e+00 -3.80273763e-04 7.21099019e-01 5.01019239e-01
5.51590741e-01 -9.45963144e-01 1.11629227e-02 2.68210649e-01
7.89986849e-01 -1.37006807e+00 3.21023971e-01 -8.14795673e-01
-9.18697596e-01 5.90534568e-01 8.24316502e-01 -1.30308703e-01
9.19772089e-01 -1.87475517e-01 -9.99229550e-02 -4.59885955e-01
-6.95490599e-01 -4.07688379e-01 8.07821035e-01 5.51835775e-01
4.81070250e-01 3.27447951e-01 -3.98105443e-01 1.35508299e+00
2.58601625e-02 5.85586624e-03 1.64265990e-01 5.20611227e-01
-1.90240741e-01 -1.10517240e+00 1.04440995e-01 1.01466644e+00
-3.91403139e-01 -4.52693224e-01 -4.68312442e-01 5.26548743e-01
5.13178743e-02 9.01681244e-01 -5.93497574e-01 -6.80196583e-01
6.09396040e-01 -2.63633460e-01 2.17083216e-01 -6.26944244e-01
-6.09001994e-01 -5.98140396e-02 2.51774788e-01 -5.43879688e-01
-2.73666590e-01 -2.13491186e-01 -1.06803834e+00 -3.71838629e-01
-8.89412224e-01 5.02726257e-01 8.46497342e-02 6.44578993e-01
8.98107648e-01 6.49278700e-01 6.88023806e-01 -3.65299731e-01
-5.17812014e-01 -9.48513389e-01 -1.00329792e+00 7.17753649e-01
-2.01967973e-02 -1.07357395e+00 -2.71293461e-01 -3.93591791e-01] | [10.218192100524902, 5.603754997253418] |
6df2aacf-826c-438a-b7b7-2f1bda33b48a | sparse-spn-depth-completion-from-sparse | 2212.00987 | null | https://arxiv.org/abs/2212.00987v1 | https://arxiv.org/pdf/2212.00987v1.pdf | Sparse SPN: Depth Completion from Sparse Keypoints | Our long term goal is to use image-based depth completion to quickly create 3D models from sparse point clouds, e.g. from SfM or SLAM. Much progress has been made in depth completion. However, most current works assume well distributed samples of known depth, e.g. Lidar or random uniform sampling, and perform poorly on uneven samples, such as from keypoints, due to the large unsampled regions. To address this problem, we extend CSPN with multiscale prediction and a dilated kernel, leading to much better completion of keypoint-sampled depth. We also show that a model trained on NYUv2 creates surprisingly good point clouds on ETH3D by completing sparse SfM points. | ['Derek Hoiem', 'Jae Yong Lee', 'Yuqun Wu'] | 2022-12-02 | null | null | null | null | ['depth-completion'] | ['computer-vision'] | [ 1.21776342e-01 8.94109756e-02 -1.67888597e-01 -3.29196334e-01
-9.02543604e-01 -4.76096481e-01 5.17004550e-01 -2.31440607e-02
-2.74305373e-01 8.77142310e-01 6.62710965e-02 4.24290039e-02
3.18553805e-01 -8.56963098e-01 -9.21280324e-01 -2.35125035e-01
-7.11667985e-02 1.07728028e+00 4.63063687e-01 1.95482969e-01
6.40244424e-01 7.29105413e-01 -1.57676363e+00 -1.04287222e-01
9.08829808e-01 7.24055111e-01 4.43736464e-01 7.48116136e-01
-3.27440977e-01 3.11581403e-01 -2.92833090e-01 -3.45232785e-02
4.96921659e-01 -3.14798094e-02 -5.95750690e-01 2.00737402e-01
1.08868110e+00 -8.92221034e-01 -1.65281773e-01 8.14353108e-01
1.49695769e-01 7.09146112e-02 6.05288148e-01 -1.28076422e+00
-8.86748433e-02 -1.34020910e-01 -7.42838800e-01 -4.20475125e-01
4.92582828e-01 1.00371882e-01 6.08986139e-01 -1.14631295e+00
8.62825155e-01 1.40722752e+00 8.94913793e-01 4.33432251e-01
-1.57058930e+00 -6.66910887e-01 -4.12214324e-02 -7.76488185e-02
-1.33786547e+00 -4.16757077e-01 5.78034401e-01 -2.69828856e-01
9.91026998e-01 -3.31383497e-02 9.20425713e-01 8.13710511e-01
3.06736529e-01 6.11483276e-01 1.04606485e+00 -1.99081361e-01
5.32256901e-01 -1.34770930e-01 -3.13563317e-01 4.58163053e-01
3.66562843e-01 6.07281663e-02 -6.27737582e-01 -6.89307451e-01
1.34025800e+00 3.90860945e-01 -2.29347467e-01 -7.97185421e-01
-1.28950477e+00 8.91133010e-01 6.05949283e-01 -4.49478567e-01
-3.97516251e-01 4.07817334e-01 -5.24186678e-02 9.64137316e-02
6.43131733e-01 2.72713751e-01 -5.55053473e-01 -1.96201503e-01
-1.32823527e+00 8.01289082e-01 7.13600278e-01 1.35116553e+00
1.51243424e+00 -2.42612839e-01 5.40555060e-01 4.81814235e-01
2.48564735e-01 9.90404308e-01 -1.25106528e-01 -1.80016708e+00
4.42287922e-01 1.82078093e-01 4.34834570e-01 -5.64693689e-01
-4.18626424e-03 2.61815209e-02 -7.02179372e-01 6.36190951e-01
3.07357222e-01 1.67224705e-01 -1.18219292e+00 1.23335671e+00
4.96798217e-01 4.72736090e-01 3.26708034e-02 7.17124283e-01
2.18725190e-01 5.58613837e-01 -3.40825111e-01 7.59054646e-02
7.00665832e-01 -6.24217212e-01 -1.56372592e-01 -6.34094894e-01
4.75054026e-01 -6.65559113e-01 8.37643325e-01 6.78249836e-01
-1.13386321e+00 -3.39419991e-01 -7.35604465e-01 -3.62944573e-01
1.80390969e-01 -4.31529015e-01 8.06166112e-01 4.22814637e-02
-1.13866127e+00 9.82175827e-01 -1.16579998e+00 -3.58314306e-01
5.37061095e-01 2.29145214e-01 -4.82776374e-01 -6.73017681e-01
-4.21573043e-01 6.91850007e-01 -9.09855068e-02 -1.30843759e-01
-1.06507921e+00 -1.02150023e+00 -8.36904764e-01 -4.25949782e-01
4.58535179e-02 -1.05647445e+00 1.34139788e+00 -3.96366239e-01
-1.22833037e+00 8.08182538e-01 -6.51880741e-01 -6.63664997e-01
6.12666845e-01 -3.56954068e-01 5.74078262e-01 3.22055697e-01
4.32932496e-01 1.34337401e+00 8.52169693e-01 -1.43651354e+00
-5.78184664e-01 -7.78319776e-01 -1.74078926e-01 2.55191177e-01
3.73592675e-01 -5.08994520e-01 -2.01194569e-01 -1.44473335e-03
8.56957018e-01 -1.06682181e+00 -7.31798172e-01 5.63945651e-01
-1.52556747e-01 1.11641973e-01 7.59800792e-01 -5.18491328e-01
1.59636840e-01 -1.90527594e+00 8.62913355e-02 7.86169022e-02
4.33769673e-01 -3.28765243e-01 -1.75876886e-01 5.94005525e-01
3.76153827e-01 -1.06433474e-01 -3.82904857e-01 -1.07225168e+00
-2.28051022e-01 5.85070491e-01 -4.77172107e-01 6.50949359e-01
-1.44169889e-02 5.81435084e-01 -8.94287765e-01 -5.07614374e-01
5.14674485e-01 6.53337657e-01 -7.38755345e-01 -5.12480084e-03
-5.73072970e-01 5.75317025e-01 -3.84610623e-01 7.86483645e-01
1.28191423e+00 -1.35748297e-01 -3.65364134e-01 1.81106150e-01
-2.74148196e-01 3.14204276e-01 -1.04735100e+00 2.43824983e+00
-5.98601699e-01 5.43034136e-01 5.14621496e-01 -3.31782818e-01
8.07194114e-01 8.06553885e-02 5.56701005e-01 -8.41487646e-02
-3.60039204e-01 4.37789828e-01 -6.08418286e-01 1.09115869e-01
9.08127069e-01 -3.53313118e-01 4.27828878e-01 9.23060551e-02
-1.42803878e-01 -1.16625261e+00 -5.49659550e-01 4.03811306e-01
1.25052440e+00 4.28604454e-01 1.45647023e-02 3.77380592e-03
-1.75482526e-01 5.88083029e-01 5.99978507e-01 5.80809474e-01
-5.87544590e-02 1.15449238e+00 2.17147753e-01 -2.77792275e-01
-1.43239152e+00 -1.29694068e+00 -5.02737045e-01 -7.57738808e-03
2.67516226e-01 -4.62746978e-01 -3.32448572e-01 -5.02729356e-01
4.25479650e-01 4.55132067e-01 -2.30651125e-01 4.03723776e-01
-3.11174661e-01 -8.81845951e-02 7.25802854e-02 4.99094516e-01
4.63184923e-01 -7.28466451e-01 -5.81422567e-01 2.82232165e-01
-7.58018345e-02 -1.16103351e+00 6.29351987e-03 2.64133722e-01
-1.62404358e+00 -8.62132251e-01 -8.16621840e-01 -4.45004761e-01
7.00470984e-01 7.65731931e-01 1.30442762e+00 -2.19142616e-01
-4.62420247e-02 5.36092877e-01 -1.82709768e-01 -4.35049117e-01
-2.37819236e-02 -1.57925650e-01 7.09269047e-02 -7.11302638e-01
3.89138728e-01 -1.03920031e+00 -7.06616759e-01 2.98265755e-01
-6.67308033e-01 8.10395107e-02 4.21710819e-01 5.34935474e-01
1.11786830e+00 -2.12573871e-01 -1.20088318e-02 -5.47228098e-01
6.16101883e-02 -3.27396899e-01 -8.68140757e-01 -4.59579140e-01
-3.21353793e-01 -8.59017074e-02 3.07715684e-01 -2.70626575e-01
-5.79068124e-01 3.73470157e-01 -1.86611265e-01 -9.62695897e-01
-3.40040028e-01 8.14004764e-02 1.49872407e-01 -3.80265206e-01
8.01544070e-01 1.88721925e-01 2.62025803e-01 -4.92137849e-01
2.94256032e-01 4.79606837e-01 4.67221826e-01 -7.23522365e-01
9.00501311e-01 1.10797203e+00 3.39605361e-01 -1.20553684e+00
-4.67220455e-01 -6.35590971e-01 -6.99428976e-01 1.95880651e-01
5.81733167e-01 -1.32052970e+00 -3.23348939e-01 3.49497050e-01
-1.40691507e+00 -6.89191103e-01 -5.58192253e-01 6.40478373e-01
-9.00631607e-01 5.55885673e-01 -6.57691538e-01 -8.96392524e-01
-1.21951714e-01 -9.51133966e-01 1.78870463e+00 -2.16478884e-01
-2.09458932e-01 -6.90340042e-01 2.60376096e-01 1.49942219e-01
2.00710356e-01 3.65488648e-01 3.45863998e-01 2.30528399e-01
-1.33201647e+00 -2.68547684e-01 -1.32688209e-01 3.94717634e-01
-8.65420327e-02 -2.55226926e-03 -1.01898766e+00 -3.12646955e-01
1.62019596e-01 -5.60699582e-01 9.13149476e-01 5.93090296e-01
8.30015302e-01 2.85615742e-01 -6.42488599e-01 8.72927547e-01
1.58081138e+00 -3.16601217e-01 8.08524430e-01 1.01034664e-01
6.56928360e-01 5.86308479e-01 8.82021546e-01 4.85625952e-01
5.92587054e-01 4.49828118e-01 8.16097796e-01 1.84912935e-01
3.33263702e-03 -7.10310340e-01 8.60356912e-02 3.71623784e-01
4.28657653e-03 2.24258110e-01 -1.06983006e+00 5.33197045e-01
-1.65269196e+00 -5.55108905e-01 -3.41211587e-01 2.33889890e+00
4.93151307e-01 1.87182456e-01 -1.99214607e-01 1.27417287e-02
1.48607224e-01 9.84426960e-02 -7.17715979e-01 8.94477740e-02
7.22166896e-02 4.57246125e-01 8.81081343e-01 8.84580135e-01
-6.16410375e-01 1.07645202e+00 6.30358744e+00 5.80247164e-01
-7.42049038e-01 4.23656590e-03 7.41649270e-02 -2.03810886e-01
-7.20266342e-01 4.64314401e-01 -1.00483537e+00 1.63472846e-01
5.89638829e-01 1.05350874e-01 4.02910680e-01 1.13336873e+00
3.72144312e-01 -7.75716126e-01 -1.03386676e+00 1.36366904e+00
-2.74472702e-02 -1.25825107e+00 -1.29590675e-01 6.93308115e-01
9.34092283e-01 6.59135520e-01 -3.38040560e-01 -2.20102351e-02
6.10135317e-01 -7.89882898e-01 5.38407207e-01 3.33648384e-01
9.47303772e-01 -6.02653325e-01 2.52917200e-01 8.34409058e-01
-1.08032048e+00 4.75091904e-01 -1.03879929e+00 -4.46489841e-01
2.84003854e-01 1.10327673e+00 -1.07390666e+00 2.98921764e-01
5.62369287e-01 9.48674142e-01 -2.07406834e-01 1.16665959e+00
-3.88237566e-01 3.10240179e-01 -1.06357944e+00 2.11916983e-01
9.41838324e-02 -4.22350317e-01 6.42772496e-01 5.11710405e-01
6.08872533e-01 -1.31028399e-01 2.48344734e-01 1.02410424e+00
7.52196386e-02 -2.89940268e-01 -1.10485697e+00 4.82420623e-01
5.65061390e-01 1.05319107e+00 -5.36863327e-01 -3.21392119e-01
-2.87358373e-01 1.15065634e+00 3.07334483e-01 3.65291327e-01
-5.01713514e-01 -5.03341481e-02 1.06252837e+00 6.07115209e-01
6.17437772e-02 -1.02780735e+00 -3.73330921e-01 -1.30084574e+00
1.29796371e-01 -3.21629941e-01 -3.40369850e-01 -1.15212619e+00
-1.11558366e+00 2.67240494e-01 3.14562693e-02 -1.31081486e+00
-3.80259007e-01 -2.79878587e-01 -4.03542936e-01 1.19381166e+00
-1.66919720e+00 -1.07792497e+00 -6.13282919e-01 5.93179107e-01
5.05116999e-01 5.12150824e-01 7.64061153e-01 -6.63238391e-02
3.43700051e-01 -3.00513238e-01 2.02210136e-02 -4.83837873e-01
7.33418941e-01 -1.32727051e+00 9.06254470e-01 5.67019343e-01
1.49227157e-01 4.49084461e-01 7.02700078e-01 -1.11555612e+00
-1.59821689e+00 -1.00446451e+00 7.34210730e-01 -7.42551386e-01
2.87635177e-01 -5.02473891e-01 -8.41359258e-01 8.67969751e-01
-2.34204352e-01 3.14314753e-01 5.23438351e-03 1.70449335e-02
-1.78007647e-01 7.41654113e-02 -1.37707484e+00 3.92250419e-01
1.15568674e+00 -5.62825084e-01 -5.10948777e-01 4.50525045e-01
7.23782241e-01 -6.35347545e-01 -6.68863297e-01 4.41133797e-01
4.84610558e-01 -1.18965507e+00 1.11942327e+00 -4.85483073e-02
5.09075999e-01 -4.59124744e-01 -4.82637495e-01 -1.36613381e+00
-7.54007092e-03 -4.91453677e-01 -4.71732654e-02 7.79667139e-01
-2.44955197e-02 -5.63548386e-01 1.57529259e+00 6.11682415e-01
-1.97961628e-01 -3.12841028e-01 -9.91409719e-01 -8.45261872e-01
1.82557210e-01 -7.37799764e-01 4.73277718e-01 6.65393233e-01
-5.02370358e-01 1.87331438e-01 -2.18641728e-01 2.37895280e-01
1.21220291e+00 1.79573387e-01 1.43213499e+00 -1.48435187e+00
-8.41304064e-02 2.48181701e-01 -5.26130021e-01 -1.74157441e+00
1.25845924e-01 -5.56441247e-01 2.00472891e-01 -1.76979506e+00
-1.41327590e-01 -8.73614132e-01 7.70177543e-01 1.34155974e-01
1.77538916e-01 3.54134053e-01 8.18923116e-03 5.04837096e-01
-2.54917413e-01 6.19067252e-01 1.42342579e+00 1.73739031e-01
-2.13473707e-01 -7.65770152e-02 -6.49292171e-02 9.05780435e-01
6.02060318e-01 -5.10499239e-01 -4.39946622e-01 -7.02957809e-01
3.10551673e-01 5.40613472e-01 6.43481076e-01 -1.41016173e+00
8.88054445e-02 -3.14275444e-01 5.89908659e-01 -1.18309820e+00
9.85826075e-01 -9.23872948e-01 2.03181669e-01 2.39756778e-01
2.42170021e-01 -1.74484655e-01 -1.09505164e-03 7.32261121e-01
-1.99344695e-01 -2.39310071e-01 6.27632380e-01 -5.56125939e-01
-4.89389718e-01 6.83389843e-01 1.53002754e-01 -1.78398654e-01
7.01818883e-01 -5.42438984e-01 -4.33023721e-02 -4.86327708e-01
-5.86067438e-01 1.78583220e-01 1.37134981e+00 -1.04594968e-01
1.01177263e+00 -1.33985412e+00 -6.98902726e-01 4.26144391e-01
1.07271925e-01 1.16490304e+00 4.19829972e-02 6.46028936e-01
-9.57282722e-01 1.76762149e-01 6.95390105e-02 -1.12728214e+00
-9.04696107e-01 1.81100428e-01 1.10276990e-01 1.11172557e-01
-1.01986837e+00 7.95060992e-01 1.99550226e-01 -5.50363183e-01
2.45574638e-01 -7.65679777e-01 6.67195737e-01 -3.25409055e-01
4.42089021e-01 3.68033320e-01 5.28499596e-02 -2.28697345e-01
-2.36844495e-01 9.44255352e-01 3.69048774e-01 -3.57756048e-01
1.28707528e+00 -9.47778597e-02 -1.12320602e-01 4.15272504e-01
1.14706755e+00 2.29810476e-01 -1.78943789e+00 -1.06930524e-01
-4.30530727e-01 -1.00269353e+00 3.20503600e-02 -7.30406046e-02
-6.15597010e-01 1.16278136e+00 1.85535878e-01 -1.78958297e-01
5.49689710e-01 -9.63866059e-03 1.03055334e+00 4.39298868e-01
1.21398461e+00 -5.01453638e-01 -1.99255690e-01 6.50108397e-01
7.09326982e-01 -1.39028335e+00 2.72222340e-01 -4.71810371e-01
-2.23749444e-01 1.03262413e+00 3.29970181e-01 -4.53957438e-01
5.32479882e-01 3.75778079e-01 -1.45239532e-01 -1.18683606e-01
-5.84101439e-01 -1.01690322e-01 -3.66843730e-01 9.68516707e-01
-7.62899965e-02 1.78145934e-02 1.75392732e-01 -2.78235823e-01
-3.48937929e-01 3.74308944e-01 5.90758860e-01 1.09470379e+00
-7.99973190e-01 -1.05546987e+00 -7.14228809e-01 5.14423430e-01
1.27509594e-01 -1.86827928e-01 -1.76786616e-01 6.27854466e-01
-9.38525498e-02 6.97593927e-01 3.02275360e-01 -2.91982800e-01
1.70172304e-01 -1.46992996e-01 8.04863632e-01 -8.86752725e-01
7.47576579e-02 1.59628734e-01 -2.12597072e-01 -8.38071525e-01
-4.24846381e-01 -9.00231540e-01 -1.28643382e+00 -5.47726452e-01
-8.53567421e-02 1.38801187e-01 9.86865938e-01 4.04572070e-01
3.00318122e-01 -2.96296954e-01 3.87727410e-01 -1.58773625e+00
-5.05787015e-01 -9.91088033e-01 -8.22489738e-01 6.61459789e-02
6.21904790e-01 -4.99702245e-01 -7.26675153e-01 -3.26391190e-01] | [8.459550857543945, -2.9710702896118164] |
15c75391-d312-4540-bc2d-95ef74c6ae4d | macro-action-based-multi-agent-robot-deep | 2209.10003 | null | https://arxiv.org/abs/2209.10003v2 | https://arxiv.org/pdf/2209.10003v2.pdf | Macro-Action-Based Multi-Agent/Robot Deep Reinforcement Learning under Partial Observability | The state-of-the-art multi-agent reinforcement learning (MARL) methods have provided promising solutions to a variety of complex problems. Yet, these methods all assume that agents perform synchronized primitive-action executions so that they are not genuinely scalable to long-horizon real-world multi-agent/robot tasks that inherently require agents/robots to asynchronously reason about high-level action selection at varying time durations. The Macro-Action Decentralized Partially Observable Markov Decision Process (MacDec-POMDP) is a general formalization for asynchronous decision-making under uncertainty in fully cooperative multi-agent tasks. In this thesis, we first propose a group of value-based RL approaches for MacDec-POMDPs, where agents are allowed to perform asynchronous learning and decision-making with macro-action-value functions in three paradigms: decentralized learning and control, centralized learning and control, and centralized training for decentralized execution (CTDE). Building on the above work, we formulate a set of macro-action-based policy gradient algorithms under the three training paradigms, where agents are allowed to directly optimize their parameterized policies in an asynchronous manner. We evaluate our methods both in simulation and on real robots over a variety of realistic domains. Empirical results demonstrate the superiority of our approaches in large multi-agent problems and validate the effectiveness of our algorithms for learning high-quality and asynchronous solutions with macro-actions. | ['Yuchen Xiao'] | 2022-09-20 | null | null | null | null | ['decision-making-under-uncertainty', 'decision-making-under-uncertainty'] | ['medical', 'reasoning'] | [-1.44193128e-01 2.08579436e-01 -2.85569042e-01 -1.01866849e-01
-9.24570799e-01 -5.10306776e-01 9.91999984e-01 3.75731021e-01
-8.23688745e-01 1.37410176e+00 -5.09826355e-02 -2.42071733e-01
-4.26941186e-01 -4.69993353e-01 -5.62432587e-01 -1.03140318e+00
-8.32846045e-01 1.23752522e+00 2.88921446e-01 -2.58889794e-01
1.25054643e-01 3.06886762e-01 -1.36924589e+00 -8.13330859e-02
4.94519025e-01 6.89493775e-01 4.27617460e-01 1.12523854e+00
3.61135244e-01 1.30168474e+00 -7.74845004e-01 3.48100603e-01
3.83904189e-01 -1.15128212e-01 -8.02383423e-01 2.91672558e-01
-5.76634407e-01 -6.02104723e-01 5.10608554e-02 7.18891561e-01
7.48337150e-01 3.11012983e-01 6.71270669e-01 -1.68191862e+00
1.99089259e-01 8.51111948e-01 -5.65191567e-01 -4.63663526e-02
4.57697451e-01 6.79530382e-01 7.35898733e-01 -1.19434722e-01
5.15538812e-01 1.83353019e+00 2.55596757e-01 6.66599035e-01
-9.90743816e-01 -3.20177019e-01 5.82273483e-01 1.13979518e-01
-8.77646387e-01 -1.81206241e-01 1.52833670e-01 -8.62621665e-02
1.25509417e+00 -3.32016289e-01 7.02575088e-01 9.37115908e-01
9.51445580e-01 1.07014048e+00 1.46012998e+00 -4.22150224e-01
9.53004003e-01 -4.80186760e-01 -7.05381989e-01 5.69351673e-01
2.58694649e-01 4.02417183e-01 -2.66209304e-01 -6.50025427e-01
7.80760646e-01 -1.07500464e-01 4.73119557e-01 -4.57827091e-01
-1.53277993e+00 8.77398431e-01 -2.24834397e-01 -1.76017687e-01
-9.92279589e-01 5.44225156e-01 4.84195441e-01 9.05918062e-01
1.22555062e-01 5.77964306e-01 -8.14769745e-01 -2.03501746e-01
-1.00062378e-01 7.19649494e-01 1.25522637e+00 8.63521516e-01
6.88879311e-01 2.89398730e-01 -1.10349625e-01 2.48873249e-01
6.04523063e-01 7.80685186e-01 4.79041874e-01 -1.72728229e+00
4.28033739e-01 -1.83862243e-02 8.79871249e-01 -2.12934256e-01
-7.87602842e-01 2.30859339e-01 -4.74679470e-01 9.94832695e-01
1.99932382e-01 -9.41247702e-01 -5.98949611e-01 1.51931405e+00
6.91274047e-01 -1.43307885e-02 9.52302873e-01 6.07936800e-01
-2.59427994e-01 7.48249710e-01 1.85444012e-01 -8.45710337e-01
1.07399261e+00 -1.06346369e+00 -6.82351768e-01 -2.75154263e-01
6.25874281e-01 -2.63372928e-01 4.54248428e-01 5.96966982e-01
-1.16633034e+00 -1.01435959e-01 -1.03906679e+00 8.38299513e-01
7.71782696e-02 -5.37601948e-01 7.40746200e-01 1.81705922e-01
-1.20849466e+00 6.78523719e-01 -1.45742357e+00 -1.42529175e-01
-7.59650990e-02 7.28243649e-01 1.17845781e-01 1.57195002e-01
-8.17363262e-01 1.17845082e+00 5.47242284e-01 -2.30966598e-01
-2.05345130e+00 1.58582062e-01 -5.37167549e-01 -3.09269339e-01
1.08839738e+00 -6.92070723e-01 2.28620672e+00 -9.90651250e-01
-2.34571815e+00 6.47579134e-02 4.04166013e-01 -7.24641323e-01
4.87061948e-01 -2.30955735e-01 -5.33746704e-02 3.56236905e-01
3.13285917e-01 6.00777507e-01 9.19036031e-01 -1.50051916e+00
-1.40368724e+00 -1.21193349e-01 4.58988249e-01 1.01899791e+00
2.34921023e-01 -1.22312024e-01 4.08515483e-01 -4.47720103e-02
-3.67499322e-01 -1.28163040e+00 -9.51609552e-01 -7.27261722e-01
2.03155518e-01 -6.02119803e-01 6.66684568e-01 3.01642597e-01
4.41063166e-01 -1.67919731e+00 4.27494347e-01 -9.84271690e-02
3.32571417e-02 -2.05605492e-01 -3.59489650e-01 1.00657535e+00
6.63151681e-01 -2.73297250e-01 1.24045722e-01 -3.97368342e-01
4.41879511e-01 8.61114860e-01 -1.39277890e-01 5.76882184e-01
-1.73422992e-01 5.30189991e-01 -1.48643494e+00 -7.11151779e-01
1.82856098e-01 -1.48887277e-01 -1.45830661e-01 3.78771752e-01
-8.77152741e-01 6.26229227e-01 -1.16511726e+00 6.70743942e-01
-1.04623074e-02 1.14849629e-02 8.17613125e-01 9.65287149e-01
-4.02545631e-01 3.43896858e-02 -1.44957936e+00 1.46473014e+00
-3.92128080e-01 1.32313296e-01 7.56576240e-01 -8.78636122e-01
3.97713184e-01 8.78249228e-01 1.06172812e+00 -4.91186082e-01
1.32152393e-01 2.23428980e-01 1.47007957e-01 -2.77777851e-01
5.49339056e-01 -1.17811210e-01 -4.10699546e-01 8.62489939e-01
-4.96818945e-02 -7.09242940e-01 4.33470547e-01 -2.56987400e-02
1.53656173e+00 2.98813611e-01 5.67182958e-01 -1.28749758e-01
3.32793206e-01 3.53684664e-01 7.51241386e-01 1.39493549e+00
-5.23668289e-01 -3.98039758e-01 5.39560080e-01 -4.95658636e-01
-7.88046241e-01 -9.13489699e-01 6.34123981e-01 1.48091197e+00
3.56678188e-01 -6.83879629e-02 -3.65139544e-01 -4.71979052e-01
1.32089779e-01 6.85221076e-01 -3.12018603e-01 3.27803761e-01
-7.41585016e-01 -7.03582704e-01 1.66182950e-01 8.74391869e-02
3.80408525e-01 -1.57929850e+00 -1.52748895e+00 8.63555372e-01
4.19301510e-01 -1.01790023e+00 -3.88066797e-03 5.55990756e-01
-8.01722884e-01 -1.13110578e+00 -3.65603834e-01 -6.28687263e-01
3.72134775e-01 2.34952897e-01 9.89691556e-01 -2.44036570e-01
1.53433755e-01 1.27014756e+00 -4.79181409e-01 -8.48537028e-01
-7.39497840e-01 -8.59873369e-02 5.68112969e-01 -2.58586496e-01
-3.17506760e-01 -4.07872051e-01 -4.12921757e-01 2.80283928e-01
-8.24332416e-01 -6.42596781e-02 9.06988263e-01 7.99383581e-01
6.17118597e-01 3.75247121e-01 4.31162685e-01 -4.55554217e-01
1.14970028e+00 -4.61121023e-01 -1.05753934e+00 3.45645636e-01
-6.32508278e-01 2.53615022e-01 7.14830935e-01 -6.82533443e-01
-1.10149503e+00 1.95323989e-01 5.11000574e-01 -9.75798070e-02
-1.87402919e-01 4.39890534e-01 2.06062660e-01 8.20364505e-02
4.06205207e-01 2.04665974e-01 2.84780324e-01 1.75283745e-01
2.47179583e-01 5.06586969e-01 5.93069196e-03 -1.23181844e+00
4.31929231e-01 4.19359595e-01 2.86255598e-01 -6.48784697e-01
-3.22299331e-01 -2.50113130e-01 -2.07889006e-01 -3.68138194e-01
7.34132826e-01 -1.06237757e+00 -1.49499905e+00 5.33884406e-01
-9.12676811e-01 -1.25672340e+00 -4.01114672e-01 6.46142781e-01
-1.42704713e+00 1.85063571e-01 -7.73062825e-01 -1.29451537e+00
-1.23869285e-01 -1.37940300e+00 1.00963557e+00 4.48088884e-01
3.00504386e-01 -1.05046248e+00 5.71827054e-01 -3.62959474e-01
3.04706097e-01 2.70990461e-01 4.97612774e-01 -5.95731318e-01
-6.09470427e-01 2.90536821e-01 6.87201619e-01 -1.89821512e-01
-1.07520126e-01 -3.18815708e-01 -3.65526944e-01 -7.76356757e-01
-1.94041561e-02 -9.32497144e-01 1.43985171e-02 5.65280914e-01
2.31646746e-01 -5.19100785e-01 -4.65215951e-01 -1.20147303e-01
1.30927098e+00 6.92332804e-01 -2.92547382e-02 9.19562757e-01
-2.96247937e-02 2.89540231e-01 1.30956292e+00 1.17036879e+00
7.68260717e-01 4.33785319e-01 9.22003210e-01 5.62892616e-01
6.77514911e-01 9.04732421e-02 1.04553735e+00 3.71697485e-01
-2.48403862e-01 -3.76764804e-01 -8.00307930e-01 3.63319159e-01
-2.62435985e+00 -1.06406510e+00 5.27404904e-01 1.98901498e+00
9.01119828e-01 2.77848542e-01 3.70211154e-01 -3.91430050e-01
4.53783244e-01 1.46112129e-01 -9.58813965e-01 -4.80253607e-01
1.61739334e-01 -1.91174403e-01 8.17043424e-01 6.35795772e-01
-1.19337463e+00 1.10517371e+00 6.23753309e+00 4.31066781e-01
-6.58578098e-01 3.56764972e-01 1.49480626e-01 -8.07823539e-02
3.34047049e-01 1.73237100e-01 -8.30199718e-01 -1.09669268e-02
1.19947827e+00 -2.22625479e-01 8.49685073e-01 1.03413570e+00
4.98566091e-01 -7.81232417e-01 -1.13084364e+00 6.41686320e-01
-5.78035116e-01 -9.74968255e-01 -4.91683811e-01 1.26750037e-01
1.09578872e+00 4.31203634e-01 -2.88249910e-01 6.47156775e-01
1.71285069e+00 -6.53360724e-01 9.48688745e-01 2.58421510e-01
1.45084858e-01 -8.39772701e-01 7.21423268e-01 8.49949360e-01
-1.26722467e+00 -7.24979758e-01 -4.15848166e-01 -6.05356455e-01
3.20910782e-01 -1.73610434e-01 -1.08763242e+00 5.88464558e-01
5.14692545e-01 3.39144558e-01 1.65756598e-01 7.59414196e-01
-4.04976666e-01 2.53628522e-01 -4.47979599e-01 -5.77436924e-01
7.66344607e-01 -1.52412474e-01 5.82135797e-01 6.72729492e-01
6.37834473e-03 2.34881073e-01 1.20953882e+00 -3.87688838e-02
6.15467668e-01 -3.55761051e-01 -5.05833507e-01 -1.01408660e-01
6.81442440e-01 1.13206542e+00 -8.74283373e-01 -3.99915248e-01
-1.82762071e-01 6.45794034e-01 3.52951616e-01 4.17397916e-01
-4.67087328e-01 -2.93469783e-02 7.92595148e-01 -6.35581851e-01
1.58209354e-01 -7.66882181e-01 3.23131591e-01 -7.49652028e-01
-4.65527594e-01 -1.24151874e+00 4.56542552e-01 -3.85903448e-01
-1.15806735e+00 2.62576789e-01 2.46418446e-01 -1.16648984e+00
-1.05304492e+00 -4.34518933e-01 -3.75954300e-01 8.97333473e-02
-1.50603783e+00 -8.29661191e-01 3.91157657e-01 7.11105049e-01
9.31326151e-01 -5.24391413e-01 9.86270666e-01 -5.98421037e-01
-2.28890091e-01 -3.30797285e-01 5.23144543e-01 -3.69244814e-01
4.50179905e-01 -1.54896212e+00 -1.07159607e-01 2.82199979e-01
-4.90488231e-01 -1.29027963e-01 8.96847129e-01 -7.53643215e-01
-2.05177474e+00 -7.17602253e-01 -1.07615374e-01 -2.34909117e-01
8.97108853e-01 6.58534691e-02 -1.24947228e-01 9.26151037e-01
6.18497193e-01 -2.65472203e-01 7.15600029e-02 -1.28226817e-01
4.81109738e-01 -1.36851281e-01 -9.19368267e-01 7.31008172e-01
5.13700604e-01 9.95702818e-02 -6.39157355e-01 6.88964069e-01
6.72711909e-01 -5.15461862e-01 -8.27669799e-01 2.05131233e-01
3.23749274e-01 -6.73155129e-01 5.71475148e-01 -5.49429655e-01
-1.15030527e-01 -3.47048044e-01 -1.31948888e-01 -1.69771707e+00
-1.85534015e-01 -1.28748596e+00 -2.42065459e-01 5.18173575e-01
1.51388511e-01 -8.44747365e-01 6.30418420e-01 3.16191077e-01
-2.41962686e-01 -5.37502408e-01 -1.22002101e+00 -8.94388914e-01
-2.04463932e-03 -6.81542456e-02 3.58476579e-01 5.15315711e-01
1.44731686e-01 4.01973128e-01 -4.01676148e-01 4.56693500e-01
8.54336143e-01 -2.25038659e-02 9.67405498e-01 -8.12269151e-01
-6.63215995e-01 -3.27025950e-01 2.26703793e-01 -8.31571460e-01
4.27915096e-01 -1.58847496e-01 7.27572381e-01 -1.70664322e+00
-2.15793923e-01 -7.15822816e-01 -1.45408809e-01 6.00475550e-01
2.44401053e-01 -9.20899212e-01 4.39656258e-01 5.01642048e-01
-1.54362869e+00 7.83885598e-01 1.34039450e+00 -1.87424704e-01
-5.82492113e-01 3.56976151e-01 5.10308035e-02 8.25885355e-01
9.89489853e-01 -6.49680197e-01 -6.42810464e-01 -3.83851796e-01
2.55492538e-01 9.70847547e-01 -1.00206003e-01 -1.06737530e+00
4.17480618e-01 -1.11522245e+00 -1.13622122e-01 -2.47336343e-01
4.30355221e-01 -6.91338718e-01 -6.64549768e-02 1.11802840e+00
-3.97420704e-01 7.64016390e-01 -9.99819487e-02 8.61140430e-01
5.42094707e-02 -2.93657273e-01 4.61890668e-01 -7.40156710e-01
-9.60459352e-01 1.72313780e-01 -1.18509126e+00 9.79556069e-02
1.58100283e+00 2.83868164e-01 -1.86845362e-01 -7.41807044e-01
-6.97247386e-01 9.96651232e-01 2.91445136e-01 3.63488421e-02
4.35183197e-01 -7.08185911e-01 -7.37723947e-01 -4.50258076e-01
-3.90641242e-01 6.04856499e-02 -2.75813550e-01 6.76254332e-01
-3.52158397e-01 2.59802669e-01 -5.17293096e-01 -5.11463583e-01
-8.81148815e-01 4.88543928e-01 3.95501018e-01 -8.73908997e-01
-4.37337428e-01 2.16225788e-01 -2.50904948e-01 -5.15627563e-01
3.85998935e-01 -5.63891567e-02 -1.46974921e-01 -1.62440881e-01
3.87296915e-01 4.93085980e-01 -5.84583461e-01 -4.45520990e-02
-6.34754151e-02 2.97323093e-02 1.28018288e-02 -1.04609227e+00
1.47163701e+00 -2.78874189e-01 3.08727124e-03 6.95285082e-01
1.01603856e-02 -5.06687582e-01 -1.88301718e+00 -1.87078401e-01
3.40133935e-01 1.07779220e-01 -2.94719070e-01 -5.83594203e-01
-3.10953468e-01 2.80964375e-01 4.16832387e-01 4.03718561e-01
6.24066651e-01 -2.69409835e-01 1.32791400e-01 1.16135383e+00
1.24291265e+00 -1.72773516e+00 4.77612227e-01 9.77299452e-01
5.89806199e-01 -1.22766817e+00 2.81182647e-01 3.23756307e-01
-1.19644821e+00 1.14213085e+00 7.12359726e-01 -2.82175928e-01
3.46960157e-01 5.04308701e-01 1.94260210e-01 -3.69504616e-02
-1.56816244e+00 -4.15322274e-01 -8.56183052e-01 9.01295543e-01
-4.23790395e-01 4.33123499e-01 -1.31902277e-01 -1.34906977e-01
2.63222247e-01 -1.32495537e-01 8.34624887e-01 1.88979530e+00
-9.87620652e-01 -1.08779430e+00 -6.69171333e-01 1.27832741e-01
-3.56078297e-01 6.16080344e-01 -2.35477425e-02 9.66793001e-01
-2.68662155e-01 1.25852835e+00 -9.56294686e-02 1.40170351e-01
1.33410543e-02 -1.68408021e-01 6.75753415e-01 -7.42511034e-01
-5.75489044e-01 3.39681923e-01 1.89362675e-01 -6.78827941e-01
-7.51040995e-01 -9.43400145e-01 -1.77066398e+00 2.96545750e-03
1.91776961e-01 3.36872101e-01 7.17776835e-01 1.24200833e+00
1.93827957e-01 4.13117379e-01 7.86748111e-01 -1.49319828e+00
-1.58477330e+00 -9.30867553e-01 -5.68672359e-01 -1.77227929e-01
5.10945439e-01 -8.53427947e-01 -1.29724964e-01 -3.76950443e-01] | [3.8937315940856934, 2.0011274814605713] |
11ccdc41-701e-40ed-a721-9c9aa1f38495 | robust-policy-optimization-in-deep | 2212.07536 | null | https://arxiv.org/abs/2212.07536v1 | https://arxiv.org/pdf/2212.07536v1.pdf | Robust Policy Optimization in Deep Reinforcement Learning | The policy gradient method enjoys the simplicity of the objective where the agent optimizes the cumulative reward directly. Moreover, in the continuous action domain, parameterized distribution of action distribution allows easy control of exploration, resulting from the variance of the representing distribution. Entropy can play an essential role in policy optimization by selecting the stochastic policy, which eventually helps better explore the environment in reinforcement learning (RL). However, the stochasticity often reduces as the training progresses; thus, the policy becomes less exploratory. Additionally, certain parametric distributions might only work for some environments and require extensive hyperparameter tuning. This paper aims to mitigate these issues. In particular, we propose an algorithm called Robust Policy Optimization (RPO), which leverages a perturbed distribution. We hypothesize that our method encourages high-entropy actions and provides a way to represent the action space better. We further provide empirical evidence to verify our hypothesis. We evaluated our methods on various continuous control tasks from DeepMind Control, OpenAI Gym, Pybullet, and IsaacGym. We observed that in many settings, RPO increases the policy entropy early in training and then maintains a certain level of entropy throughout the training period. Eventually, our agent RPO shows consistently improved performance compared to PPO and other techniques: entropy regularization, different distributions, and data augmentation. Furthermore, in several settings, our method stays robust in performance, while other baseline mechanisms fail to improve and even worsen the performance. | ['Yexiang Xue', 'Md Masudur Rahman'] | 2022-12-14 | null | null | null | null | ['continuous-control'] | ['playing-games'] | [-1.28120616e-01 1.67907491e-01 -7.55758822e-01 1.39941216e-01
-4.95295674e-01 -4.46900308e-01 5.67105114e-01 6.09868877e-02
-8.01422656e-01 1.27441096e+00 3.91222596e-01 -2.71181554e-01
-3.29461366e-01 -5.48791885e-01 -6.96089327e-01 -9.66450155e-01
-1.39556780e-01 3.15716594e-01 -1.09266847e-01 -3.06713521e-01
2.82385767e-01 2.31909633e-01 -1.32247436e+00 -4.20023948e-01
1.17567027e+00 7.62688100e-01 2.94401199e-01 3.80564004e-01
3.14475894e-01 8.94002676e-01 -5.92709541e-01 2.00293779e-01
5.22153199e-01 -4.20927793e-01 -5.74780822e-01 -6.05604574e-02
5.97378835e-02 -5.11321068e-01 -2.62548596e-01 1.06259882e+00
5.22058785e-01 8.14999163e-01 4.77713674e-01 -1.18041050e+00
-5.07938862e-01 8.00121725e-01 -4.61986899e-01 1.64972674e-02
1.43043578e-01 7.16703057e-01 9.46500599e-01 -1.62758917e-01
4.60063189e-01 1.27403712e+00 3.77057046e-01 7.07043231e-01
-1.56018293e+00 -4.54722464e-01 4.04605359e-01 -1.59985617e-01
-9.17960584e-01 -8.57200474e-02 6.60830140e-01 -1.24258377e-01
7.05494523e-01 8.74291211e-02 9.85308468e-01 1.24458504e+00
1.46694303e-01 1.08698559e+00 1.45699000e+00 -2.57719606e-01
6.54574692e-01 2.02187911e-01 -3.46751630e-01 5.17371297e-01
2.40536273e-01 7.01285124e-01 -4.00936842e-01 -2.90711880e-01
8.76856625e-01 -8.96048993e-02 -3.93447042e-01 -5.94744682e-01
-9.72302616e-01 9.41481352e-01 1.96953923e-01 8.42266381e-02
-5.67223549e-01 3.91098857e-01 3.21467131e-01 4.30637002e-01
2.95794278e-01 1.21262228e+00 -6.02466404e-01 -8.18002284e-01
-6.84778512e-01 7.42221594e-01 8.72991562e-01 5.06968975e-01
4.41736341e-01 2.03831345e-01 -4.76820379e-01 7.66663671e-01
1.19986989e-01 3.42529953e-01 6.40457511e-01 -1.54692698e+00
4.83687878e-01 3.35629135e-01 5.04525304e-01 -6.15232468e-01
-4.28005338e-01 -4.72249508e-01 -3.62402022e-01 6.90301478e-01
6.41685724e-01 -6.43512249e-01 -7.12882221e-01 2.11912513e+00
3.00427496e-01 -1.94108069e-01 5.26987426e-02 9.39772964e-01
-1.50098249e-01 4.29363459e-01 2.33471110e-01 -2.83985406e-01
9.26314414e-01 -1.01794648e+00 -8.90251160e-01 -3.73314679e-01
5.33752322e-01 -1.99273437e-01 1.77876031e+00 4.32002127e-01
-1.04522014e+00 -2.21795365e-01 -9.78359163e-01 4.06114727e-01
-5.88765219e-02 -8.79156664e-02 6.82692230e-01 4.98655558e-01
-8.42717111e-01 1.17302299e+00 -1.08960593e+00 -1.80128962e-01
4.66734052e-01 2.19042882e-01 1.20420553e-01 4.21219140e-01
-1.13295031e+00 9.88727808e-01 6.70605421e-01 -4.88819778e-01
-8.82303357e-01 -7.96815932e-01 -6.69147789e-01 2.38784850e-01
7.17233479e-01 -5.41786075e-01 1.56508338e+00 -8.01660359e-01
-2.21997523e+00 1.06479563e-01 3.68312657e-01 -7.08003998e-01
7.93158352e-01 -6.11996353e-01 2.51176298e-01 -6.01687469e-02
-2.23954558e-01 8.30545247e-01 9.18766201e-01 -1.17447662e+00
-4.48914409e-01 -1.49365306e-01 3.61113310e-01 6.64092302e-01
-5.47547638e-01 -4.00247574e-01 6.91395700e-02 -6.32715642e-01
-4.39686030e-01 -1.22379637e+00 -5.29280961e-01 -1.90766633e-01
-3.09169650e-01 -1.14264682e-01 5.72158039e-01 -2.36790523e-01
1.38729739e+00 -2.13338923e+00 1.72400340e-01 1.60049513e-01
3.73489857e-02 1.48351505e-01 -3.69776576e-03 2.41446108e-01
1.59038797e-01 4.17462528e-01 -1.57890350e-01 -2.72768617e-01
2.76920229e-01 4.17034507e-01 -3.86876941e-01 3.13184142e-01
-1.24159537e-01 8.33195984e-01 -1.05760777e+00 -3.41684490e-01
1.23634309e-01 1.82721585e-01 -9.55101609e-01 1.12578169e-01
-6.63446248e-01 7.69505084e-01 -7.58715451e-01 3.54037672e-01
2.08099604e-01 -2.41440922e-01 2.58385241e-01 4.04705048e-01
-2.07538813e-01 3.65455717e-01 -1.13936734e+00 1.41381705e+00
-3.95857096e-01 4.58077252e-01 -7.44348988e-02 -9.42242265e-01
6.75957620e-01 1.17198117e-02 7.45214999e-01 -6.08767092e-01
1.16420500e-01 -9.98506788e-03 1.19822614e-01 -4.13217187e-01
6.65601790e-01 8.95034298e-02 1.90019980e-01 4.68306869e-01
-4.49020654e-01 -2.93744534e-01 1.58823043e-01 -7.02274665e-02
1.07027614e+00 5.22303343e-01 3.37129354e-01 -2.85301298e-01
-8.06875229e-02 -4.58520241e-02 5.58809280e-01 1.11436594e+00
-5.41523933e-01 2.31515225e-02 8.06308627e-01 -3.06860767e-02
-1.00041342e+00 -8.05550754e-01 -1.58376366e-01 1.28463030e+00
1.20747544e-01 -2.95043886e-01 -6.57151580e-01 -7.81981885e-01
3.47887188e-01 9.85410929e-01 -6.68985128e-01 -3.93107146e-01
-6.13969445e-01 -7.83209443e-01 3.29971671e-01 4.22388226e-01
6.83600247e-01 -1.22360182e+00 -8.61701846e-01 3.20666820e-01
-3.30702960e-02 -7.76057482e-01 -6.05613232e-01 2.82883316e-01
-1.14281487e+00 -8.14911127e-01 -5.13159692e-01 -1.47845447e-01
4.35871184e-01 -1.94053248e-01 9.22006190e-01 -1.48130983e-01
1.75730199e-01 4.25846994e-01 -1.70108631e-01 -3.68742287e-01
-3.36462826e-01 2.40220875e-01 3.25147241e-01 -5.75509965e-01
1.48570696e-02 -7.87265837e-01 -6.85249329e-01 9.54940915e-02
-6.46554172e-01 -3.14168930e-01 4.61787373e-01 9.97940719e-01
5.06523252e-01 4.25413698e-02 5.58616817e-01 -5.65154791e-01
1.18056023e+00 -4.81226027e-01 -7.72006929e-01 -6.76677376e-02
-1.07529306e+00 6.17621958e-01 8.61688614e-01 -8.90530825e-01
-9.86881196e-01 -1.52282447e-01 2.60114372e-01 -5.30591011e-01
8.74068961e-02 3.75024021e-01 1.41172111e-01 1.05930679e-01
8.88818681e-01 8.75326768e-02 3.42115372e-01 -3.84099275e-01
3.61545354e-01 2.04420000e-01 1.54140675e-02 -9.54649389e-01
5.34489810e-01 2.59311616e-01 -9.89212543e-02 -4.88490731e-01
-7.51063526e-01 2.04140887e-01 3.23604904e-02 -4.76696938e-02
6.32710040e-01 -5.92386127e-01 -1.10442877e+00 4.39393334e-02
-4.66778249e-01 -1.06312132e+00 -7.80805409e-01 7.22410738e-01
-9.63168561e-01 1.64285481e-01 -5.41339338e-01 -1.05394530e+00
-1.49286926e-01 -1.22850370e+00 7.55791843e-01 5.73620737e-01
-2.10807443e-01 -1.04787290e+00 2.46395513e-01 -6.79286793e-02
4.81504887e-01 3.20732385e-01 6.99848115e-01 -6.15677655e-01
-4.00336355e-01 2.54400522e-01 4.36061352e-01 3.88595253e-01
1.67399645e-01 -1.53419986e-01 -6.55983090e-01 -5.58188975e-01
-2.91524418e-02 -7.35067666e-01 8.59372377e-01 5.76375782e-01
1.54268718e+00 -5.80941319e-01 -1.45062760e-01 6.57285213e-01
1.12885785e+00 3.50668192e-01 3.91366661e-01 8.44035864e-01
2.48169318e-01 2.44442701e-01 8.70372474e-01 8.42344046e-01
1.59401432e-01 5.31065166e-01 4.76471215e-01 5.74727207e-02
4.08742040e-01 -4.97855157e-01 6.99404061e-01 1.20170385e-01
-1.90163866e-01 -1.15329385e-01 -6.10301316e-01 1.61850154e-01
-1.96146834e+00 -9.82216537e-01 7.32382953e-01 2.30622005e+00
1.39389384e+00 3.10963243e-01 4.26268578e-01 -1.59794495e-01
2.44642809e-01 3.24629456e-01 -1.12830853e+00 -3.90378237e-01
3.74321230e-02 -3.57544422e-02 8.20548654e-01 5.68298817e-01
-1.01775885e+00 9.02237415e-01 6.79901600e+00 8.05777073e-01
-1.07069385e+00 -2.35074669e-01 6.89259350e-01 -5.44041157e-01
-1.57643914e-01 -9.22390670e-02 -6.96337283e-01 6.63245678e-01
7.57121742e-01 -2.37636179e-01 1.04642296e+00 1.26860046e+00
5.80412388e-01 -2.98836499e-01 -9.99330521e-01 5.65838456e-01
-5.54121375e-01 -1.11929333e+00 -5.86931884e-01 3.76174748e-01
9.39201832e-01 1.34296492e-01 3.48622501e-01 8.51857305e-01
8.84375095e-01 -9.31780815e-01 5.67253292e-01 4.15899128e-01
4.70003515e-01 -8.41671944e-01 3.04046661e-01 5.89920521e-01
-5.59049368e-01 -5.22407055e-01 -3.50616306e-01 -1.52827337e-01
-1.28829613e-01 2.17800632e-01 -7.17030585e-01 -1.00517258e-01
5.37588954e-01 3.84600282e-01 -2.67956436e-01 8.98544908e-01
-4.41247374e-01 7.38169670e-01 -4.90428865e-01 -3.43844503e-01
4.94602799e-01 -4.16061044e-01 8.74189854e-01 6.05694354e-01
1.60583816e-02 -1.51930330e-02 6.25087023e-01 9.68685389e-01
-2.56482605e-02 2.08026432e-02 -6.44185185e-01 -3.11800659e-01
6.94934666e-01 8.86486351e-01 -4.25037146e-01 -1.63676307e-01
7.77952969e-02 5.27857244e-01 4.72663701e-01 6.66739941e-01
-9.71590996e-01 -2.44326741e-01 9.77873206e-01 -1.12266228e-01
1.53421670e-01 -3.36427569e-01 -2.87770271e-01 -1.09141123e+00
-1.29776215e-02 -1.11839855e+00 1.57655075e-01 -1.63049102e-01
-9.67950225e-01 7.42041320e-02 2.38187075e-01 -1.07774472e+00
-5.23993134e-01 -2.09291771e-01 -4.30479288e-01 4.91654426e-01
-1.50615072e+00 -1.46889240e-01 9.65777561e-02 2.81579345e-01
5.10195196e-01 -1.07180603e-01 5.99582613e-01 -5.05114645e-02
-7.71162212e-01 6.40954375e-01 6.00945652e-01 -2.90806562e-01
6.71716392e-01 -1.60676265e+00 -7.33302087e-02 3.16725552e-01
-2.99167126e-01 6.42876863e-01 9.73791361e-01 -6.73740983e-01
-1.26065493e+00 -8.14346373e-01 -1.58674851e-01 -2.17703789e-01
7.22887993e-01 -3.97702027e-03 -7.79149950e-01 6.06366813e-01
1.54436007e-01 -2.90195823e-01 3.43496174e-01 9.96020585e-02
1.97914839e-01 9.52922255e-02 -1.14539671e+00 1.04215240e+00
9.61918652e-01 -2.27874368e-01 -4.52458441e-01 4.64820981e-01
9.34638917e-01 -6.75953925e-01 -9.27749455e-01 1.63639262e-01
4.01155114e-01 -7.58620858e-01 7.60320365e-01 -8.81198943e-01
4.10097986e-01 7.76660219e-02 -1.41100325e-02 -1.82201350e+00
-1.40360549e-01 -1.07833385e+00 -6.36237085e-01 8.84157240e-01
4.08906460e-01 -7.47296691e-01 9.39763963e-01 9.71812665e-01
8.68170261e-02 -1.23772085e+00 -6.50749087e-01 -1.09204888e+00
2.82014221e-01 -2.03547940e-01 6.11912251e-01 8.22947919e-01
2.29835197e-01 -1.35872990e-01 -4.35974956e-01 -1.49591118e-01
5.81512690e-01 -8.47531203e-03 7.11435914e-01 -7.29919672e-01
-6.28815949e-01 -7.87208557e-01 1.92283034e-01 -1.34114611e+00
2.68472940e-01 -4.90071148e-01 1.24985963e-01 -1.23318386e+00
-1.05652869e-01 -6.41789496e-01 -3.33908975e-01 6.36826634e-01
-4.13764775e-01 -5.36110103e-01 4.01260257e-01 1.14529066e-01
-4.93086934e-01 1.16186404e+00 1.68949509e+00 6.07295819e-02
-9.20880616e-01 1.48440853e-01 -7.21012890e-01 7.19253123e-01
1.33069479e+00 -4.34319347e-01 -7.21199930e-01 -1.01433046e-01
7.33191594e-02 1.74166914e-02 9.30666700e-02 -8.05053473e-01
-1.87983423e-01 -8.27001512e-01 2.42031321e-01 7.00418651e-02
2.79248357e-01 -5.83746195e-01 -5.17893255e-01 5.82779348e-01
-7.25823462e-01 -7.97542334e-02 2.55208284e-01 5.91923356e-01
1.77252188e-01 -9.82024893e-02 9.28819299e-01 -9.03340727e-02
-2.91064084e-01 2.63341576e-01 -3.94255430e-01 5.12877822e-01
9.11358595e-01 -1.18241690e-01 -2.06096292e-01 -5.96040785e-01
-6.90632701e-01 6.89400136e-01 6.28868937e-01 3.05158079e-01
3.08663964e-01 -1.28772628e+00 -2.73981571e-01 -1.25156686e-01
-3.74909818e-01 -5.94208241e-02 -2.41907552e-01 9.46963310e-01
-6.69760555e-02 1.81260630e-01 -4.67828847e-02 -3.94975394e-01
-6.94854975e-01 3.76960367e-01 6.21609867e-01 -5.51164210e-01
-7.70019829e-01 5.53944826e-01 -1.66895866e-01 -3.72717410e-01
6.64169550e-01 -6.34646237e-01 -2.11094990e-01 -5.49573600e-02
3.60716701e-01 4.22397912e-01 -5.48955023e-01 2.58015037e-01
4.28559482e-02 7.22915903e-02 -2.98591629e-02 -3.31197143e-01
1.36116278e+00 1.47169512e-02 4.02041495e-01 3.24269503e-01
7.95320570e-01 -3.81745286e-02 -2.09515643e+00 -7.35857571e-03
8.41249451e-02 -5.61338544e-01 1.84417531e-01 -7.96275973e-01
-9.00412858e-01 3.63943160e-01 5.79151809e-01 2.73088425e-01
9.58202899e-01 -3.55659604e-01 5.08110881e-01 7.12182224e-01
2.98347354e-01 -1.59055829e+00 3.13263714e-01 6.10638618e-01
7.79140592e-01 -1.37954140e+00 1.87731594e-01 4.20537472e-01
-1.14256549e+00 8.85894358e-01 8.86453032e-01 -1.24262020e-01
4.11282539e-01 1.04219176e-01 -2.98054785e-01 5.79300448e-02
-9.48885381e-01 -2.43766025e-01 -9.29660350e-02 3.91755462e-01
1.90488040e-01 8.05846304e-02 -3.42059761e-01 3.07518989e-01
-4.09878641e-01 -4.21091206e-02 3.91289711e-01 1.01894045e+00
-6.67692900e-01 -1.12093413e+00 -3.63576293e-01 3.87147129e-01
-4.89481658e-01 4.52815294e-02 -1.64402544e-01 1.08644187e+00
-2.31212258e-01 7.21636951e-01 -1.51951134e-01 -1.92733660e-01
3.23143423e-01 4.72867563e-02 3.93527120e-01 -2.68392026e-01
-4.97650564e-01 9.11829546e-02 -2.10826378e-02 -8.98545921e-01
-1.41889855e-01 -8.04640651e-01 -1.52512991e+00 -2.66806901e-01
-2.97503233e-01 2.26537749e-01 5.78551412e-01 8.19550931e-01
3.21472496e-01 6.19754553e-01 8.08710396e-01 -6.86127126e-01
-1.39979589e+00 -9.02200520e-01 -5.14783204e-01 2.81082392e-01
5.76967478e-01 -1.09025049e+00 -5.47603965e-01 -4.92777765e-01] | [4.122490882873535, 2.1387715339660645] |
56fcfdd5-124f-4e19-98b2-e8d21463b7a6 | botartist-twitter-bot-detection-machine | 2306.00037 | null | https://arxiv.org/abs/2306.00037v2 | https://arxiv.org/pdf/2306.00037v2.pdf | BotArtist: Twitter bot detection Machine Learning model based on Twitter suspension | Twitter as one of the most popular social networks, offers a means for communication and online discourse, which unfortunately has been the target of bots and fake accounts, leading to the manipulation and spreading of false information. Towards this end, we gather a challenging, multilingual dataset of social discourse on Twitter, originating from 9M users regarding the recent Russo-Ukrainian war, in order to detect the bot accounts and the conversation involving them. We collect the ground truth for our dataset through the Twitter API suspended accounts collection, containing approximately 343K of bot accounts and 8M of normal users. Additionally, we use a dataset provided by Botometer-V3 with 1,777 Varol, 483 German accounts, and 1,321 US accounts. Besides the publicly available datasets, we also manage to collect 2 independent datasets around popular discussion topics of the 2022 energy crisis and the 2022 conspiracy discussions. Both of the datasets were labeled according to the Twitter suspension mechanism. We build a novel ML model for bot detection using the state-of-the-art XGBoost model. We combine the model with a high volume of labeled tweets according to the Twitter suspension mechanism ground truth. This requires a limited set of profile features allowing labeling of the dataset in different time periods from the collection, as it is independent of the Twitter API. In comparison with Botometer our methodology achieves an average 11% higher ROC-AUC score over two real-case scenario datasets. | ['Alexander Shevtsov', 'Sotiris Ioannidis', 'Polyvios Pratikakis', 'Ioannis Lamprou', 'Despoina Antonakaki'] | 2023-05-31 | null | null | null | null | ['twitter-bot-detection'] | ['miscellaneous'] | [-3.37936252e-01 3.87119442e-01 -5.03798604e-01 2.37991795e-01
-6.11069322e-01 -8.96616697e-01 1.43915641e+00 5.71214318e-01
-6.30521357e-01 8.94037485e-01 2.95001030e-01 -3.21381062e-01
4.49031651e-01 -1.00780010e+00 -3.95675153e-01 -3.95905465e-01
1.03434324e-01 7.50060499e-01 3.76231283e-01 -5.41759908e-01
4.60275948e-01 2.85206765e-01 -1.21896732e+00 4.83167589e-01
5.33384800e-01 8.15618455e-01 -3.01299214e-01 5.66809595e-01
-2.54408091e-01 1.08305264e+00 -1.17096460e+00 -6.66331947e-01
9.88396481e-02 -1.08791843e-01 -1.04701614e+00 -3.37185115e-02
3.91359143e-02 -2.38839507e-01 -2.44295090e-01 1.00712752e+00
2.95259863e-01 -3.37101281e-01 3.85009348e-01 -1.54268205e+00
4.41458747e-02 1.09391201e+00 -5.85016012e-01 3.62278730e-01
3.49248141e-01 4.24495846e-01 9.86625195e-01 -3.61484468e-01
1.12253094e+00 1.25856113e+00 6.39308989e-01 1.79671407e-01
-1.05296516e+00 -8.54645967e-01 -2.60607809e-01 2.74195105e-01
-1.03235090e+00 -1.22012079e-01 6.39383554e-01 -8.80563736e-01
3.50201517e-01 4.90167469e-01 7.72119939e-01 1.72666049e+00
-2.35393435e-01 5.90879142e-01 1.54186428e+00 -9.92758498e-02
1.36415139e-01 7.41393626e-01 2.99971431e-01 3.50992292e-01
4.05660331e-01 -2.69170403e-01 -5.44428229e-01 -7.15349853e-01
-1.39785767e-01 -2.80486315e-01 2.06946097e-02 5.61115503e-01
-1.01581824e+00 1.29724967e+00 3.77041966e-01 6.14161909e-01
-3.56687158e-01 -1.52734175e-01 7.75163949e-01 1.90590054e-01
1.04991806e+00 4.40253586e-01 -2.48536885e-01 -3.22108209e-01
-8.67227435e-01 4.77012455e-01 1.07412326e+00 6.10248983e-01
8.89160335e-01 -3.38239640e-01 -1.28918767e-01 4.20953304e-01
2.00705603e-01 6.29508674e-01 4.74668294e-01 -4.99260485e-01
7.42470980e-01 9.13393140e-01 4.11285639e-01 -1.58452082e+00
-6.46659791e-01 -4.00191426e-01 -4.94116098e-01 -2.52619684e-01
7.50514805e-01 -3.45591664e-01 -1.29244298e-01 1.40341318e+00
6.02429152e-01 -5.51522151e-02 -4.43766147e-01 5.78183293e-01
8.48703086e-01 6.19289637e-01 -1.65128559e-01 -4.33518887e-01
1.49979889e+00 -5.90454996e-01 -7.51281559e-01 -9.48943943e-02
7.59823918e-01 -6.38597786e-01 7.42285669e-01 3.25405240e-01
-6.26616240e-01 5.49217425e-02 -7.29223490e-01 3.08914244e-01
-8.29997540e-01 -2.56495982e-01 1.84214175e-01 7.79213250e-01
-4.97712404e-01 5.84391057e-01 -3.83786589e-01 -5.51908672e-01
5.15034735e-01 -1.00848883e-01 -2.16978490e-01 4.81031954e-01
-1.39061379e+00 9.74650443e-01 5.11400521e-01 -1.84008911e-01
-6.18276179e-01 -4.68701899e-01 -2.98045725e-01 -4.32211488e-01
5.24376810e-01 2.16250420e-01 1.00878358e+00 -4.35682774e-01
-1.32765472e+00 1.26159704e+00 2.09174931e-01 -6.78889871e-01
1.29619491e+00 1.43642714e-02 -3.01155955e-01 1.30045414e-01
5.11593103e-01 4.22051065e-02 8.12696099e-01 -1.07448888e+00
-6.90358996e-01 -3.76185983e-01 9.60536525e-02 -7.02569902e-01
-4.88225490e-01 3.03678751e-01 4.39524353e-02 -3.72386009e-01
-2.91260839e-01 -1.08135319e+00 3.17170471e-01 -9.38812554e-01
-1.11221242e+00 -3.22299600e-01 1.23679924e+00 -8.68910968e-01
1.36783981e+00 -1.84896398e+00 -1.55760169e-01 1.73624560e-01
6.17346644e-01 3.86081189e-01 4.82526004e-01 8.78134072e-01
2.73553729e-01 5.25029361e-01 1.51872441e-01 -6.34389877e-01
2.59558409e-02 -1.40158042e-01 -5.15468955e-01 9.83351469e-01
7.46067520e-03 6.79371476e-01 -1.22269595e+00 -3.94742370e-01
1.09461829e-01 -7.95456115e-03 -3.18112910e-01 -5.73200919e-02
-3.89136732e-01 7.75137186e-01 -5.25404036e-01 4.44688350e-01
6.11443698e-01 -3.46709341e-01 3.04981351e-01 -8.92633796e-02
-4.55203176e-01 5.07473648e-01 -7.14576304e-01 7.87834585e-01
-1.85213894e-01 1.07875264e+00 1.57799736e-01 -5.52708626e-01
8.53192687e-01 2.40061387e-01 6.05842590e-01 -4.16025549e-01
1.02261734e+00 3.68580192e-01 -4.86111015e-01 -3.02862376e-01
6.18159115e-01 1.85630098e-01 -4.04365152e-01 7.90789604e-01
-1.44687563e-01 -1.11904889e-01 4.29419041e-01 2.92693436e-01
1.34136105e+00 -5.58481514e-01 2.49554649e-01 -2.74654508e-01
6.04366243e-01 4.79350090e-01 2.58270115e-01 6.89902067e-01
-3.55722755e-01 1.05572328e-01 9.83002424e-01 -5.47335923e-01
-9.84347641e-01 -3.85500342e-01 -1.57911330e-01 7.74195075e-01
-3.34420949e-02 -6.08761728e-01 -7.81357050e-01 -8.82136464e-01
1.05623389e-03 9.03452754e-01 -6.37026072e-01 2.66314626e-01
-7.30751693e-01 -1.03461301e+00 8.27511787e-01 -7.01290607e-01
8.41663778e-01 -9.43249285e-01 -4.88930434e-01 4.04905438e-01
-8.29457104e-01 -1.61846006e+00 1.25649393e-01 -3.20267767e-01
-7.99017847e-02 -1.72961020e+00 -1.10152490e-01 1.74711108e-01
6.91981986e-02 -2.82991957e-02 1.06605637e+00 5.77688515e-02
1.09141074e-01 -1.74997076e-01 -4.43232119e-01 -5.65133154e-01
-1.21543837e+00 4.85273957e-01 4.16045859e-02 3.60108405e-01
2.82218963e-01 -4.56274182e-01 -2.78055847e-01 4.29285079e-01
-6.17978752e-01 -1.65545538e-01 -2.81794459e-01 4.44270760e-01
-3.49161804e-01 -6.91124350e-02 6.53808594e-01 -1.42247295e+00
6.08584464e-01 -1.17282176e+00 -7.99321473e-01 -5.85123003e-01
-5.20553470e-01 -7.54877210e-01 4.67472106e-01 -4.03330445e-01
-4.26756084e-01 -5.48774958e-01 9.33790952e-02 -4.33513410e-02
4.38818634e-02 3.47423226e-01 4.69561517e-01 3.35567325e-01
1.09859312e+00 -2.57054508e-01 1.17615543e-01 -4.41806614e-01
3.70123178e-01 1.31508720e+00 1.84942752e-01 -9.33703035e-02
1.08248758e+00 8.08435082e-01 -2.97214210e-01 -1.11270010e+00
-1.01210856e+00 -7.08190739e-01 -3.70534599e-01 -5.25975347e-01
7.46735692e-01 -8.09328914e-01 -1.35107136e+00 9.49409068e-01
-1.60223711e+00 -1.52803913e-01 1.81529466e-02 2.99995653e-02
-2.45127827e-01 2.78635025e-01 -8.75863314e-01 -1.08979058e+00
-3.01586866e-01 -9.10544336e-01 5.85796893e-01 -2.43124649e-01
-6.85371697e-01 -7.69672632e-01 1.04255766e-01 6.14817142e-01
5.21709442e-01 9.86095250e-01 4.28519517e-01 -1.12877166e+00
-2.90486753e-01 -6.10786080e-01 -3.90168995e-01 1.96855366e-01
1.27104744e-01 4.77037691e-02 -1.10184908e+00 -3.16919118e-01
1.03247993e-01 -3.39988142e-01 7.43641794e-01 -3.26509655e-01
5.00055432e-01 -8.70947778e-01 -4.11252588e-01 -3.03717136e-01
1.10474122e+00 -3.60357493e-01 3.79656523e-01 7.90736258e-01
5.94957113e-01 7.19772637e-01 3.99167985e-01 7.98443198e-01
3.87910575e-01 5.79527438e-01 8.28743339e-01 2.79063731e-01
1.76616296e-01 -1.63381666e-01 3.86742741e-01 7.15194881e-01
1.98695641e-02 -3.30371708e-01 -1.14516914e+00 5.11563301e-01
-1.78161907e+00 -1.08524966e+00 -6.84167683e-01 2.00284410e+00
7.58930922e-01 4.06579822e-01 8.82208824e-01 3.86184037e-01
1.07152689e+00 4.35050070e-01 5.14527038e-02 -1.51017657e-03
-1.16193891e-01 -1.55853659e-01 8.65646601e-01 6.98419929e-01
-1.15257800e+00 8.76903236e-01 5.74459505e+00 8.93242240e-01
-1.21901894e+00 7.75192022e-01 4.84531224e-01 -1.80515781e-01
2.16545939e-01 -1.81529045e-01 -9.73333597e-01 1.09724033e+00
1.30856299e+00 -1.12319134e-01 3.70283544e-01 5.14234066e-01
5.55430532e-01 -1.80240750e-01 -3.59848946e-01 6.95889473e-01
-2.24899184e-02 -1.79276764e+00 -4.62620616e-01 5.24284363e-01
6.71358407e-01 6.72951102e-01 -4.34828043e-01 2.44465768e-01
3.22150201e-01 -6.72078550e-01 1.06349051e+00 2.35875130e-01
4.52178448e-01 -4.61130947e-01 9.15175915e-01 9.30963337e-01
-4.90407646e-01 -3.56727868e-01 2.12006330e-01 -8.65242258e-02
2.72274613e-01 1.07528555e+00 -1.29821706e+00 4.72390324e-01
5.10976911e-01 7.32692659e-01 -5.64880729e-01 4.72148567e-01
-2.78007716e-01 9.70564842e-01 -5.83706558e-01 -4.84976530e-01
4.11813945e-01 -3.49494219e-02 1.10654652e+00 1.19504845e+00
-6.72369227e-02 -4.33541447e-01 1.79317459e-01 9.66441691e-01
-3.23944181e-01 7.95438141e-02 -6.49291933e-01 -3.08743238e-01
6.25585377e-01 1.39620388e+00 -6.86699629e-01 -4.52385694e-01
-4.41068830e-03 3.54776680e-01 3.79374862e-01 -9.30561200e-02
-9.63916838e-01 2.12563649e-01 3.04156840e-01 7.97253907e-01
-2.82148331e-01 -1.60483226e-01 -1.06560059e-01 -1.02323508e+00
-2.73251534e-01 -9.95847046e-01 2.18545049e-01 -1.14646479e-01
-1.51104355e+00 5.57831705e-01 2.39264861e-01 -1.06313801e+00
-3.74945104e-01 -4.53776419e-01 -5.50253630e-01 6.99045002e-01
-1.29660344e+00 -1.28195751e+00 -2.80340701e-01 2.27861375e-01
1.69685811e-01 -1.81456864e-01 4.18190509e-01 4.47968245e-01
-6.30088568e-01 1.25697061e-01 -2.74029467e-02 4.05722976e-01
6.17645144e-01 -8.92317057e-01 3.60499799e-01 2.73950100e-01
-3.04233372e-01 2.26948336e-01 1.09361398e+00 -9.38861847e-01
-9.36492205e-01 -1.33295202e+00 1.14017892e+00 -8.97992015e-01
1.40178823e+00 -5.70834219e-01 -5.53203821e-01 6.06073201e-01
1.71565503e-01 -3.92539561e-01 3.65002155e-01 3.76934744e-02
-2.59503961e-01 2.57298678e-01 -1.42243874e+00 2.63747096e-01
5.58634102e-01 -4.86393809e-01 -2.97872633e-01 1.09016633e+00
4.26012456e-01 -3.46329898e-01 -6.04361117e-01 -6.30209520e-02
2.89495558e-01 -1.13434207e+00 4.02311981e-01 -4.72075045e-01
4.87907678e-01 -1.52684346e-01 -2.28399336e-02 -1.09478652e+00
1.30410358e-01 -9.43796337e-01 -3.27120312e-02 1.47315037e+00
4.23005700e-01 -1.10468757e+00 8.00823331e-01 7.90863410e-02
4.43799794e-01 -3.09254646e-01 -1.24969137e+00 -6.39571905e-01
6.66669756e-02 -4.84306365e-01 4.57277209e-01 1.34614742e+00
1.95144355e-01 2.74048150e-01 -4.54362422e-01 -9.23777446e-02
5.64586222e-01 -1.74614370e-01 1.21621311e+00 -1.41637003e+00
-1.11255683e-01 -4.79388326e-01 -3.86760026e-01 -3.00273657e-01
2.08500117e-01 -7.95138240e-01 -5.12715697e-01 -1.00517380e+00
1.53704450e-01 -4.14329022e-01 4.66634363e-01 2.99453199e-01
2.56627679e-01 6.18041098e-01 3.35978270e-01 6.24377489e-01
-2.53920227e-01 2.91631907e-01 7.91347086e-01 -4.11646903e-01
-1.22191243e-01 8.63623060e-03 -3.62311274e-01 9.45922971e-01
8.02601278e-01 -8.79936755e-01 3.10989290e-01 2.80538529e-01
6.21399403e-01 -2.40643114e-01 5.58362186e-01 -7.43371844e-01
-1.45935923e-01 1.61128223e-03 -2.75924951e-01 -8.05131257e-01
3.39488983e-01 -5.46458602e-01 1.19117849e-01 7.57975161e-01
1.47377312e-01 -1.19668446e-01 6.05489537e-02 6.18836343e-01
3.44484821e-02 -1.73854485e-01 8.66239727e-01 -2.84845233e-01
1.25291213e-01 1.03339972e-02 -6.80022061e-01 3.02227199e-01
1.03456116e+00 2.28331536e-01 -1.10885751e+00 -3.30957234e-01
-6.20681286e-01 5.68521172e-02 4.00157392e-01 3.62785250e-01
-4.53032792e-01 -9.94435966e-01 -1.05809569e+00 -2.32733592e-01
-4.50517386e-02 -5.75650811e-01 -2.49579236e-01 1.04793346e+00
-5.59578598e-01 2.37445697e-01 -4.76527214e-02 -6.30700052e-01
-1.04583049e+00 3.41027081e-01 -3.42388870e-04 -5.98957002e-01
-3.45080346e-01 1.95127979e-01 -7.19485283e-01 -5.61181068e-01
-3.09217870e-01 9.28771347e-02 -4.45241123e-01 8.92826617e-01
6.09697580e-01 8.93920898e-01 1.47045329e-01 -1.08058727e+00
-4.07471776e-01 -2.36965746e-01 1.12274855e-01 -2.11598814e-01
1.16497445e+00 -1.85512885e-01 -4.96579498e-01 8.25294971e-01
1.23933136e+00 5.52057803e-01 -3.09833109e-01 -1.38718292e-01
1.67441249e-01 -5.33291161e-01 -8.00527781e-02 -5.41478157e-01
-6.70056224e-01 4.25653309e-01 8.89170840e-02 1.28009212e+00
1.97483554e-01 1.27415076e-01 7.52705455e-01 -6.09712712e-02
5.86828947e-01 -1.03825152e+00 4.85960441e-03 7.42785215e-01
8.30691397e-01 -1.30183864e+00 6.19476922e-02 -6.33713007e-01
-1.48289517e-01 8.31387639e-01 1.75257102e-01 -3.94117832e-02
6.68832362e-01 -2.06214190e-01 6.11629747e-02 -4.99763727e-01
-3.91046911e-01 -4.07644594e-03 -2.58678645e-01 8.95664543e-02
5.87207377e-02 2.92068809e-01 -6.67803645e-01 4.02760923e-01
-5.85603714e-01 -4.39512193e-01 8.65898073e-01 6.12531424e-01
-3.80847603e-01 -9.05417383e-01 -5.35003364e-01 3.99799109e-01
-7.31307447e-01 2.27229506e-01 -8.22435260e-01 1.04702544e+00
1.45307630e-01 1.41855335e+00 -1.36043340e-01 -5.60435891e-01
2.01907635e-01 -2.12464072e-02 -2.27206618e-01 -4.20440257e-01
-1.13117635e+00 -4.18017775e-01 7.51716256e-01 -4.36796069e-01
-2.99681157e-01 -7.27647841e-01 -8.24093461e-01 -1.17988431e+00
-5.74982882e-01 2.73713976e-01 8.48740578e-01 1.16545117e+00
2.31810838e-01 8.55370015e-02 1.02004302e+00 -9.21846330e-01
-5.15847027e-01 -1.52054429e+00 -4.91802841e-01 7.66031384e-01
3.53251010e-01 -7.02773929e-01 -8.69944036e-01 -4.71272260e-01] | [8.150965690612793, 10.198779106140137] |
33428b91-13fc-40a5-a25a-33c37764d5e9 | ganmex-one-vs-one-attributions-using-gan | 2011.06015 | null | https://arxiv.org/abs/2011.06015v4 | https://arxiv.org/pdf/2011.06015v4.pdf | GANMEX: One-vs-One Attributions Guided by GAN-based Counterfactual Explanation Baselines | Attribution methods have been shown as promising approaches for identifying key features that led to learned model predictions. While most existing attribution methods rely on a baseline input for performing feature perturbations, limited research has been conducted to address the baseline selection issues. Poor choices of baselines limit the ability of one-vs-one (1-vs-1) explanations for multi-class classifiers, which means the attribution methods were not able to explain why an input belongs to its original class but not the other specified target class. 1-vs-1 explanation is crucial when certain classes are more similar than others, e.g. two bird types among multiple animals, by focusing on key differentiating features rather than shared features across classes. In this paper, we present GAN-based Model EXplainability (GANMEX), a novel approach applying Generative Adversarial Networks (GAN) by incorporating the to-be-explained classifier as part of the adversarial networks. Our approach effectively selects the counterfactual baseline as the closest realistic sample belong to the target class, which allows attribution methods to provide true 1-vs-1 explanations. We showed that GANMEX baselines improved the saliency maps and led to stronger performance on perturbation-based evaluation metrics over the existing baselines. Existing attribution results are known for being insensitive to model randomization, and we demonstrated that GANMEX baselines led to better outcome under the cascading randomization of the model. | ['Zohar Karnin', 'Pin-Ju Tien', 'Sheng-Min Shih'] | 2020-11-11 | null | null | null | null | ['counterfactual-explanation'] | ['miscellaneous'] | [ 6.36931121e-01 5.10441184e-01 -3.62451643e-01 -5.05625248e-01
-3.46306354e-01 -6.07277632e-01 9.45202589e-01 -1.20947950e-01
-1.29512133e-04 9.38339531e-01 2.28573248e-01 -1.97372243e-01
-2.21650764e-01 -7.50631392e-01 -9.80858445e-01 -6.83649004e-01
1.42426282e-01 4.13748056e-01 2.45310664e-02 -2.50182778e-01
3.93087268e-01 2.76115239e-01 -1.72011876e+00 5.25607884e-01
7.92747140e-01 6.29437685e-01 -2.47025229e-02 5.45050263e-01
1.33207142e-01 5.93685746e-01 -1.02462006e+00 -6.57594562e-01
4.16860461e-01 -9.64227438e-01 -6.11308813e-01 -4.20927405e-01
6.42954886e-01 -1.66588143e-01 -5.04220165e-02 1.06086278e+00
2.81028599e-01 7.72802159e-02 1.04859209e+00 -2.15873528e+00
-1.10383248e+00 9.17588055e-01 -3.57177764e-01 1.64332658e-01
2.85440773e-01 3.24950546e-01 1.03040814e+00 -5.31073272e-01
4.34847742e-01 1.45215142e+00 7.24877656e-01 1.00836062e+00
-1.53967071e+00 -1.14426780e+00 2.73606777e-01 5.26908576e-01
-1.04769850e+00 -1.50864631e-01 8.21606219e-01 -4.03050303e-01
6.85069799e-01 6.64976835e-01 3.72368455e-01 1.65339005e+00
3.72739583e-01 4.72638845e-01 1.41179073e+00 -2.39403948e-01
3.22226107e-01 4.50056285e-01 -8.96250382e-02 1.55606598e-01
3.44560117e-01 6.50373042e-01 -6.86245263e-01 -1.70721948e-01
5.10392964e-01 -2.47718077e-02 -3.47926527e-01 -2.21704602e-01
-1.29579341e+00 1.23297358e+00 9.33991134e-01 -8.53434801e-02
-5.47843874e-01 3.55553180e-01 9.25548524e-02 2.34871894e-01
3.06659222e-01 1.20664239e+00 -4.43073124e-01 1.44265309e-01
-9.50191021e-01 5.09465873e-01 4.82150465e-01 8.38890851e-01
7.66452789e-01 1.91569433e-01 -7.52926171e-01 3.58800322e-01
-3.10484953e-02 4.48195130e-01 5.60792208e-01 -9.03276145e-01
1.18985184e-01 5.94891787e-01 4.74413224e-02 -1.07994127e+00
-1.66821465e-01 -6.11295342e-01 -7.64099240e-01 6.03733957e-01
2.63595253e-01 -9.11186170e-03 -1.02974415e+00 2.22673416e+00
1.76604778e-01 1.91347376e-01 2.40070239e-01 1.04345918e+00
7.88981616e-01 2.70418942e-01 1.69955224e-01 1.36792302e-01
1.16498768e+00 -1.00207162e+00 -5.06999552e-01 -4.35172141e-01
1.83608234e-01 -5.59646428e-01 1.36877871e+00 -1.52835608e-01
-5.65724611e-01 -5.13697803e-01 -1.03172326e+00 5.49744844e-01
-5.64593911e-01 -2.68074125e-01 7.45453119e-01 7.79036462e-01
-9.01627004e-01 7.88390279e-01 -4.01687890e-01 -2.86247253e-01
6.81223750e-01 3.58148098e-01 -3.12442601e-01 2.08954602e-01
-1.40195918e+00 1.13762856e+00 1.91938654e-01 -3.12238067e-01
-1.36218309e+00 -1.12111950e+00 -6.12383008e-01 3.07942778e-01
1.32622331e-01 -8.53336394e-01 1.06664765e+00 -1.52481639e+00
-1.06626666e+00 6.13601565e-01 -8.33765715e-02 -8.34124506e-01
6.98260844e-01 -7.71238282e-02 -3.06946099e-01 -1.62619337e-01
3.34014535e-01 1.16814137e+00 1.18191016e+00 -1.61051035e+00
-1.77885920e-01 -3.31889801e-02 3.43840003e-01 1.49492696e-01
-5.03012240e-02 -9.90661904e-02 4.79605496e-01 -8.29535007e-01
-2.14063823e-01 -1.02732587e+00 -2.18002349e-02 3.56066041e-02
-9.38101053e-01 1.61803558e-01 8.82424116e-01 -2.88901746e-01
7.79349208e-01 -1.75109434e+00 -1.49291292e-01 -1.02895781e-01
2.58852065e-01 1.81216046e-01 -1.38957575e-01 2.47091144e-01
-6.02729082e-01 5.24807632e-01 -2.07290724e-01 -2.53719121e-01
1.58755586e-01 1.27882496e-01 -6.95549786e-01 1.80448860e-01
4.58818763e-01 8.52298975e-01 -9.18372929e-01 -1.68568432e-01
1.80665031e-01 4.20647353e-01 -5.36163628e-01 3.22335839e-01
-1.76383987e-01 4.69816029e-01 -1.82485953e-02 4.25238252e-01
6.04120195e-01 5.73913269e-02 -4.35099036e-01 -6.60886988e-02
3.61968040e-01 1.49771482e-01 -8.32240164e-01 8.21784198e-01
-2.66820967e-01 7.87565768e-01 -6.03983283e-01 -7.05966711e-01
9.49730098e-01 9.12752375e-02 -7.81345069e-02 -3.71229947e-01
-8.98653716e-02 3.93909290e-02 4.58478600e-01 9.40968990e-02
3.22458178e-01 -3.86348277e-01 -4.72523831e-02 4.17085558e-01
-1.52026683e-01 -1.50570258e-01 -3.19852501e-01 2.15561107e-01
9.66298342e-01 1.50049731e-01 4.19443429e-01 -3.59023184e-01
7.15229586e-02 2.08323255e-01 7.32624114e-01 1.11572206e+00
-3.05520236e-01 9.88367260e-01 5.71648359e-01 -3.79264832e-01
-9.13459599e-01 -1.06936347e+00 -1.21650107e-01 9.27939653e-01
3.19939375e-01 -4.81816456e-02 -8.34485352e-01 -9.79959548e-01
1.63643628e-01 1.64932156e+00 -1.43871236e+00 -8.46852422e-01
5.64208142e-02 -6.64721191e-01 7.09929645e-01 5.23436248e-01
4.96586084e-01 -1.29423904e+00 -9.26563382e-01 -1.35398611e-01
-2.65218746e-02 -5.30983984e-01 -5.47298968e-01 3.30134332e-01
-5.30131638e-01 -1.23164630e+00 -5.33132493e-01 -1.21716343e-01
7.91694582e-01 1.61668763e-01 1.16958702e+00 8.93040597e-02
8.23316276e-02 8.70546177e-02 -2.96612322e-01 -9.50052559e-01
-7.03694105e-01 -8.27084929e-02 2.78590471e-01 -1.59229804e-02
3.34289819e-01 -5.14886439e-01 -5.98094523e-01 6.39950335e-01
-7.21223414e-01 3.79987508e-01 4.81873155e-01 1.01295650e+00
3.18407446e-01 -2.44374037e-01 9.05481279e-01 -1.03566635e+00
5.91390669e-01 -6.42703950e-01 -9.26462337e-02 2.49225721e-01
-9.68750775e-01 2.68904597e-01 8.95854592e-01 -8.05594683e-01
-8.28621566e-01 -8.61841515e-02 4.19924796e-01 -8.64710987e-01
-2.62672007e-01 7.80442879e-02 -2.98694164e-01 6.11087941e-02
9.52547967e-01 1.14967756e-01 4.07130569e-02 -5.24214581e-02
3.37455124e-01 3.00187379e-01 5.47636330e-01 -3.69959474e-01
9.99331594e-01 1.79999888e-01 -4.83392552e-03 -1.17842883e-01
-8.85462701e-01 3.40213209e-01 -4.85000730e-01 -2.46751159e-01
7.53751516e-01 -6.18001223e-01 -4.14013475e-01 2.49760240e-01
-1.04618680e+00 -2.95682818e-01 -4.74098265e-01 2.79920697e-01
-4.73145276e-01 -3.18768024e-01 1.61252022e-01 -6.17566645e-01
-2.37428889e-01 -1.28296947e+00 7.42947340e-01 4.64499831e-01
-6.88567698e-01 -8.62175822e-01 -1.32398590e-01 2.81969070e-01
8.04398000e-01 6.04607046e-01 9.68695998e-01 -1.16282296e+00
-3.53721082e-01 -1.87198296e-01 -1.49913445e-01 2.18406036e-01
3.32699448e-01 9.44956616e-02 -1.24948251e+00 -1.48155242e-01
-1.99569508e-01 -1.86030373e-01 8.18175793e-01 3.64602178e-01
1.22712266e+00 -6.55704439e-01 -4.11534458e-01 5.08595824e-01
1.05117452e+00 6.75703213e-02 6.57509804e-01 5.47913551e-01
6.91162944e-01 5.29511988e-01 6.31063104e-01 6.65350705e-02
2.45329931e-01 7.73441017e-01 1.01911545e+00 -3.78001392e-01
-4.26091075e-01 -6.11593246e-01 3.83396506e-01 -2.38389879e-01
1.87003002e-01 -2.56362289e-01 -7.23281264e-01 6.43539548e-01
-1.60374057e+00 -1.19867826e+00 -1.29774407e-01 2.28669715e+00
7.21024215e-01 1.90170735e-01 -7.57237375e-02 -4.78555150e-02
1.05709243e+00 -1.66398787e-03 -1.06383181e+00 -5.25729954e-01
-3.52413654e-01 -2.87274439e-02 4.79818434e-01 3.25503677e-01
-9.26122725e-01 8.06283712e-01 6.70020151e+00 6.49252594e-01
-1.14219117e+00 6.15511350e-02 7.22911835e-01 -2.79815197e-01
-6.80421174e-01 2.55555063e-01 -6.22900128e-01 6.28036380e-01
8.64693701e-01 -4.88299817e-01 4.91500318e-01 9.97226417e-01
1.72109425e-01 8.98743197e-02 -1.40222931e+00 6.34003818e-01
1.78969011e-01 -1.23521328e+00 5.49609184e-01 1.35766864e-01
9.54568505e-01 -3.08492392e-01 2.94855803e-01 3.39263052e-01
4.69775170e-01 -1.43307817e+00 1.08857071e+00 5.02951324e-01
7.37292945e-01 -7.26590037e-01 8.95632744e-01 1.27289027e-01
-6.69033766e-01 -1.28887817e-02 -3.66669446e-01 -2.34941289e-01
-2.94912636e-01 3.18144947e-01 -1.20555079e+00 3.60084355e-01
6.81233525e-01 5.03518164e-01 -9.42814350e-01 8.26730251e-01
-5.40051401e-01 9.19862747e-01 -4.55483273e-02 -8.61639008e-02
2.20649410e-02 3.38357031e-01 9.21378195e-01 7.70171463e-01
1.68162122e-01 -3.12240213e-01 -1.19398400e-01 1.49178159e+00
-1.58340961e-01 -3.83702189e-01 -6.64999902e-01 1.58488244e-01
7.11232841e-01 9.24732327e-01 -6.74945235e-01 -2.82466024e-01
6.36368021e-02 9.38028216e-01 1.01316907e-01 2.21963435e-01
-1.33339894e+00 -3.87910515e-01 1.05163980e+00 1.29569501e-01
2.03979760e-01 6.07628107e-01 -7.02910125e-01 -9.03650463e-01
-2.21830547e-01 -1.12999201e+00 2.72538632e-01 -1.03118598e+00
-1.40291893e+00 6.69838190e-01 2.35294372e-01 -1.13720012e+00
-2.59549141e-01 -2.25665480e-01 -1.09646499e+00 1.06126261e+00
-1.31443930e+00 -1.36711836e+00 -4.66528982e-01 4.53844070e-01
4.91849035e-01 -3.17204833e-01 1.08975136e+00 -3.11918944e-01
-4.73651260e-01 9.94348586e-01 -1.79812722e-02 -2.06889465e-01
8.92840743e-01 -1.43956733e+00 3.18702579e-01 1.02339327e+00
1.35042682e-01 7.78561652e-01 1.25075722e+00 -6.71404779e-01
-7.07248390e-01 -1.21657717e+00 6.46831691e-01 -6.86754227e-01
2.98698694e-01 -2.70909369e-01 -9.38901901e-01 7.09706962e-01
2.14145184e-01 -2.53454894e-01 7.01214254e-01 -3.50667797e-02
-4.90725070e-01 5.41543309e-03 -1.50219297e+00 6.81333423e-01
9.05789256e-01 -3.91709775e-01 -6.93931162e-01 1.23935163e-01
9.04344916e-01 -2.39721239e-01 -4.73253787e-01 3.80958945e-01
4.58507419e-01 -1.04666007e+00 9.17801023e-01 -9.19757545e-01
1.00760829e+00 -5.95504224e-01 -2.65811831e-01 -1.79440963e+00
-4.49791551e-01 -3.36827308e-01 4.39920500e-02 1.49274063e+00
6.89614773e-01 -8.85152459e-01 5.43424368e-01 8.53935421e-01
-1.34006903e-01 -6.28174603e-01 -8.85917783e-01 -7.91781962e-01
1.79503143e-01 -1.82707518e-01 1.34157050e+00 1.16270196e+00
-4.94300574e-01 1.23689264e-01 -5.10370612e-01 2.77465612e-01
6.20622218e-01 2.64630139e-01 8.59813690e-01 -1.02618062e+00
-2.46334508e-01 -5.88586211e-01 -7.26634443e-01 -1.13787539e-01
3.04911405e-01 -9.74058986e-01 2.64638960e-02 -1.40711105e+00
3.66548032e-01 -3.63647461e-01 -3.83563280e-01 9.28458691e-01
-6.93820536e-01 3.42758089e-01 4.65212137e-01 2.77422220e-01
1.30021507e-02 6.40026689e-01 1.04041445e+00 -3.74319881e-01
4.15158793e-02 3.01270783e-02 -1.38824511e+00 4.77609962e-01
8.97822678e-01 -8.86224151e-01 -5.42705357e-01 -1.12018906e-01
-1.55146644e-01 -4.31618690e-01 1.00262368e+00 -1.00594604e+00
-7.68269673e-02 -3.61353159e-01 7.13145971e-01 -9.00053829e-02
2.28212848e-01 -6.39367104e-01 6.23902619e-01 5.83916843e-01
-7.79815435e-01 2.86678299e-02 4.06786978e-01 5.30536115e-01
7.55787715e-02 -1.72564179e-01 7.63289154e-01 5.89577183e-02
-5.65116644e-01 -9.04629752e-02 -3.65465283e-02 2.31476054e-02
1.23005199e+00 -4.15810138e-01 -6.16667449e-01 -5.69385946e-01
-4.27948624e-01 -2.72883661e-02 8.16052973e-01 7.78747320e-01
5.10135770e-01 -1.39185214e+00 -9.18867111e-01 6.62723854e-02
2.66151756e-01 -3.63440841e-01 1.88452266e-02 3.82339239e-01
4.29804549e-02 1.13385953e-01 -6.17247880e-01 -5.73486269e-01
-1.28769374e+00 6.38144195e-01 5.68213940e-01 7.37243192e-03
-3.31456959e-01 9.44646776e-01 6.64568722e-01 -5.47038436e-01
-9.94414911e-02 -4.15656418e-02 -1.72078490e-01 -2.07891643e-01
4.47757632e-01 2.47869909e-01 -2.23999873e-01 -6.85804844e-01
-6.19713068e-01 3.31598893e-02 2.79152989e-02 6.23672865e-02
1.23223507e+00 1.99370220e-01 2.29802385e-01 4.70572412e-01
8.21604669e-01 -2.47463807e-01 -1.43101108e+00 2.64736921e-01
-4.21556920e-01 -7.63195395e-01 -3.11039351e-02 -1.34645236e+00
-1.10658419e+00 7.55510390e-01 6.47215188e-01 1.30465522e-01
1.08391500e+00 -1.01603664e-01 1.65961683e-01 -1.29639521e-01
2.97532082e-01 -4.45237577e-01 -8.08523223e-02 2.09036339e-02
1.36853302e+00 -1.35666966e+00 -7.97480866e-02 -2.75262356e-01
-1.07455528e+00 8.87452424e-01 9.69791949e-01 -3.02500166e-02
1.51304856e-01 -3.01673621e-01 4.86562848e-02 -8.35278332e-02
-8.22006047e-01 9.57819670e-02 4.95919228e-01 1.01730943e+00
1.66927472e-01 4.27953959e-01 1.85879260e-01 8.58347833e-01
-7.97316313e-01 -4.29021776e-01 6.88195527e-01 3.76017809e-01
1.16184369e-01 -8.16031277e-01 -6.18390262e-01 7.99530029e-01
-1.74117729e-01 -2.21478507e-01 -8.46141815e-01 9.00064528e-01
2.29311585e-01 1.04765522e+00 -2.63196565e-02 -7.16545224e-01
3.29468489e-01 7.64094144e-02 1.21853113e-01 -5.32857418e-01
-9.57500994e-01 -8.01484764e-01 -1.34500787e-01 -5.22259474e-01
-2.96031892e-01 -7.44777441e-01 -1.01060462e+00 -4.25603867e-01
-5.01851976e-01 5.86696826e-02 5.06741405e-01 8.60420406e-01
5.59008718e-01 8.34880590e-01 6.27281010e-01 -9.29442942e-01
-5.69381177e-01 -1.06064200e+00 -2.09095523e-01 7.69048512e-01
3.88945401e-01 -1.19778788e+00 -7.64781535e-01 6.04314543e-02] | [8.803476333618164, 5.533175468444824] |
7b1437f5-a2db-4864-b245-9e8620df3574 | a-rich-morphological-tagger-for-english | null | null | https://aclanthology.org/E17-2018 | https://aclanthology.org/E17-2018.pdf | A Rich Morphological Tagger for English: Exploring the Cross-Linguistic Tradeoff Between Morphology and Syntax | A traditional claim in linguistics is that all human languages are equally expressive{---}able to convey the same wide range of meanings. Morphologically rich languages, such as Czech, rely on overt inflectional and derivational morphology to convey many semantic distinctions. Languages with comparatively limited morphology, such as English, should be able to accomplish the same using a combination of syntactic and contextual cues. We capitalize on this idea by training a tagger for English that uses syntactic features obtained by automatic parsing to recover complex morphological tags projected from Czech. The high accuracy of the resulting model provides quantitative confirmation of the underlying linguistic hypothesis of equal expressivity, and bodes well for future improvements in downstream HLT tasks including machine translation. | ['John Sylak-Glassman', 'Rebecca Knowles', 'Matt Post', 'Ryan Cotterell', 'Christo Kirov'] | 2017-04-01 | null | null | null | eacl-2017-4 | ['morphological-tagging'] | ['natural-language-processing'] | [-1.41029991e-02 1.06130660e-01 -3.02267194e-01 -7.01746523e-01
-6.02657497e-01 -1.01535428e+00 6.20438159e-01 5.48608363e-01
-6.65002406e-01 6.31157756e-01 6.38352096e-01 -8.19998860e-01
-1.86317608e-01 -7.89307177e-01 -2.75923940e-03 -2.99871832e-01
3.49739432e-01 3.92237872e-01 1.52381614e-01 -7.28031933e-01
2.42226437e-01 3.88849169e-01 -1.02327514e+00 4.29266363e-01
6.82346523e-01 4.96148050e-01 2.60310858e-01 6.56338111e-02
-6.27278328e-01 6.12785518e-01 -3.03860962e-01 -9.62067306e-01
-1.99705623e-02 -3.08316469e-01 -9.43165183e-01 -1.76935017e-01
-1.01592839e-02 1.57227069e-01 2.63989240e-01 1.06055498e+00
2.79344112e-01 -1.59292430e-01 7.03002036e-01 -4.89478648e-01
-6.30004942e-01 9.77736890e-01 -1.93042651e-01 2.09276438e-01
4.92762208e-01 -4.61208709e-02 1.65697968e+00 -9.84732509e-01
8.08567286e-01 1.25991690e+00 6.45053685e-01 5.58866084e-01
-1.40811050e+00 -2.59316891e-01 3.75427268e-02 -3.96283031e-01
-1.17630720e+00 -6.48679018e-01 7.31805205e-01 -3.76628280e-01
1.25924289e+00 2.54679024e-01 4.85709369e-01 7.79185593e-01
4.59069423e-02 4.66876626e-01 1.60046864e+00 -7.90965497e-01
-1.20833792e-01 3.36652070e-01 -1.45256734e-02 5.33569753e-01
3.51922899e-01 1.36229554e-02 -3.77269894e-01 -1.41248265e-02
4.91638452e-01 -3.92507702e-01 -1.53986573e-01 3.21116328e-01
-1.36627352e+00 9.68496323e-01 1.95629895e-02 9.80189681e-01
-1.03123918e-01 -1.45081878e-01 7.60888219e-01 5.69298744e-01
2.79578000e-01 7.34054089e-01 -8.20968390e-01 -8.30869377e-02
-6.46019399e-01 -1.69860478e-02 8.94896328e-01 7.52505600e-01
6.12796724e-01 6.81626145e-03 3.54142636e-01 1.16064715e+00
3.20620716e-01 4.81664151e-01 6.34528518e-01 -1.03829217e+00
4.54076588e-01 6.11076057e-01 -4.12498228e-02 -7.96605647e-01
-3.56092811e-01 -1.12922251e-01 -2.84338623e-01 1.25617966e-01
5.73586226e-01 -1.14995457e-01 -5.73610604e-01 2.08627200e+00
2.61005193e-01 -1.04550683e+00 4.90118146e-01 5.68828404e-01
5.00991583e-01 5.75339556e-01 6.54046357e-01 -8.27872455e-02
1.59836340e+00 -2.42684875e-02 -5.13775766e-01 -3.99329990e-01
9.94316399e-01 -1.02067053e+00 1.54774153e+00 6.05945438e-02
-1.37976265e+00 -1.46007806e-01 -7.91052282e-01 -3.12508047e-01
-4.74744618e-01 -4.30993736e-02 1.17465794e+00 9.84887719e-01
-1.04781187e+00 4.65259194e-01 -7.82752335e-01 -4.60714161e-01
1.30085601e-02 2.94681132e-01 -6.16952240e-01 2.55267531e-01
-1.17854333e+00 1.08465362e+00 6.54381633e-01 1.70875624e-01
2.05747694e-01 -2.71054655e-01 -1.13159025e+00 -1.18328720e-01
-3.19356844e-02 -6.37885094e-01 1.28738832e+00 -1.33386695e+00
-1.20726895e+00 1.75198424e+00 -1.38249800e-01 -3.03961746e-02
-8.47341493e-02 1.67483464e-01 -4.50443536e-01 3.64347212e-02
3.13570410e-01 4.92794096e-01 1.77723125e-01 -1.06950319e+00
-9.13924336e-01 -7.13864148e-01 9.02994722e-02 1.87137708e-01
-2.74953812e-01 9.51552927e-01 3.16022076e-02 -7.47836709e-01
3.59919637e-01 -8.18429112e-01 -5.31622358e-02 -4.74610001e-01
-9.45699960e-02 -4.67980325e-01 1.77503318e-01 -8.37605417e-01
1.20706177e+00 -1.94987261e+00 -1.43554419e-01 2.92182386e-01
-1.32048145e-01 -6.77726269e-02 2.41445825e-01 5.97670257e-01
-2.89250235e-03 4.74888206e-01 -2.47681454e-01 -4.97035384e-02
4.60512131e-01 6.89578354e-01 -2.69083202e-01 2.38276809e-01
5.31441867e-01 1.26153898e+00 -6.71891332e-01 -5.04089057e-01
8.04004669e-02 1.08640015e-01 -3.30202401e-01 -2.31216058e-01
-6.26324266e-02 2.09393695e-01 -4.11157906e-01 6.81116164e-01
2.02750310e-01 1.56038314e-01 7.88991511e-01 4.35813725e-01
-3.25792104e-01 1.21656537e+00 -8.58089209e-01 1.41804802e+00
-5.89067042e-01 2.95912921e-01 3.04540306e-01 -9.49535012e-01
6.74230874e-01 4.25506413e-01 -3.22542600e-02 -7.12502182e-01
3.12190861e-01 7.65888512e-01 5.53484857e-01 -3.01274747e-01
5.34870803e-01 -9.50852692e-01 -7.34379828e-01 2.83003777e-01
-1.64190620e-01 -3.10517818e-01 2.53096316e-02 -2.60085855e-02
8.21718931e-01 1.73366413e-01 8.10182154e-01 -8.25343728e-01
5.20864606e-01 2.58215755e-01 8.29819262e-01 1.42209589e-01
2.01279558e-02 1.41425192e-01 4.36607093e-01 -1.82479471e-01
-1.13128364e+00 -1.07245541e+00 -5.28631747e-01 1.32525337e+00
-2.08033472e-01 -2.00066239e-01 -5.59317708e-01 -5.61987758e-01
-7.59971887e-02 1.04860973e+00 -2.76845217e-01 2.58067936e-01
-9.97302353e-01 -7.09795654e-01 8.66556346e-01 6.58720553e-01
-1.05411716e-01 -1.40405071e+00 -5.43506682e-01 5.08745909e-01
-5.53907491e-02 -1.22126615e+00 5.20437583e-02 4.38270956e-01
-7.80564904e-01 -6.25681162e-01 -1.29601032e-01 -1.17464101e+00
2.62335241e-01 -1.83653012e-01 1.23594189e+00 3.21872890e-01
2.47388363e-01 5.81529811e-02 -2.64723986e-01 -4.63012069e-01
-9.73390162e-01 -8.48321244e-02 -2.47491285e-01 -5.10647476e-01
7.12255478e-01 -6.46549821e-01 -2.28820276e-02 -1.00899667e-01
-9.79490697e-01 -2.11555243e-01 3.75305206e-01 6.12159193e-01
3.77355456e-01 -2.84903437e-01 5.15081644e-01 -1.29474628e+00
5.67515254e-01 -3.49559844e-01 -2.63665795e-01 5.05968668e-02
-4.37139094e-01 2.44022846e-01 6.53349698e-01 -2.24002033e-01
-1.36503005e+00 -2.29177371e-01 -8.05365384e-01 6.24081731e-01
-4.24355924e-01 6.12364888e-01 -7.11779118e-01 2.86894858e-01
4.88509178e-01 -6.78791432e-03 -2.61855751e-01 -4.08497155e-01
4.81411457e-01 7.79440880e-01 7.11695075e-01 -9.70023572e-01
7.14764833e-01 3.83363843e-01 -6.31172210e-02 -9.11819041e-01
-7.85932362e-01 -4.61660504e-01 -6.05372488e-01 4.62928146e-01
1.04734945e+00 -8.38518500e-01 -4.15638298e-01 2.34160021e-01
-1.23552442e+00 -3.23043048e-01 -4.13778692e-01 5.45756459e-01
-5.05812228e-01 3.94589931e-01 -1.06874967e+00 -5.64668238e-01
-2.12612525e-01 -6.38218582e-01 9.96582270e-01 -2.60921091e-01
-7.04640090e-01 -1.48581958e+00 -6.80892766e-02 3.70846651e-02
2.57061988e-01 2.64368594e-01 1.62591469e+00 -1.03481162e+00
7.78354378e-03 4.87024933e-02 -2.24138442e-02 2.30733663e-01
1.77089214e-01 -2.35758409e-01 -7.33190775e-01 6.43104389e-02
2.53928244e-01 -3.81241262e-01 6.84315741e-01 2.71424539e-02
1.49610937e-01 -2.87415028e-01 -6.64786920e-02 3.28847378e-01
1.64079177e+00 -3.50096449e-03 5.69040298e-01 3.56286675e-01
4.16144729e-01 1.11409497e+00 4.44216162e-01 -6.95453137e-02
8.24570954e-01 4.72410113e-01 -7.33977780e-02 -8.79551545e-02
-1.22137465e-01 -2.43236050e-01 5.38742602e-01 1.10135853e+00
9.98708457e-02 2.84176203e-03 -1.06391132e+00 8.54834020e-01
-1.45595837e+00 -8.31883073e-01 -4.96409416e-01 2.03708315e+00
1.31139743e+00 3.78880918e-01 -9.32981968e-02 2.90928990e-01
6.08809888e-01 1.46172464e-01 4.36146587e-01 -9.81587887e-01
-4.84070241e-01 5.91817796e-01 2.89496690e-01 7.23024249e-01
-8.31074417e-01 1.70672131e+00 6.53907919e+00 7.42374837e-01
-1.09197366e+00 1.49111241e-01 2.36403033e-01 5.95740020e-01
-8.55527103e-01 2.17808455e-01 -8.35349381e-01 2.43152142e-01
8.86610985e-01 -2.86125913e-02 2.48101324e-01 5.03573120e-01
2.51093864e-01 -1.89302176e-01 -1.12125742e+00 5.49802840e-01
-1.54072016e-01 -8.94517541e-01 -6.16049841e-02 6.95631430e-02
4.65249270e-01 -2.01173276e-02 -1.17626429e-01 7.47281164e-02
6.43563867e-01 -9.52155352e-01 9.95576620e-01 -5.46557717e-02
9.47388947e-01 -6.27372086e-01 7.57901311e-01 3.23896736e-01
-1.03633428e+00 1.63062140e-01 -5.63205659e-01 -4.88355964e-01
3.81056666e-01 3.20498109e-01 -6.21406913e-01 3.11991274e-01
1.37439147e-01 1.84361696e-01 -4.01448220e-01 3.85520369e-01
-8.91241193e-01 9.34012413e-01 -3.42288136e-01 -9.52085927e-02
3.75317544e-01 -3.18758100e-01 5.90158820e-01 1.73270953e+00
2.18130693e-01 2.66420811e-01 1.35568455e-01 6.06294096e-01
1.14211813e-01 7.96791017e-01 -8.19137394e-01 -1.63607717e-01
4.73843813e-01 1.27259028e+00 -1.03677762e+00 -3.59402031e-01
-7.63216078e-01 9.35927212e-01 6.51240110e-01 3.42308544e-02
-2.77540922e-01 -3.18440884e-01 7.01155305e-01 2.34831199e-01
1.75643444e-01 -6.49668396e-01 -5.57907939e-01 -1.00168765e+00
1.87221050e-01 -8.03763270e-01 4.54414636e-01 -5.09583771e-01
-1.39690781e+00 4.81980830e-01 -2.55992144e-01 -5.18153071e-01
-4.81381625e-01 -1.02520585e+00 -4.44728494e-01 1.07978153e+00
-1.48733521e+00 -1.62382782e+00 5.93283117e-01 5.27960062e-01
3.46321374e-01 1.57786101e-01 1.10495293e+00 -8.77049640e-02
-8.55341330e-02 5.14311671e-01 -2.90442079e-01 3.29835743e-01
6.09988511e-01 -1.65111518e+00 4.26020652e-01 7.55040705e-01
5.19716859e-01 9.83102679e-01 7.37922907e-01 -5.19969940e-01
-1.08909082e+00 -9.02861893e-01 1.86613774e+00 -7.12939739e-01
1.11172438e+00 -3.36720377e-01 -8.57497454e-01 8.74053836e-01
3.84486735e-01 -3.12278450e-01 9.27397192e-01 5.02652824e-01
-6.16910517e-01 2.74268538e-01 -1.03007364e+00 5.63983202e-01
1.24396968e+00 -7.69065917e-01 -1.30706561e+00 -6.57011941e-02
7.77511418e-01 2.10954528e-02 -9.48662519e-01 1.74574211e-01
5.73165238e-01 -8.74697208e-01 4.83557492e-01 -8.88588190e-01
3.25827181e-01 -1.96388122e-02 -5.89600265e-01 -1.31338990e+00
-4.67850566e-01 -5.81773400e-01 8.34007502e-01 1.50025558e+00
7.79120028e-01 -8.75952005e-01 3.75012696e-01 6.58241928e-01
-2.87039101e-01 -4.34657693e-01 -7.73380041e-01 -8.28729212e-01
6.89504206e-01 -9.07093823e-01 5.01533091e-01 1.15090013e+00
5.83504260e-01 8.63731503e-01 4.06895578e-01 -1.09876551e-01
2.31729850e-01 1.49361297e-01 2.32059792e-01 -1.61867213e+00
-3.61104339e-01 -5.69102585e-01 -5.68232894e-01 -6.50527120e-01
6.14250600e-01 -1.33726907e+00 -1.47270009e-01 -1.26922309e+00
9.16874409e-02 -8.45414460e-01 -1.02632016e-01 7.11896360e-01
-8.02978203e-02 5.55648148e-01 2.04725012e-01 1.46997228e-01
-2.54718632e-01 6.06439151e-02 6.97331607e-01 4.87436265e-01
-2.87335336e-01 -2.80862093e-01 -1.22864902e+00 1.20206475e+00
8.28357041e-01 -5.41968644e-01 1.43364742e-01 -5.68658710e-01
7.80788183e-01 -1.09103881e-02 7.88613781e-02 -1.65446654e-01
-2.19412863e-01 -4.69094276e-01 3.07837784e-01 8.85114670e-02
5.82798012e-02 -7.94550955e-01 -1.21341676e-01 2.78104752e-01
-1.71278879e-01 3.95711154e-01 2.53414363e-01 -1.78081587e-01
-2.43806243e-01 -5.12149572e-01 6.40910208e-01 -4.05873448e-01
-7.18818128e-01 -1.16838641e-01 -5.08951008e-01 4.99213338e-01
5.90166509e-01 -4.57863390e-01 -9.28510502e-02 -4.70136069e-02
-7.18381822e-01 -3.56001854e-01 8.78540039e-01 2.95685351e-01
2.21758902e-01 -1.20540154e+00 -7.73152292e-01 1.35833964e-01
1.83847949e-01 -6.40706182e-01 -5.52194178e-01 7.91106045e-01
-3.78209114e-01 4.22833025e-01 -4.99079376e-02 -1.05222594e-02
-1.09814095e+00 4.96288151e-01 8.97144526e-02 -2.14618757e-01
-5.49690247e-01 6.18849337e-01 2.61436045e-01 -6.66722417e-01
-5.68163335e-01 -2.84553587e-01 -1.01591818e-01 8.28112885e-02
2.06242472e-01 -3.19260880e-02 9.22657102e-02 -1.25858176e+00
-5.82980037e-01 3.55330229e-01 2.15172991e-01 -6.73676908e-01
1.25922430e+00 -4.24161881e-01 -4.68878329e-01 5.22347391e-01
1.00046885e+00 1.11278427e+00 -4.81860131e-01 -5.09531982e-02
6.65666997e-01 -2.44400814e-01 -2.98213214e-01 -8.75008285e-01
-3.15098852e-01 8.73941898e-01 -2.66800016e-01 2.11748183e-01
8.47997665e-01 3.89474422e-01 7.97622144e-01 9.86762345e-02
4.72419411e-01 -1.20151436e+00 -6.61549211e-01 6.57625914e-01
5.76812744e-01 -8.02835464e-01 -5.17422616e-01 -5.29105425e-01
-7.24074125e-01 8.96353960e-01 -1.24537144e-02 1.96957719e-02
3.08723599e-01 6.07519984e-01 2.48990729e-01 -4.02315527e-01
-6.59553945e-01 -7.05899060e-01 1.08339183e-01 5.22621036e-01
1.03551924e+00 4.78062242e-01 -1.18445289e+00 7.75699437e-01
-7.92961419e-01 -6.51807666e-01 3.46836835e-01 8.09273899e-01
-6.51900649e-01 -1.60692012e+00 -2.01700985e-01 1.56656340e-01
-1.09350395e+00 -5.89279056e-01 -7.08286226e-01 1.14787710e+00
3.30808818e-01 1.01531601e+00 1.53802082e-01 6.93965703e-02
1.96822703e-01 4.86295581e-01 7.08671749e-01 -1.00102365e+00
-7.81652272e-01 3.72851670e-01 6.50033534e-01 -2.08469912e-01
-3.21006775e-01 -8.61907661e-01 -1.71789384e+00 -2.96783924e-01
-1.98161170e-01 2.56255716e-01 5.02388954e-01 1.06732452e+00
-2.60831863e-01 -1.64884046e-01 2.85571426e-01 -3.41011882e-01
-6.61796451e-01 -8.42858970e-01 -8.77126217e-01 5.75718880e-01
-2.40391716e-01 -1.68692857e-01 -3.84673357e-01 3.16365138e-02] | [10.428389549255371, 9.90538501739502] |
65572ec8-bbac-439c-a9b6-ac8fc1e1e8fd | hierarchical-feature-alignment-network-for | 2207.08485 | null | https://arxiv.org/abs/2207.08485v2 | https://arxiv.org/pdf/2207.08485v2.pdf | Hierarchical Feature Alignment Network for Unsupervised Video Object Segmentation | Optical flow is an easily conceived and precious cue for advancing unsupervised video object segmentation (UVOS). Most of the previous methods directly extract and fuse the motion and appearance features for segmenting target objects in the UVOS setting. However, optical flow is intrinsically an instantaneous velocity of all pixels among consecutive frames, thus making the motion features not aligned well with the primary objects among the corresponding frames. To solve the above challenge, we propose a concise, practical, and efficient architecture for appearance and motion feature alignment, dubbed hierarchical feature alignment network (HFAN). Specifically, the key merits in HFAN are the sequential Feature AlignMent (FAM) module and the Feature AdaptaTion (FAT) module, which are leveraged for processing the appearance and motion features hierarchically. FAM is capable of aligning both appearance and motion features with the primary object semantic representations, respectively. Further, FAT is explicitly designed for the adaptive fusion of appearance and motion features to achieve a desirable trade-off between cross-modal features. Extensive experiments demonstrate the effectiveness of the proposed HFAN, which reaches a new state-of-the-art performance on DAVIS-16, achieving 88.7 $\mathcal{J}\&\mathcal{F}$ Mean, i.e., a relative improvement of 3.5% over the best published result. | ['Jinhui Tang', 'Zhenmin Tang', 'Fumin Shen', 'Guo-Sen Xie', 'Yazhou Yao', 'Gensheng Pei'] | 2022-07-18 | null | null | null | null | ['video-salient-object-detection', 'unsupervised-video-object-segmentation'] | ['computer-vision', 'computer-vision'] | [ 1.83190063e-01 -4.07122850e-01 -4.11093533e-01 -3.80724341e-01
-3.22080344e-01 -3.83997232e-01 4.10527349e-01 -2.18671620e-01
-3.77651185e-01 3.31204683e-01 -7.49773160e-03 1.81161121e-01
-1.84444606e-01 -5.07135391e-01 -5.52212894e-01 -8.71118128e-01
1.49032876e-01 -1.88008267e-02 4.27641571e-01 5.79962432e-02
1.98394164e-01 4.53122199e-01 -1.75484896e+00 -1.41487837e-01
1.07165098e+00 1.37310708e+00 2.95164555e-01 5.47805905e-01
-2.10272416e-01 6.59273803e-01 -1.70421883e-01 -2.14082122e-01
2.77539134e-01 -3.85219753e-01 -9.24093246e-01 5.13189793e-01
7.62059510e-01 -4.86043155e-01 -5.79106927e-01 1.07486725e+00
1.23149000e-01 3.61028373e-01 6.76098049e-01 -1.50297236e+00
-5.86600065e-01 1.60862029e-01 -7.36586452e-01 5.32874227e-01
1.10742129e-01 4.01883125e-01 1.14369559e+00 -8.85062337e-01
5.06626606e-01 1.09915423e+00 2.07320854e-01 4.74324495e-01
-9.18720245e-01 -5.16887546e-01 4.31883365e-01 4.77085829e-01
-1.31659913e+00 -5.21683037e-01 7.37515748e-01 -7.18155563e-01
6.44903660e-01 2.40923181e-01 7.99846411e-01 6.86106920e-01
1.46854864e-02 1.09631670e+00 6.85440063e-01 -1.16142944e-01
3.98122845e-03 -1.80815607e-01 2.54185677e-01 8.38427484e-01
3.20199817e-01 5.04956208e-02 -6.17141366e-01 2.86524773e-01
1.04488659e+00 1.77417040e-01 -3.08271796e-01 -3.65040511e-01
-1.20415902e+00 5.81282556e-01 4.77448970e-01 1.44501597e-01
-3.55171651e-01 3.06580275e-01 2.23696798e-01 -1.17601097e-01
2.34006628e-01 4.83722836e-02 -2.88857400e-01 -1.39476717e-01
-9.40932989e-01 9.21146348e-02 1.96808279e-01 1.07358015e+00
9.15373325e-01 2.11197197e-01 -4.30102825e-01 7.52007008e-01
5.43600142e-01 5.10216236e-01 3.62707227e-01 -1.22402668e+00
3.22559446e-01 7.20720291e-01 1.55185089e-01 -1.22250295e+00
-4.10169214e-01 -3.17852825e-01 -6.72134042e-01 -1.47504598e-01
4.72880036e-01 -5.56983380e-03 -9.39890683e-01 1.84076500e+00
6.57501280e-01 4.46720511e-01 2.71403082e-02 1.18449354e+00
8.43843460e-01 7.25171149e-01 2.66005278e-01 -3.51726592e-01
1.27335644e+00 -1.24157763e+00 -6.89578235e-01 -4.20067668e-01
2.64514536e-01 -8.36597443e-01 7.78784752e-01 -7.69956037e-02
-1.17205048e+00 -9.24136579e-01 -1.05805206e+00 -1.17891058e-01
1.44322962e-01 7.24896938e-02 7.72591114e-01 2.88750410e-01
-7.66729236e-01 4.88527417e-01 -9.15539861e-01 -2.11925387e-01
6.82129681e-01 5.22441745e-01 -2.02594981e-01 9.44174826e-03
-9.20566916e-01 2.17634320e-01 4.13900822e-01 4.05790061e-01
-7.44417250e-01 -6.06088817e-01 -7.79218197e-01 -5.18371060e-04
3.91684592e-01 -8.61630619e-01 9.84513342e-01 -1.15444767e+00
-1.54477108e+00 5.79603970e-01 -4.69784498e-01 -1.52891934e-01
4.47481662e-01 -4.13589925e-01 -4.46714699e-01 5.38784623e-01
8.16113502e-02 1.00414228e+00 1.11457419e+00 -1.16810560e+00
-1.06043887e+00 -2.79595762e-01 5.87321445e-02 3.03313881e-01
-6.70344174e-01 -8.15199018e-02 -8.93353343e-01 -6.98787332e-01
2.63906389e-01 -8.68016243e-01 -1.22704633e-01 2.51101285e-01
-1.62438154e-01 -3.58916491e-01 9.12840366e-01 -4.74039406e-01
1.28952658e+00 -2.14535093e+00 3.19550693e-01 1.63893346e-02
2.67745197e-01 5.16264081e-01 -1.50605217e-01 -1.17000803e-01
1.95326090e-01 -1.12839125e-01 -1.85966715e-01 -1.36514574e-01
-1.74300388e-01 1.44194484e-01 5.67145199e-02 7.09487438e-01
3.67432028e-01 1.14537215e+00 -1.03801036e+00 -6.93357766e-01
4.04545456e-01 3.77148390e-01 -6.22737885e-01 2.42917478e-01
-1.51247874e-01 6.71446085e-01 -8.15501809e-01 6.59042120e-01
6.14415467e-01 -3.83116782e-01 -1.05239809e-01 -4.57807392e-01
-5.56000620e-02 -1.36236832e-01 -1.31510317e+00 1.71477282e+00
6.47505745e-02 5.92998266e-01 -1.00319823e-02 -8.11537266e-01
8.88143897e-01 9.94901806e-02 1.07226789e+00 -6.04436755e-01
3.55715454e-01 1.45946279e-01 -2.12795958e-02 -6.64241195e-01
5.26166379e-01 2.69821435e-01 1.61366180e-01 2.03148112e-01
3.28457057e-01 1.31383359e-01 3.75455976e-01 6.60899095e-03
7.22564101e-01 3.00300241e-01 1.25460058e-01 -3.21971208e-01
9.52636719e-01 -2.74711043e-01 9.63133216e-01 5.04013717e-01
-6.42554283e-01 5.90590656e-01 9.50930119e-02 -3.73228729e-01
-8.52592230e-01 -1.09436059e+00 -6.13938235e-02 9.83547449e-01
8.69895160e-01 -3.09665620e-01 -7.74073303e-01 -8.31942677e-01
2.69641280e-02 2.18408614e-01 -3.57133806e-01 -2.50087231e-01
-6.73448026e-01 -4.97310758e-01 6.52847886e-02 7.51732111e-01
7.63534665e-01 -9.66485441e-01 -6.43764138e-01 1.80843294e-01
-3.95442814e-01 -1.51394498e+00 -8.63029718e-01 -2.70253360e-01
-8.04750264e-01 -9.08356130e-01 -6.13108158e-01 -7.65834749e-01
6.98187888e-01 8.02360415e-01 6.14250839e-01 2.90339738e-01
-2.10824475e-01 4.66265798e-01 -4.37000722e-01 -6.53412752e-03
3.98114137e-02 7.29270056e-02 1.49971068e-01 5.80400050e-01
2.50844836e-01 -4.86301988e-01 -1.08896530e+00 4.93444324e-01
-1.13968098e+00 1.25375122e-01 5.82247257e-01 8.10479462e-01
6.23377621e-01 -3.98321539e-01 4.38118786e-01 -4.27570105e-01
-2.06245407e-01 -2.37019986e-01 -5.97153962e-01 1.75638705e-01
-3.76905710e-01 -9.37724188e-02 4.32103932e-01 -2.87604004e-01
-9.57428634e-01 2.78176814e-01 7.62436586e-03 -6.38801992e-01
-9.41839367e-02 -5.36856754e-03 -3.48308086e-01 -2.03275874e-01
1.59109458e-01 2.63594121e-01 -4.80611809e-02 -1.68766528e-01
5.39216459e-01 6.10300064e-01 7.54562378e-01 -4.94966954e-01
9.74301755e-01 7.06055939e-01 -6.72557652e-02 -8.26171756e-01
-1.01738703e+00 -8.80654097e-01 -9.38208163e-01 -5.39328992e-01
1.03811765e+00 -8.32199574e-01 -8.35175455e-01 7.14833319e-01
-9.58976626e-01 -1.06773347e-01 -2.69683301e-01 5.53836346e-01
-7.24610925e-01 5.65008283e-01 -4.93117601e-01 -6.10108972e-01
-3.81484419e-01 -1.29591012e+00 1.01712215e+00 8.20494235e-01
-1.27695277e-01 -8.61050606e-01 -4.84051883e-01 6.23881042e-01
2.26664796e-01 1.78156734e-01 6.44611597e-01 -2.28489265e-01
-8.15882504e-01 1.23470701e-01 -6.22431874e-01 4.27820504e-01
3.09916884e-01 2.61456341e-01 -9.17236090e-01 -3.58997554e-01
-4.78033721e-02 -1.46096751e-01 8.56018543e-01 5.25284946e-01
1.09035540e+00 -3.91819514e-02 -3.25949430e-01 7.72188187e-01
1.35037279e+00 3.12593848e-01 5.09366632e-01 2.80689418e-01
1.23763883e+00 5.41356266e-01 8.91119838e-01 4.53085452e-01
4.20550793e-01 9.19182479e-01 4.89756644e-01 -2.30455510e-02
-3.16597939e-01 1.58552043e-02 4.20401126e-01 9.42467272e-01
-3.18803228e-02 -9.49269012e-02 -6.25298321e-01 6.32260084e-01
-2.06632423e+00 -8.21647823e-01 -1.64071143e-01 1.93784940e+00
6.91144466e-01 8.99175033e-02 1.89222157e-01 -1.25852555e-01
8.38708282e-01 3.68220568e-01 -6.81507647e-01 5.22952378e-02
-6.22917339e-02 -6.80318475e-02 4.50664282e-01 3.72267008e-01
-1.43352079e+00 1.09816837e+00 5.19039869e+00 8.31902206e-01
-1.09590745e+00 -9.95195471e-03 5.48552752e-01 -7.83364922e-02
-1.22134097e-01 3.59183475e-02 -8.68128955e-01 5.13169765e-01
5.66444278e-01 -1.54254049e-01 2.52449542e-01 6.44965231e-01
3.03648680e-01 -5.24778559e-04 -9.53751504e-01 1.02988791e+00
3.88723798e-03 -1.30894852e+00 9.73479301e-02 -1.27436161e-01
9.54292834e-01 -2.35452995e-01 1.38367772e-01 -1.17855743e-01
-1.18492283e-01 -7.31103659e-01 1.08036947e+00 4.86476660e-01
6.15982831e-01 -6.92317247e-01 4.87610549e-01 -4.16424647e-02
-1.62590897e+00 -2.66220093e-01 -9.04446989e-02 2.07582116e-01
5.15884638e-01 4.43526983e-01 -1.34534955e-01 9.07572269e-01
7.50267446e-01 1.18774152e+00 -4.28538710e-01 1.16310322e+00
1.65741763e-03 4.59503353e-01 -2.37097323e-01 1.94641009e-01
5.81922591e-01 -3.17916989e-01 6.58533156e-01 1.02433395e+00
1.80481419e-01 1.10071242e-01 4.01940793e-01 6.64677799e-01
3.78439203e-02 1.38437688e-01 -9.26649868e-02 -5.15220948e-02
3.76834333e-01 1.45764291e+00 -9.41330314e-01 -3.42726201e-01
-6.17179513e-01 9.78774548e-01 1.78269967e-01 3.78611833e-01
-9.98591900e-01 -1.68196708e-01 9.67699409e-01 -1.21536590e-01
6.36361241e-01 -3.61740500e-01 -1.15951397e-01 -1.15633762e+00
2.17925042e-01 -5.29747546e-01 3.85237128e-01 -4.23601121e-01
-1.18180943e+00 4.56335276e-01 -8.47068653e-02 -1.50998843e+00
2.13802494e-02 -5.19519925e-01 -4.01886493e-01 5.28453708e-01
-1.61285269e+00 -1.17634833e+00 -6.66319668e-01 5.45383811e-01
7.50025570e-01 -5.23082167e-02 2.16530874e-01 4.90489513e-01
-9.21724200e-01 4.73500758e-01 -2.35894565e-02 1.30741715e-01
5.30625522e-01 -9.79126632e-01 1.77649304e-01 1.17685044e+00
2.42529482e-01 2.91447103e-01 4.06456530e-01 -4.90679294e-01
-1.53123951e+00 -1.43222749e+00 4.12664205e-01 -2.02498242e-01
5.62570095e-01 -9.40244794e-02 -9.18146908e-01 4.16592151e-01
6.65185926e-03 3.99291217e-01 4.00708437e-01 -5.60444534e-01
-1.53279424e-01 -2.83397824e-01 -7.09080875e-01 5.53242207e-01
1.32875669e+00 -2.63400972e-01 -2.87036568e-01 1.08982757e-01
7.53807545e-01 -4.52899396e-01 -9.99800980e-01 6.46467865e-01
6.86834455e-01 -8.52600336e-01 1.07408977e+00 -5.45879006e-01
3.98623407e-01 -6.41568780e-01 -2.80472606e-01 -6.93626881e-01
-5.11157334e-01 -8.22068632e-01 -4.00826871e-01 1.50130689e+00
-1.33509748e-02 -3.40350598e-01 8.37352276e-01 3.98233324e-01
-2.83535689e-01 -9.07993376e-01 -8.94826889e-01 -6.44042730e-01
-2.67466247e-01 -4.18862909e-01 3.76610488e-01 7.89984643e-01
-4.68050659e-01 3.03335190e-01 -3.01729500e-01 1.65701374e-01
4.77890700e-01 2.96490222e-01 6.81761563e-01 -1.11675251e+00
-1.54895186e-01 -6.77992880e-01 -6.82887673e-01 -1.53942573e+00
1.75019577e-01 -8.28576148e-01 1.82461306e-01 -1.47216702e+00
2.28853852e-01 -4.54770505e-01 -4.94282544e-01 2.60992020e-01
-5.98733664e-01 2.32626647e-01 4.86997277e-01 3.36078316e-01
-9.10772681e-01 8.26421857e-01 1.54841983e+00 -1.99578658e-01
-3.31248313e-01 -9.32756439e-03 -6.61663771e-01 8.32380056e-01
4.74868000e-01 -1.75792053e-01 -4.44768667e-01 -5.04033864e-01
-4.02869403e-01 -1.55119583e-01 3.63658309e-01 -9.73548651e-01
2.75286824e-01 -4.23857361e-01 4.69605148e-01 -6.46204650e-01
2.91422725e-01 -7.95347750e-01 -2.59009916e-02 3.94690633e-01
-5.81459589e-02 5.33941276e-02 4.36484180e-02 7.00604141e-01
-3.26376140e-01 -1.20981917e-01 9.29365993e-01 2.69558966e-01
-1.28265405e+00 7.50201404e-01 -1.45930484e-01 1.21423803e-01
1.32977378e+00 -4.34457958e-01 -2.74527431e-01 -7.76132941e-02
-3.96846831e-01 4.68150884e-01 2.96083212e-01 6.94972336e-01
6.26340389e-01 -1.37720871e+00 -5.91195107e-01 2.83789456e-01
1.25932306e-01 2.01802611e-01 5.89809954e-01 1.18681419e+00
-5.73970914e-01 3.22186917e-01 -2.94000357e-01 -1.02310896e+00
-9.98337805e-01 3.91599536e-01 2.07884654e-01 8.07536095e-02
-6.32743180e-01 9.11417603e-01 3.69346887e-01 2.04985946e-01
1.16162807e-01 -2.01856345e-01 -3.26785028e-01 1.84851348e-01
4.97247010e-01 5.51496625e-01 -2.33758941e-01 -1.11408246e+00
-3.93545240e-01 9.15531993e-01 -1.96300805e-01 2.22302064e-01
1.20380116e+00 -5.58265924e-01 -9.53303874e-02 2.76406646e-01
1.29708207e+00 -3.18897069e-01 -1.82988882e+00 -4.23997611e-01
-6.38774633e-02 -6.63793802e-01 -4.59814779e-02 -1.49383739e-01
-1.55441999e+00 7.70103455e-01 6.45411670e-01 2.75550410e-02
1.16928482e+00 -1.80937294e-02 1.08816683e+00 -7.95985758e-02
1.55295268e-01 -9.89028156e-01 3.49881679e-01 4.61166561e-01
3.58840376e-01 -1.03445613e+00 3.65424599e-03 -7.15354621e-01
-7.29802310e-01 1.10383248e+00 8.69511902e-01 -7.94205666e-02
4.50486958e-01 -1.36451885e-01 1.96509976e-02 -6.76940158e-02
-3.83883774e-01 -2.77301967e-01 7.03079879e-01 3.23065847e-01
2.02343494e-01 -1.45132571e-01 -2.75505245e-01 4.68438655e-01
1.98154762e-01 -1.37128860e-01 1.06142811e-01 8.26585770e-01
-4.29883242e-01 -8.19142818e-01 -1.12517931e-01 4.40168828e-01
-3.49183768e-01 2.61296451e-01 5.19356411e-03 6.55968368e-01
2.06636101e-01 9.45918381e-01 1.83928356e-01 -3.78701210e-01
2.77910233e-01 -2.02111468e-01 4.82457876e-01 -3.12912315e-01
-2.27366835e-01 2.95883417e-01 -2.65395373e-01 -9.04005826e-01
-8.18692446e-01 -8.17785084e-01 -1.46493125e+00 -1.46777540e-01
-3.12471211e-01 -1.41091943e-01 1.84014484e-01 1.18635452e+00
3.34775627e-01 6.11095786e-01 9.63683844e-01 -1.08652461e+00
-2.27994651e-01 -5.85129142e-01 -3.62164259e-01 6.14027619e-01
3.40528250e-01 -9.88119662e-01 -8.05771351e-02 3.53867769e-01] | [9.236352920532227, -0.22447825968265533] |
5c72773f-ccf3-4f49-aec0-7e3b6956efa4 | acrnet-attention-cube-regression-network-for | 2210.05130 | null | https://arxiv.org/abs/2210.05130v1 | https://arxiv.org/pdf/2210.05130v1.pdf | ACRNet: Attention Cube Regression Network for Multi-view Real-time 3D Human Pose Estimation in Telemedicine | Human pose estimation (HPE) for 3D skeleton reconstruction in telemedicine has long received attention. Although the development of deep learning has made HPE methods in telemedicine simpler and easier to use, addressing low accuracy and high latency remains a big challenge. In this paper, we propose a novel multi-view Attention Cube Regression Network (ACRNet), which regresses the 3D position of joints in real time by aggregating informative attention points on each cube surface. More specially, a cube whose each surface contains uniformly distributed attention points with specific coordinate values is first created to wrap the target from the main view. Then, our network regresses the 3D position of each joint by summing and averaging the coordinates of attention points on each surface after being weighted. To verify our method, we first tested ACRNet on the open-source ITOP dataset; meanwhile, we collected a new multi-view upper body movement dataset (UBM) on the trunk support trainer (TruST) to validate the capability of our model in real rehabilitation scenarios. Experimental results demonstrate the superiority of ACRNet compared with other state-of-the-art methods. We also validate the efficacy of each module in ACRNet. Furthermore, Our work analyzes the performance of ACRNet under the medical monitoring indicator. Because of the high accuracy and running speed, our model is suitable for real-time telemedicine settings. The source code is available at https://github.com/BoceHu/ACRNet | ['Sunil K. Agrawal', 'Xupeng Ai', 'Chenfei Zhu', 'Boce Hu'] | 2022-10-11 | null | null | null | null | ['3d-human-pose-estimation'] | ['computer-vision'] | [-2.62475789e-01 1.18200794e-01 -1.24848597e-01 -1.04643553e-01
-8.28444839e-01 -2.12599356e-02 -1.85840860e-01 -4.02618170e-01
-3.63858223e-01 5.43368638e-01 5.07912934e-01 6.80633113e-02
-2.36548826e-01 -4.41062182e-01 -8.66926193e-01 -6.67825580e-01
-9.19808671e-02 7.18946874e-01 5.99972419e-02 -2.64411539e-01
-2.32908189e-01 3.53805393e-01 -1.26969874e+00 2.05626637e-01
8.87703776e-01 8.90222907e-01 1.63577601e-01 3.10408533e-01
6.81718051e-01 4.51193184e-01 -5.11084199e-01 -1.81804106e-01
8.93170312e-02 -2.71757245e-01 -6.91961110e-01 -1.17646933e-01
2.72660911e-01 -5.07309079e-01 -5.46106696e-01 6.87848508e-01
1.25606585e+00 1.51837528e-01 2.69014925e-01 -1.14011121e+00
-3.59194249e-01 3.51843208e-01 -7.08665550e-01 2.72680938e-01
2.83170968e-01 9.15811658e-02 7.29533195e-01 -9.64483202e-01
4.51795578e-01 9.24469292e-01 8.27910244e-01 6.97302699e-01
-6.80775344e-01 -7.39456654e-01 2.41717428e-01 2.98679620e-01
-1.20935750e+00 -3.44094299e-02 7.12851167e-01 -2.55704343e-01
7.64314175e-01 2.70300597e-01 1.12544155e+00 1.26144505e+00
6.52375460e-01 1.06461620e+00 7.21775472e-01 3.34641337e-02
-5.88661991e-02 -4.59312290e-01 1.79198757e-02 7.35621870e-01
1.36218846e-01 -1.08070523e-01 -4.96040940e-01 1.49175733e-01
1.19654858e+00 3.87241632e-01 -5.75835109e-01 -3.77900541e-01
-1.40233946e+00 5.79842865e-01 8.85084331e-01 1.62844121e-01
-7.54950464e-01 4.16485578e-01 5.03184915e-01 -2.21680045e-01
4.78098899e-01 2.02505589e-02 -4.97682691e-01 -3.39690089e-01
-3.54096621e-01 2.48830929e-01 4.60475147e-01 9.63353634e-01
-1.55615598e-01 -2.14192554e-01 -4.29766357e-01 6.76236272e-01
1.52595267e-01 2.98753083e-01 6.51229560e-01 -7.78859794e-01
7.21683621e-01 6.31681621e-01 -2.18540490e-01 -8.39466870e-01
-9.34853673e-01 -7.41651714e-01 -8.85965705e-01 4.95301373e-02
3.27497065e-01 -3.90645981e-01 -8.02648008e-01 1.56883430e+00
6.57649159e-01 1.04309268e-01 -5.16545475e-01 1.63637877e+00
9.08237457e-01 2.41069362e-01 -9.06931236e-02 1.62901729e-01
1.50921285e+00 -1.10902703e+00 -6.30563140e-01 2.12859213e-02
6.21997833e-01 -2.90122896e-01 1.29011655e+00 5.03133357e-01
-1.02167594e+00 -5.72147131e-01 -6.74017727e-01 -1.56916991e-01
2.15230733e-01 3.77812028e-01 5.92401087e-01 2.96625316e-01
-5.39242685e-01 7.37149715e-01 -1.33671272e+00 -1.64044544e-01
5.18776119e-01 4.37344283e-01 -4.48193371e-01 2.12712139e-02
-1.23367131e+00 7.40035415e-01 -1.17054969e-01 5.24544895e-01
-7.26377904e-01 -8.05105329e-01 -5.78540146e-01 -4.29455265e-02
5.35497606e-01 -1.10908544e+00 1.17533016e+00 -6.66596591e-01
-1.54697585e+00 5.87072432e-01 1.83353499e-01 -8.27315599e-02
8.23250175e-01 -9.35659766e-01 4.11982834e-02 2.18521327e-01
5.32058813e-02 4.48749453e-01 5.54715574e-01 -8.54316950e-01
-3.40886891e-01 -9.42825615e-01 5.53897135e-02 5.45903504e-01
-2.24252284e-01 -3.47494781e-01 -7.11352229e-01 -8.01150441e-01
3.69383842e-01 -1.20870662e+00 -1.16532132e-01 1.80616260e-01
-3.77397656e-01 -2.16447160e-01 3.93365711e-01 -1.03926301e+00
1.21903312e+00 -1.95245767e+00 7.11065650e-01 7.97242150e-02
3.22211176e-01 5.90921100e-03 2.18636647e-01 1.30436271e-01
-3.06179464e-01 -1.53669357e-01 -2.18093791e-03 -1.46007344e-01
-3.68333489e-01 2.45306581e-01 1.74966261e-01 7.25360155e-01
-1.80473581e-01 1.06475770e+00 -7.69828618e-01 -5.93318462e-01
2.99291432e-01 7.67089009e-01 -9.14214671e-01 1.64741814e-01
1.14446059e-01 9.26986396e-01 -7.37897754e-01 6.73502326e-01
2.13974461e-01 -4.12393987e-01 7.14446381e-02 -4.86431301e-01
2.91196823e-01 -3.51811908e-02 -7.99235582e-01 2.17082047e+00
-4.29753423e-01 1.01578586e-01 -1.75694138e-01 -7.74255931e-01
5.47117651e-01 4.59326029e-01 1.13335013e+00 -5.23734331e-01
4.96196210e-01 -5.46836853e-02 4.95750122e-02 -9.98109519e-01
3.01173124e-02 -4.33656760e-02 -2.30529886e-02 2.44695783e-01
-3.50196175e-02 3.63644481e-01 -4.28642720e-01 -4.97100018e-02
9.68695760e-01 5.55715501e-01 1.06409304e-01 3.64534035e-02
1.98225066e-01 -2.02876002e-01 6.28604591e-01 2.52343476e-01
-2.80420452e-01 1.01240838e+00 4.11742508e-01 -5.78229308e-01
-7.89007604e-01 -1.08177590e+00 4.53021191e-02 9.80551302e-01
6.68449625e-02 -4.05789167e-01 -8.52597773e-01 -9.55822289e-01
8.64483193e-02 2.36683622e-01 -8.53846192e-01 -1.84202865e-01
-9.53526258e-01 -6.92824244e-01 3.92924845e-01 1.06589556e+00
4.44190472e-01 -1.19994700e+00 -1.02755797e+00 1.28660470e-01
-8.28196108e-01 -8.19214344e-01 -6.68710887e-01 -2.25873962e-01
-1.19958496e+00 -1.24678075e+00 -1.18597603e+00 -5.72941124e-01
6.64413869e-01 8.07446837e-02 8.43457580e-01 -1.15147028e-02
-3.54673564e-01 4.93147492e-01 -5.05097926e-01 -3.91069174e-01
3.29036474e-01 3.01371783e-01 1.24343596e-01 -1.84302583e-01
-1.83989983e-02 -5.95277607e-01 -9.72664952e-01 4.14713442e-01
-4.30895627e-01 1.02213189e-01 6.55083477e-01 8.63758743e-01
8.24075282e-01 -3.80672276e-01 4.69894618e-01 -6.33779049e-01
4.92367476e-01 -4.06245917e-01 -1.22726463e-01 5.81548214e-02
-2.42980838e-01 -2.78031409e-01 1.90067500e-01 -3.73688877e-01
-6.97609961e-01 4.41869348e-02 -4.54764158e-01 -7.33921349e-01
5.08405343e-02 6.54026985e-01 -2.69369613e-02 2.58189112e-01
3.49670351e-01 -5.77878952e-03 2.94162929e-01 -6.47699296e-01
5.88510670e-02 5.34051538e-01 5.09323597e-01 -4.80758160e-01
2.86907434e-01 4.44505692e-01 -9.71643999e-02 -5.90038300e-01
-8.85867178e-01 -2.89257050e-01 -6.52328372e-01 -6.84445620e-01
9.98595357e-01 -9.71512139e-01 -1.12873721e+00 3.93298239e-01
-1.02443624e+00 -2.89312214e-01 -1.95312545e-01 8.82245481e-01
-7.87979424e-01 4.37958062e-01 -6.66746557e-01 -6.24616325e-01
-8.53817165e-01 -1.08837295e+00 1.27572334e+00 -8.66364986e-02
-3.55269939e-01 -5.91012836e-01 4.59291898e-02 7.99540043e-01
1.83140524e-02 6.28049612e-01 6.43359959e-01 -2.93703049e-01
-1.37619600e-01 -3.59703690e-01 5.04609942e-02 7.57960677e-02
-9.42387357e-02 -4.45492685e-01 -6.08590007e-01 -3.52877736e-01
1.19134009e-01 -2.51811564e-01 5.46491921e-01 7.69986868e-01
1.66715240e+00 -2.09834024e-01 -3.20085585e-01 7.04466164e-01
9.35653210e-01 -2.71401614e-01 7.65562117e-01 4.73347396e-01
1.00206733e+00 5.58923304e-01 8.28544557e-01 5.54766059e-01
5.34100652e-01 8.91873002e-01 7.25486398e-01 -2.86262929e-01
-2.17120517e-02 -2.07187533e-01 1.86157256e-01 1.14209878e+00
-9.16593850e-01 -8.45227297e-03 -8.55868280e-01 2.99351007e-01
-2.01674509e+00 -6.94956422e-01 -1.71681076e-01 2.11856174e+00
5.39210260e-01 7.21395109e-03 3.53665888e-01 5.41247241e-02
5.46517372e-01 -1.96326956e-01 -6.62789762e-01 3.18746597e-01
5.26704013e-01 2.51926959e-01 3.96668851e-01 -5.35133481e-03
-1.02104664e+00 4.52363670e-01 4.94885302e+00 6.04080796e-01
-1.15322673e+00 3.11984688e-01 3.02340448e-01 -5.19361377e-01
1.45510390e-01 -6.14205658e-01 -4.43836778e-01 5.35037220e-01
4.48181748e-01 2.16468051e-01 1.44069850e-01 8.60630155e-01
4.09784496e-01 9.87296849e-02 -9.09777701e-01 9.49124694e-01
9.07482356e-02 -8.22384179e-01 -1.74465716e-01 3.67909968e-02
2.97523856e-01 3.88375819e-01 -6.54791296e-02 2.55508512e-01
-2.41071686e-01 -8.41233313e-01 6.21509910e-01 6.68846607e-01
7.83408165e-01 -6.41197622e-01 9.57325339e-01 4.44100887e-01
-1.02422631e+00 -1.37483105e-01 -5.99735156e-02 3.60889137e-02
3.29653770e-01 2.39358917e-01 -5.80360889e-01 8.82485509e-01
1.12818980e+00 8.30496788e-01 -3.07557493e-01 1.11051810e+00
-4.30835485e-01 4.47519064e-01 -2.86124676e-01 1.37149349e-01
5.07618263e-02 -6.31470233e-02 5.72096050e-01 6.85289681e-01
4.20348644e-01 2.54902035e-01 6.35895804e-02 4.86546993e-01
1.44074976e-01 1.90611839e-01 -2.84578949e-01 4.71274763e-01
1.12397028e-02 1.21324193e+00 -4.95216012e-01 -9.10350233e-02
-2.29345113e-01 9.45230126e-01 2.90965557e-01 3.02733094e-01
-1.46772897e+00 -1.38573676e-01 5.04782021e-01 4.30676281e-01
1.18827038e-01 -7.74480402e-02 -3.14006716e-01 -1.25663137e+00
3.25873584e-01 -8.86536121e-01 6.08556271e-01 -1.04152811e+00
-1.13485992e+00 5.17280400e-01 3.12094148e-02 -1.65581083e+00
4.24093269e-02 -4.61473674e-01 -3.85416687e-01 6.30349576e-01
-8.25546324e-01 -1.05240285e+00 -7.39237607e-01 7.77989864e-01
6.12085283e-01 2.39210695e-01 8.40375900e-01 4.63247240e-01
-7.95682788e-01 6.16117954e-01 -3.18204433e-01 3.01747471e-01
6.48468912e-01 -8.93516600e-01 2.97544479e-01 2.40670204e-01
-1.29063830e-01 6.61608338e-01 4.04112041e-01 -7.65146255e-01
-1.42865074e+00 -9.87135112e-01 4.14593428e-01 -6.56598151e-01
2.39074051e-01 -1.50528356e-01 -7.47776568e-01 1.03407741e+00
-2.25845382e-01 5.99422567e-02 5.27282178e-01 5.64492457e-02
1.63429022e-01 -1.18030272e-01 -9.37618256e-01 5.76112330e-01
1.27873349e+00 1.18806273e-01 -6.02414727e-01 4.66572434e-01
6.38683796e-01 -1.09935689e+00 -1.19958937e+00 6.93894148e-01
1.09824145e+00 -6.36355400e-01 9.71898258e-01 -7.86742866e-01
7.34966159e-01 -1.55879661e-01 1.79431364e-01 -1.29835403e+00
-3.31331342e-01 -3.17564994e-01 -2.02426285e-01 4.97278929e-01
1.18864879e-01 -6.97462440e-01 9.17742610e-01 4.42328542e-01
-4.63060498e-01 -1.39661443e+00 -1.27756560e+00 -3.82370085e-01
-8.55577961e-02 -3.66384834e-01 2.86944598e-01 6.76139474e-01
1.63410485e-01 4.59825009e-01 -6.13839686e-01 1.35668546e-01
5.01691580e-01 6.03365675e-02 8.50030780e-01 -1.12180912e+00
-3.74163985e-01 -5.16892746e-02 -5.35664737e-01 -9.70648885e-01
-2.69925565e-01 -8.51611078e-01 -9.80277061e-02 -1.78749049e+00
1.53928667e-01 -2.39959672e-01 -2.75444359e-01 5.09507596e-01
-2.05153227e-01 2.47519895e-01 1.72658831e-01 1.46638408e-01
-4.20509040e-01 8.24225843e-01 1.85019946e+00 1.45618901e-01
-1.84764728e-01 3.66152138e-01 -5.14724433e-01 7.82141626e-01
8.10137928e-01 -5.42150438e-01 -2.03826889e-01 -5.01615584e-01
-2.91293245e-02 2.78953403e-01 6.32492125e-01 -1.05408788e+00
1.66754946e-01 3.04028749e-01 6.79013073e-01 -9.10675883e-01
4.18030560e-01 -1.03426671e+00 1.68382928e-01 7.53500342e-01
4.83830972e-03 1.80144221e-01 3.54475938e-02 4.07895058e-01
1.22239470e-01 3.54103416e-01 4.13500160e-01 -2.99200058e-01
-3.19720358e-01 6.98762119e-01 -7.50256106e-02 1.35429561e-01
1.05978048e+00 -2.73664713e-01 6.94946498e-02 -3.81012112e-01
-1.08385563e+00 4.96945560e-01 5.24901003e-02 5.15300274e-01
7.99977362e-01 -1.37387538e+00 -6.56113684e-01 1.77066952e-01
7.08043575e-02 3.92153621e-01 7.45129526e-01 1.53269446e+00
-5.60224533e-01 3.09536755e-01 -3.84669632e-01 -8.65872443e-01
-1.27639294e+00 4.41412210e-01 6.07426107e-01 -2.13188827e-01
-1.29624724e+00 5.55738866e-01 2.00066984e-01 -4.94220644e-01
4.47034270e-01 -4.90257204e-01 -3.02565604e-01 -1.91831708e-01
1.44009173e-01 6.64595425e-01 1.06330670e-01 -4.88372207e-01
-4.74703550e-01 8.93298805e-01 1.57352626e-01 8.33670795e-02
1.58207810e+00 3.26227732e-02 1.15517154e-01 3.32870454e-01
9.34921563e-01 -2.75822431e-01 -1.21700680e+00 1.24784581e-01
-5.00584543e-01 -3.44819874e-01 3.75677906e-02 -6.70996785e-01
-1.57726216e+00 9.59738731e-01 6.84453487e-01 -4.93216515e-01
1.13415277e+00 -5.45998327e-02 9.64156806e-01 7.63959885e-02
5.09393632e-01 -9.60663259e-01 3.25484186e-01 1.61595449e-01
1.34796238e+00 -1.05252194e+00 1.62962750e-01 -3.95408988e-01
-9.26174343e-01 9.81758296e-01 9.45630312e-01 -2.80868351e-01
7.30720162e-01 6.44256920e-02 1.74292237e-01 -5.15880942e-01
-3.43432903e-01 1.19836167e-01 3.61691773e-01 3.73056084e-01
7.27293789e-01 6.79552108e-02 -5.65261066e-01 9.68364358e-01
-2.44002283e-01 3.68425936e-01 1.13449298e-01 8.95774305e-01
-6.65217787e-02 -5.83220899e-01 -3.24759841e-01 4.11085576e-01
-7.08324254e-01 3.08784366e-01 -7.89102986e-02 1.06840205e+00
-6.24040253e-02 4.65625852e-01 -8.49631131e-02 -5.82970977e-01
8.98124218e-01 -1.04722999e-01 6.90024793e-01 -4.91943866e-01
-8.05821419e-01 2.64903784e-01 -1.22344285e-01 -9.10931110e-01
-2.54718363e-01 -6.28117502e-01 -1.50813031e+00 -1.00432351e-01
-2.63856888e-01 -1.16594553e-01 4.75086540e-01 7.18526959e-01
4.07738805e-01 1.14001381e+00 3.42219830e-01 -1.07612598e+00
-7.22203314e-01 -1.21212924e+00 -4.65113401e-01 4.13254321e-01
-4.74547688e-03 -9.61100221e-01 5.98835722e-02 -2.09542930e-01] | [7.1663055419921875, -0.552754819393158] |
29de8135-4dd5-46c5-8792-b627be193060 | direct-and-indirect-transactions-and | 1911.11569 | null | https://arxiv.org/abs/1911.11569v5 | https://arxiv.org/pdf/1911.11569v5.pdf | Direct and indirect transactions and requirements | The indirect transactions between sectors of an economic system has been a long-standing open problem. There have been numerous attempts to conceptually define and mathematically formulate this notion in various other scientific fields in literature as well. The existing direct and indirect effects formulations, however, can neither determine the direct and indirect transactions separately nor quantify these transactions between two individual sectors of interest in a multisectoral economic system. The novel concepts of the direct, indirect and transfer (total) transactions between any two sectors are introduced, and the corresponding requirements matrices and coefficients are systematically formulated relative to both final demands and gross outputs based on the system decomposition theory in the present manuscript. It is demonstrated theoretically and through illustrative examples that the proposed requirements matrices accurately define and correctly quantify the corresponding direct, indirect, and total interactions and relationships. The proposed requirements matrices for the US economy using aggregated input-output tables for multiple years are then presented and briefly analyzed. | ['Huseyin Coskun', 'Husna Betul Coskun'] | 2019-11-23 | null | null | null | null | ['novel-concepts'] | ['reasoning'] | [ 1.27638653e-01 -3.39887813e-02 -3.41094434e-01 2.02889666e-01
9.80530307e-02 -8.70912492e-01 1.08056986e+00 -5.83297536e-02
-6.31921813e-02 1.04160190e+00 1.16778053e-01 -4.09979522e-01
-1.12971103e+00 -7.91532040e-01 -2.59703070e-01 -9.42945242e-01
-3.85371596e-01 2.10922867e-01 -2.58920759e-01 -7.54563391e-01
3.16938281e-01 3.58767390e-01 -1.43467081e+00 -2.14638606e-01
9.40689325e-01 1.18400717e+00 1.81251913e-01 9.54582095e-02
1.56981692e-01 4.07840014e-01 -4.07480627e-01 -2.04200223e-01
4.75180089e-01 -5.22320092e-01 -4.77845222e-01 1.60498336e-01
-9.38107848e-01 -2.05494255e-01 4.01592165e-01 7.44472861e-01
7.87329748e-02 4.26671989e-02 1.03320956e+00 -1.46465564e+00
-4.41322744e-01 6.99628890e-01 -6.95092916e-01 3.27581726e-02
3.84272859e-02 1.77498728e-01 1.01102495e+00 -1.09500968e+00
2.37911269e-01 1.31082952e+00 4.13944572e-01 -3.45821559e-01
-1.40877688e+00 -6.61285341e-01 2.03236774e-01 -2.63292551e-01
-1.11041081e+00 -4.03072864e-01 6.00531697e-01 -5.65972507e-01
1.55204487e+00 6.59087956e-01 1.04134560e+00 1.14477053e-01
7.64950275e-01 1.70588493e-01 1.33287203e+00 -5.11629999e-01
-1.60309538e-01 1.86834812e-01 9.02310535e-02 -3.26416433e-01
9.37439322e-01 4.08376664e-01 2.76478343e-02 -1.68770179e-02
8.86449039e-01 -1.20247953e-01 -2.51747370e-01 -4.95301127e-01
-1.20377398e+00 5.07858098e-01 1.33446250e-02 6.48469865e-01
-6.74878001e-01 -4.62626368e-02 2.46652603e-01 3.37867826e-01
3.37363005e-01 2.80769825e-01 -6.68944299e-01 4.07392919e-01
-5.68634629e-01 1.24559879e-01 9.38155293e-01 8.01872909e-01
6.23277485e-01 1.49375960e-01 3.93634170e-01 4.86513317e-01
4.10290182e-01 1.11738133e+00 -3.09533685e-01 -9.49216783e-01
6.97607696e-01 6.55312061e-01 7.63428211e-01 -1.58482993e+00
-4.22481865e-01 -6.94601655e-01 -7.90791988e-01 2.07219645e-01
4.55342680e-01 -5.15761256e-01 -2.17600353e-02 1.50798786e+00
1.09858938e-01 -5.15864015e-01 4.24629211e-01 5.77549994e-01
8.83966759e-02 9.67986405e-01 2.53437549e-01 -1.16103268e+00
1.17543113e+00 -3.58346522e-01 -1.26740396e+00 7.15391338e-02
3.81507158e-01 -9.33826923e-01 3.31387758e-01 9.71751213e-02
-1.38371086e+00 3.57928537e-02 -5.34149051e-01 6.34801090e-01
-7.06427693e-01 2.39636615e-01 5.42497575e-01 5.08039296e-01
-9.11974132e-01 1.29692644e-01 -7.32139766e-01 -2.09718332e-01
-3.72200847e-01 6.25829518e-01 9.49270651e-02 7.65005350e-01
-1.50117493e+00 1.29889905e+00 7.44977742e-02 6.17468357e-01
-2.23719701e-01 -5.12394011e-01 -5.58595061e-01 4.31459308e-01
3.80918503e-01 -5.92749476e-01 1.02274716e+00 -8.48363519e-01
-1.28974056e+00 1.90324351e-01 3.89543116e-01 -9.59780905e-03
2.98623949e-01 2.85242379e-01 -5.15840828e-01 -1.55158401e-01
8.35151374e-02 -1.97027847e-01 -9.10209045e-02 -1.60444641e+00
-9.48377013e-01 -3.91315877e-01 2.18166709e-01 6.74922884e-01
-2.96636641e-01 1.59846038e-01 5.00764906e-01 -7.35716641e-01
7.99960867e-02 -7.26041853e-01 -3.24995965e-01 -5.43913126e-01
-1.76580593e-01 -4.59776856e-02 6.60050094e-01 -6.62593842e-01
1.46917570e+00 -1.62962735e+00 2.27271661e-01 4.62254345e-01
-4.45656270e-01 -6.30337819e-02 2.83019245e-01 1.16993749e+00
-3.58883232e-01 1.40441269e-01 -5.08580148e-01 3.59435737e-01
3.05303961e-01 3.46075684e-01 -2.44650930e-01 4.45737720e-01
-1.02677561e-01 5.23007989e-01 -8.53337646e-01 -1.07150219e-01
5.28768122e-01 9.00330469e-02 -2.50508487e-01 -3.39486063e-01
4.81502861e-01 1.24580048e-01 -6.17703676e-01 5.68171382e-01
9.23640192e-01 1.93044931e-01 8.81431580e-01 -4.74573791e-01
-8.34725261e-01 9.57273990e-02 -1.44532788e+00 7.32885301e-01
-6.21289492e-01 9.32833850e-02 6.86033368e-01 -1.56442976e+00
6.38480008e-01 5.86074889e-01 1.08053970e+00 -7.07696319e-01
1.51840165e-01 6.94607913e-01 1.98491678e-01 -3.53265673e-01
5.54341137e-01 -6.35248125e-01 -5.51718771e-01 5.59318960e-01
-5.50556462e-03 -1.93454009e-02 4.11194354e-01 -2.05319196e-01
1.95386991e-01 4.45773304e-02 9.46118712e-01 -1.16013980e+00
7.26738036e-01 1.60904169e-01 6.97218776e-01 -1.68690458e-01
-5.03681749e-02 -6.24792695e-01 6.21664524e-01 -1.24257961e-02
-6.73031747e-01 -7.59817898e-01 -2.51654625e-01 6.68927312e-01
6.52337015e-01 2.68344104e-01 -4.79197174e-01 5.77573525e-03
4.00038928e-01 6.45580053e-01 -4.10869509e-01 2.04687163e-01
-2.14487135e-01 -1.17296541e+00 1.11903526e-01 5.65046212e-04
4.58175242e-01 -4.81348693e-01 -9.80386078e-01 3.54869545e-01
-4.16941673e-01 -6.99792087e-01 -3.85865718e-02 2.11677596e-01
-8.64047527e-01 -1.27180994e+00 -7.25381911e-01 -4.65808243e-01
6.97984636e-01 4.45115298e-01 7.44229376e-01 -3.13829035e-01
3.11450928e-01 1.64963916e-01 7.27607235e-02 -4.10477549e-01
-1.53295130e-01 -3.96857768e-01 6.78592265e-01 3.32773551e-02
-5.22157252e-02 -5.68174839e-01 -5.73797166e-01 9.04225051e-01
-7.69771397e-01 6.73503727e-02 5.43970048e-01 6.96136415e-01
3.60333890e-01 4.03279334e-01 1.07261491e+00 -2.44035438e-01
9.13873613e-01 -6.57109618e-01 -9.28095281e-01 3.40391278e-01
-9.97794867e-01 -2.63328284e-01 2.33612150e-01 1.44216657e-01
-1.65848172e+00 -9.73559394e-02 4.99318570e-01 3.53990555e-01
1.70321956e-01 1.00273681e+00 -3.80056798e-01 -7.23766536e-02
-8.78730118e-02 -4.20668442e-03 1.21372454e-01 -4.01841134e-01
1.76565841e-01 4.70376104e-01 2.28081957e-01 -6.66830540e-01
8.47089589e-01 2.38778815e-01 3.94375086e-01 -4.01321381e-01
1.80130258e-01 -3.67276162e-01 -5.41042864e-01 -6.02482855e-01
2.44455054e-01 -9.45234299e-01 -9.70855892e-01 5.18287718e-01
-9.52370703e-01 -1.58995062e-01 -4.17330325e-01 7.12615728e-01
-7.56878078e-01 1.20765977e-01 -2.06039518e-01 -1.49861300e+00
4.89013307e-02 -1.25122714e+00 4.58030134e-01 -1.03321657e-01
1.09076627e-01 -1.35465360e+00 -1.15746044e-01 -7.81682357e-02
6.01590395e-01 5.66687882e-01 6.01651907e-01 2.44573712e-01
-5.26553690e-01 3.37107658e-01 -1.53831527e-01 1.98390543e-01
7.31741548e-01 2.35504314e-01 -2.52897054e-01 -2.21053705e-01
6.12995103e-02 5.25734723e-01 1.90390542e-01 7.63505876e-01
2.17434447e-02 -8.39424491e-01 -5.55493176e-01 6.52464256e-02
2.00712514e+00 7.86036611e-01 6.52749538e-01 4.86202747e-01
-2.45982543e-01 1.25979519e+00 1.08720207e+00 7.03833461e-01
2.06304595e-01 5.19010782e-01 6.11432791e-01 -3.31292748e-01
5.56434393e-01 4.24438447e-01 4.38369244e-01 1.04101479e+00
-8.65437686e-01 -4.31821406e-01 -7.82604218e-01 8.08318317e-01
-1.92354512e+00 -1.12966573e+00 -5.69909275e-01 2.00878167e+00
5.19170463e-01 -2.17442468e-01 1.62791595e-01 3.21311891e-01
7.21982718e-01 -1.81654602e-01 -1.36205822e-01 -6.96479619e-01
-3.68839025e-01 -1.30069315e-01 1.23854661e+00 4.29191500e-01
-6.52281523e-01 2.75200069e-01 7.64176989e+00 5.53342402e-01
-6.02014601e-01 -1.61150888e-01 7.68755794e-01 1.90165684e-01
-4.63275939e-01 4.53873545e-01 -4.35651213e-01 2.15853542e-01
9.42395747e-01 -9.02256846e-01 3.35109532e-01 3.79779041e-02
1.14189494e+00 -6.06485605e-01 -3.92746627e-01 2.85381019e-01
-8.23562622e-01 -9.08375442e-01 1.28118604e-01 7.64693201e-01
9.83241260e-01 -6.12206757e-01 2.10206017e-01 -1.92049563e-01
3.54005486e-01 -3.62495512e-01 7.13094532e-01 8.00515294e-01
8.51289570e-01 -9.72831249e-01 7.06483066e-01 5.07972777e-01
-1.75147045e+00 -6.48089528e-01 3.53398733e-02 -7.19312251e-01
6.13039672e-01 5.84005356e-01 8.38615373e-02 1.20951128e+00
5.54913640e-01 5.23747802e-01 3.50179106e-01 4.68918085e-01
4.24981505e-01 2.39058182e-01 -4.65784401e-01 8.27116892e-02
2.77825147e-01 -9.87707078e-01 4.70762819e-01 8.41215491e-01
7.83927500e-01 5.91262400e-01 -4.64435965e-01 9.74257290e-01
5.12424052e-01 3.50323886e-01 -7.90370941e-01 -1.23956367e-01
3.26492131e-01 1.03412461e+00 -4.97921914e-01 -2.11041614e-01
-5.91326118e-01 -6.37675971e-02 -8.21010947e-01 6.44060850e-01
-8.24745834e-01 -4.65816826e-01 1.08221245e+00 1.67102858e-01
-2.15808913e-01 -2.76033700e-01 -6.23632908e-01 -9.35538888e-01
-1.07035495e-01 -4.93403763e-01 1.16597131e-01 -3.89015526e-01
-8.30009460e-01 -2.45474622e-01 7.92692006e-01 -1.77057648e+00
-2.95926303e-01 -5.31655669e-01 -1.60567373e-01 1.31595194e+00
-1.58949077e+00 -8.81201208e-01 3.82354349e-01 4.35706317e-01
2.46648192e-01 -2.91683584e-01 7.09724426e-01 2.81528711e-01
-7.43699431e-01 -1.71362311e-02 7.95905173e-01 -5.01073778e-01
1.01984635e-01 -8.24385822e-01 -2.51574337e-01 9.62234974e-01
-1.08294201e+00 7.38364160e-01 9.12918448e-01 -5.86511433e-01
-1.41624522e+00 -5.42997777e-01 1.12142289e+00 4.33078855e-02
1.09995139e+00 2.32681856e-01 -1.50927559e-01 5.95472932e-01
9.53605294e-01 -4.83849496e-01 3.91648799e-01 -3.88521850e-01
5.46684086e-01 -4.26418096e-01 -1.18426204e+00 4.12626505e-01
9.99011219e-01 -4.51085269e-02 -4.95388031e-01 6.89283311e-02
1.83538496e-01 1.85638815e-01 -1.30251622e+00 4.90386546e-01
8.50910544e-01 -7.57032156e-01 1.21568441e+00 -1.69704318e-01
1.85172662e-01 -2.01926902e-01 -3.05988699e-01 -1.51180863e+00
-3.90833318e-01 -5.47379971e-01 2.66470402e-01 1.26057732e+00
5.21314621e-01 -1.34621441e+00 -3.06885064e-01 6.01019144e-01
-1.30745515e-01 -4.51687157e-01 -7.75813043e-01 -1.17969882e+00
3.11539948e-01 2.00865075e-01 9.17775214e-01 1.28188860e+00
9.09113705e-01 -1.93560135e-03 -2.63377488e-01 1.88723817e-01
7.59050131e-01 3.24979782e-01 4.54721123e-01 -1.18507779e+00
7.76129887e-02 -7.34481037e-01 1.96828201e-01 -6.59255803e-01
-4.66271162e-01 -3.93942535e-01 -3.70228648e-01 -1.83577132e+00
-4.57440875e-02 -7.71523535e-01 -4.97749418e-01 3.18040550e-01
5.98145545e-01 -1.00045018e-01 2.29957148e-01 5.48435509e-01
6.07435048e-01 6.44289613e-01 1.55408049e+00 -3.33111472e-02
-1.95586041e-01 8.70357156e-02 -8.72668862e-01 5.77887893e-01
6.47636950e-01 -9.79330465e-02 -6.54998183e-01 -2.59952128e-01
4.05773401e-01 7.69173503e-01 1.96428016e-01 -5.47693670e-01
-1.74700856e-01 -1.35737002e+00 -3.96731287e-01 -6.93166375e-01
1.68482825e-01 -1.42707753e+00 1.14641559e+00 7.32832849e-01
9.78227779e-02 3.64207715e-01 2.02741474e-01 4.83758926e-01
-3.17343891e-01 3.33822280e-01 4.03422654e-01 4.50030193e-02
-2.56904572e-01 -2.06650898e-01 -7.16963232e-01 -7.53451645e-01
1.51483822e+00 -4.57407713e-01 -3.61699730e-01 -3.23785484e-01
-9.78205621e-01 4.25425828e-01 1.96582064e-01 -5.94221093e-02
9.10611544e-03 -1.40611696e+00 -4.71057802e-01 -2.09990501e-01
-4.80949849e-01 -5.48152983e-01 1.02212861e-01 1.37681961e+00
-3.28056335e-01 1.13331556e+00 -7.16672897e-01 -2.85108209e-01
-8.63946974e-01 3.11400115e-01 6.46329224e-01 -5.17350674e-01
2.02119231e-01 5.20152077e-02 8.15354705e-01 -3.53404433e-01
-5.33846736e-01 -7.51984894e-01 -4.53384310e-01 5.19209802e-01
-2.88265735e-01 7.10250914e-01 -2.96486884e-01 -1.28605235e+00
-5.21338999e-01 7.63454795e-01 1.07083058e+00 -3.78268272e-01
1.42307591e+00 -9.64046121e-01 -3.27005565e-01 1.70434415e-01
7.37443447e-01 -3.02200973e-01 -9.51807022e-01 7.51775727e-02
-1.66201979e-01 -3.74193758e-01 2.99248900e-02 -8.79361212e-01
-1.32862020e+00 4.19275939e-01 2.31475413e-01 9.36682582e-01
1.66103566e+00 -6.27966404e-01 3.37108403e-01 2.20873743e-01
4.56063181e-01 -1.46936798e+00 -6.59878373e-01 2.54675478e-01
1.27884138e+00 -3.22503537e-01 7.91298375e-02 -9.50475037e-01
-2.51976401e-01 7.94771791e-01 8.64883326e-03 -3.62229981e-02
1.06321537e+00 6.05594277e-01 -4.05560315e-01 4.28470597e-02
-6.50731564e-01 -1.69577882e-01 -2.48392791e-01 3.83067548e-01
5.31435728e-01 5.41448474e-01 -1.49314368e+00 7.81386733e-01
2.48991549e-01 -4.30254526e-02 7.28305340e-01 8.97835910e-01
-2.18007818e-01 -1.09881282e+00 -7.72097707e-01 2.57748306e-01
-3.91394824e-01 2.64131188e-01 -1.02050595e-01 1.34435177e+00
9.55094174e-02 1.13044524e+00 1.65354580e-01 -9.92695466e-02
8.94575894e-01 -3.96555781e-01 3.80323604e-02 1.26555398e-01
-5.60164690e-01 5.02757192e-01 3.30950141e-01 -9.71725583e-02
-1.13884318e+00 -1.23746336e+00 -1.13749659e+00 -5.87038636e-01
-8.17262232e-01 4.13781941e-01 6.70093477e-01 9.63389337e-01
8.49739760e-02 4.75542605e-01 1.07416928e+00 -1.09359396e+00
-6.39047623e-01 -9.46853817e-01 -1.09681439e+00 -7.86711574e-02
1.01789959e-01 -1.18959498e+00 -7.60439098e-01 2.78476387e-01] | [5.627202033996582, 3.501335382461548] |
9a542ab4-d507-4239-8fa8-12f3b6a87128 | simple-vs-oversampling-based-classification | null | null | https://aclanthology.org/2020.wanlp-1.24 | https://aclanthology.org/2020.wanlp-1.24.pdf | Simple vs Oversampling-based Classification Methods for Fine Grained Arabic Dialect Identification in Twitter | In this paper, we present a description of our experiments on country-level Arabic dialect identification. A comparison study between a set of classifiers has been carried out. The best results were achieved using the Linear Support Vector Classification (LSVC) model by applying a Random Over Sampling (ROS) process yielding an F1-score of 18.74% in the post-evaluation phase.In the evaluation phase, our best submitted system has achieved an F1-score of 18.27%, very close to the average F1-score (18.80%) obtained for all the submitted systems. | ['Mourad Abbas', 'Mohamed Lichouri'] | null | null | null | null | coling-wanlp-2020-12 | ['dialect-identification'] | ['natural-language-processing'] | [-2.42805764e-01 -1.16542399e-01 7.00810701e-02 -6.03083491e-01
-7.98000276e-01 -8.17672372e-01 1.12863505e+00 4.33553278e-01
-5.79162121e-01 7.02660263e-01 -1.77103847e-01 -2.81561583e-01
4.44810465e-02 -5.72995484e-01 -3.80344950e-02 -7.67919302e-01
-1.45443991e-01 6.75016761e-01 3.54811698e-01 -8.44612777e-01
6.35652184e-01 4.74187970e-01 -1.78749037e+00 5.89729071e-01
1.23487818e+00 5.87656140e-01 -1.38832316e-01 7.96874106e-01
-5.89912608e-02 4.29697305e-01 -1.16319871e+00 -7.31831431e-01
-6.71557114e-02 -4.69917595e-01 -1.10031867e+00 -1.74882650e-01
7.36769795e-01 2.74947453e-02 1.17271863e-01 1.15673530e+00
4.01303798e-01 -1.48376197e-01 1.03766966e+00 -1.14106286e+00
-3.85762841e-01 8.48151088e-01 -3.41212571e-01 1.02345021e-02
8.11532021e-01 -3.55515391e-01 8.41035903e-01 -1.00882244e+00
6.22000813e-01 1.16925538e+00 7.22377837e-01 3.70484620e-01
-1.24005592e+00 -4.12983447e-01 -1.30971059e-01 1.85606331e-01
-1.67369390e+00 -4.26581413e-01 5.52226067e-01 -4.16025311e-01
8.73432934e-01 4.35237050e-01 4.90555167e-01 5.87214231e-01
-1.61492929e-01 7.37765431e-01 1.76320517e+00 -1.27463770e+00
2.12259799e-01 8.18919480e-01 7.20831215e-01 7.27100611e-01
1.36971340e-01 -3.91921073e-01 -1.54673755e-01 -1.70264199e-01
3.74903172e-01 -8.71776283e-01 1.56135075e-02 3.33842069e-01
-9.07298207e-01 1.03188670e+00 2.19702303e-01 8.67918074e-01
-2.62768418e-01 -8.16395938e-01 5.08923888e-01 7.45123506e-01
2.81959593e-01 2.75947630e-01 -5.70870519e-01 -5.21247797e-02
-1.06165588e+00 4.12737340e-01 1.00919437e+00 6.04706705e-01
1.93468347e-01 3.69377315e-01 -9.22593102e-03 1.28495955e+00
1.66009665e-01 6.19565308e-01 7.64071643e-01 -4.88891095e-01
3.77012521e-01 4.52465534e-01 1.96489543e-01 -6.60302639e-01
-2.97213525e-01 -1.88363031e-01 -4.42806154e-01 3.70347381e-01
1.04217458e+00 -2.35655248e-01 -8.58830690e-01 1.21768832e+00
-1.35827595e-02 -6.42527461e-01 3.10087919e-01 6.46640539e-01
8.80066395e-01 8.65755677e-01 -1.10836945e-01 -3.94417524e-01
1.13292861e+00 -9.06495512e-01 -7.59233057e-01 3.46420288e-01
6.47873878e-01 -1.36685884e+00 1.12734735e+00 9.72261250e-01
-8.49820256e-01 -6.64005637e-01 -1.21700251e+00 6.71209574e-01
-5.91903746e-01 5.06255805e-01 3.78488928e-01 1.61782753e+00
-1.26342309e+00 3.11371654e-01 -3.06991935e-01 -6.49136007e-01
-2.09076032e-01 4.98489887e-01 -6.90539479e-01 -6.27877712e-02
-1.00592923e+00 1.15804732e+00 3.00791413e-01 -4.95497063e-02
-3.12563062e-01 -1.41240647e-02 -5.78928649e-01 -4.36880738e-01
-3.55285078e-01 4.85003948e-01 9.77656662e-01 -1.14180994e+00
-1.69736111e+00 1.19742358e+00 -2.01371908e-01 -2.97132045e-01
5.41191280e-01 -1.34109989e-01 -8.40664208e-01 1.67315137e-02
-5.54807782e-02 4.27301228e-01 5.60420215e-01 -1.42474067e+00
-7.54906237e-01 -2.23717943e-01 -1.61224186e-01 -9.53984931e-02
-8.49480107e-02 6.69009566e-01 1.87513456e-02 -6.75771654e-01
2.93971509e-01 -1.05226374e+00 1.18697390e-01 -1.02069843e+00
-6.84470590e-03 -4.02479649e-01 4.00179118e-01 -1.20752275e+00
1.46961391e+00 -1.78036189e+00 -6.30133525e-02 6.48922324e-01
-5.18548667e-01 7.87500024e-01 1.58481643e-01 6.09805703e-01
-3.61735731e-01 -1.69757992e-01 -3.15331370e-01 -1.41599521e-01
-9.60668847e-02 -9.15196463e-02 -2.57047176e-01 5.38789332e-01
9.91310030e-02 5.28740697e-02 -7.10558236e-01 -5.43196380e-01
1.72342569e-01 3.45228612e-01 -1.66047707e-01 6.31217211e-02
2.45195627e-01 -1.27877995e-01 1.84614196e-01 1.03207016e+00
1.03246450e+00 6.16455436e-01 3.92167509e-01 8.73357877e-02
-3.02301228e-01 8.83883610e-02 -1.50998747e+00 9.69717205e-01
-3.47474515e-01 8.27200770e-01 3.70667763e-02 -8.77566576e-01
1.29384482e+00 3.89136940e-01 -1.06889466e-02 -3.51901889e-01
1.24741517e-01 6.85493231e-01 3.59898210e-01 -1.87765017e-01
5.64369023e-01 1.34934047e-02 -2.64182597e-01 3.92581731e-01
3.97806317e-01 7.16301380e-03 4.10010368e-01 2.17794701e-02
1.59762204e-01 7.18178004e-02 3.37354481e-01 -7.42404640e-01
1.32311022e+00 2.66658276e-01 1.02739222e-03 6.38718367e-01
-5.99210560e-01 5.57852447e-01 4.42366093e-01 -3.62333477e-01
-7.52650678e-01 -1.17899382e+00 -6.01551712e-01 1.09160483e+00
-4.52683687e-01 -3.41434300e-01 -1.33800387e+00 -8.00921082e-01
-2.77933329e-01 7.58390129e-01 -4.42032725e-01 4.13352907e-01
-8.50150764e-01 -1.20223308e+00 8.22909355e-01 1.22888714e-01
3.60544950e-01 -8.87312889e-01 1.79780379e-01 1.38602167e-01
-2.62747616e-01 -7.82420576e-01 1.79308802e-02 -7.81071484e-02
-5.35141289e-01 -8.22814822e-01 -9.10737336e-01 -1.46069288e+00
4.84153390e-01 -1.83173701e-01 1.07676220e+00 1.30626008e-01
1.82205200e-01 -1.70046166e-01 -7.07610726e-01 -4.87757683e-01
-8.53094935e-01 3.66125017e-01 2.16492921e-01 2.00471595e-01
6.72878981e-01 2.97752798e-01 -2.10195240e-02 4.70488191e-01
-2.56561458e-01 -6.56861961e-01 1.19330108e-01 1.05038035e+00
4.80783209e-02 2.27311179e-02 7.16452241e-01 -9.25347447e-01
7.13599026e-01 5.24548441e-02 -6.78332031e-01 5.76150596e-01
-7.99508035e-01 -3.30181032e-01 6.57326579e-01 -4.03690755e-01
-1.00934875e+00 1.76089242e-01 -7.01954782e-01 4.81776237e-01
-4.20030683e-01 1.46747172e-01 -3.91647406e-02 -3.22488874e-01
9.02389050e-01 3.27182800e-01 2.64863610e-01 -3.34946662e-01
3.76178622e-02 1.55005753e+00 2.91781634e-01 -4.37589198e-01
3.44838381e-01 -6.71211481e-02 -4.57473934e-01 -1.41351867e+00
-4.64743465e-01 -5.40428579e-01 -9.58310008e-01 -4.27404821e-01
3.20328414e-01 -8.45411956e-01 -4.55732018e-01 1.24178934e+00
-7.25044072e-01 -1.51231140e-01 3.16519260e-01 5.61484754e-01
-3.12429398e-01 3.25014859e-01 -5.94217181e-01 -1.04109073e+00
-3.66892248e-01 -1.33557713e+00 5.84650993e-01 1.08315676e-01
-5.47428250e-01 -1.01403666e+00 -5.95273413e-02 2.47274444e-01
3.65213484e-01 1.55782714e-01 5.51898718e-01 -8.84486318e-01
4.62205142e-01 -4.35758770e-01 6.29390627e-02 4.60703254e-01
9.28604677e-02 2.93335080e-01 -1.41785276e+00 -5.28270304e-01
-1.81177095e-01 -2.21353769e-01 5.65327048e-01 3.25509489e-01
4.94778097e-01 4.79127765e-02 -1.67966448e-02 -1.13564194e-03
1.48922026e+00 3.43688905e-01 5.66505551e-01 6.47120833e-01
2.66471386e-01 8.98545682e-01 9.15435493e-01 2.15274185e-01
4.40976143e-01 8.75667155e-01 -1.60114512e-01 6.20807335e-02
-1.12320162e-01 1.60321832e-01 5.52382410e-01 7.32819498e-01
-2.63192922e-01 8.24670494e-02 -1.27413881e+00 3.84278029e-01
-1.44543779e+00 -9.71779704e-01 -6.40534580e-01 2.34966445e+00
8.69155765e-01 2.18433529e-01 9.38021958e-01 9.80006635e-01
8.38223457e-01 -1.30917341e-01 4.76872563e-01 -1.19312334e+00
-4.50483739e-01 4.55748409e-01 3.55637550e-01 1.29384708e+00
-1.33854568e+00 1.25308061e+00 7.43565989e+00 9.55285430e-01
-1.14312053e+00 -7.46405497e-02 3.73436302e-01 4.98631001e-01
2.23502353e-01 -2.67155409e-01 -1.03260243e+00 5.10264993e-01
1.30678809e+00 -1.88839853e-01 4.58751261e-01 4.89952385e-01
-3.81618440e-01 -2.77504385e-01 -4.16643322e-01 8.08026075e-01
4.81440276e-01 -7.03639388e-01 2.55751591e-02 -1.21774256e-01
7.26639152e-01 -1.94491729e-01 -1.70249175e-02 3.56483161e-01
1.19197302e-01 -8.61677706e-01 8.52600098e-01 1.00844808e-01
8.88736486e-01 -1.15772521e+00 1.13903856e+00 2.91490436e-01
-8.53333473e-01 -1.24158286e-01 -1.24388501e-01 -8.37378725e-02
-3.10566217e-01 2.79539078e-01 -1.13044035e+00 5.43862581e-01
6.02743387e-01 2.44183570e-01 -6.44652367e-01 8.44121039e-01
-1.91266164e-01 1.10488474e+00 -3.36095870e-01 -4.71139818e-01
3.85298401e-01 -3.93125921e-01 2.76116133e-01 1.77084398e+00
1.42777180e-02 -3.45369965e-01 -1.61788717e-01 -2.02552095e-01
7.06562757e-01 7.57967174e-01 -2.16412410e-01 3.51111174e-01
3.77141863e-01 9.25270319e-01 -8.85703385e-01 -3.51728827e-01
-3.54564130e-01 1.01196194e+00 1.15235552e-01 -6.75954297e-02
-6.28915429e-01 -9.86605883e-01 3.47850680e-01 -1.75160572e-01
2.74650156e-01 -5.68715893e-02 -5.02903342e-01 -9.60992575e-01
5.19061415e-03 -1.08072615e+00 7.33594298e-01 -1.82644576e-01
-1.07689953e+00 9.84506547e-01 1.93328387e-03 -1.16078079e+00
-4.99639332e-01 -9.26120520e-01 -3.63633752e-01 1.39975202e+00
-9.27818120e-01 -1.25456643e+00 8.33693668e-02 3.18139941e-01
4.11294818e-01 -1.09011316e+00 1.30301297e+00 1.66837305e-01
-4.12555248e-01 1.22655106e+00 2.83467561e-01 2.76439130e-01
9.17944133e-01 -1.60328484e+00 -3.04791946e-02 8.12296748e-01
1.41815439e-01 6.27153873e-01 8.26522470e-01 -1.77939102e-01
-9.44370985e-01 -5.21373630e-01 1.54119551e+00 -4.96832699e-01
4.92879987e-01 -1.77616522e-01 -9.50565100e-01 3.70478094e-01
4.71799582e-01 -6.36762619e-01 9.00453269e-01 4.61157680e-01
-4.75332946e-01 -1.07389219e-01 -1.74242449e+00 3.31896216e-01
1.89985290e-01 -3.48535478e-01 -7.66761243e-01 2.42752030e-01
4.88282479e-02 -1.98030800e-01 -1.12829411e+00 5.85248917e-02
9.05462027e-01 -9.46616709e-01 9.65071738e-01 -2.78612822e-01
4.81096096e-03 -3.15549314e-01 -4.60918903e-01 -1.50242150e+00
-1.90771833e-01 -2.83968925e-01 4.27641213e-01 1.42281914e+00
7.37628460e-01 -8.75261247e-01 4.95214999e-01 6.36517107e-02
2.75801510e-01 -3.59137923e-01 -7.30151415e-01 -8.07043910e-01
6.09335124e-01 -2.99180865e-01 4.45267916e-01 1.33289039e+00
3.37495685e-01 2.55650550e-01 -6.56121299e-02 1.90940529e-01
6.79592848e-01 1.03623047e-01 4.06828314e-01 -1.13387060e+00
4.14083600e-02 -7.44051695e-01 -4.96726573e-01 -1.24705285e-02
2.39864305e-01 -9.85781550e-01 -1.11907706e-01 -1.08164763e+00
-2.11863562e-01 -4.51029986e-01 -1.89493448e-01 3.12071741e-01
-2.03854531e-01 5.49962461e-01 1.97205231e-01 1.21720366e-01
2.66354848e-02 -3.15776497e-01 6.26999378e-01 -3.23627740e-02
-3.83666784e-01 2.48505950e-01 -6.21606112e-01 5.20847082e-01
1.31368244e+00 -1.88300028e-01 1.16960041e-01 -1.23202736e-02
-3.17834795e-01 -2.42571995e-01 -4.70705360e-01 -1.00034440e+00
-1.56018808e-01 8.47388152e-03 4.18403000e-01 -7.06191301e-01
1.90869674e-01 -4.76779789e-01 -3.00138146e-01 9.63577569e-01
-6.22832514e-02 4.56585854e-01 2.37957314e-01 -2.16325834e-01
-5.40742815e-01 -6.12690270e-01 1.00283837e+00 2.13813677e-01
-7.43013501e-01 -4.76642936e-01 -8.28797400e-01 -3.08189183e-01
9.67938244e-01 -3.84070247e-01 6.68566748e-02 -9.14452374e-02
-8.73115242e-01 -7.56654739e-02 4.71790731e-01 3.02972317e-01
3.52254212e-01 -1.10301936e+00 -1.25409508e+00 4.12250668e-01
3.22234660e-01 -1.12585199e+00 -9.96443927e-02 4.00734633e-01
-1.13233268e+00 4.84487921e-01 -6.24869645e-01 -6.32800460e-01
-1.89334369e+00 -4.40586507e-02 2.44055122e-01 1.13659138e-02
-1.84835792e-02 7.83165872e-01 -8.65880191e-01 -1.01596904e+00
1.29791498e-01 4.44903553e-01 -9.86131370e-01 4.93207067e-01
6.44795716e-01 7.84149110e-01 5.65605581e-01 -1.30111325e+00
-6.18238270e-01 5.58874130e-01 -2.87326902e-01 -6.19500279e-01
1.05912077e+00 2.85988659e-01 -3.45747977e-01 5.44384122e-01
9.90057290e-01 4.65719938e-01 -1.81483015e-01 9.60952342e-02
1.62946403e-01 -5.26142418e-01 -7.30791464e-02 -1.10724831e+00
-5.05422056e-01 4.91104305e-01 1.02512550e+00 4.84419048e-01
8.64610970e-01 -3.97872597e-01 1.91871986e-01 3.00585717e-01
4.86685067e-01 -1.44009948e+00 -5.83478630e-01 8.52111876e-01
9.23438609e-01 -1.27807903e+00 -2.88818814e-02 -6.98940575e-01
-8.44457865e-01 1.23434675e+00 3.21203291e-01 -3.13251019e-01
1.04158902e+00 1.04415178e-01 6.41292930e-01 3.91305059e-01
-3.28939736e-01 -1.02125406e-01 2.94684082e-01 9.07045245e-01
1.03041792e+00 6.73210621e-01 -9.78881955e-01 4.90892947e-01
-6.81432784e-01 -3.82366747e-01 7.40818560e-01 7.83694446e-01
-4.59497660e-01 -1.37738645e+00 -8.26648772e-01 2.71752566e-01
-5.58494866e-01 2.14067101e-01 -6.84517980e-01 9.43580031e-01
-4.52817008e-02 1.32044947e+00 5.29708900e-02 -6.60570145e-01
4.23950940e-01 3.25929970e-01 8.30384433e-01 -1.77157640e-01
-1.08514953e+00 -7.11793229e-02 5.00041246e-01 5.24453558e-02
-4.41727221e-01 -1.27887106e+00 -9.32528436e-01 -6.29593790e-01
-4.69531715e-01 3.73900503e-01 9.56783950e-01 6.56417310e-01
-5.07299364e-01 -2.58009672e-01 1.04668689e+00 -4.57213581e-01
-6.01640522e-01 -1.27492535e+00 -8.02165926e-01 2.93758839e-01
9.25458446e-02 -3.33769262e-01 -3.50082040e-01 7.39902034e-02] | [10.20732307434082, 10.703264236450195] |
f8db18ea-7320-4a05-bc54-11d426e5d51e | end-to-end-one-shot-human-parsing | 2105.01241 | null | https://arxiv.org/abs/2105.01241v2 | https://arxiv.org/pdf/2105.01241v2.pdf | End-to-end One-shot Human Parsing | Previous human parsing models are limited to parsing humans into pre-defined classes, which is inflexible for practical fashion applications that often have new fashion item classes. In this paper, we define a novel one-shot human parsing (OSHP) task that requires parsing humans into an open set of classes defined by any test example. During training, only base classes are exposed, which only overlap with part of the test-time classes. To address three main challenges in OSHP, i.e., small sizes, testing bias, and similar parts, we devise an End-to-end One-shot human Parsing Network (EOP-Net). Firstly, an end-to-end human parsing framework is proposed to parse the query image into both coarse-grained and fine-grained human classes, which builds a strong embedding network with rich semantic information shared across different granularities, facilitating identifying small-sized human classes. Then, we propose learning momentum-updated prototypes by gradually smoothing the training time static prototypes, which helps stabilize the training and learn robust features. Moreover, we devise a dual metric learning scheme which encourages the network to enhance features' both representational capability and transferability. Therefore, our EOP-Net can learn representative features that can quickly adapt to the novel classes and mitigate the testing bias issue. In addition, we employ a contrastive loss at the prototype level, thereby enforcing the distances among the classes in the fine-grained metric space to discriminate similar parts. We tailor three existing popular human parsing benchmarks to the OSHP task. Experiments on the new benchmarks demonstrate that EOP-Net outperforms representative one-shot segmentation models by large margins, which serves as a strong baseline for further research on this new task. The source code is available at https://github.com/Charleshhy/One-shot-Human-Parsing. | ['Bohan Zhuang', 'DaCheng Tao', 'Jianfei Cai', 'Jing Zhang', 'Haoyu He'] | 2021-05-04 | null | null | null | null | ['one-shot-segmentation', 'human-parsing'] | ['computer-vision', 'computer-vision'] | [ 4.06582467e-02 2.38724694e-01 -3.38898540e-01 -7.32542753e-01
-8.54207039e-01 -5.32652140e-01 1.54937521e-01 3.12391692e-03
-5.68780243e-01 3.66066426e-01 2.76234001e-03 1.08465672e-01
3.94759886e-02 -1.01016939e+00 -6.60917580e-01 -6.07898772e-01
1.56111851e-01 5.26649117e-01 5.72291613e-01 -5.33760823e-02
-2.31229350e-01 -3.47760953e-02 -1.27934945e+00 1.73803598e-01
8.78334522e-01 1.01842654e+00 3.17199558e-01 5.63072920e-01
-2.19610538e-02 2.60982215e-01 -3.69285673e-01 -7.50002563e-01
3.51378530e-01 -4.57096457e-01 -7.99715579e-01 1.05259903e-01
2.35141024e-01 -3.59717220e-01 -1.76787898e-01 1.21915579e+00
5.74407935e-01 3.44486624e-01 2.41671398e-01 -1.11590755e+00
-1.07351506e+00 6.23638332e-01 -5.31552732e-01 3.76879014e-02
3.56674701e-01 4.04994488e-01 1.22824764e+00 -8.31732988e-01
6.41021073e-01 1.38776469e+00 7.68208146e-01 1.01767504e+00
-9.73291636e-01 -7.44941473e-01 5.02375364e-01 6.22095838e-02
-8.99671972e-01 -2.01998986e-02 8.97374690e-01 -2.33427390e-01
4.27290887e-01 1.53135836e-01 5.47000647e-01 1.16766977e+00
-8.90214965e-02 1.02265811e+00 9.59103405e-01 -2.98467308e-01
3.07579845e-01 1.13404199e-01 7.18775034e-01 9.30548012e-01
1.81262925e-01 1.09088153e-01 -2.19889492e-01 1.85768768e-01
6.51428044e-01 8.76417905e-02 4.37770505e-03 -6.35626495e-01
-1.06552315e+00 1.06626332e+00 7.39969611e-01 2.37602651e-01
-9.18247104e-02 -2.49221727e-01 4.96735483e-01 8.01689625e-02
2.79771984e-01 3.14232051e-01 -7.37398565e-01 3.15686758e-03
-4.80813473e-01 1.88641518e-01 6.55053616e-01 1.08512163e+00
8.26311707e-01 -5.37561774e-01 -4.03001457e-01 1.08998919e+00
1.82792053e-01 1.66783497e-01 7.22865045e-01 -9.52943385e-01
6.49695814e-01 7.87153900e-01 -2.40390211e-01 -7.38822103e-01
-4.21135932e-01 -2.63739526e-01 -8.03821921e-01 -1.57478422e-01
4.51922059e-01 -2.40549475e-01 -1.12130761e+00 2.05161524e+00
7.79671669e-01 -2.62718706e-04 -1.11545391e-01 9.89418507e-01
9.16170299e-01 5.94176233e-01 3.48848581e-01 1.04257055e-01
1.69603324e+00 -1.37855065e+00 -3.47418845e-01 -5.05413055e-01
5.52231312e-01 -2.96723515e-01 1.67589283e+00 1.00113586e-01
-8.88983250e-01 -9.14328218e-01 -1.01364052e+00 -2.96188563e-01
-5.63846290e-01 1.32998437e-01 7.26548374e-01 6.85471237e-01
-5.60152531e-01 7.21211195e-01 -8.89641762e-01 -4.22527611e-01
7.82544255e-01 2.67658174e-01 -1.90222919e-01 -3.68613809e-01
-1.26699424e+00 2.15034589e-01 5.24041653e-01 1.15682013e-01
-5.32312155e-01 -5.61552584e-01 -1.17556298e+00 2.40727067e-02
5.61366618e-01 -6.46623254e-01 1.22801304e+00 -8.62205029e-01
-1.57701242e+00 1.10056627e+00 1.33262515e-01 -1.13387465e-01
6.44086301e-01 -1.87510997e-01 -2.49252632e-01 3.20024155e-02
5.61524570e-01 8.45351756e-01 6.23819113e-01 -1.02175593e+00
-7.45214999e-01 -4.62126583e-01 2.36704201e-01 9.10772681e-02
-3.62532049e-01 3.52422299e-04 -8.65901232e-01 -7.48342454e-01
1.12521924e-01 -9.49016213e-01 -4.11434978e-01 -4.11148556e-02
-6.45211995e-01 -4.92444366e-01 3.50897640e-01 -5.62806368e-01
1.18442070e+00 -2.34907222e+00 2.20635355e-01 -7.88198113e-02
1.99514180e-01 1.73188448e-01 -2.86142737e-01 -2.25041986e-01
8.90523568e-02 6.67689666e-02 -5.19543827e-01 -2.85892993e-01
2.99707025e-01 3.01311553e-01 -1.06879026e-02 1.55356079e-01
3.82229656e-01 1.06440020e+00 -1.03306019e+00 -5.04378796e-01
-4.32699285e-02 1.54650241e-01 -6.17153585e-01 4.42049861e-01
-8.78715590e-02 4.35657382e-01 -7.08300471e-01 8.63483429e-01
6.34391069e-01 -3.94413143e-01 2.14587077e-02 -1.03351124e-01
3.11375499e-01 -1.21345364e-01 -1.04657686e+00 1.85034800e+00
-3.32761168e-01 -1.42274320e-01 -1.82466328e-01 -9.16502416e-01
7.40989566e-01 -3.15975428e-01 1.85582221e-01 -7.79221952e-01
3.05863202e-01 6.49531707e-02 -5.59987314e-02 -3.10325116e-01
1.83788374e-01 -1.12436200e-02 -6.44222558e-01 8.05324763e-02
3.36895525e-01 2.56945133e-01 3.99247885e-01 2.50420533e-02
1.01226366e+00 3.14878792e-01 2.77033269e-01 -1.82927236e-01
4.02563781e-01 -5.14406376e-02 1.23679590e+00 7.03673661e-01
-7.46246397e-01 7.18228102e-01 6.27773166e-01 -5.58413923e-01
-8.27844560e-01 -1.29427838e+00 -1.64391801e-01 1.64594376e+00
5.98302126e-01 -3.00198674e-01 -1.11104202e+00 -1.25154138e+00
-6.82946369e-02 4.63439733e-01 -8.71565163e-01 -3.24405313e-01
-8.43870342e-01 -9.84122455e-01 2.13675171e-01 9.18385208e-01
6.52282894e-01 -1.22497904e+00 -6.74881876e-01 2.56506771e-01
3.83598683e-03 -1.02701223e+00 -6.98872745e-01 2.67511189e-01
-6.37316644e-01 -1.22099006e+00 -9.28813338e-01 -1.35457110e+00
6.74044847e-01 -2.28619780e-02 1.14180923e+00 -6.03191629e-02
-3.89586657e-01 2.16434553e-01 -5.16459823e-01 -1.23670124e-01
-8.99086297e-02 9.82159972e-02 -2.39007324e-01 -1.87801152e-01
6.95148289e-01 -1.15883656e-01 -7.27925539e-01 5.15487850e-01
-6.32935345e-01 6.20544031e-02 5.66679835e-01 9.61791515e-01
9.28149104e-01 -2.84096133e-02 5.97395837e-01 -1.24037623e+00
4.84353453e-01 -5.64549506e-01 -4.26086515e-01 4.50204968e-01
-4.14970785e-01 -1.79606825e-01 7.77392268e-01 -6.76204383e-01
-1.05997658e+00 1.59738928e-01 -1.30728066e-01 -1.93921119e-01
-2.00206622e-01 1.02489136e-01 -7.34066725e-01 2.16521144e-01
3.43312889e-01 -7.48213455e-02 -2.78847545e-01 -4.36317980e-01
5.88653266e-01 4.67251658e-01 6.32602990e-01 -7.93680787e-01
7.69741714e-01 1.76029906e-01 -5.31700373e-01 -3.02065074e-01
-1.21061051e+00 -5.11500835e-01 -7.37829089e-01 2.23790795e-01
1.24527955e+00 -8.23127806e-01 -3.26148778e-01 5.34440696e-01
-9.08530474e-01 -5.95997095e-01 -5.41588545e-01 2.25979105e-01
-5.67855775e-01 3.37959975e-01 -1.12313688e+00 -2.70803511e-01
-3.67817998e-01 -1.17144048e+00 1.26873457e+00 6.16587639e-01
-1.42186448e-01 -6.91681623e-01 8.49361047e-02 4.93036956e-01
-8.11205506e-02 1.83105990e-01 9.19280350e-01 -8.74440372e-01
-3.73114169e-01 -1.92982107e-01 -3.20957541e-01 4.35961157e-01
4.08082362e-03 -3.19683492e-01 -9.00524735e-01 -3.93360496e-01
-5.50013445e-02 -5.80190301e-01 1.03054988e+00 2.80523181e-01
1.42373931e+00 -7.19713047e-02 -3.24076474e-01 8.30911219e-01
1.10293579e+00 8.24445188e-02 3.58137459e-01 4.41573024e-01
8.37443888e-01 6.91228032e-01 9.56881225e-01 1.54589102e-01
5.39040864e-01 4.28161472e-01 7.01031089e-02 -1.52413800e-01
-6.12560958e-02 -4.60632712e-01 1.88572690e-01 8.85067165e-01
3.42485458e-01 -8.97366628e-02 -6.45284653e-01 4.59598005e-01
-1.87380397e+00 -5.08788049e-01 4.83695418e-01 2.08416581e+00
9.00082409e-01 4.66468841e-01 1.96292445e-01 -2.76249468e-01
1.07366478e+00 1.28074780e-01 -8.80204499e-01 -1.92217931e-01
1.68541446e-01 4.18847889e-01 2.31402442e-01 1.49751976e-01
-1.61118317e+00 1.16409183e+00 4.55889368e+00 7.49552846e-01
-7.59209275e-01 2.36687347e-01 8.73289824e-01 5.20893857e-02
7.58663639e-02 -1.21157944e-01 -1.04342437e+00 7.66934872e-01
6.37765586e-01 -1.51642486e-02 1.60286531e-01 1.25728953e+00
-3.04577768e-01 3.95080000e-01 -1.16395736e+00 9.06563878e-01
-9.02302340e-02 -7.96710134e-01 -7.98086300e-02 -2.92845279e-01
4.66131896e-01 -1.18589215e-01 -1.44608719e-02 1.02367067e+00
4.27269548e-01 -7.15876102e-01 6.39089584e-01 -1.10524762e-02
7.94225276e-01 -6.24534786e-01 7.23151684e-01 2.58346289e-01
-1.48385274e+00 -3.87881756e-01 -7.13667750e-01 3.23244721e-01
1.46900266e-01 4.05720890e-01 -1.88212961e-01 3.94736379e-01
9.66979980e-01 6.07083261e-01 -7.33488262e-01 7.79412925e-01
-4.00332987e-01 4.67290044e-01 -1.47081062e-01 -4.09414507e-02
3.16067398e-01 -1.52481452e-01 1.12417527e-01 1.09415019e+00
1.73956871e-01 2.34371573e-01 6.30151510e-01 7.93650746e-01
-3.37640136e-01 1.54976234e-01 -1.26050830e-01 1.28160074e-01
4.62839216e-01 1.33791709e+00 -1.01824856e+00 -4.19343948e-01
-5.47903299e-01 1.22924125e+00 5.33147335e-01 1.90688014e-01
-9.89042938e-01 -7.41578400e-01 5.20556986e-01 -6.21103756e-02
4.69148397e-01 2.29733646e-01 -1.72188506e-01 -1.32323468e+00
2.18405709e-01 -7.95755267e-01 8.68348122e-01 -2.42627382e-01
-1.71192551e+00 6.27143741e-01 -1.70152068e-01 -1.13950324e+00
-1.17044067e-02 -8.38713586e-01 -7.19946146e-01 6.04403794e-01
-1.43009257e+00 -1.41292536e+00 -4.34973300e-01 4.29563731e-01
8.61771524e-01 3.65610011e-02 6.08321786e-01 3.00176144e-01
-9.34988976e-01 9.56301689e-01 -2.06856355e-01 5.71507692e-01
7.28391349e-01 -1.34993744e+00 6.76680982e-01 8.28864932e-01
-3.91685106e-02 7.02644467e-01 2.52331614e-01 -6.46850049e-01
-1.02246571e+00 -1.46301425e+00 6.24346614e-01 -4.85252321e-01
5.18542826e-01 -6.42039180e-01 -1.10972404e+00 7.81520009e-01
-3.04734379e-01 5.16528010e-01 7.47615337e-01 2.44887516e-01
-5.03660679e-01 -1.07173145e-01 -1.20240915e+00 4.50019419e-01
1.24694920e+00 -2.95408487e-01 -7.53414452e-01 3.55115861e-01
1.12201607e+00 -4.58156198e-01 -8.00052404e-01 4.81818378e-01
5.01264751e-01 -8.17458510e-01 8.92573535e-01 -7.66488671e-01
4.90448236e-01 4.22115438e-03 -8.35970119e-02 -1.11168957e+00
-7.45092928e-01 -3.52063119e-01 -2.38489397e-02 1.40337455e+00
3.28234524e-01 -5.56888163e-01 1.08728206e+00 7.54751444e-01
-1.54453501e-01 -1.09019005e+00 -8.96787405e-01 -8.83837044e-01
4.41907853e-01 -1.85271487e-01 7.06897438e-01 8.76259804e-01
-2.25258484e-01 4.19212848e-01 -7.22397119e-02 7.69493431e-02
7.76603162e-01 3.08584481e-01 7.18772650e-01 -1.23539710e+00
-6.08559310e-01 -3.85066509e-01 -3.73627096e-01 -1.09609246e+00
2.13333115e-01 -9.49321806e-01 2.72364914e-01 -1.19190061e+00
5.75519264e-01 -5.59426904e-01 -7.21564293e-01 5.35865545e-01
-7.11855412e-01 -9.04390123e-03 2.17330858e-01 6.62665740e-02
-1.00627899e+00 4.81629848e-01 1.30337393e+00 -1.37640119e-01
-2.73831606e-01 4.33037020e-02 -9.35151279e-01 8.18372548e-01
8.71914804e-01 -4.58597898e-01 -3.87021363e-01 -4.31439042e-01
-5.57801537e-02 -3.09105009e-01 3.32332075e-01 -9.06125486e-01
1.09811164e-01 -7.11196586e-02 5.79953253e-01 -2.29800656e-01
8.45633149e-02 -4.91396338e-01 -4.11122024e-01 4.14086908e-01
-3.75599504e-01 -1.95309132e-01 -1.33864880e-01 7.59756684e-01
-2.04142909e-02 -2.51326352e-01 1.01711786e+00 -3.82132709e-01
-1.16442096e+00 7.52281904e-01 4.97104079e-01 5.44802904e-01
1.12566233e+00 -3.07466358e-01 -2.24587694e-01 3.77481669e-01
-7.65619278e-01 6.41856015e-01 5.39598405e-01 6.42721176e-01
3.88538986e-01 -1.30110145e+00 -4.42128301e-01 2.76903301e-01
3.46409470e-01 3.14839751e-01 4.21156526e-01 3.61318171e-01
-2.63063192e-01 -5.26358671e-02 -1.82097867e-01 -5.70522904e-01
-1.01375663e+00 8.09906781e-01 4.34120037e-02 -4.23290253e-01
-7.76821554e-01 1.26984107e+00 5.36153257e-01 -1.02950215e+00
3.30320954e-01 -3.55473548e-01 -1.86029524e-01 -5.43870740e-02
4.79909807e-01 2.28789002e-01 -3.13378245e-01 -4.29922104e-01
-3.17340136e-01 7.02466309e-01 -2.35698730e-01 4.24301654e-01
1.28220689e+00 -1.12249717e-01 1.98502183e-01 3.89242619e-01
1.39556611e+00 -1.80938244e-01 -1.54060054e+00 -3.08515996e-01
1.37680918e-01 -3.08853894e-01 -4.09122139e-01 -7.72914648e-01
-1.08901966e+00 8.56203616e-01 8.28715086e-01 -5.46284355e-02
1.05939770e+00 3.21977496e-01 1.41073430e+00 2.37035736e-01
3.66556942e-01 -1.34718001e+00 1.78673625e-01 4.12973076e-01
3.62561554e-01 -1.55462837e+00 -3.00106704e-01 -6.64363325e-01
-6.59355760e-01 1.00507700e+00 1.03827155e+00 -1.41623780e-01
5.01266301e-01 -2.24176757e-02 4.51522321e-02 -9.19515416e-02
-3.57315779e-01 -1.89435765e-01 1.40093267e-01 6.18480563e-01
2.66889274e-01 3.45766038e-01 -3.50159079e-01 1.55698574e+00
-2.09703356e-01 -1.26909614e-01 1.77975446e-02 9.69539702e-01
-5.33740342e-01 -1.07368815e+00 -5.68701141e-03 4.59101319e-01
-3.60709608e-01 1.05143398e-01 -2.45854601e-01 8.69616807e-01
4.69662189e-01 6.52079582e-01 -7.19058036e-04 -3.09219390e-01
6.62447870e-01 -4.33396772e-02 3.35092664e-01 -9.16677654e-01
-4.32736874e-01 -2.10658014e-01 -1.52736217e-01 -8.09994519e-01
4.01105322e-02 -4.59934026e-01 -1.36279297e+00 1.96127057e-01
-2.67448276e-01 3.80642712e-02 1.51072532e-01 6.62633657e-01
2.07836881e-01 5.49583137e-01 6.20214045e-01 -6.26476169e-01
-9.77075219e-01 -7.04734802e-01 -5.63404739e-01 8.43922973e-01
-1.32530257e-01 -7.10194290e-01 -2.87100941e-01 -1.38598755e-01] | [8.96111011505127, 0.384874165058136] |
64d17180-51a7-4e89-b4f6-6e8be0e11876 | deepa2-a-modular-framework-for-deep-argument | 2110.01509 | null | https://arxiv.org/abs/2110.01509v3 | https://arxiv.org/pdf/2110.01509v3.pdf | DeepA2: A Modular Framework for Deep Argument Analysis with Pretrained Neural Text2Text Language Models | In this paper, we present and implement a multi-dimensional, modular framework for performing deep argument analysis (DeepA2) using current pre-trained language models (PTLMs). ArgumentAnalyst -- a T5 model (Raffel et al. 2020) set up and trained within DeepA2 -- reconstructs argumentative texts, which advance an informal argumentation, as valid arguments: It inserts, e.g., missing premises and conclusions, formalizes inferences, and coherently links the logical reconstruction to the source text. We create a synthetic corpus for deep argument analysis, and evaluate ArgumentAnalyst on this new dataset as well as on existing data, specifically EntailmentBank (Dalvi et al. 2021). Our empirical findings vindicate the overall framework and highlight the advantages of a modular design, in particular its ability to emulate established heuristics (such as hermeneutic cycles), to explore the model's uncertainty, to cope with the plurality of correct solutions (underdetermination), and to exploit higher-order evidence. | ['Kyle Richardson', 'Gregor Betz'] | 2021-10-04 | null | https://aclanthology.org/2022.starsem-1.2 | https://aclanthology.org/2022.starsem-1.2.pdf | sem-naacl-2022-7 | ['logical-reasoning-reading-comprehension'] | ['natural-language-processing'] | [-1.22332731e-02 1.15612543e+00 -2.95653850e-01 -1.92639083e-01
-9.38217938e-01 -1.09232461e+00 1.13293517e+00 5.32701075e-01
-2.50059575e-01 9.23354626e-01 8.77621889e-01 -1.24176955e+00
-3.68577719e-01 -9.59118366e-01 -7.09983647e-01 -1.23275116e-01
2.26226345e-01 9.56807554e-01 -1.43985495e-01 -3.19994330e-01
4.84442592e-01 2.78973337e-02 -1.18225610e+00 8.70152354e-01
8.59244287e-01 6.69678986e-01 -3.15651983e-01 6.07217669e-01
-3.10498416e-01 1.45345008e+00 -6.10377133e-01 -1.33848166e+00
-2.69770890e-01 -3.10389042e-01 -1.55792212e+00 -4.54903364e-01
6.64275438e-02 -4.93003458e-01 2.56491546e-02 7.40726173e-01
2.99515247e-01 -5.90210259e-01 6.98704123e-01 -8.37052882e-01
-5.97490668e-01 1.65178919e+00 -3.70810926e-02 1.67910218e-01
3.48756164e-01 2.23989319e-02 1.42466760e+00 -9.16522503e-01
1.13181341e+00 1.74597251e+00 9.51247990e-01 4.11013871e-01
-1.20296919e+00 4.77514230e-02 -6.84814667e-03 2.56595045e-01
-3.38115990e-01 -6.75007105e-01 8.73435199e-01 -2.61292368e-01
1.02759242e+00 3.58991623e-01 8.22159410e-01 1.63460541e+00
1.77702963e-01 1.19038534e+00 1.10232770e+00 -7.07650065e-01
3.61128479e-01 -2.52134427e-02 2.86707163e-01 2.97655135e-01
7.29254246e-01 -1.43713793e-02 -2.90440083e-01 -4.73408133e-01
1.94429487e-01 -7.94680655e-01 2.77473122e-01 -1.04297385e-01
-1.03191066e+00 1.15700614e+00 9.27654747e-03 4.11401570e-01
-5.16954780e-01 7.88102746e-02 8.12755167e-01 3.71412873e-01
2.57316291e-01 5.50215185e-01 -5.59958577e-01 -3.27036649e-01
-8.01566422e-01 6.32682085e-01 1.18317640e+00 2.49844819e-01
1.43764541e-01 -1.71739399e-01 -1.37524530e-01 5.43086827e-01
6.58176541e-01 2.11862043e-01 3.21764737e-01 -1.39057553e+00
8.17350209e-01 6.42195761e-01 7.12986737e-02 -8.80230725e-01
-2.73707390e-01 -4.85909760e-01 -3.58741015e-01 5.64894080e-02
6.09477878e-01 -2.75716245e-01 -1.32566705e-01 1.66504812e+00
3.37589175e-01 -4.03359205e-01 7.87330687e-01 5.88811278e-01
9.25997198e-01 3.36709082e-01 1.15475513e-01 -1.39832050e-01
1.37869155e+00 -5.74081600e-01 -7.74071097e-01 -4.88711268e-01
1.01337540e+00 -7.42044449e-01 1.02489209e+00 4.85411644e-01
-1.84547997e+00 1.95568446e-02 -1.01897526e+00 -5.46782792e-01
-1.75649524e-01 4.95498590e-02 8.77086043e-01 4.49310750e-01
-8.32740307e-01 6.93887413e-01 -5.38329542e-01 2.05114618e-01
7.49101579e-01 -2.66596019e-01 -9.01974365e-02 2.02636719e-01
-1.62946057e+00 1.18879128e+00 7.73897767e-01 5.23948550e-01
-7.69512177e-01 -3.14657599e-01 -9.30615246e-01 -7.23542199e-02
6.10425651e-01 -7.48093843e-01 1.48349893e+00 -7.71490872e-01
-1.50248730e+00 1.30973804e+00 -3.08417194e-02 -1.01596689e+00
9.40375686e-01 -4.06420410e-01 -1.03024304e-01 7.62698352e-02
3.46254885e-01 3.02392006e-01 3.58482629e-01 -1.25906622e+00
-2.55289823e-01 -3.74109685e-01 6.31048203e-01 -1.07245352e-02
2.07365349e-01 9.35465693e-02 5.18197060e-01 -4.20486182e-01
2.85316885e-01 -6.34197593e-01 1.26492213e-02 -3.81036371e-01
-8.32670927e-01 -5.91917157e-01 3.11812341e-01 -7.45318592e-01
1.34780431e+00 -1.76520884e+00 2.84257472e-01 2.39697605e-01
3.75588417e-01 3.37639570e-01 2.33291388e-01 7.62401402e-01
-9.32600126e-02 5.48731327e-01 -3.72930735e-01 -3.57456803e-01
5.82214952e-01 6.13610327e-01 -7.57358193e-01 1.63212046e-01
3.42180818e-01 1.41282964e+00 -7.59890854e-01 -7.03228414e-01
1.42272666e-01 1.19054370e-01 -6.24412417e-01 -2.39948377e-01
-6.27071023e-01 -1.66045189e-01 -4.25460756e-01 5.28845549e-01
6.19853735e-01 -3.22466671e-01 9.46268678e-01 -1.83908224e-01
-3.39871913e-01 1.28447294e+00 -1.06423807e+00 1.60620868e+00
-4.45727408e-01 7.68245757e-01 1.78040683e-01 -1.11119437e+00
6.33905411e-01 3.38765800e-01 -3.46255869e-01 -7.01659560e-01
5.25638223e-01 7.24818289e-01 2.66124219e-01 -4.59283829e-01
2.45210335e-01 -4.06172276e-02 -2.47855276e-01 8.64071786e-01
-1.86944887e-01 -3.02553624e-01 4.50967431e-01 5.62589049e-01
6.76895201e-01 4.14721310e-01 5.64269841e-01 -6.49883628e-01
7.69283056e-01 2.73432106e-01 3.34904432e-01 7.85538971e-01
1.24980599e-01 2.59786770e-02 1.27914667e+00 -9.17425632e-01
-1.27100492e+00 -1.02697515e+00 -4.22854960e-01 5.14859855e-01
-4.58112240e-01 -5.39438367e-01 -8.34274232e-01 -8.29493403e-01
3.55956778e-02 1.44979239e+00 -7.91904151e-01 1.93027183e-01
-1.01568270e+00 -4.98958200e-01 9.49393451e-01 4.19168532e-01
4.27062571e-01 -1.27850282e+00 -1.20474577e+00 6.36255026e-01
-7.03772664e-01 -8.01634610e-01 8.85160267e-01 4.27474499e-01
-6.67185903e-01 -1.48506522e+00 2.45512262e-01 -4.39209372e-01
1.97974935e-01 -6.45892262e-01 1.41216350e+00 2.80224770e-01
5.09587266e-02 -9.62720141e-02 -1.51708856e-01 -5.69741249e-01
-1.05271637e+00 -1.77341267e-01 -5.12387991e-01 -6.54643297e-01
3.86084616e-01 -4.66748744e-01 -2.61838764e-01 -3.02699685e-01
-9.13837850e-01 9.95654911e-02 3.05763453e-01 9.96594250e-01
1.69705138e-01 -2.89295703e-01 6.57402635e-01 -1.12293851e+00
1.20391953e+00 -4.45890546e-01 -4.12681341e-01 5.26515722e-01
-6.97452188e-01 4.32406813e-01 4.71762985e-01 5.11425547e-02
-1.39986980e+00 -1.20510542e+00 -3.72147471e-01 4.58457083e-01
2.33013958e-01 8.89679790e-01 -4.35028113e-02 7.26347744e-01
9.30283368e-01 -2.49369413e-01 1.34330511e-01 -2.29773775e-01
7.85866141e-01 4.55925077e-01 6.25845075e-01 -1.51300728e+00
6.31167293e-01 5.85036874e-01 3.99543438e-03 -1.97196752e-01
-1.28511095e+00 5.29771268e-01 -5.78540981e-01 8.05825144e-02
4.13056165e-01 -7.05070555e-01 -8.28682065e-01 7.57161677e-02
-1.64587867e+00 -4.74571705e-01 -5.57378590e-01 3.02171528e-01
-5.54552317e-01 6.24359012e-01 -9.24623489e-01 -8.95231366e-01
-5.92071950e-01 -9.23543930e-01 7.50124276e-01 -1.05133802e-01
-9.79935467e-01 -1.41958332e+00 1.31371453e-01 5.10532916e-01
1.76890448e-01 6.07829809e-01 1.47733855e+00 -9.46407378e-01
-7.68167749e-02 1.18936598e-01 -2.49160349e-01 3.16638321e-01
-4.04920012e-01 1.78573817e-01 -9.67360139e-01 3.25681269e-01
3.40845227e-01 -9.39838588e-01 6.21516049e-01 2.41894066e-01
7.74290085e-01 -6.68257713e-01 3.95510234e-02 1.28490642e-01
8.43300343e-01 -2.49025717e-01 9.97327864e-01 1.11600184e+00
-7.02168494e-02 8.93148541e-01 4.54938322e-01 3.37530941e-01
7.99630702e-01 4.46047299e-02 4.91561651e-01 3.13036472e-01
-1.17395833e-01 -4.63485926e-01 5.42022400e-02 4.76897120e-01
-2.92267017e-02 -2.62007952e-01 -1.01587605e+00 5.39753675e-01
-2.13997149e+00 -1.13041794e+00 -4.99569774e-01 1.68256235e+00
1.09632075e+00 6.78720713e-01 -2.85959512e-01 6.26845717e-01
2.07149386e-01 2.96888053e-01 -2.52979219e-01 -9.82554436e-01
-7.57887542e-01 5.22489727e-01 -2.34628618e-01 9.64302182e-01
-7.12734401e-01 1.16056597e+00 6.06616688e+00 5.24934530e-01
-5.12205720e-01 -1.77935094e-01 7.92606473e-01 6.11201599e-02
-1.24167049e+00 5.84859192e-01 -3.27237874e-01 1.03212260e-01
8.10028613e-01 -1.34112462e-01 3.81310731e-01 6.39891207e-01
-1.13321707e-01 -7.90888816e-02 -1.21725225e+00 2.79569268e-01
-9.62514579e-02 -1.98387790e+00 1.38901711e-01 6.41067028e-02
3.60416830e-01 -3.16202432e-01 9.78487656e-02 3.59629810e-01
7.34972775e-01 -9.91372705e-01 1.50604701e+00 -4.46342528e-02
3.53299469e-01 -5.43048799e-01 1.04607618e+00 4.62849885e-01
-3.88942242e-01 -3.71548951e-01 -1.17628142e-01 -4.78586167e-01
3.02572846e-01 5.75800478e-01 -6.98222458e-01 7.50400901e-01
2.68643379e-01 4.91465777e-01 -3.36028457e-01 -2.14352265e-01
-9.71372604e-01 6.25614345e-01 -3.24421287e-01 -1.70184031e-01
6.85999691e-01 -1.60696834e-01 5.93616605e-01 1.03801441e+00
-5.67073114e-02 1.05725244e-01 -4.51622516e-01 1.13260961e+00
-1.97480153e-02 -4.26765792e-02 -4.17248815e-01 -1.85938045e-01
6.89969480e-01 1.04035115e+00 -4.91733134e-01 -6.07982457e-01
4.29385751e-02 3.38571906e-01 6.00829542e-01 4.53318842e-02
-6.56784236e-01 1.96625173e-01 1.53380126e-01 6.36134436e-03
-3.95559818e-02 3.44342031e-02 -7.29150414e-01 -1.15417266e+00
3.40715587e-01 -1.35048342e+00 5.92119932e-01 -7.37141669e-01
-1.11563885e+00 4.25160736e-01 1.76613703e-01 -3.27990144e-01
-8.09488714e-01 -7.40741670e-01 -8.48714411e-01 7.46748388e-01
-1.56508064e+00 -1.45344508e+00 5.29145360e-01 1.73272222e-01
3.09538245e-01 1.40336096e-01 9.43503261e-01 -2.24275768e-01
-2.99568981e-01 2.43882611e-01 -2.48782277e-01 2.21946582e-01
2.89567769e-01 -1.11670113e+00 7.49667048e-01 7.74916232e-01
5.54972701e-02 1.11025226e+00 7.79133081e-01 -7.42496848e-01
-1.33364522e+00 -2.76632965e-01 1.55069757e+00 -9.03565407e-01
1.26663637e+00 -2.12741226e-01 -8.62176239e-01 8.26630771e-01
5.38228095e-01 -4.35757071e-01 6.67332172e-01 5.06577790e-01
-6.17078125e-01 3.63985240e-01 -1.20343733e+00 9.12150860e-01
1.07817054e+00 -6.55231297e-01 -1.61285317e+00 1.86379656e-01
4.99446154e-01 -5.27466118e-01 -5.24887264e-01 3.50339353e-01
7.22458780e-01 -1.30379820e+00 9.22675073e-01 -9.46691930e-01
1.09233892e+00 1.23442739e-01 -2.81324983e-01 -9.63349223e-01
6.10163854e-03 -8.49472523e-01 -3.90000612e-01 1.18969893e+00
6.27565563e-01 -8.05461645e-01 3.99095446e-01 6.91068053e-01
-2.42521435e-01 -7.21459687e-01 -1.18831849e+00 -1.08589135e-01
6.20531201e-01 -7.22357035e-01 7.92577267e-01 1.03949428e+00
3.48372102e-01 4.44376767e-01 3.05034310e-01 -3.83639365e-01
8.44497323e-01 3.64095181e-01 7.18494534e-01 -1.65879869e+00
-1.47338420e-01 -7.16642916e-01 4.25194532e-01 -5.85147202e-01
5.72837889e-01 -1.07037020e+00 -4.68835086e-01 -1.92517769e+00
-2.25533471e-01 -3.34825754e-01 2.03972906e-01 2.91615546e-01
2.19837278e-01 -4.43470120e-01 -7.06353113e-02 1.21852972e-01
-3.44976127e-01 3.03823680e-01 8.78866613e-01 1.02622278e-01
8.97761285e-02 -4.67928439e-01 -1.24949658e+00 1.33509898e+00
7.40371406e-01 -3.28875095e-01 -1.78530365e-01 -8.05966258e-01
1.14565957e+00 1.04347438e-01 7.68929362e-01 -1.87655419e-01
4.59674746e-02 4.89390790e-02 2.59797662e-01 -7.33292878e-01
-8.88217390e-02 -4.10743296e-01 -9.68572870e-02 5.91400146e-01
-7.77268469e-01 2.19875082e-01 3.34446937e-01 1.89431846e-01
-2.59158369e-02 -6.03223920e-01 2.43729532e-01 -3.72252941e-01
-1.84348479e-01 -7.41439044e-01 -1.89672828e-01 6.20565712e-01
5.43398440e-01 -1.02670297e-01 -1.01905417e+00 6.34159818e-02
-6.02641881e-01 2.06641197e-01 2.97338426e-01 7.27830082e-02
4.70550179e-01 -1.11445010e+00 -1.13114178e+00 -1.44632459e-01
-1.76579803e-01 2.12468565e-01 8.76219645e-02 5.89983582e-01
-7.45639086e-01 3.92964691e-01 2.12700963e-02 -2.21184433e-01
-6.93179011e-01 2.44002879e-01 1.22543372e-01 -7.27394760e-01
-7.14692950e-01 6.65398419e-01 -2.78191626e-01 -5.41797042e-01
-2.27122575e-01 -5.81656337e-01 -2.93061852e-01 3.80694509e-01
2.46085152e-01 2.45315552e-01 -1.29530191e-01 -2.48508140e-01
-4.37573671e-01 9.44744796e-02 -1.62584230e-01 -4.86382693e-01
1.45794940e+00 -7.83872008e-02 -6.85955226e-01 3.96582931e-01
5.54065526e-01 3.77774447e-01 -8.54713559e-01 -2.41384760e-01
4.02631670e-01 4.97927293e-02 -1.57086968e-01 -1.25240636e+00
3.48627158e-02 8.60594571e-01 -4.87026155e-01 5.37000418e-01
1.19348973e-01 1.71264201e-01 4.84935492e-01 6.18026912e-01
-5.47173433e-02 -1.39938581e+00 -5.16031012e-02 7.42053986e-01
1.03541613e+00 -8.99682224e-01 1.15251273e-01 -1.66639499e-02
-6.24477744e-01 1.51367855e+00 2.08382934e-01 5.94883785e-02
1.48264989e-01 4.35087383e-01 1.55280650e-01 -6.71127379e-01
-1.07111692e+00 1.96128860e-01 -2.46069655e-01 1.94673330e-01
5.62345028e-01 2.54311264e-01 -7.20402360e-01 7.74492502e-01
-6.58721089e-01 -2.49828145e-01 5.77311873e-01 7.81663358e-01
-1.41029492e-01 -1.09953654e+00 -3.29142243e-01 2.12583449e-02
-6.33351445e-01 -3.80887985e-01 -7.88366973e-01 1.03075409e+00
9.43466797e-02 1.16409039e+00 -8.52572769e-02 3.27487379e-01
6.26194179e-02 1.74252927e-01 5.05199134e-01 -1.99134797e-01
-7.95478880e-01 8.33644811e-03 1.11289239e+00 -1.90025479e-01
-4.16694850e-01 -7.57575154e-01 -1.31588387e+00 -4.95328039e-01
-8.08182284e-02 5.21933436e-01 7.42724895e-01 1.39678669e+00
2.13927776e-01 4.09488589e-01 8.18749368e-02 -3.19315821e-01
-9.81224358e-01 -8.13706875e-01 1.63925096e-01 2.25569844e-01
9.16673057e-03 -4.17713195e-01 -6.77462697e-01 -3.12440693e-01] | [9.66491985321045, 9.567083358764648] |
8fba1199-5ff2-415e-a372-9c05e3d85859 | a-3d-object-detection-and-pose-estimation | 1703.03940 | null | http://arxiv.org/abs/1703.03940v1 | http://arxiv.org/pdf/1703.03940v1.pdf | A 3D Object Detection and Pose Estimation Pipeline Using RGB-D Images | 3D object detection and pose estimation has been studied extensively in
recent decades for its potential applications in robotics. However, there still
remains challenges when we aim at detecting multiple objects while retaining
low false positive rate in cluttered environments. This paper proposes a robust
3D object detection and pose estimation pipeline based on RGB-D images, which
can detect multiple objects simultaneously while reducing false positives.
Detection begins with template matching and yields a set of template matches. A
clustering algorithm then groups templates of similar spatial location and
produces multiple-object hypotheses. A scoring function evaluates the
hypotheses using their associated templates and non-maximum suppression is
adopted to remove duplicate results based on the scores. Finally, a combination
of point cloud processing algorithms are used to compute objects' 3D poses.
Existing object hypotheses are verified by computing the overlap between model
and scene points. Experiments demonstrate that our approach provides
competitive results comparable to the state-of-the-arts and can be applied to
robot random bin-picking. | ['Juan Rojas', 'Yisheng Guan', 'Ruotao He'] | 2017-03-11 | null | null | null | null | ['robust-3d-object-detection'] | ['computer-vision'] | [ 2.98804045e-01 -2.86509007e-01 1.60823017e-01 -2.14777276e-01
-7.48526156e-01 -6.39306724e-01 4.35810655e-01 4.21500504e-01
-6.30059361e-01 7.87115693e-02 -6.34946823e-01 1.17775761e-01
-1.61135122e-01 -5.74647963e-01 -6.03208125e-01 -5.67025185e-01
-2.10686773e-01 1.11084485e+00 1.00034690e+00 7.98255354e-02
6.15522265e-01 1.14602077e+00 -1.91661310e+00 -1.09197028e-01
4.80583966e-01 1.18228722e+00 7.97025084e-01 4.67527837e-01
-5.86536154e-02 2.87385006e-02 -7.15188503e-01 2.14468203e-02
7.42202461e-01 -4.91515845e-02 -2.39243254e-01 4.83195245e-01
4.35179114e-01 -2.88374722e-01 2.09559277e-01 1.15247011e+00
3.30198854e-01 9.60924923e-02 6.35986090e-01 -1.37816548e+00
2.86998451e-01 -1.78651109e-01 -6.41965151e-01 -2.68383287e-02
6.14008307e-01 1.61378890e-01 5.43048382e-01 -1.33383584e+00
4.50941324e-01 1.44268548e+00 5.32279015e-01 1.71443492e-01
-1.22482145e+00 -7.23739147e-01 -5.57323247e-02 1.64269581e-01
-1.60518098e+00 -1.20971203e-01 5.97757518e-01 -4.16697234e-01
1.08182943e+00 2.56252527e-01 7.98634350e-01 1.57395333e-01
8.67225826e-02 6.32252216e-01 7.64710426e-01 -5.42978406e-01
3.58303636e-01 -3.40419970e-02 -1.39430642e-01 5.84691882e-01
5.92801154e-01 -4.30841409e-02 -5.86645961e-01 -8.32159668e-02
7.15660751e-01 1.72983825e-01 2.32372716e-01 -9.70085621e-01
-1.45837522e+00 4.82955456e-01 5.74593842e-01 -2.07348932e-02
-7.06407189e-01 3.86852548e-02 3.98988351e-02 -2.03928366e-01
4.23728861e-02 4.38477576e-01 -1.86613366e-01 1.97669402e-01
-5.93122005e-01 6.56826138e-01 4.77988631e-01 1.35750651e+00
9.43306088e-01 -4.19202924e-01 2.62414545e-01 6.46329999e-01
4.99080896e-01 6.77702010e-01 4.04220857e-02 -9.43406940e-01
1.66629732e-01 1.03291619e+00 3.18289399e-01 -1.04459298e+00
-5.17928541e-01 -2.44321465e-01 -2.54825741e-01 6.81014836e-01
1.73661947e-01 5.64859986e-01 -1.05877745e+00 8.48844409e-01
8.24438751e-01 -3.47480550e-03 -1.47261679e-01 1.19063711e+00
6.07743442e-01 2.18008831e-01 -3.84977698e-01 -2.83277389e-02
1.29001570e+00 -4.49676514e-01 -4.11832601e-01 -5.91336608e-01
4.12119813e-02 -1.08172011e+00 4.99403477e-01 4.69549388e-01
-1.00437558e+00 -6.44577324e-01 -1.09642553e+00 1.24286532e-01
-2.28944823e-01 1.93810791e-01 3.03268135e-01 4.73555118e-01
-6.42008662e-01 3.31265509e-01 -1.02883172e+00 -6.51572764e-01
3.77431333e-01 8.73887479e-01 -3.14289272e-01 -8.63542706e-02
-2.17948794e-01 1.31511092e+00 7.41635501e-01 7.31915012e-02
-7.51890361e-01 -2.82958478e-01 -7.75341511e-01 -4.20731366e-01
6.13448381e-01 -4.54367906e-01 1.31297278e+00 -2.18358889e-01
-1.37567723e+00 9.60602164e-01 -1.14319660e-01 -2.56385654e-01
5.53585410e-01 -3.50328356e-01 7.49205649e-02 1.88835844e-01
3.90432477e-01 7.61498690e-01 7.07909882e-01 -1.35107315e+00
-1.19272554e+00 -7.72502601e-01 -4.30967033e-01 5.14565170e-01
3.12012672e-01 1.80639282e-01 -9.77947414e-01 3.45416483e-03
1.17224896e+00 -1.10043001e+00 -5.26432276e-01 2.33506754e-01
-1.76040962e-01 -2.19654217e-01 9.39158797e-01 -6.16814755e-02
3.92141134e-01 -2.07010531e+00 5.11671640e-02 5.62746644e-01
-8.14140067e-02 -1.34388953e-01 1.28536999e-01 2.91592062e-01
4.29122150e-01 -4.21680421e-01 -5.97126735e-03 -3.82717460e-01
-1.94140114e-02 1.92064971e-01 1.31933196e-02 8.74073803e-01
3.87776643e-01 4.31786865e-01 -8.65639150e-01 -5.97813308e-01
8.19871306e-01 1.67463183e-01 -3.53435308e-01 2.12562054e-01
-1.67204916e-01 1.04664214e-01 -3.97907436e-01 1.00802803e+00
9.51197743e-01 1.14792332e-01 7.68258497e-02 -2.30618685e-01
-3.77473235e-01 1.78781807e-01 -1.71970582e+00 1.51729155e+00
1.84190542e-01 2.54747778e-01 2.07576692e-01 -6.04350865e-01
1.23607898e+00 -2.49857247e-01 6.12027705e-01 -3.04746747e-01
3.96852225e-01 5.32973588e-01 8.94281268e-03 -2.46769220e-01
7.82288671e-01 8.76208246e-02 -5.91889396e-02 -6.34891959e-03
-1.27716199e-01 -8.09041500e-01 2.68010616e-01 -1.01085737e-01
1.10844445e+00 3.28573048e-01 5.05897343e-01 2.56693810e-02
2.50736862e-01 6.68453932e-01 2.91916192e-01 7.65113175e-01
-3.16966712e-01 5.79802990e-01 -3.27497721e-01 -2.32126698e-01
-8.44129741e-01 -1.29727697e+00 -1.51695594e-01 6.18315160e-01
1.00328302e+00 -1.06290787e-01 -3.08099180e-01 -3.86955827e-01
4.91748780e-01 4.65619355e-01 -2.96040535e-01 4.67684343e-02
-5.43753743e-01 -4.49518204e-01 -5.91640621e-02 3.85018915e-01
1.40086487e-01 -7.98990607e-01 -1.47718823e+00 2.96774089e-01
1.50890484e-01 -1.12964082e+00 1.25687569e-01 5.12292325e-01
-9.99395907e-01 -1.19645941e+00 -3.51279706e-01 -8.01030517e-01
8.22351038e-01 9.90486443e-01 7.99075425e-01 -7.71266520e-02
-5.73902190e-01 3.35132450e-01 -5.77167928e-01 -9.08148587e-01
-3.04652631e-01 -2.85541832e-01 3.90702844e-01 -3.10934573e-01
6.44062817e-01 -7.20160231e-02 -4.82662171e-01 7.97213495e-01
-5.80293000e-01 -1.74265474e-01 9.57200110e-01 3.42147529e-01
1.01896107e+00 -1.22098522e-02 -3.62216868e-02 7.31579773e-03
1.07018165e-01 -3.45642082e-02 -1.10907125e+00 -6.76906779e-02
-2.45562449e-01 -1.39673889e-01 -2.18345299e-01 -6.56325817e-01
-5.50590992e-01 8.59965861e-01 3.50337774e-01 -7.37009823e-01
-3.82104993e-01 1.32943541e-01 -7.22657219e-02 -3.96868795e-01
6.50185704e-01 -2.35778447e-02 1.85179070e-01 -4.83627051e-01
1.50324792e-01 6.19875610e-01 7.06924796e-01 -1.19337089e-01
9.84054923e-01 6.20388508e-01 2.99240232e-01 -7.50575185e-01
-2.34657928e-01 -1.18430662e+00 -1.03328001e+00 -7.52367795e-01
5.80868721e-01 -8.23179901e-01 -8.28131974e-01 1.84058249e-01
-1.29897332e+00 3.48284811e-01 -6.17876202e-02 8.43681633e-01
-4.98818576e-01 1.64810136e-01 -7.89275467e-02 -1.22202837e+00
-4.94506583e-02 -1.22340012e+00 1.26653981e+00 1.87712997e-01
-1.41108215e-01 1.43189460e-01 -1.52668282e-01 4.39275466e-02
-2.12863728e-01 1.73951492e-01 2.75248468e-01 -6.32365108e-01
-1.02609313e+00 -7.15735316e-01 -6.77116588e-02 -3.09770972e-01
1.05683766e-01 -8.80406201e-02 -7.00146437e-01 -1.62588343e-01
-2.10135713e-01 2.03836471e-01 5.44498861e-01 3.76269430e-01
5.55693269e-01 3.29935223e-01 -8.24155569e-01 3.45163941e-02
1.37221658e+00 3.33595663e-01 3.03242683e-01 6.87897980e-01
3.17901134e-01 6.35864019e-01 1.29177558e+00 4.84029472e-01
6.44725561e-02 7.75340259e-01 1.04883933e+00 1.83933079e-01
2.17875183e-01 1.06349304e-01 1.10481203e-01 2.25157201e-01
-4.16778261e-03 1.03968561e-01 -1.18610156e+00 5.57218850e-01
-1.92652440e+00 -7.22063005e-01 -5.08888006e-01 2.34324908e+00
2.26654917e-01 5.61363041e-01 4.09165353e-01 3.05679500e-01
1.00259447e+00 -6.27627671e-01 -7.54764795e-01 2.60525435e-01
7.67421424e-02 5.60644791e-02 9.09921169e-01 2.24780262e-01
-1.11907387e+00 9.27820921e-01 5.83767891e+00 3.14645797e-01
-7.15390801e-01 -2.27108985e-01 -2.06814289e-01 -1.78982347e-01
4.75771368e-01 -1.11161791e-01 -1.04178774e+00 4.69069220e-02
1.64825439e-01 8.26464891e-02 -5.88116720e-02 1.18530941e+00
-5.74887171e-02 -5.88624477e-01 -1.06292856e+00 1.14064991e+00
2.40325257e-01 -9.02358234e-01 -2.02435330e-01 1.34621710e-01
3.47782612e-01 2.53573447e-01 -1.35266945e-01 -1.00122439e-02
2.69836217e-01 -4.26598161e-01 1.29001546e+00 3.28853637e-01
1.69361636e-01 -7.67404318e-01 7.90482104e-01 6.56746566e-01
-1.29024088e+00 -1.45253181e-01 -6.56486630e-01 -1.07020788e-01
1.72060862e-01 4.68542546e-01 -1.69802392e+00 4.98286843e-01
9.51440334e-01 2.77108550e-01 -5.93122303e-01 1.76023793e+00
5.59497289e-02 -2.68163651e-01 -8.57591867e-01 -4.90808249e-01
-2.46894062e-02 -2.06574917e-01 7.92248130e-01 1.06177223e+00
6.05874121e-01 9.29563493e-02 7.31298685e-01 7.60347486e-01
4.36879039e-01 9.57273468e-02 -4.43200201e-01 5.28508008e-01
8.61524343e-01 1.26613545e+00 -1.36637044e+00 -1.40600041e-01
-1.34147912e-01 8.76920879e-01 8.59971792e-02 -2.38148242e-01
-5.93120396e-01 -2.36678496e-01 4.45915788e-01 1.42323986e-01
5.08864284e-01 -7.40792811e-01 -2.90436357e-01 -5.47635734e-01
1.09698594e-01 -5.01753211e-01 2.77302682e-01 -8.77639711e-01
-1.04145372e+00 3.90202254e-01 2.30444431e-01 -1.74046111e+00
-2.72901863e-01 -8.80639315e-01 -5.93076944e-02 6.77843869e-01
-1.18449008e+00 -1.07643080e+00 -5.62945127e-01 3.28810066e-01
7.10077584e-01 1.33198410e-01 5.45492470e-01 -3.09227947e-02
-9.13362280e-02 -3.64112295e-02 -1.10949211e-01 -1.28470808e-01
5.05855739e-01 -1.11322415e+00 1.80474639e-01 8.42176318e-01
6.92833886e-02 4.32689458e-01 9.26685691e-01 -1.07324624e+00
-1.76024783e+00 -8.92017007e-01 4.32736069e-01 -4.04052556e-01
2.03009412e-01 -4.60780889e-01 -6.07470572e-01 3.08106214e-01
-4.00308132e-01 2.16670539e-02 3.30244422e-01 -2.42170945e-01
-1.82411056e-02 3.32591683e-02 -1.32584453e+00 3.93798977e-01
1.00067389e+00 1.20821528e-01 -6.54241085e-01 3.60844344e-01
3.09476316e-01 -8.13550770e-01 -4.53951448e-01 7.51069725e-01
6.37093782e-01 -8.24970245e-01 1.14623809e+00 6.64164200e-02
-3.64422083e-01 -1.04406273e+00 -3.25499177e-01 -8.32823932e-01
-4.47138518e-01 -2.47544959e-01 1.42564505e-01 7.78027594e-01
4.71581481e-02 -1.56738479e-02 9.81333673e-01 4.09336388e-01
-3.99513662e-01 -5.22577018e-02 -1.19045973e+00 -1.08538473e+00
-7.24114537e-01 -4.89025325e-01 3.26294899e-01 3.38018268e-01
-1.96575686e-01 1.35978118e-01 2.46957317e-01 7.06161141e-01
9.44648862e-01 3.97538006e-01 1.25593936e+00 -1.62106895e+00
1.82816952e-01 -5.78877687e-01 -9.12557065e-01 -9.69914019e-01
-2.01791719e-01 -6.46694958e-01 7.53640413e-01 -1.55042863e+00
3.14594731e-02 -4.68190312e-01 4.64576781e-02 3.71194601e-01
3.56485769e-02 5.74686289e-01 3.36597502e-01 4.72058535e-01
-8.15948725e-01 2.20410869e-01 7.27790177e-01 1.32749096e-01
-3.43267411e-01 1.40947299e-02 9.91800874e-02 7.25629687e-01
6.91211522e-01 -5.49951792e-01 3.30438912e-01 -1.25683293e-01
-9.13831517e-02 -3.32703888e-01 5.57102621e-01 -1.50863457e+00
4.89692271e-01 -2.17807710e-01 6.94579899e-01 -1.58682263e+00
7.16640651e-01 -1.17705452e+00 2.10475445e-01 7.72023618e-01
2.28076205e-01 1.27415344e-01 3.04001212e-01 5.86679399e-01
1.73179895e-01 -4.72222894e-01 7.98747241e-01 -3.70640546e-01
-9.81767535e-01 -1.97413284e-02 -4.75278854e-01 -8.74523044e-01
1.41521430e+00 -7.42660522e-01 3.23038727e-01 9.59904864e-02
-4.65947181e-01 1.87703937e-01 6.43669188e-01 4.36113745e-01
9.72116828e-01 -1.21430111e+00 -5.99425614e-01 3.05472642e-01
4.12098289e-01 3.98055494e-01 -2.98034877e-01 6.38523161e-01
-6.23230577e-01 2.04745844e-01 -1.81687504e-01 -1.36634290e+00
-1.63991725e+00 5.37051618e-01 1.32707939e-01 4.11452234e-01
-3.89127016e-01 8.64686370e-01 -3.26815248e-01 -3.88143212e-01
4.04435754e-01 -4.29025084e-01 4.72623967e-02 -3.59560549e-02
3.82913351e-01 4.29063529e-01 3.68459105e-01 -8.01618099e-01
-8.38648915e-01 7.91827738e-01 5.29524535e-02 -2.84451157e-01
1.38336277e+00 -1.10689290e-01 -2.06774231e-02 3.48648041e-01
6.88170135e-01 -1.64952859e-01 -1.24096465e+00 -2.31483117e-01
3.66411120e-01 -8.42551053e-01 -7.96774849e-02 -5.50709188e-01
-3.66344780e-01 5.08369744e-01 1.02130270e+00 6.55568913e-02
8.08617890e-01 4.69192147e-01 9.80291963e-02 5.88668346e-01
7.99210429e-01 -9.95494604e-01 2.83777356e-01 5.97220540e-01
9.38780963e-01 -1.29471362e+00 3.33378255e-01 -6.51443183e-01
-1.51859552e-01 1.20768607e+00 6.50469780e-01 -3.47632617e-01
1.74032107e-01 4.25527632e-01 -5.00088520e-02 -2.23954603e-01
-7.12286457e-02 -7.07250476e-01 3.15751374e-01 7.26250827e-01
-1.95101738e-01 -1.06477380e-01 -8.14592652e-03 1.15690060e-01
-2.60982811e-01 -3.17957431e-01 1.95270985e-01 1.47740304e+00
-1.09577501e+00 -8.97291064e-01 -1.19531202e+00 3.26802999e-01
-8.24492127e-02 4.35263544e-01 -5.98468423e-01 9.03059304e-01
2.25663349e-01 1.05876410e+00 3.45882714e-01 -5.14367223e-01
7.17922866e-01 -1.92274526e-01 7.78066933e-01 -8.38580549e-01
-5.18861651e-01 7.14878738e-01 -2.68499702e-01 -6.53851748e-01
-5.69003284e-01 -1.09251142e+00 -1.51987934e+00 3.26001942e-01
-9.28609729e-01 -1.52031228e-01 1.15972137e+00 6.80295229e-01
3.04855436e-01 2.07453236e-01 3.37580174e-01 -1.56364977e+00
-4.27870303e-01 -8.28981817e-01 -3.52741331e-01 2.99410313e-01
1.11672357e-01 -1.19124687e+00 -1.10570766e-01 -5.14993854e-02] | [7.360309600830078, -2.3249313831329346] |
0491f76e-6864-4d27-9f3a-8ca710da005b | end-to-end-trainable-trident-person-search | null | null | http://openaccess.thecvf.com//content/ICCV2021/html/Han_End-to-End_Trainable_Trident_Person_Search_Network_Using_Adaptive_Gradient_Propagation_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Han_End-to-End_Trainable_Trident_Person_Search_Network_Using_Adaptive_Gradient_Propagation_ICCV_2021_paper.pdf | End-to-End Trainable Trident Person Search Network Using Adaptive Gradient Propagation | Person search suffers from the conflicting objectives of commonness and uniqueness between the person detection and re-identification tasks that make the end-to-end training of person search networks difficult. In this paper, we propose a trident network for person search that performs detection, re-identification, and part classification together. We also devise a novel end-to-end training method using adaptive gradient weighting that controls the flow of back-propagated gradients through the re-identification and part classification networks according to the quality of the person detection. The proposed method not only prevents the over-fitting but encourages to exploit fine-grained features by incorporating the part classification branch into the person search framework. Experimental results on the CUHK-SYSU and PRW datasets demonstrate that the proposed method achieves the best performance among the state-of-the-art end-to-end person search methods. | ['Jae-Young Sim', 'Kuhyeun Ko', 'Byeong-Ju Han'] | 2021-01-01 | null | null | null | iccv-2021-1 | ['person-search'] | ['computer-vision'] | [-2.35773072e-01 -4.78795171e-01 -3.86619046e-02 -5.25216162e-01
-2.88691670e-01 -3.47650319e-01 5.12176454e-01 -3.40831280e-01
-1.02837193e+00 4.96083826e-01 2.77384728e-01 2.63776898e-01
-2.73129046e-01 -4.49251354e-01 -2.99398929e-01 -4.20368612e-01
1.66760057e-01 7.46005774e-01 1.22685961e-01 4.99048978e-02
-1.39496595e-01 4.17587757e-01 -1.55935919e+00 -1.77899301e-01
8.53259087e-01 1.06902528e+00 -6.34359494e-02 5.72030365e-01
3.35187018e-01 1.90044791e-01 -5.84010541e-01 -8.08156490e-01
5.16654730e-01 -1.71235681e-01 -6.71201348e-01 1.15281574e-01
8.21853638e-01 -3.74681652e-01 -6.58718467e-01 1.00705278e+00
1.17974210e+00 5.90085089e-01 5.13591230e-01 -1.11238742e+00
-8.28070402e-01 2.48060673e-01 -5.36763728e-01 5.15425205e-01
3.53074700e-01 1.97040543e-01 7.06424654e-01 -9.40124929e-01
2.34365061e-01 1.43796337e+00 9.82206881e-01 1.07678497e+00
-1.02711177e+00 -8.50937963e-01 3.34260672e-01 3.65364254e-01
-1.73871744e+00 -4.57338810e-01 5.58667541e-01 -2.32387051e-01
7.62461543e-01 2.07468227e-01 7.94246256e-01 1.06604934e+00
-4.93615836e-01 1.03998029e+00 6.81975245e-01 -3.64021271e-01
-1.22954287e-01 2.20585167e-01 6.00278199e-01 1.04432571e+00
3.26422125e-01 5.09239614e-01 -5.47775269e-01 -4.31504488e-01
7.46918976e-01 1.01960383e-01 -2.37062067e-01 -2.33103096e-01
-8.61528754e-01 5.63343823e-01 6.28740430e-01 8.15860778e-02
-3.94894660e-01 9.31691080e-02 5.65981030e-01 -1.62657481e-02
3.47258151e-01 1.15045972e-01 -2.28119537e-01 9.91774648e-02
-1.15652394e+00 6.13705516e-01 6.38854146e-01 8.07861269e-01
2.78051883e-01 -2.37390861e-01 -9.35795665e-01 1.23548257e+00
3.65186095e-01 4.84982252e-01 7.39125550e-01 -5.61523080e-01
4.37374234e-01 6.58336163e-01 4.87152427e-01 -8.03812683e-01
-4.42369163e-01 -1.04028046e+00 -9.02802169e-01 -7.81056210e-02
6.01479411e-01 -1.48809776e-01 -1.25154388e+00 1.86857772e+00
4.25828606e-01 1.91398859e-01 -2.86747456e-01 1.37349629e+00
1.03569901e+00 1.16132908e-01 3.49363625e-01 2.52353787e-01
1.70198810e+00 -1.50305891e+00 -5.33314824e-01 -3.32509160e-01
2.87319720e-02 -3.38675678e-01 6.91414416e-01 -6.96785748e-02
-1.20098186e+00 -1.05566931e+00 -9.05602694e-01 1.43474080e-02
-3.12918007e-01 6.00369990e-01 4.31252807e-01 9.03583109e-01
-9.04336393e-01 3.50602657e-01 -4.61515516e-01 -4.55416918e-01
4.59311664e-01 5.58461845e-01 -1.71369150e-01 9.94382128e-02
-1.17160964e+00 6.36503160e-01 4.17227060e-01 4.38811153e-01
-5.00122070e-01 -4.99279797e-01 -5.30453384e-01 3.78813177e-01
3.75529051e-01 -1.22137725e+00 1.12828219e+00 -7.54275739e-01
-1.30815911e+00 1.04274118e+00 -2.47747108e-01 -4.25982058e-01
9.47539508e-01 -3.38845313e-01 -5.26368737e-01 -9.22705233e-02
2.64046192e-01 6.76937103e-01 7.84241915e-01 -9.28758562e-01
-1.05590093e+00 -6.48431301e-01 -3.23218524e-01 4.85982776e-01
-4.75243509e-01 2.45135456e-01 -1.04856431e+00 -7.14479089e-01
-2.26076186e-01 -8.04491520e-01 -3.12470011e-02 2.23172978e-01
-4.77175087e-01 -7.28397787e-01 4.98522192e-01 -7.85705030e-01
1.27725208e+00 -1.83892250e+00 1.07400663e-01 2.91868985e-01
1.90624043e-01 4.71128345e-01 -1.63074136e-01 -5.63686043e-02
1.21081278e-01 -2.36763313e-01 1.61734670e-01 -8.76701117e-01
2.86945671e-01 -3.15678954e-01 2.12758467e-01 6.15678549e-01
-4.65647876e-01 1.01524866e+00 -6.11471832e-01 -5.59107065e-01
1.09073967e-02 4.22757208e-01 -2.54951447e-01 2.26154625e-01
3.75277281e-01 1.32840186e-01 -4.71462220e-01 7.80244946e-01
6.19248033e-01 -1.78607121e-01 -4.46728021e-01 -1.21307023e-01
5.99526316e-02 -1.40363127e-01 -1.25274348e+00 1.54625785e+00
9.82772633e-02 1.48017898e-01 2.21379414e-01 -6.14399195e-01
6.76278114e-01 1.59301087e-01 1.30693987e-01 -7.36209929e-01
1.28844559e-01 2.29110554e-01 -9.93840098e-02 -2.63193339e-01
6.02188766e-01 2.89518088e-01 -8.67355168e-02 3.62576514e-01
-5.66069484e-02 9.05541003e-01 3.89572561e-01 -1.69786572e-01
5.30639112e-01 1.92945860e-02 5.86193353e-02 -2.37402469e-01
1.04825640e+00 -3.38273525e-01 5.68537235e-01 1.20750928e+00
-6.98154986e-01 3.60764265e-01 -4.31065440e-01 -7.04759538e-01
-9.58961666e-01 -8.36839795e-01 -1.07919537e-02 1.39749932e+00
3.66725326e-01 -4.88734022e-02 -9.80876982e-01 -8.57826889e-01
3.05474430e-01 2.01951042e-01 -6.19375765e-01 -1.96050391e-01
-6.18925631e-01 -8.44866157e-01 1.05379248e+00 6.24836802e-01
1.17711151e+00 -1.13896239e+00 -4.19353515e-01 1.02996878e-01
-3.72402966e-01 -1.01583350e+00 -1.23542130e+00 -2.94409424e-01
-3.12776417e-01 -9.68258917e-01 -1.56981111e+00 -1.14178371e+00
7.94056714e-01 2.57220387e-01 8.63919735e-01 2.87285060e-01
-3.62607747e-01 5.24304330e-01 -8.08315352e-02 -1.38089031e-01
1.83253095e-01 2.58754998e-01 3.61117125e-01 2.06620365e-01
6.50430560e-01 -1.48068696e-01 -1.00175047e+00 6.91838086e-01
-2.07411021e-01 -3.78425598e-01 3.29048842e-01 9.86782253e-01
5.86262763e-01 1.76833689e-01 2.76178390e-01 -1.54272765e-01
1.06975102e+00 6.77756295e-02 -5.40631413e-01 7.57045150e-01
-9.97331262e-01 -1.43711030e-01 2.67343909e-01 -6.68043673e-01
-1.03877354e+00 3.38836387e-02 -4.51701172e-02 -3.26150179e-01
-1.48855194e-01 -8.77760202e-02 3.59779485e-02 -4.39915031e-01
5.70287764e-01 7.00098276e-01 -1.94579393e-01 -6.69487834e-01
1.46881297e-01 7.11830080e-01 9.56162333e-01 -3.50227058e-01
8.91831398e-01 3.92605752e-01 -3.27783614e-01 -4.99368191e-01
-6.67697787e-01 -1.05804873e+00 -5.22531629e-01 -2.16210559e-01
8.92208338e-01 -1.10208690e+00 -1.16408587e+00 8.24800968e-01
-1.05545378e+00 2.54864823e-02 -1.04251884e-01 2.40229547e-01
-1.57571852e-01 6.52342141e-01 -5.76468349e-01 -1.00961161e+00
-1.19244289e+00 -8.38398576e-01 1.04734361e+00 7.91459918e-01
-1.49357736e-01 -6.81243539e-01 1.60182975e-02 5.15679657e-01
2.92257726e-01 -3.44503313e-01 2.46130139e-01 -6.43168271e-01
-2.81490058e-01 -6.68560445e-01 -5.41817307e-01 6.08897861e-03
-1.08022839e-01 -8.74834359e-01 -8.18290532e-01 -6.53416514e-01
-3.33668113e-01 -2.29856744e-01 1.23607159e+00 3.90739650e-01
1.05696344e+00 -3.38105649e-01 -5.91386437e-01 9.24173295e-01
1.21497059e+00 -7.26427212e-02 3.50153118e-01 5.07247567e-01
7.09119499e-01 4.73503441e-01 2.44732618e-01 3.77672970e-01
4.86887604e-01 1.05815196e+00 -1.61740243e-01 -9.95865092e-02
-3.74882847e-01 -3.79404604e-01 -9.11892653e-02 -1.52742252e-01
-3.80951345e-01 -3.42706650e-01 -5.45343935e-01 5.60643852e-01
-2.23979139e+00 -1.39429963e+00 2.81821996e-01 2.24583936e+00
6.56628311e-01 -2.83545554e-02 7.20757782e-01 -1.68628514e-01
1.06027567e+00 -1.44504313e-03 -6.19701326e-01 3.62645715e-01
-3.23246308e-02 -1.57844275e-01 6.47498846e-01 3.81959289e-01
-1.45617843e+00 1.00606728e+00 6.24462366e+00 9.80194688e-01
-5.90536475e-01 2.42439657e-01 4.72010761e-01 -1.78538173e-01
4.61270750e-01 -6.01366699e-01 -1.54392612e+00 6.32747412e-01
2.24758849e-01 1.48153543e-01 5.94166696e-01 9.39064682e-01
4.26715054e-02 1.27351463e-01 -9.85554278e-01 1.54723918e+00
3.32990706e-01 -8.98749530e-01 -1.35898650e-01 -1.30875900e-01
4.49397773e-01 -1.88833833e-01 9.11098570e-02 4.43758845e-01
3.33376266e-02 -7.85489738e-01 9.87201691e-01 7.12145925e-01
4.77258295e-01 -8.46684575e-01 6.78969085e-01 5.21590889e-01
-1.47862768e+00 -4.15716887e-01 -3.73462081e-01 1.78296894e-01
3.64570618e-01 8.98028612e-02 -5.95735550e-01 1.98214680e-01
8.92038822e-01 2.31154442e-01 -8.37410748e-01 1.55306160e+00
-9.13270190e-02 1.11019313e-01 -4.14593041e-01 -2.40488812e-01
3.09850648e-02 7.64532164e-02 7.40000904e-01 1.43749475e+00
3.94187160e-02 -6.12512752e-02 5.09463608e-01 1.07233512e+00
-1.32722601e-01 -1.99442595e-01 2.21324757e-01 2.23085672e-01
4.34460819e-01 1.18079174e+00 -5.74228942e-01 -4.40082729e-01
-9.57935303e-02 1.44235671e+00 4.59951401e-01 6.05817556e-01
-7.02608049e-01 -2.63665080e-01 5.43380976e-01 6.19505793e-02
3.74837399e-01 8.90620127e-02 -1.28379688e-01 -1.06225419e+00
2.76200235e-01 -7.64300525e-01 7.42725134e-01 -3.15011710e-01
-1.65172744e+00 5.63986719e-01 1.05539653e-02 -7.83499300e-01
-2.75511116e-01 -4.90821749e-01 -5.58108509e-01 1.20047915e+00
-1.43383241e+00 -1.41862988e+00 -5.65570176e-01 7.60635614e-01
5.58303893e-01 -6.78071439e-01 6.43542290e-01 6.69903040e-01
-7.91415513e-01 1.45351768e+00 -1.65169418e-01 5.88293016e-01
5.87353706e-01 -9.13870513e-01 5.91175735e-01 9.64920998e-01
-2.18132183e-01 8.13421607e-01 5.68674743e-01 -8.48709762e-01
-7.97049403e-01 -8.41563404e-01 1.05242944e+00 -1.81177691e-01
1.30916201e-02 -4.06044811e-01 -4.30216640e-01 2.88729817e-01
-2.17947766e-01 -5.94225852e-03 4.30279553e-01 4.57349390e-01
-3.94960105e-01 -1.39772147e-01 -1.32897758e+00 5.72603464e-01
1.46275973e+00 -5.11092067e-01 -3.68105471e-01 3.11136454e-01
3.02218705e-01 -3.96285981e-01 -3.35792333e-01 2.03725353e-01
9.05871153e-01 -7.26420164e-01 1.57935011e+00 -4.87898350e-01
-4.07362968e-01 -2.41184041e-01 2.15909615e-01 -9.76222217e-01
-9.98754561e-01 -5.08318603e-01 -2.68962264e-01 1.13916349e+00
1.92318097e-01 -5.92184126e-01 1.09012556e+00 1.02036536e+00
3.48285407e-01 -6.44274294e-01 -1.04641843e+00 -9.80131984e-01
-3.76493573e-01 1.49178565e-01 6.12598181e-01 3.25937480e-01
-2.80329525e-01 3.27138275e-01 -7.22410083e-01 1.79094106e-01
1.18776488e+00 -1.48766534e-02 6.28512800e-01 -1.45481074e+00
-5.86055815e-01 -8.49787116e-01 -3.88787836e-01 -1.30016398e+00
1.21002518e-01 -7.97466695e-01 5.24417125e-02 -1.35216331e+00
6.91644073e-01 -3.71206015e-01 -3.53908002e-01 4.45511967e-01
-6.11366212e-01 1.54443681e-01 2.85894364e-01 5.56745231e-01
-8.67347538e-01 6.57707453e-01 1.04616821e+00 -3.09312046e-01
-3.83624136e-01 6.07290506e-01 -6.45475686e-01 3.80241096e-01
4.63666648e-01 -1.49713457e-01 -1.25507802e-01 -4.53862727e-01
-2.97077566e-01 -3.29440594e-01 8.19431365e-01 -1.21636975e+00
7.22444654e-01 4.28040564e-01 9.83889699e-01 -7.72819161e-01
6.50778711e-01 -5.96447527e-01 -2.13185623e-01 6.18661940e-01
-4.07111824e-01 8.62845629e-02 2.63237823e-02 7.83999324e-01
2.13128217e-02 -2.63642669e-01 8.55467081e-01 -3.61014724e-01
-7.50660717e-01 7.34229386e-01 1.17923610e-01 -4.14760262e-02
6.47051692e-01 -5.82059324e-01 -1.37423113e-01 -3.42517793e-01
-7.01230347e-01 6.69863880e-01 1.16319858e-01 4.73426819e-01
5.85347950e-01 -1.40097976e+00 -7.31733203e-01 1.96510330e-01
-1.16049089e-02 -4.12156671e-01 3.36405069e-01 4.65904117e-01
-4.68987040e-02 6.19826198e-01 2.06771772e-02 -1.94577783e-01
-1.66754723e+00 5.07061124e-01 8.79512548e-01 -5.22621334e-01
-4.27746207e-01 1.34685934e+00 6.80172443e-02 -5.79885423e-01
9.89919364e-01 4.24611419e-01 -3.77422869e-01 -1.09382130e-01
7.48012125e-01 6.76718652e-01 -4.15523261e-01 -8.58133614e-01
-4.99588728e-01 6.73381269e-01 -4.51400071e-01 -1.10925250e-01
9.11948144e-01 -3.20202440e-01 2.37110347e-01 -4.14683312e-01
9.31043148e-01 -4.18451428e-01 -1.10991478e+00 -4.94320333e-01
-2.43921220e-01 -3.94746870e-01 -5.65833002e-02 -1.21677709e+00
-9.02018249e-01 3.07199806e-01 1.22103965e+00 -1.65341064e-01
8.67418766e-01 -1.19065478e-01 1.06823719e+00 5.46976864e-01
3.17539990e-01 -1.50258243e+00 -4.93068323e-02 4.62444514e-01
7.07751989e-01 -1.23467565e+00 1.70631349e-01 -2.84585446e-01
-2.78882533e-01 7.22153366e-01 8.35767090e-01 -8.22996423e-02
4.62016791e-01 -4.33968723e-01 -1.70645878e-01 -1.36304915e-01
-1.16117284e-01 -4.69648957e-01 8.22140515e-01 6.67910695e-01
-1.00720949e-01 4.69790027e-02 -4.68436778e-01 8.09220374e-01
-1.09806083e-01 2.02136934e-01 -5.06241679e-01 5.36722720e-01
-4.13567007e-01 -9.91206169e-01 -6.35296941e-01 2.53734916e-01
-4.22718018e-01 -2.94546019e-02 -6.29242182e-01 5.27029872e-01
4.53602821e-01 8.75241756e-01 -1.11227244e-01 -2.53721297e-01
4.06423956e-01 1.73003212e-01 4.19477165e-01 -3.94407660e-02
-9.06957984e-01 -4.29187864e-02 7.33093992e-02 -3.79332513e-01
-2.93078750e-01 -5.28118432e-01 -9.39831734e-01 -3.09686273e-01
-3.55356723e-01 1.33194387e-01 3.85094613e-01 8.88973296e-01
4.34748173e-01 2.83219874e-01 2.40149111e-01 -8.94875765e-01
-9.50981677e-01 -1.11075592e+00 -4.38144058e-01 5.96698046e-01
3.87495846e-01 -7.07774758e-01 -1.43020689e-01 -3.37872833e-01] | [14.825587272644043, 0.8201420903205872] |
ae71f43f-62a5-450f-9640-b91838e1a520 | learning-to-quantize-vulnerability-patterns | 2306.06109 | null | https://arxiv.org/abs/2306.06109v1 | https://arxiv.org/pdf/2306.06109v1.pdf | Learning to Quantize Vulnerability Patterns and Match to Locate Statement-Level Vulnerabilities | Deep learning (DL) models have become increasingly popular in identifying software vulnerabilities. Prior studies found that vulnerabilities across different vulnerable programs may exhibit similar vulnerable scopes, implicitly forming discernible vulnerability patterns that can be learned by DL models through supervised training. However, vulnerable scopes still manifest in various spatial locations and formats within a program, posing challenges for models to accurately identify vulnerable statements. Despite this challenge, state-of-the-art vulnerability detection approaches fail to exploit the vulnerability patterns that arise in vulnerable programs. To take full advantage of vulnerability patterns and unleash the ability of DL models, we propose a novel vulnerability-matching approach in this paper, drawing inspiration from program analysis tools that locate vulnerabilities based on pre-defined patterns. Specifically, a vulnerability codebook is learned, which consists of quantized vectors representing various vulnerability patterns. During inference, the codebook is iterated to match all learned patterns and predict the presence of potential vulnerabilities within a given program. Our approach was extensively evaluated on a real-world dataset comprising more than 188,000 C/C++ functions. The evaluation results show that our approach achieves an F1-score of 94% (6% higher than the previous best) and 82% (19% higher than the previous best) for function and statement-level vulnerability identification, respectively. These substantial enhancements highlight the effectiveness of our approach to identifying vulnerabilities. The training code and pre-trained models are available at https://github.com/optimatch/optimatch. | ['Dinh Phung', 'Chakkrit Tantithamthavorn', 'Van Nguyen', 'Trung Le', 'Michael Fu'] | 2023-05-26 | null | null | null | null | ['vulnerability-detection'] | ['miscellaneous'] | [-9.12757665e-02 -2.45212689e-01 -5.18756270e-01 -2.99471349e-01
-1.00164175e+00 -1.01346540e+00 8.09256956e-02 6.60258651e-01
2.09817767e-01 1.57338515e-01 2.00419962e-01 -8.81557882e-01
1.71064198e-01 -9.76451933e-01 -7.17063785e-01 -2.21174657e-01
-4.84151095e-01 -4.97522593e-01 3.32520545e-01 -8.17573443e-02
6.62636220e-01 2.91697592e-01 -1.48858666e+00 6.13229930e-01
8.45521629e-01 6.93595588e-01 -8.71571228e-02 6.07363641e-01
-3.64232391e-01 7.83888996e-01 -7.13393033e-01 -4.08520073e-01
1.29001662e-01 2.22615108e-01 -9.18311000e-01 -8.03284407e-01
3.73667836e-01 -3.39489460e-01 -2.37931356e-01 1.42067146e+00
1.29846334e-01 -3.85538936e-01 2.44945839e-01 -1.23106778e+00
-6.90084100e-01 9.69219863e-01 -6.40422761e-01 4.45740253e-01
5.13435662e-01 1.62296906e-01 1.32513368e+00 -7.00366735e-01
2.82281190e-01 1.02934074e+00 1.01382208e+00 4.04170454e-01
-9.69300926e-01 -6.94564581e-01 6.12986609e-02 4.05798070e-02
-1.37534285e+00 -1.70307547e-01 7.92604446e-01 -6.97010994e-01
1.42637908e+00 3.11875105e-01 1.56478405e-01 9.41327870e-01
5.62626004e-01 4.88293350e-01 5.50320268e-01 -4.10545915e-01
2.00073585e-01 -5.86061887e-02 6.08278811e-01 6.38011217e-01
2.29844302e-01 5.83935902e-02 -8.23255107e-02 -8.81108224e-01
1.70743942e-01 2.64938653e-01 -1.41800150e-01 2.30195615e-02
-8.34277093e-01 8.94416451e-01 5.64650059e-01 4.17475045e-01
-1.16454124e-01 8.21730345e-02 8.45162094e-01 3.36654902e-01
1.63393140e-01 5.25462151e-01 -5.08181036e-01 -3.85488987e-01
-9.18323100e-01 2.46429160e-01 6.78097785e-01 7.65145838e-01
1.03188586e+00 1.52975366e-01 -4.16467041e-02 4.34670031e-01
3.47801536e-01 1.94925040e-01 3.21612865e-01 -6.30642116e-01
7.42682517e-01 1.27981818e+00 -3.18480819e-01 -1.34877121e+00
-8.15827772e-02 -2.22838446e-01 -4.60408926e-01 2.24345952e-01
1.01288192e-01 -1.06012248e-01 -5.55783927e-01 1.64621341e+00
4.26901467e-02 1.74214482e-01 1.51367769e-01 1.69360042e-01
5.62266290e-01 5.84353328e-01 1.56169400e-01 2.77442604e-01
1.03301191e+00 -3.22813541e-01 -2.26485148e-01 -2.12902233e-01
9.25277650e-01 -6.23077214e-01 1.27263856e+00 1.86429635e-01
-6.07554436e-01 -4.54150707e-01 -1.13660431e+00 3.07813048e-01
-4.73141760e-01 -1.88745826e-01 8.21598649e-01 1.11311245e+00
-1.12707031e+00 5.14046252e-01 -8.70789707e-01 -4.95614521e-02
4.79281038e-01 2.01639414e-01 -2.10423693e-01 3.00699659e-02
-1.14640164e+00 3.24816614e-01 5.23087204e-01 -2.13671267e-01
-1.20803154e+00 -1.10369873e+00 -1.00729871e+00 3.56943667e-01
3.86403590e-01 1.72585011e-01 9.67695296e-01 -9.08333242e-01
-5.96424162e-01 5.35010040e-01 -1.26640499e-01 -1.86842620e-01
-1.83702141e-01 -2.65068680e-01 -7.16931105e-01 -9.13794935e-02
9.89727452e-02 -1.33315427e-02 4.26209301e-01 -1.13566017e+00
-5.65163910e-01 -1.08718194e-01 6.12690628e-01 -5.93585372e-01
-9.52499926e-01 4.54838306e-01 -2.55373776e-01 -4.89306301e-01
-3.19891542e-01 -6.08440220e-01 -1.70049429e-01 -3.17212343e-01
-6.72362864e-01 -1.21658303e-01 8.35738778e-01 -8.51208568e-01
2.05982184e+00 -2.15772009e+00 -1.96799576e-01 4.72502410e-01
4.09438670e-01 4.26143438e-01 -2.89071679e-01 6.17189825e-01
-3.86702925e-01 5.43814182e-01 -5.47491193e-01 1.06831312e-01
1.22766338e-01 -1.05861969e-01 -8.15908313e-01 2.41204396e-01
2.56063014e-01 7.47099221e-01 -8.95267606e-01 -2.36295387e-01
-1.34394810e-01 2.90568829e-01 -8.24361384e-01 2.77261019e-01
-2.77311206e-01 -2.99279153e-01 -6.13951087e-01 1.14832854e+00
6.41489804e-01 -8.83288980e-02 3.66563797e-01 2.05943465e-01
-1.48130417e-01 3.46909791e-01 -9.08725858e-01 1.32315361e+00
-5.31743944e-01 5.48358500e-01 -1.44385532e-01 -7.97841072e-01
9.92734015e-01 3.40560466e-01 1.78433046e-01 -4.96956110e-01
-2.61806130e-01 2.45310724e-01 -5.34164459e-02 -4.98564154e-01
4.75866646e-01 4.57229167e-01 -6.17924571e-01 6.00780725e-01
-3.20091486e-01 4.05737132e-01 -8.70476365e-02 2.52346128e-01
1.63414931e+00 -2.08703130e-01 3.03858906e-01 -4.32085186e-01
7.24719644e-01 1.99183449e-02 7.84766972e-01 5.16617596e-01
-2.03919068e-01 2.33230591e-01 1.15920734e+00 -6.55883372e-01
-8.50768626e-01 -9.49985802e-01 -7.32596144e-02 9.20679688e-01
-3.24315906e-01 -8.21781754e-01 -9.59996819e-01 -1.15646613e+00
7.12024570e-02 5.76477766e-01 -7.75801361e-01 -3.65055740e-01
-8.18693340e-01 -7.51058161e-01 9.73666668e-01 8.78499746e-01
1.23983137e-01 -9.89000201e-01 -7.23339140e-01 4.87957112e-02
-5.14039993e-02 -5.47590613e-01 -4.66644704e-01 -8.18701927e-03
-6.29713118e-01 -1.33760142e+00 -1.08165272e-01 -5.87743521e-01
6.34525776e-01 3.65981483e-03 1.27274239e+00 7.00545311e-01
-4.05205637e-01 8.95143673e-02 -3.90130073e-01 -2.14566309e-02
-7.89576888e-01 1.05119511e-01 -1.48136586e-01 -4.29798245e-01
5.20097733e-01 -5.89165330e-01 -3.26878816e-01 1.50324389e-01
-1.03344476e+00 -7.72906184e-01 3.56033325e-01 6.45554066e-01
4.25617576e-01 3.46060604e-01 5.50388575e-01 -8.32182169e-01
7.05476582e-01 -9.54445839e-01 -9.38749135e-01 3.71011823e-01
-4.94190872e-01 -1.86870806e-02 7.61788547e-01 -3.83220106e-01
-8.31918836e-01 -6.65404946e-02 -3.73452157e-01 -3.56854945e-01
-6.93552271e-02 9.39779580e-01 -5.27157426e-01 -2.33534589e-01
6.94176495e-01 2.00040098e-02 -3.27731788e-01 -4.34558630e-01
9.01878253e-02 6.21831000e-01 3.78254741e-01 -8.75550747e-01
8.90552461e-01 -5.03597669e-02 -4.92227495e-01 -2.79923528e-01
-3.22321415e-01 -1.93828419e-01 -5.60005307e-01 1.11647554e-01
5.72410882e-01 -5.76822996e-01 -5.77059686e-01 4.23839867e-01
-1.00090671e+00 -3.28864932e-01 2.28143170e-01 -2.66032368e-01
9.36174020e-02 7.88582504e-01 -5.58832705e-01 -6.35233760e-01
-5.18846154e-01 -1.49473763e+00 6.77373827e-01 7.36810341e-02
-5.44777274e-01 -1.15939498e+00 3.24881941e-01 -1.06396034e-01
5.93229711e-01 5.19473612e-01 1.76499295e+00 -6.20315611e-01
-5.21154761e-01 -3.75010759e-01 -1.51794061e-01 2.44581789e-01
2.82422423e-01 4.33978796e-01 -9.92178559e-01 -4.11168426e-01
-2.36537665e-01 -1.02194071e-01 5.87607801e-01 -7.59623274e-02
1.48431361e+00 -3.60399157e-01 -5.56153238e-01 6.54585779e-01
1.75308919e+00 3.64839852e-01 6.57423079e-01 5.50248981e-01
8.95044327e-01 6.09884739e-01 3.64659876e-01 4.22827095e-01
3.42955142e-01 5.11809170e-01 9.09774005e-01 2.24327520e-01
2.97413260e-01 -1.97943226e-01 6.51481807e-01 4.56694961e-01
4.01076376e-01 5.33471517e-02 -1.72965741e+00 9.00523782e-01
-1.40826583e+00 -9.55923140e-01 -6.48401827e-02 2.33949375e+00
9.89429116e-01 3.47310424e-01 2.35174093e-02 2.61556774e-01
5.12655258e-01 2.46583268e-01 -6.17622375e-01 -7.88057208e-01
1.51754245e-01 2.47629374e-01 2.77801454e-01 4.59609032e-01
-1.11968184e+00 8.24046016e-01 6.52547026e+00 7.11984158e-01
-1.29156911e+00 -1.16128817e-01 5.28176963e-01 1.49714634e-01
-8.98184717e-01 1.88213632e-01 -9.37260389e-01 4.85255569e-01
1.22607660e+00 -4.00434792e-01 9.52298641e-02 1.10764372e+00
-3.15844953e-01 2.33919144e-01 -1.10689390e+00 3.67675155e-01
-1.29348874e-01 -1.41342175e+00 -1.34626776e-01 1.33625910e-01
6.33244634e-01 -6.95543066e-02 3.50281328e-01 5.58711410e-01
3.72335196e-01 -1.08994079e+00 6.57474875e-01 2.98697352e-01
7.01421380e-01 -1.08013594e+00 7.46768773e-01 1.08046263e-01
-1.50395393e+00 -6.76882625e-01 -3.44509602e-01 5.27828783e-02
-3.98508579e-01 7.47388899e-01 -6.97220981e-01 5.14416456e-01
9.40288484e-01 7.41262496e-01 -1.07204485e+00 7.43088007e-01
-1.21814184e-01 8.57923627e-01 7.33302012e-02 1.55321613e-01
2.19568580e-01 5.34734368e-01 3.18341106e-01 1.34099483e+00
4.10877258e-01 -4.06775087e-01 6.94447756e-02 1.32268643e+00
1.45150190e-02 -1.18737526e-01 -5.93734503e-01 -1.34870037e-01
7.78550208e-01 1.08238614e+00 -5.01613379e-01 -1.75212085e-01
-6.66082025e-01 3.59944254e-01 3.76188785e-01 2.53329903e-01
-9.54105496e-01 -7.12040186e-01 1.11361539e+00 -2.78600063e-02
1.63242459e-01 -2.22964678e-02 -4.02915865e-01 -1.04319763e+00
3.59500587e-01 -1.30205393e+00 4.86543953e-01 -2.31626369e-02
-1.17681360e+00 9.20533359e-01 6.51465654e-02 -1.08729482e+00
-2.86263585e-01 -4.58738059e-01 -1.24485326e+00 1.12159562e+00
-1.35716319e+00 -9.94877875e-01 -1.05335131e-01 3.43108088e-01
2.18398601e-01 -3.60868484e-01 9.71159875e-01 2.95927048e-01
-8.85860026e-01 1.20834219e+00 5.87545782e-02 3.75656128e-01
4.71616834e-01 -1.26601553e+00 1.11811280e+00 1.37933064e+00
-2.93301195e-01 1.31508243e+00 2.36568138e-01 -9.21615124e-01
-1.51189339e+00 -1.27226222e+00 8.68113697e-01 -8.11750710e-01
9.61371779e-01 -3.83813173e-01 -1.55632210e+00 6.98869348e-01
9.98257287e-03 -5.28934353e-04 9.89355683e-01 1.94678038e-01
-1.06071627e+00 9.49125662e-02 -1.15099800e+00 5.70737004e-01
6.53046787e-01 -9.56141829e-01 -4.16554958e-01 -1.23025097e-01
7.46666014e-01 -2.15696558e-01 -1.05002844e+00 3.36715132e-01
4.70836639e-01 -1.12722623e+00 1.19188106e+00 -3.99659187e-01
7.15696156e-01 -1.08670153e-01 -5.16144991e-01 -8.40265214e-01
-1.96663097e-01 -4.08025056e-01 -2.51168191e-01 1.49664342e+00
5.80081880e-01 -7.24842489e-01 6.38240159e-01 7.33837187e-01
-1.72319725e-01 -8.74601185e-01 -7.20554173e-01 -5.45200527e-01
5.45038939e-01 -6.10915780e-01 1.22217023e+00 1.26006722e+00
3.60568643e-01 -4.45741653e-01 1.45483002e-01 6.59088612e-01
6.74709499e-01 2.43290469e-01 4.04152662e-01 -1.01547313e+00
-4.16086882e-01 -6.68508053e-01 -3.45156491e-01 -2.73072243e-01
6.82795584e-01 -9.54008579e-01 -2.48117000e-01 -9.39735532e-01
3.39149475e-01 -6.51942194e-01 -4.99280870e-01 1.14832675e+00
-5.93681216e-01 -9.06301811e-02 -7.76583105e-02 1.43155679e-01
-1.48199484e-01 -3.42024826e-02 3.18637013e-01 -4.12606508e-01
-1.39685243e-01 2.48998543e-03 -9.39959764e-01 7.00974643e-01
9.74771023e-01 -4.95998979e-01 -3.35541010e-01 -5.51522017e-01
4.39260453e-01 5.24883866e-02 4.08436298e-01 -1.09559953e+00
-1.45407105e-02 -1.28845081e-01 7.33221695e-02 -4.05165374e-01
-3.23038220e-01 -5.07894993e-01 9.11343321e-02 5.80311477e-01
-3.83358806e-01 4.61197644e-01 8.16139877e-01 2.32386217e-01
-2.30137974e-01 -5.06985545e-01 3.53364527e-01 3.03990487e-02
-1.13481462e+00 2.42084309e-01 -3.55779201e-01 -4.92825024e-02
1.01367342e+00 -7.88153484e-02 -4.55386639e-01 1.64560050e-01
-1.63366124e-01 9.66549367e-02 7.98552334e-01 7.21938610e-01
7.86611199e-01 -1.16505873e+00 -4.09202576e-01 4.95475680e-01
4.02235657e-01 -3.93774390e-01 4.07147050e-01 2.11768955e-01
-3.52408201e-01 3.63524079e-01 -3.07554692e-01 -2.20622733e-01
-1.40898228e+00 8.08258951e-01 3.17133486e-01 -3.33436608e-01
-5.29823959e-01 8.04499567e-01 1.44923165e-01 -6.29865408e-01
1.90464541e-01 -5.15116870e-01 -2.28248522e-01 -2.88966507e-01
8.73373806e-01 2.97993034e-01 9.64000002e-02 -4.94892895e-01
-6.37112737e-01 6.18079960e-01 -3.03426504e-01 4.67230678e-01
1.19319403e+00 4.38625574e-01 -4.77683634e-01 4.08318490e-02
1.48937321e+00 2.40787968e-01 -1.05424082e+00 -1.18155614e-01
3.60918969e-01 -5.97785473e-01 -1.29881306e-02 -7.90537298e-01
-1.17709982e+00 1.12373340e+00 5.01104295e-01 2.76743323e-01
1.26069748e+00 -1.42855585e-01 9.05636847e-01 1.66691214e-01
5.43713570e-01 -2.45116636e-01 1.26461247e-02 5.70721924e-01
5.08509398e-01 -1.04155052e+00 -3.68993282e-01 -2.69902706e-01
-2.19534487e-01 1.14000392e+00 7.37046659e-01 -4.58989218e-02
6.00705683e-01 7.71037102e-01 2.24424303e-02 -7.91208223e-02
-6.56123102e-01 3.99936318e-01 1.32178515e-01 7.38231242e-01
6.19429052e-01 1.27773002e-01 1.53872177e-01 8.20875347e-01
7.34708086e-02 -4.44799244e-01 6.10263705e-01 1.22319150e+00
-4.42920029e-01 -1.47002578e+00 -5.42982101e-01 3.44030321e-01
-8.59895527e-01 -4.09595758e-01 -2.78555840e-01 4.78058964e-01
-8.27426929e-03 9.01396811e-01 -1.77658752e-01 -8.10355842e-01
3.76238257e-01 -3.25775407e-02 -1.84467033e-01 -8.15387309e-01
-1.14952075e+00 -4.72934067e-01 -2.34085739e-01 -8.98152471e-01
3.83349180e-01 -5.69084585e-01 -1.25034511e+00 -4.77306753e-01
9.13687125e-02 2.02672631e-01 1.76582679e-01 4.39786792e-01
5.25944293e-01 6.33618653e-01 6.62733853e-01 -5.06924570e-01
-7.31418610e-01 -5.04793346e-01 -1.92010120e-01 1.67164460e-01
5.20216763e-01 -4.10375834e-01 -4.75130290e-01 1.26458276e-02] | [7.0875701904296875, 7.773055553436279] |
dc4d4d19-4014-41f3-ae39-d24a5f3ff6b6 | neural-fine-gray-monotonic-neural-networks | 2305.06703 | null | https://arxiv.org/abs/2305.06703v1 | https://arxiv.org/pdf/2305.06703v1.pdf | Neural Fine-Gray: Monotonic neural networks for competing risks | Time-to-event modelling, known as survival analysis, differs from standard regression as it addresses censoring in patients who do not experience the event of interest. Despite competitive performances in tackling this problem, machine learning methods often ignore other competing risks that preclude the event of interest. This practice biases the survival estimation. Extensions to address this challenge often rely on parametric assumptions or numerical estimations leading to sub-optimal survival approximations. This paper leverages constrained monotonic neural networks to model each competing survival distribution. This modelling choice ensures the exact likelihood maximisation at a reduced computational cost by using automatic differentiation. The effectiveness of the solution is demonstrated on one synthetic and three medical datasets. Finally, we discuss the implications of considering competing risks when developing risk scores for medical practice. | ['Jessica Barrett', 'Brian Tom', 'Chang Ho Yoon', 'Vincent Jeanselme'] | 2023-05-11 | null | null | null | null | ['survival-analysis'] | ['miscellaneous'] | [ 3.34005266e-01 1.62708610e-01 -6.34672165e-01 -5.61451972e-01
-9.31119204e-01 -1.51898533e-01 3.92446250e-01 4.95584637e-01
-7.42442131e-01 1.16812396e+00 3.85880649e-01 -8.48790705e-01
-5.54857314e-01 -5.74084222e-01 -1.95264176e-01 -6.92249954e-01
-4.11043644e-01 5.72671056e-01 -3.87117088e-01 2.63941735e-01
2.19902292e-01 5.11333704e-01 -1.06620562e+00 -1.23718858e-01
7.98590541e-01 7.69802868e-01 -4.83882517e-01 7.72759259e-01
1.83644742e-01 8.26622963e-01 -5.40426850e-01 -3.46113771e-01
-3.13418508e-02 -4.07124043e-01 -3.23618501e-01 -4.05704260e-01
-5.08615784e-02 -5.89688957e-01 8.31566900e-02 4.35638040e-01
6.32563949e-01 -2.85105612e-02 1.10023165e+00 -1.32078838e+00
-3.16249371e-01 3.26390475e-01 -1.62457854e-01 1.16729759e-01
1.34729475e-01 -1.30382359e-01 6.76087916e-01 -3.95499974e-01
2.25146174e-01 9.18795705e-01 1.14642406e+00 5.39060533e-01
-1.39824617e+00 -3.88316423e-01 -3.62252623e-01 -2.98962921e-01
-1.17704964e+00 -3.81280690e-01 4.36130702e-01 -5.03249049e-01
7.47193396e-01 4.72257167e-01 5.79267442e-01 1.11246705e+00
8.90770316e-01 3.26887399e-01 1.19303823e+00 -3.63620728e-01
2.72613764e-01 -1.24681756e-01 2.59205550e-01 -1.50852967e-02
4.03320670e-01 9.32946920e-01 7.61675611e-02 -7.15066075e-01
7.81018674e-01 2.83873618e-01 -5.12069976e-03 -2.29628697e-01
-9.79395926e-01 1.13947713e+00 -6.98475838e-02 -7.83414021e-02
-5.41995883e-01 2.73198843e-01 4.05965626e-01 3.00035417e-01
4.94325787e-01 8.18053782e-02 -6.34731889e-01 -8.89268294e-02
-1.41506314e+00 2.03938633e-01 1.19379485e+00 3.51558656e-01
-1.56376228e-01 7.20890239e-02 -3.60203087e-01 5.14595270e-01
3.54818642e-01 -1.92207359e-02 2.60092854e-01 -1.03843045e+00
-9.65737626e-02 4.06534135e-01 3.86692017e-01 -4.18094665e-01
-9.38917875e-01 -9.42467511e-01 -1.05979788e+00 6.22191668e-01
9.49974120e-01 -4.35360372e-01 -9.33860242e-01 1.69340026e+00
2.39403814e-01 -1.35307416e-01 2.28412524e-01 4.90366727e-01
1.61933631e-01 3.15962642e-01 6.26090229e-01 -6.99467599e-01
1.10209858e+00 -2.81002671e-01 -6.63123906e-01 -1.81147773e-02
9.72321928e-01 -3.34836274e-01 5.05725741e-01 4.91162419e-01
-1.22473085e+00 2.47479111e-01 -7.77432323e-01 1.51198342e-01
-4.55223024e-02 -2.48643965e-01 6.67281747e-01 9.32890773e-01
-7.62340963e-01 8.99643660e-01 -8.19430053e-01 -2.57319868e-01
4.94784087e-01 6.07665122e-01 3.53533067e-02 4.01551425e-01
-1.31579041e+00 1.16199505e+00 3.10962409e-01 8.80545229e-02
-5.61114132e-01 -9.99499142e-01 -7.15196609e-01 2.10466430e-01
2.22823501e-01 -1.13134646e+00 1.14727294e+00 -1.27626002e+00
-1.09828913e+00 7.08611012e-01 7.45760575e-02 -6.64286673e-01
1.02113247e+00 -9.25940424e-02 -3.89517277e-01 -3.35447758e-01
-1.07845455e-01 2.20911369e-01 3.67358655e-01 -6.84275866e-01
-6.46270335e-01 -1.81370959e-01 -4.16824043e-01 2.91199591e-02
2.90033873e-02 3.75906140e-01 1.52050391e-01 -9.32863295e-01
-3.10639113e-01 -5.99850237e-01 -6.41869068e-01 -8.78320113e-02
-1.31524608e-01 -4.47429381e-02 -3.27789038e-02 -7.95663595e-01
1.51557279e+00 -1.92962253e+00 -1.13306567e-01 1.29990950e-01
9.19100828e-03 -3.36133093e-01 4.22676772e-01 4.81235087e-01
-3.68833870e-01 1.14001840e-01 -5.09328604e-01 -3.23492616e-01
-2.47102872e-01 6.08895496e-02 3.21644098e-02 7.35545397e-01
1.35974184e-01 8.63677979e-01 -5.42469978e-01 -7.24776804e-01
1.75965920e-01 7.15278506e-01 -4.63014901e-01 6.80075362e-02
3.00371557e-01 4.57281619e-01 -9.93579328e-02 6.03140414e-01
3.12852710e-01 -3.81887734e-01 2.10317373e-01 5.60442507e-01
-1.12379663e-01 9.43234414e-02 -7.73513973e-01 1.06123161e+00
-3.00588846e-01 1.55973718e-01 -1.89559788e-01 -1.01121628e+00
6.42356336e-01 6.04165554e-01 5.12147963e-01 -2.18441129e-01
3.23813409e-01 3.66906852e-01 1.30569607e-01 -8.02339911e-02
-5.37564568e-02 -1.10180008e+00 -2.68021107e-01 2.16841921e-01
-2.36359745e-01 3.36907387e-01 -3.18541288e-01 -2.81933069e-01
8.82163644e-01 2.20037043e-01 8.70992601e-01 -3.88122112e-01
2.59773999e-01 6.40817881e-02 7.91432679e-01 8.38670909e-01
-3.70235085e-01 7.41097629e-01 9.35437381e-01 -5.25019705e-01
-8.09355319e-01 -1.17752945e+00 -7.53407121e-01 6.37770593e-01
-4.52068955e-01 3.28425378e-01 -3.72816473e-01 -7.29346156e-01
1.51209757e-01 1.41231847e+00 -9.00855064e-01 -2.58438826e-01
-4.26431984e-01 -1.41425395e+00 6.37989998e-01 5.16468704e-01
-5.20098090e-01 -9.82981563e-01 -8.84927213e-01 5.28433383e-01
1.59433633e-01 -2.01674640e-01 -2.67640322e-01 5.54982722e-01
-1.34355783e+00 -1.10856271e+00 -1.23737454e+00 -3.12448025e-01
4.34635073e-01 -6.34392381e-01 1.24255097e+00 1.41039237e-01
-1.38276398e-01 2.20108375e-01 1.74555331e-01 -4.55383837e-01
-6.73248947e-01 -1.86335593e-01 -1.71142802e-01 -3.05603325e-01
4.82568920e-01 -5.10060370e-01 -8.63356471e-01 1.05472237e-01
-8.59447181e-01 -1.95080623e-01 5.65977216e-01 1.07688940e+00
4.40646261e-01 -1.97858065e-01 1.08377922e+00 -9.08467412e-01
6.32944942e-01 -9.44442868e-01 -5.93923330e-01 1.39362454e-01
-1.10090005e+00 -8.02042931e-02 5.31804979e-01 -6.25433803e-01
-9.62695360e-01 4.74947393e-02 3.86447422e-02 -2.52382811e-02
-3.65791023e-01 8.08474720e-01 9.08463746e-02 3.68845463e-01
4.52829748e-01 -6.37423173e-02 1.19956374e-01 -2.52992213e-01
-1.91238850e-01 4.12360638e-01 2.90439934e-01 -1.20481692e-01
1.57961950e-01 3.61777723e-01 4.52701956e-01 -4.22918379e-01
-6.17445827e-01 -1.72706798e-01 -3.35865110e-01 -1.18103579e-01
7.16347516e-01 -7.20660329e-01 -9.39791799e-01 2.25565895e-01
-7.90867031e-01 -3.03111970e-01 -4.65217650e-01 7.70935953e-01
-9.95264173e-01 1.90350533e-01 -4.37066972e-01 -1.37574732e+00
-2.73024470e-01 -6.77892089e-01 5.17415226e-01 1.32012427e-01
-7.46401548e-01 -1.41926885e+00 2.54919052e-01 5.83578013e-02
4.27346379e-01 8.80416870e-01 1.10734463e+00 -8.95426750e-01
2.19990481e-02 -7.75843620e-01 3.73517238e-02 6.48684893e-03
1.17903516e-01 -1.09366097e-01 -6.31971002e-01 -3.13868076e-01
1.54773086e-01 6.20120317e-02 7.19388843e-01 1.20902193e+00
7.87945569e-01 -4.03916538e-01 -4.35516953e-01 6.52243435e-01
1.38519645e+00 5.08070946e-01 5.71067870e-01 4.37006325e-01
1.64650843e-01 1.28104639e+00 5.50196052e-01 4.35268641e-01
4.24540460e-01 3.80066812e-01 3.57586205e-01 -2.49905929e-01
4.15664017e-01 -2.20650155e-03 1.11875668e-01 3.30579206e-02
1.11579545e-01 -3.69673640e-01 -1.18505609e+00 7.79283047e-01
-1.75077415e+00 -6.45492494e-01 -4.84241515e-01 2.71996379e+00
8.61215651e-01 4.13715988e-01 4.51919645e-01 1.87622160e-01
5.38784921e-01 -4.42897618e-01 -5.87527990e-01 -5.59919536e-01
-7.21783340e-02 -6.14735521e-02 8.48109245e-01 5.95809817e-01
-9.26398277e-01 1.44148663e-01 7.38050365e+00 4.40590560e-01
-7.54064798e-01 3.17929722e-02 1.31068349e+00 -3.87114853e-01
-4.83353324e-02 3.25410038e-01 -4.98666644e-01 4.63023305e-01
1.48112428e+00 -2.16688886e-01 -1.48746833e-01 4.19493437e-01
5.62232792e-01 -2.50902772e-01 -1.17471361e+00 3.70312214e-01
-3.52592379e-01 -8.99903119e-01 -3.42488050e-01 2.16551781e-01
2.55954981e-01 -4.58310634e-01 8.03953707e-02 2.28420958e-01
5.34716666e-01 -1.33359158e+00 6.00673199e-01 1.04967630e+00
1.04453707e+00 -1.14386022e+00 9.59347308e-01 3.80251974e-01
-2.74166495e-01 -1.56060740e-01 2.07507923e-01 -3.13291103e-01
6.40948534e-01 6.98607445e-01 -6.44229889e-01 4.36696708e-01
3.92512947e-01 2.43046150e-01 -2.46871248e-01 1.30189550e+00
1.36037424e-01 9.48722064e-01 -3.48820835e-01 1.27570212e-01
1.41807541e-01 3.43076438e-02 5.34772813e-01 1.09705853e+00
5.46935856e-01 -6.57365024e-02 -4.47567552e-01 8.09985995e-01
5.80864012e-01 6.97835758e-02 -3.71562123e-01 8.97389799e-02
-4.63266112e-02 8.45661163e-01 -7.98257649e-01 -7.53492191e-02
-4.90994960e-01 6.95466220e-01 -6.44374862e-02 3.64794075e-01
-6.65986001e-01 -1.69597730e-01 2.08470613e-01 4.60266769e-01
-1.54275686e-01 6.29416853e-02 -7.98579991e-01 -7.05289066e-01
-3.85973722e-01 -5.30727446e-01 1.00352883e+00 -2.79666573e-01
-1.43067527e+00 2.21070811e-01 2.50311643e-01 -1.03659272e+00
-6.35651290e-01 -3.92063677e-01 -6.71173871e-01 1.13866913e+00
-1.33595121e+00 -9.99172926e-01 3.14968795e-01 -1.60386518e-03
5.78118861e-02 2.57129192e-01 8.12390387e-01 1.61065936e-01
-5.88091075e-01 6.24511123e-01 3.26380461e-01 -3.32069635e-01
8.00234020e-01 -1.51953328e+00 1.90672368e-01 3.65889698e-01
-6.62929535e-01 4.32415277e-01 8.77521992e-01 -1.08461726e+00
-5.43171883e-01 -8.16782951e-01 1.35120213e+00 -5.68803728e-01
4.86715823e-01 9.27180648e-02 -9.55533504e-01 5.67955494e-01
-1.57019243e-01 -3.14871728e-01 1.02529573e+00 1.82276502e-01
2.66723841e-01 3.66300762e-01 -1.33054709e+00 6.69250846e-01
5.93687236e-01 6.94703087e-02 -5.51119149e-01 1.21979825e-02
2.80070931e-01 3.80407600e-03 -1.06875575e+00 4.61324811e-01
8.91061008e-01 -1.11564541e+00 1.16604161e+00 -1.06420040e+00
5.24841189e-01 1.96735740e-01 2.14271862e-02 -8.75175536e-01
-2.80849844e-01 -6.68742359e-01 -1.48409307e-01 9.71098661e-01
4.38617796e-01 -7.76523590e-01 8.04173231e-01 9.73119795e-01
3.45055670e-01 -9.39780235e-01 -1.45849931e+00 -8.04535627e-01
6.83427095e-01 -5.82158446e-01 4.40983832e-01 9.47440743e-01
-1.32657364e-01 -1.39302522e-01 -6.31813586e-01 6.67941272e-02
1.07565677e+00 -2.82401860e-01 1.37027711e-01 -1.60870123e+00
-2.79517472e-01 -5.50056994e-01 -1.22383192e-01 -3.15940857e-01
1.12247340e-01 -5.43630660e-01 -2.01533467e-01 -1.40088308e+00
2.13884264e-01 -4.92581427e-01 -6.37289226e-01 3.22002470e-01
-3.75678569e-01 -8.83157998e-02 -3.06105316e-01 9.84556749e-02
6.33919388e-02 2.92491287e-01 6.50756419e-01 4.22610223e-01
-3.68056148e-01 4.02081579e-01 -8.35119665e-01 8.40447545e-01
9.63596821e-01 -9.17705357e-01 -2.83689741e-02 3.24418873e-01
3.86197597e-01 1.05452049e+00 7.20600486e-01 -5.45925558e-01
3.68585959e-02 -4.77128416e-01 6.29273355e-01 -5.49401939e-01
-2.00718082e-02 -1.02203226e+00 6.88599944e-01 9.01339352e-01
-5.38916647e-01 -2.74076313e-02 1.68865606e-01 6.74176157e-01
1.21731378e-01 -4.33757901e-01 8.53036642e-01 2.52146482e-01
2.56462932e-01 -6.03741296e-02 -6.66324973e-01 -9.01336819e-02
1.22410297e+00 -5.81304431e-01 2.76342154e-01 -7.08155692e-01
-1.16295850e+00 3.43518645e-01 5.18848538e-01 6.12486228e-02
4.47083026e-01 -1.20367789e+00 -1.02444351e+00 -7.25927502e-02
-6.20840602e-02 -4.23573762e-01 3.49555016e-01 1.27034569e+00
-4.14729536e-01 4.82659936e-01 -3.69359576e-03 -1.13840386e-01
-1.13664091e+00 7.57652104e-01 7.44710565e-01 -6.58242583e-01
-5.18657327e-01 3.66141170e-01 3.51683378e-01 -1.99347422e-01
7.80623406e-02 1.72256324e-02 -2.81871527e-01 2.69654065e-01
3.45764309e-01 8.92660916e-01 -2.97149301e-01 -4.34819788e-01
-3.90684247e-01 4.74071205e-02 -4.34002765e-02 -3.09380472e-01
1.21140397e+00 -2.75262177e-01 1.13299228e-01 7.67669141e-01
9.12559986e-01 -3.17766339e-01 -1.45983303e+00 2.64195800e-01
4.88981903e-01 -1.23499848e-01 2.68060386e-01 -1.21697474e+00
-5.56543410e-01 6.15577340e-01 6.80876851e-01 1.66750610e-01
1.25426197e+00 -4.18560058e-01 3.82460594e-01 -3.17347109e-01
-6.79042563e-02 -6.62632048e-01 -6.81134880e-01 -2.94441022e-02
7.10172057e-01 -1.15196013e+00 1.33750856e-01 4.63638417e-02
-4.27589118e-01 1.16440213e+00 -2.98685618e-02 -1.60684645e-01
7.58425117e-01 2.60142297e-01 1.67723015e-01 1.01988517e-01
-7.63900936e-01 3.56668800e-01 2.63455302e-01 3.98276895e-01
6.21904910e-01 9.09470096e-02 -9.04089153e-01 9.63675976e-01
9.42490697e-02 4.74943310e-01 6.30586386e-01 9.91082907e-01
1.93315521e-01 -1.12858295e+00 -5.50891101e-01 9.30487335e-01
-1.28224230e+00 -3.78829002e-01 -1.89610600e-01 9.46263313e-01
-2.68125743e-01 9.14321780e-01 5.38135357e-02 3.41962039e-01
1.65297747e-01 3.01166981e-01 1.58380076e-01 -1.78285912e-01
-8.74079287e-01 3.66301328e-01 2.21837014e-01 -2.24999815e-01
-2.59445488e-01 -1.01873529e+00 -1.09353626e+00 -2.42826521e-01
-4.92581248e-01 7.68193379e-02 3.07151377e-01 6.96233034e-01
-6.95304619e-03 7.13788092e-01 2.67642140e-01 -3.80241305e-01
-9.25515592e-01 -5.59680939e-01 -6.79750264e-01 -6.46318868e-02
7.90630877e-01 -6.02983952e-01 -7.91225433e-01 -1.67186320e-01] | [7.847569942474365, 5.478878021240234] |
09c67701-d7f9-4ef7-991e-bfab7fcb7871 | multi-task-learning-of-cascaded-cnn-for | 1805.01290 | null | http://arxiv.org/abs/1805.01290v1 | http://arxiv.org/pdf/1805.01290v1.pdf | Multi-task Learning of Cascaded CNN for Facial Attribute Classification | Recently, facial attribute classification (FAC) has attracted significant
attention in the computer vision community. Great progress has been made along
with the availability of challenging FAC datasets. However, conventional FAC
methods usually firstly pre-process the input images (i.e., perform face
detection and alignment) and then predict facial attributes. These methods
ignore the inherent dependencies among these tasks (i.e., face detection,
facial landmark localization and FAC). Moreover, some methods using
convolutional neural network are trained based on the fixed loss weights
without considering the differences between facial attributes. In order to
address the above problems, we propose a novel multi-task learning of cas-
caded convolutional neural network method, termed MCFA, for predicting multiple
facial attributes simultaneously. Specifically, the proposed method takes
advantage of three cascaded sub-networks (i.e., S_Net, M_Net and L_Net
corresponding to the neural networks under different scales) to jointly train
multiple tasks in a coarse-to-fine manner, which can achieve end-to-end
optimization. Furthermore, the proposed method automatically assigns the loss
weight to each facial attribute based on a novel dynamic weighting scheme, thus
making the proposed method concentrate on predicting the more difficult facial
attributes. Experimental results show that the proposed method outperforms
several state-of-the-art FAC methods on the challenging CelebA and LFWA
datasets. | ['Ni Zhuang', 'Yan Yan', 'Si Chen', 'Hanzi Wang'] | 2018-05-03 | null | null | null | null | ['facial-attribute-classification'] | ['computer-vision'] | [ 0.12227622 -0.29689044 0.03770065 -0.7446982 -0.5281681 0.0579512
0.44165736 -0.00875005 -0.5008215 0.3745443 -0.11991737 0.26102033
-0.19133332 -0.7211579 -0.49484843 -0.8260844 0.22132598 0.5054006
-0.20091513 -0.02609576 0.01418687 0.6085771 -1.6325521 -0.04100299
0.9518074 1.5754902 0.05795512 0.05579469 -0.13843293 0.64158595
-0.10396 -0.62913245 0.05727628 -0.24267535 -0.5261816 0.3466961
0.76530385 -0.36675528 -0.15898181 1.1479963 0.63006693 0.09696501
0.5604365 -1.4237909 -0.5393276 0.08554169 -0.959037 -0.02070562
-0.03234644 -0.08030649 1.0404398 -1.244373 -0.01401103 1.4974308
0.6780565 0.42168716 -1.1265179 -1.0490031 0.45569322 0.5107583
-1.8767716 -0.80825806 0.88654846 -0.36500365 0.35850993 -0.11740211
0.4441635 0.64963657 -0.20932236 0.9213456 1.0216404 -0.10356543
-0.13734615 -0.14006233 -0.17879507 0.9684262 -0.05268098 -0.24821499
-0.23477726 -0.07833631 0.56707424 0.16782933 -0.11141153 -0.24350867
-0.9978313 0.6788121 0.39775157 -0.04604795 -0.6274484 0.03331942
0.4603144 0.03985625 0.68851596 -0.30786455 -0.5015391 0.21306665
-0.80019313 0.14783268 0.19511646 0.86009115 1.0216522 0.08160871
-0.28088772 1.1934023 0.45514524 0.31055206 0.23384507 -0.52095497
0.484339 0.6275332 -0.11165485 -1.3371918 -0.49991322 -0.342631
-1.0458959 0.08240603 0.35564 -0.10407087 -0.882971 1.8721699
0.5830639 0.49347618 -0.12928027 0.8816684 0.7559031 0.55797976
0.4767308 -0.22420458 1.3460394 -0.9600408 -0.8510705 -0.2317274
0.26059765 -0.90333825 0.93162453 0.2054427 -0.8559965 -0.78583664
-0.8486126 0.04781814 -0.23533772 0.77247053 0.6583209 0.39393386
-0.8610359 0.27533937 -0.7005876 -0.2568751 0.98477405 0.7157447
-0.46096724 -0.12439754 -1.0129154 0.5569808 0.35152733 0.64503235
-0.8477459 -0.43366987 -0.9734493 0.28686273 0.65448165 -0.50552297
0.97005403 -1.3613228 -1.4812635 0.70097387 -0.27055973 0.10859495
0.37575042 -0.3796362 -0.50649345 -0.06135195 0.05847271 0.6640376
0.975163 -1.0459296 -0.8978308 -0.7792649 -0.13709731 0.12266038
-0.7660521 0.2962833 -0.72485054 -0.6236576 0.04317964 -0.825436
0.0588056 0.22529618 -0.4082882 -0.60425353 0.96421176 -0.62758416
1.1733644 -2.076702 0.19947796 0.21960928 0.24758017 0.49791813
-0.31912458 -0.09530121 -0.17199026 -0.15148668 -0.27710274 -0.6437102
-0.08244013 -0.02223666 0.2620338 0.61609584 0.47845766 0.48408535
-0.64753795 -0.92387015 0.17275727 0.72288793 -0.485387 0.34104908
-0.12901425 0.50608987 -0.91238874 1.0419279 1.1048893 -0.07001929
-0.09645844 -0.44475105 -0.1461663 -0.36021397 -1.1421263 1.5032076
-0.684693 0.20666039 0.34967157 -0.9343276 1.0538208 0.39080843
0.5807659 -0.5649072 0.3260858 0.30559915 0.0279295 -0.5266762
0.0485565 -0.04595269 0.20315552 0.17464836 0.18300268 0.5569913
0.0515641 -0.21859342 0.38428777 0.31893927 0.21879184 -0.16064842
1.2932996 -0.42980197 1.026972 -0.04685582 -0.2737659 0.40823522
0.31417632 -0.71206623 -0.83377075 -0.46471894 -0.22105965 1.26807
0.19947626 -0.32044643 -0.91765887 -0.75142705 0.1406991 0.08402997
-0.7549108 -0.08257762 -0.58618987 -0.9407367 0.47761652 0.54904324
0.8960841 -1.1308919 -0.06789674 0.25951636 -0.22898147 -1.2242377
-0.6665109 -0.42607287 -0.4995832 -0.9479691 -0.60049736 -0.8360984
0.7482106 0.08233058 0.7384023 0.3426055 -0.18699658 0.07549987
-0.20282067 -0.37324196 0.16454288 0.25507018 0.06404231 1.1052203
0.5435047 -0.6381862 -0.70724905 0.5172804 -0.6973836 0.00952804
1.0033679 0.90911245 0.9019231 -0.1734607 0.6785472 -0.81590176
0.3993445 -0.5153374 -0.648443 0.38184512 -0.5486271 -0.1408503
0.8737219 -0.30022016 -1.054037 0.36682752 -0.44908482 -0.6311493
-0.24713542 0.47698614 -0.82906294 -0.31408325 0.01452447 0.2911969
0.18233581 -0.5379964 0.08410627 0.61726344 0.47265762 -0.67349863
0.77650744 0.4162369 0.21064588 -0.5071629 -1.0310198 -0.21284246
-0.81533426 -0.3644311 0.94128454 -0.99945855 -1.0109917 0.91729486
-1.1363053 0.16877763 0.39620465 0.39187342 -0.40401533 0.1942841
-0.47331408 -0.72313315 -0.450153 -1.3039262 1.0988164 0.5087851
0.28547096 -0.73345184 -0.38173428 0.22109635 0.49483427 0.40208325
0.78706104 -0.55862373 -0.4635104 -0.11468069 -0.6686121 0.35223433
0.426755 0.07832772 -1.1671265 -0.29344374 -0.358919 -0.56980276
0.7547425 0.26159236 1.7190276 -0.38066489 -0.26956126 0.89839494
1.3275418 0.16561615 0.5340799 0.38884914 1.1019392 0.6795944
0.87366825 0.5989833 0.4781143 0.9129422 0.62496144 -0.31063908
0.11421441 -0.01377474 -0.03730397 0.6044204 -0.3679966 0.05036781
-0.66944975 0.41962484 -1.9055605 -0.68140745 0.17009227 2.0907402
0.6253673 -0.21814625 0.09419625 -0.08149689 1.0057056 0.34882626
-0.64645237 -0.11026067 0.00818018 0.15813975 0.1537526 0.16799752
-1.5124009 1.0212766 4.179613 1.2092935 -1.2357476 0.12295213
0.95171726 0.1986136 0.15284298 -0.2747734 -0.95500153 0.6555186
0.38234282 0.09307407 0.35994354 0.8766965 0.13236722 0.30326897
-0.9588467 1.067524 0.18668605 -0.84028953 0.25649422 -0.04445129
0.40737465 -0.4985287 0.14565647 0.28566745 -0.09306806 -1.1468844
0.67239887 0.5965062 1.0019557 -1.1699939 0.9264977 0.05633805
-1.5922121 -0.14857516 -0.51858133 0.31824344 -0.07334317 0.44080204
-0.32266963 0.6028634 0.64191407 0.82890075 -0.5946634 0.93317896
-0.06286646 0.3140925 -0.13570146 0.22748308 0.3287317 -0.40149543
0.19614634 0.92349756 0.4432662 0.19222447 0.42703384 0.58341146
-0.28358427 0.4711119 0.01366998 0.1567521 0.5079874 1.7547123
-0.370739 -0.05034552 -0.6660806 0.87170047 0.5376345 0.09358816
-0.83571523 -0.47938558 0.85808724 0.04869733 0.39403108 0.09650997
0.08919898 -0.8160349 0.12524053 -0.8151408 0.5536719 -0.45075613
-1.3054188 0.90740156 -0.245815 -1.287453 0.04908428 -0.4593378
-0.72227484 1.0290208 -1.9582415 -1.6197114 -0.7319382 0.9734315
0.48279503 -0.3322691 0.6446341 0.8198068 -1.0455285 0.9260542
-0.10346035 0.34870097 0.93958294 -0.9011359 0.06190623 0.5623772
-0.33194992 0.42700222 0.17073393 -0.30689886 -1.1452322 -1.54599
0.88590163 0.09199607 0.40074623 -0.23437516 -1.0134674 0.6590395
-0.07912981 0.53639925 0.54826415 0.1291946 -0.2925193 -0.754743
-1.081296 0.3124161 0.90702856 -0.3083191 0.05048031 0.23371832
0.2591345 -0.22770128 -0.753802 0.7185064 0.68285006 -0.7606105
0.9407595 -0.50821304 0.4306488 -0.45361277 0.03135631 -1.300492
-0.41727164 -0.25051042 0.17360522 1.6233801 0.15119179 -0.6424697
0.6578133 0.45135555 -0.1364678 -1.1367221 -0.89932424 -0.43355694
-0.3257083 0.02653737 0.94755393 0.92986757 -0.7725362 0.37317288
-0.612187 0.3346052 0.6904699 0.254603 0.71694475 -1.5826458
0.12432399 -0.50000393 -0.54062986 -0.78658646 0.43254054 -0.7103922
-0.0482187 -1.1564121 0.2627248 -0.7819156 -0.54464483 0.76091677
-0.57596487 0.32256418 0.08755749 0.33093098 -0.66574717 0.97277004
1.3369291 -0.12055182 0.08794151 0.10223828 -0.62576073 0.9057324
0.7132946 -0.35859233 -0.23297207 -0.5323748 -0.15003271 -0.09707945
0.37905177 -0.98771507 0.2729747 -0.21374388 0.7562074 -0.5097793
0.32704338 -1.0298643 -0.16305915 0.11255644 -0.11078801 0.22765067
0.27617988 0.5401262 -0.5646588 0.12023323 1.0338405 0.05143104
-0.92358226 1.0736899 0.13142398 -0.1630485 1.1425302 -0.02356444
-0.00469778 -0.10916112 -0.68651915 0.58366674 0.12469201 0.31072554
0.5886708 -1.6536627 -0.98721564 0.1578924 0.25684857 0.1495986
0.5275217 0.97602075 -0.29700786 0.06406004 -0.5686762 -0.54813987
-1.4971479 0.4012883 0.37671167 -0.15046021 -0.21849263 0.93169224
0.37935677 -0.3837025 0.19854313 0.3670772 -0.6917926 0.36081943
0.51239365 0.1801416 0.17388523 -1.0340714 -0.548972 0.98134434
-0.33790666 0.44118783 1.4006399 -0.12684658 -0.35897428 -0.0204811
1.2905079 -0.08813593 -1.3715084 -0.5456143 -0.17768055 -0.63685745
0.12469646 -0.48317277 -1.7296449 1.0724663 0.7280805 -0.39403135
1.5697267 -0.24446878 0.85769945 0.17319898 0.23257947 -1.0575109
-0.0342538 0.18610689 0.703627 -1.2973832 0.08622646 -0.63039213
-0.5293303 1.1824962 1.0885653 0.13044043 0.55737466 -0.14469521
0.05047279 -0.10215531 -0.33730534 -0.27210566 0.2626231 0.16876362
0.34485716 -0.04011451 -0.24491893 0.76764816 0.2447071 0.09523952
-0.08710907 0.6220772 -0.26696637 -1.1063465 -0.39821017 0.5422477
-0.64105684 -0.0550751 -0.14189902 0.84174913 0.51699555 0.75855446
0.12069976 -0.51787263 0.35738283 -0.054654 0.29706773 -0.27609938
-0.36857194 0.13926235 -0.15425186 -0.51792127 -0.60885787 -0.7765414
-0.9907885 -0.37344167 -0.29706654 0.0922448 0.4713365 1.0058104
0.39243975 0.46546668 0.95232105 -0.8776523 -0.3413022 -1.0991962
-0.45038068 0.48269144 0.30603087 -1.0240017 -0.1044799 -0.03557015] | [13.511107444763184, 0.7706761360168457] |
75128921-1348-41cd-98a0-57f73a246f2e | improving-temporal-relation-extraction-with | null | null | https://aclanthology.org/W16-2914 | https://aclanthology.org/W16-2914.pdf | Improving Temporal Relation Extraction with Training Instance Augmentation | null | ['Guergana Savova', 'Dmitriy Dligach', 'Chen Lin', 'Timothy Miller', 'Steven Bethard'] | 2016-08-01 | null | null | null | ws-2016-8 | ['temporal-relation-extraction', 'temporal-information-extraction'] | ['natural-language-processing', 'natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.282594203948975, 3.763812780380249] |
8eac8bee-cf36-4312-8f70-28a0953e730d | u3e-unsupervised-and-erasure-based-evidence | 2210.02621 | null | https://arxiv.org/abs/2210.02621v3 | https://arxiv.org/pdf/2210.02621v3.pdf | U3E: Unsupervised and Erasure-based Evidence Extraction for Machine Reading Comprehension | More tasks in Machine Reading Comprehension(MRC) require, in addition to answer prediction, the extraction of evidence sentences that support the answer. However, the annotation of supporting evidence sentences is usually time-consuming and labor-intensive. In this paper, to address this issue and considering that most of the existing extraction methods are semi-supervised, we propose an unsupervised evidence extraction method (U3E). U3E takes the changes after sentence-level feature erasure in the document as input, simulating the decline in problem-solving ability caused by human memory decline. In order to make selections on the basis of fully understanding the semantics of the original text, we also propose metrics to quickly select the optimal memory model for this input changes. To compare U3E with typical evidence extraction methods and investigate its effectiveness in evidence extraction, we conduct experiments on different datasets. Experimental results show that U3E is simple but effective, not only extracting evidence more accurately, but also significantly improving model performance. | ['Chenghao Wu', 'Shumin Shi', 'Suzhe He'] | 2022-10-06 | null | null | null | null | ['machine-reading-comprehension'] | ['natural-language-processing'] | [ 3.57491404e-01 8.60124305e-02 -1.97300360e-01 -3.40825021e-01
-5.17354012e-01 -3.11870098e-01 2.67097741e-01 9.32117760e-01
-5.98757207e-01 9.67967987e-01 5.82867503e-01 -3.91748220e-01
-1.88878790e-01 -8.23856473e-01 -7.27582872e-01 -6.64017498e-02
3.91578913e-01 8.77812877e-02 5.85070491e-01 7.58395791e-02
9.99465644e-01 -2.60806438e-02 -1.67703867e+00 4.75650162e-01
1.51165068e+00 8.67663920e-01 4.41128701e-01 7.13039577e-01
-6.24970257e-01 1.15825582e+00 -7.52277911e-01 -3.83049190e-01
-3.58924180e-01 -6.60615861e-01 -1.16198683e+00 5.41078374e-02
1.38714731e-01 -3.72141838e-01 -5.23226969e-02 1.06357706e+00
3.82886827e-01 8.92226398e-02 6.57986641e-01 -9.03114378e-01
-4.67821687e-01 8.00708592e-01 -3.13882619e-01 5.68919897e-01
8.20238411e-01 -1.44900769e-01 6.97769344e-01 -1.07371879e+00
5.86956263e-01 1.05985773e+00 3.77406120e-01 2.95007765e-01
-5.33891141e-01 -5.19181311e-01 3.05548042e-01 7.70857275e-01
-9.52435911e-01 -3.90305966e-01 7.91428626e-01 -2.34813675e-01
8.10737669e-01 3.62534583e-01 4.70441699e-01 8.16069782e-01
2.09634632e-01 1.11484659e+00 1.47043419e+00 -9.39218104e-01
2.20220402e-01 2.08009481e-01 8.81577075e-01 8.96797776e-01
3.54210705e-01 -4.50180650e-01 -9.94780004e-01 -1.56626720e-02
1.45601064e-01 -2.88565964e-01 -6.30698323e-01 2.47603476e-01
-1.02506638e+00 5.07332504e-01 3.65920998e-02 2.35140666e-01
-3.79098833e-01 -2.27511585e-01 3.44782591e-01 4.25250620e-01
2.96249509e-01 2.15031922e-01 -6.00385010e-01 -3.40834767e-01
-8.32018018e-01 2.02383086e-01 8.02721083e-01 8.84153545e-01
3.02354813e-01 -5.07731378e-01 -5.58921695e-01 6.66037619e-01
2.23643541e-01 3.09672773e-01 8.37976098e-01 -6.19631290e-01
8.43990982e-01 1.11767566e+00 1.81818232e-01 -1.05102241e+00
-4.79482770e-01 -4.74112391e-01 -5.92766523e-01 -5.04173875e-01
2.81230032e-01 -3.40620056e-02 -6.34299755e-01 1.52020252e+00
2.68496990e-01 -1.13546170e-01 -2.89909020e-02 6.31223321e-01
1.05042958e+00 6.22869790e-01 4.37727682e-02 -6.53219044e-01
1.49364197e+00 -9.84493256e-01 -1.11988962e+00 -2.67384499e-01
7.83449113e-01 -6.95759654e-01 1.41035056e+00 3.06138784e-01
-1.00150144e+00 -4.41615313e-01 -1.23212743e+00 -1.22196294e-01
-2.84862220e-01 2.08464533e-01 3.37772548e-01 3.02098811e-01
-4.02963787e-01 5.05640686e-01 -3.74410421e-01 -1.45147055e-01
2.20217228e-01 1.39002595e-02 -2.39952222e-01 -1.16272002e-01
-1.42094064e+00 1.08944750e+00 5.94805658e-01 9.03809369e-02
-2.29373634e-01 -3.79133612e-01 -7.06476331e-01 2.84364015e-01
6.81199551e-01 -6.15283251e-01 1.33561325e+00 -6.67397141e-01
-1.26315987e+00 4.25027698e-01 -4.25036162e-01 -2.99763799e-01
2.65573263e-01 -5.31863391e-01 -2.53266633e-01 1.98956758e-01
4.27215360e-02 1.89196959e-01 7.12343156e-01 -8.59320164e-01
-6.26757205e-01 -4.05590415e-01 -1.13511689e-01 3.15449476e-01
-5.72533369e-01 1.39165491e-01 -4.70195472e-01 -5.98766267e-01
2.95038670e-01 -4.45449829e-01 2.21972406e-01 -4.39817846e-01
-4.32346314e-01 -6.24092996e-01 3.43100548e-01 -1.09934485e+00
1.95043325e+00 -1.91222513e+00 1.13959149e-01 2.03312084e-01
2.20259979e-01 2.94827640e-01 9.04799327e-02 3.20567280e-01
4.75419998e-01 2.21784607e-01 -3.35377514e-01 6.77238479e-02
-8.22524056e-02 3.15880738e-02 -3.25130969e-01 -1.73409685e-01
7.59085193e-02 7.93719530e-01 -8.17769647e-01 -1.05199635e+00
-2.83515334e-01 -3.40896696e-01 -2.63959318e-01 3.94213438e-01
-3.47123414e-01 1.82957470e-01 -7.38471031e-01 7.77761877e-01
3.67956161e-01 -4.20619249e-01 1.37092769e-02 7.54639432e-02
5.45000657e-02 5.19064963e-01 -1.09096396e+00 1.34951675e+00
-1.62131339e-01 5.29398441e-01 -3.76068115e-01 -7.47352242e-01
8.72343898e-01 1.37438893e-01 -2.42692932e-01 -9.01847780e-01
2.57623196e-01 5.47733903e-01 9.47236717e-02 -1.02757132e+00
4.78107899e-01 1.49691090e-01 1.02956764e-01 5.13686299e-01
-1.93142265e-01 8.23187009e-02 4.01799858e-01 3.89304936e-01
1.30987537e+00 -6.16165325e-02 5.92747986e-01 -7.57951438e-02
7.90532827e-01 -1.25540331e-01 5.48246801e-01 8.28874707e-01
-8.62712190e-02 2.27249295e-01 3.88504118e-01 3.04658804e-02
-6.37017250e-01 -6.04362726e-01 -7.07049444e-02 6.96669936e-01
1.26127020e-01 -6.28537178e-01 -1.02672160e+00 -9.28295255e-01
-1.49961263e-01 9.07444894e-01 -3.28493595e-01 -3.66751969e-01
-5.77413619e-01 -6.51317179e-01 3.25525999e-01 6.74792230e-01
9.19359624e-01 -1.38556898e+00 -8.34090173e-01 3.46923143e-01
-7.09904015e-01 -8.48617077e-01 -2.77296484e-01 6.98212087e-02
-1.12888110e+00 -1.16510355e+00 -4.86014456e-01 -6.81048810e-01
8.63252878e-01 2.88974822e-01 1.07064414e+00 9.10522997e-01
1.09948650e-01 2.35573873e-01 -7.51519978e-01 -7.18536139e-01
-2.81271666e-01 1.80579603e-01 -1.68001547e-01 -3.31767589e-01
6.36391521e-01 -1.33334547e-01 -4.90595907e-01 1.37086838e-01
-9.51485991e-01 1.93875238e-01 8.34645569e-01 7.02141225e-01
5.33506870e-01 2.68108457e-01 9.45414245e-01 -1.02363098e+00
1.30202591e+00 -5.02152562e-01 -1.72677249e-01 7.34984279e-01
-1.13508415e+00 2.35387683e-01 4.04921800e-01 -2.64918476e-01
-1.15444028e+00 -4.19092894e-01 -6.12231381e-02 1.97986722e-01
6.30729198e-02 1.09560084e+00 -1.66018322e-01 3.45296741e-01
4.04284090e-01 4.68704075e-01 -1.68752357e-01 -6.06176555e-01
4.16307338e-02 9.65278864e-01 3.56358320e-01 -5.64656854e-01
3.94303232e-01 -2.93354601e-01 -2.04860419e-01 -5.37909508e-01
-1.27574408e+00 -3.08239609e-01 -6.46252096e-01 -3.84922981e-01
6.03862941e-01 -3.93950462e-01 -3.90478671e-01 3.79952490e-01
-1.35776734e+00 4.82153744e-02 5.90760186e-02 4.55504149e-01
-2.31763020e-01 5.24816334e-01 -5.61192989e-01 -8.09783578e-01
-4.83421296e-01 -7.42791116e-01 5.36693811e-01 3.82997304e-01
-4.53443855e-01 -5.89890242e-01 3.01447208e-03 7.63080955e-01
8.83860290e-02 -1.05051249e-01 1.54929817e+00 -6.71365976e-01
-4.95723933e-01 -2.72298962e-01 -2.03829333e-01 2.93238580e-01
-1.49057657e-01 -1.43744633e-01 -6.44783914e-01 1.95529148e-01
2.90787846e-01 -4.71699953e-01 8.98957491e-01 3.59532982e-02
1.41446078e+00 -4.44534391e-01 -1.13095902e-01 -1.19077653e-01
9.18263018e-01 2.19918892e-01 6.96034014e-01 6.28684998e-01
2.78138548e-01 8.88682365e-01 1.16104364e+00 2.28548944e-01
5.56333184e-01 3.33148479e-01 2.46953256e-02 3.44774783e-01
-1.10997222e-01 -4.35139894e-01 3.96248400e-01 1.44070399e+00
2.06702594e-02 -4.57130104e-01 -6.96084678e-01 4.43401933e-01
-1.99170351e+00 -6.70996547e-01 -4.69617397e-01 1.91169786e+00
1.29398394e+00 5.31465113e-01 -2.88631260e-01 6.17253244e-01
3.97433162e-01 -2.50485271e-01 -5.42810678e-01 -3.96560848e-01
-5.60255982e-02 2.05115139e-01 -2.75851451e-02 3.63829225e-01
-5.61634183e-01 6.47876501e-01 5.99077749e+00 8.03977728e-01
-5.79578996e-01 7.62555525e-02 4.90012616e-01 1.48219869e-01
-4.63884652e-01 9.42337066e-02 -7.32889175e-01 5.83050430e-01
7.72194028e-01 -1.39118522e-01 -4.44862135e-02 3.91345263e-01
2.08402038e-01 -5.37724555e-01 -9.66887414e-01 6.10116661e-01
1.74840271e-01 -9.83435571e-01 2.37362236e-01 -4.55675662e-01
3.42951298e-01 -6.62456691e-01 -4.41389412e-01 3.80929470e-01
-3.09806943e-01 -8.24764132e-01 8.14301312e-01 9.43670571e-01
1.60150275e-01 -5.55029094e-01 1.10155165e+00 9.50056493e-01
-8.04958880e-01 -1.53104141e-01 -3.89235646e-01 -2.89945006e-01
4.77922447e-02 7.37317264e-01 -6.02786362e-01 3.21511269e-01
5.95843971e-01 2.69308060e-01 -1.04771626e+00 1.00552881e+00
-7.31292427e-01 9.79515016e-01 -4.71034199e-02 -6.23185754e-01
-2.09376872e-01 1.32558495e-01 4.85764831e-01 1.23644161e+00
3.33298564e-01 4.96571600e-01 -3.31578627e-02 6.13073528e-01
-3.59426320e-01 5.92749774e-01 -2.36259326e-01 -1.37091398e-01
8.96837890e-01 8.15386176e-01 -8.16641450e-01 -5.00182092e-01
-2.46834442e-01 8.92857492e-01 5.53493738e-01 5.47686294e-02
-5.29540777e-01 -6.47442400e-01 -4.64304000e-01 1.10980213e-01
-4.50852811e-02 -8.82335305e-02 -4.57057208e-01 -1.08843219e+00
6.92355156e-01 -1.02532196e+00 4.17051882e-01 -9.07029271e-01
-9.83044386e-01 2.69482285e-01 1.29468694e-01 -9.65372801e-01
2.21187435e-03 -3.98911864e-01 -5.51744044e-01 5.87712884e-01
-1.54280567e+00 -5.93776941e-01 -4.62285429e-01 1.94052547e-01
7.92178035e-01 -1.83753129e-02 5.76874912e-01 1.79958090e-01
-7.25344896e-01 5.31527162e-01 -3.05523336e-01 -1.49932697e-01
4.90837634e-01 -1.29457641e+00 -4.45152968e-02 1.07201207e+00
9.82666239e-02 8.79680872e-01 5.90724826e-01 -9.20620382e-01
-1.16037595e+00 -6.00447476e-01 1.45018530e+00 -4.44943696e-01
4.93924052e-01 -5.22357114e-02 -1.35760546e+00 3.77399951e-01
2.02763721e-01 -6.12977624e-01 7.41132319e-01 1.09007865e-01
-6.86584935e-02 4.40751538e-02 -9.81378257e-01 6.34434104e-01
1.06161678e+00 -4.57201540e-01 -1.25938511e+00 1.26031190e-01
7.90464163e-01 -2.92533189e-01 -7.11832047e-01 5.94004989e-01
3.07560623e-01 -6.94683552e-01 5.56653142e-01 -6.04433239e-01
7.17631459e-01 -3.50526452e-01 1.16101431e-03 -1.04078794e+00
-1.12819009e-01 -1.01561151e-01 -6.25662506e-01 1.40291953e+00
6.32883847e-01 -3.56498003e-01 4.33949262e-01 6.86763406e-01
-3.83760594e-02 -1.07522368e+00 -7.99637318e-01 -4.85352099e-01
-1.97609320e-01 -2.97798157e-01 6.74494803e-01 7.45755315e-01
3.99230242e-01 6.05718553e-01 -2.07972929e-01 -5.16109355e-02
3.41167390e-01 2.38919064e-01 5.91488719e-01 -1.22945940e+00
-8.78343135e-02 -3.71595651e-01 5.64463884e-02 -1.16186810e+00
1.34645626e-01 -7.79774249e-01 2.68822402e-01 -1.75885105e+00
5.90949893e-01 -1.41756535e-01 -3.95512402e-01 3.21102679e-01
-8.39020371e-01 -5.08327782e-01 -3.44353765e-02 3.05700779e-01
-7.44601727e-01 6.14545166e-01 1.27812743e+00 -1.23087220e-01
-1.41751289e-01 1.77637674e-02 -7.41061211e-01 8.95163476e-01
8.58595192e-01 -5.33583641e-01 -6.25577986e-01 -3.46773922e-01
5.64658940e-01 1.27031595e-01 1.01847835e-01 -8.16872239e-01
3.08428705e-01 -1.92422509e-01 4.11909580e-01 -7.90319085e-01
-2.04920411e-01 -6.17423832e-01 -5.52020431e-01 5.30666471e-01
-6.16588295e-01 2.91734129e-01 8.55409056e-02 6.45795524e-01
-3.51147324e-01 -1.12422812e+00 2.03349397e-01 -1.95890740e-02
-7.26860821e-01 -1.66124120e-01 -5.49706459e-01 1.83343485e-01
5.52300453e-01 -4.98417094e-02 -6.55673504e-01 -2.76622236e-01
-2.24982128e-01 4.34241980e-01 -6.18934520e-02 4.71943945e-01
1.04473996e+00 -1.07276094e+00 -5.93774498e-01 -1.62833214e-01
2.55048275e-01 -6.02205470e-03 1.02249580e-02 7.27092624e-01
-4.30207133e-01 4.21799272e-01 8.92827809e-02 -8.82502869e-02
-1.55945945e+00 4.14226413e-01 -1.66646019e-01 -4.03365016e-01
-5.87910593e-01 6.53695107e-01 -4.50437784e-01 -1.46661773e-01
4.34071034e-01 -5.89542150e-01 -8.75610530e-01 1.12589158e-01
6.58070803e-01 4.80857670e-01 1.69171497e-01 1.15513362e-01
-8.60243365e-02 2.92545438e-01 -2.31900990e-01 -1.15996273e-02
9.83719468e-01 -2.52533078e-01 -4.37489152e-01 6.01401627e-01
5.67552030e-01 2.65488237e-01 -6.39652848e-01 -3.84130716e-01
6.46388113e-01 -3.57035518e-01 2.89357267e-02 -9.31486666e-01
-3.98843110e-01 7.64635682e-01 4.25310731e-01 4.80296016e-02
1.19358563e+00 -2.43702546e-01 1.07603610e+00 8.12362015e-01
4.21428889e-01 -1.42535436e+00 3.23408872e-01 5.03972173e-01
9.87143576e-01 -1.11380339e+00 3.12939137e-01 -6.12296462e-01
-3.01136047e-01 1.12103808e+00 9.57248986e-01 4.50278282e-01
5.05216956e-01 6.31612912e-03 -1.67298421e-01 -3.98620725e-01
-8.52308452e-01 8.71920511e-02 5.04807830e-01 -3.17413546e-02
4.64965910e-01 -3.26574892e-02 -1.16300786e+00 9.91801679e-01
-2.57190257e-01 2.71546364e-01 5.24086893e-01 1.40264714e+00
-1.06564820e+00 -1.10129809e+00 -2.93868631e-01 8.71709645e-01
-2.77347565e-01 -2.16793954e-01 -7.61830628e-01 4.82886463e-01
4.49091867e-02 1.24227142e+00 -3.36502194e-01 -3.92679542e-01
4.52124745e-01 4.49693531e-01 5.38445473e-01 -6.09270334e-01
-4.73383218e-01 -4.68576729e-01 3.35386485e-01 -1.53816223e-01
-4.58970517e-01 -4.61440504e-01 -1.67470574e+00 -1.19672023e-01
-7.79961884e-01 3.47300589e-01 5.51215768e-01 1.48870409e+00
2.10000843e-01 6.45518839e-01 3.17781538e-01 2.64397919e-01
-6.86097622e-01 -1.39254963e+00 -2.11713269e-01 4.55881536e-01
-2.45090034e-02 -7.57839322e-01 -3.10896128e-01 -4.33540866e-02] | [11.412104606628418, 8.293479919433594] |
6c06af3c-2376-49d0-883c-4c82d0d71070 | learning-joint-embedding-with-multimodal-cues | null | null | https://dl.acm.org/citation.cfm?id=3206064 | http://www.cs.cmu.edu/~fmetze/interACT/Publications_files/publications/ICMR2018_Camera_Ready.pdf | Learning Joint Embedding with Multimodal Cues for Cross-Modal Video-Text Retrieval | Constructing a joint representation invariant across different modalities (e.g., video, language) is of significant importance in many multimedia applications. While there are a number of recent successes in developing effective image-text retrieval methods by learning
joint representations, the video-text retrieval task, in contrast, has not been explored to its fullest extent. In this paper, we study how
to effectively utilize available multi-modal cues from videos for the cross-modal video-text retrieval task. Based on our analysis,
we propose a novel framework that simultaneously utilizes multimodal features (different visual characteristics, audio inputs, and text) by a fusion strategy for efficient retrieval. Furthermore, we explore several loss functions in training the joint embedding and propose a modified pairwise ranking loss for the retrieval task. Experiments on MSVD and MSR-VTT datasets demonstrate that our method achieves significant performance gain compared to the state-of-the-art approaches. | ['Amit K. Roy-Chowdhury', 'Niluthpol Chowdhury Mithun', 'Juncheng Li', 'Florian Metze'] | 2018-06-11 | null | null | null | icmr-2018-6 | ['video-text-retrieval'] | ['computer-vision'] | [ 2.62889504e-01 -8.51824522e-01 -3.33457470e-01 -2.62994409e-01
-1.61254692e+00 -4.51462656e-01 7.73973644e-01 5.07508256e-02
-4.44857985e-01 3.71541440e-01 3.34098607e-01 2.20015511e-01
-3.55467290e-01 -2.09007949e-01 -4.71048892e-01 -7.47351527e-01
2.39327222e-01 2.74688974e-02 1.27679929e-01 1.12496883e-01
2.93338537e-01 8.14434588e-02 -1.61485350e+00 3.51559103e-01
6.31064296e-01 1.21356833e+00 2.08915830e-01 3.72636944e-01
7.28631094e-02 6.17686987e-01 -2.08161458e-01 -4.78473663e-01
2.02909738e-01 -2.86818713e-01 -4.77156579e-01 1.34568229e-01
6.61465585e-01 -5.93895555e-01 -8.80073845e-01 8.08793902e-01
7.66095102e-01 4.28768903e-01 8.68545532e-01 -1.23010957e+00
-7.72735894e-01 2.65046716e-01 -8.28312159e-01 1.48060635e-01
5.57597041e-01 -3.84661317e-01 1.45383358e+00 -1.27103055e+00
5.53435385e-01 1.40315282e+00 1.90145716e-01 1.56878039e-01
-1.04630911e+00 -6.34760141e-01 2.18638852e-01 4.55665797e-01
-1.69602966e+00 -5.20080149e-01 8.40774715e-01 -3.61678988e-01
5.88750720e-01 2.35702947e-01 2.81503350e-01 1.10667741e+00
8.82626995e-02 1.27690899e+00 7.06423521e-01 -5.17843664e-01
-2.86701053e-01 2.71538235e-02 -9.14007574e-02 7.09976017e-01
6.13023937e-02 -3.90729934e-01 -7.74911284e-01 -3.14795345e-01
6.22896910e-01 3.45311344e-01 -4.72720236e-01 -6.83665991e-01
-1.34719586e+00 8.83918583e-01 1.34342805e-01 3.64551306e-01
-3.15348148e-01 3.60086471e-01 6.96740329e-01 3.32937658e-01
4.99322444e-01 -8.50751176e-02 -1.50772743e-02 -1.47720456e-01
-1.08986545e+00 1.87348947e-01 4.43512708e-01 8.56875956e-01
6.20965779e-01 -8.73523057e-02 -2.71442324e-01 1.29550636e+00
6.04713380e-01 7.72944570e-01 4.94241357e-01 -7.02670336e-01
6.44981802e-01 3.64470690e-01 -5.78903072e-02 -1.26112509e+00
2.31387690e-01 1.66047499e-01 -7.17583835e-01 -4.85688388e-01
-6.16170429e-02 2.36946508e-01 -7.00712979e-01 1.68310893e+00
1.11089908e-01 2.10575253e-01 1.74486358e-02 9.39942300e-01
7.90775716e-01 7.96459794e-01 -4.72025052e-02 -2.34283000e-01
1.16254723e+00 -9.72643495e-01 -1.00278687e+00 1.04229987e-01
3.16805303e-01 -1.09586394e+00 6.03533566e-01 1.47403032e-01
-1.15455139e+00 -3.87539893e-01 -9.39224184e-01 -2.86043704e-01
-2.38390744e-01 2.22684532e-01 4.49525148e-01 2.68018633e-01
-9.45469975e-01 5.47916442e-02 -7.69757211e-01 -5.31345367e-01
1.22657798e-01 2.24883825e-01 -5.74719310e-01 -5.84958971e-01
-1.26563668e+00 6.64716840e-01 1.74254194e-01 -1.03603832e-01
-8.68018925e-01 -2.82814175e-01 -9.56878841e-01 -5.30543327e-02
5.63608885e-01 -8.40169728e-01 9.68097448e-01 -8.52678835e-01
-1.27807200e+00 6.21544361e-01 -2.17784226e-01 -6.08137064e-02
4.06175822e-01 -3.04282933e-01 -4.00147051e-01 7.53994167e-01
1.34443462e-01 7.10968435e-01 1.40819192e+00 -1.24305487e+00
-5.12472153e-01 -4.33411777e-01 1.09612666e-01 5.27624846e-01
-7.81386614e-01 1.91852123e-01 -1.17565334e+00 -8.85372102e-01
1.45642295e-01 -1.00769281e+00 1.72534421e-01 3.55065614e-01
-1.78466871e-01 -3.35767895e-01 9.52658653e-01 -6.63902879e-01
1.39648736e+00 -2.27350450e+00 6.23791695e-01 1.41795561e-01
-1.82266459e-02 1.95136577e-01 -3.44647229e-01 9.08067584e-01
1.47257432e-01 5.37689812e-02 -7.73140863e-02 -7.07044303e-01
2.86610901e-01 9.52678472e-02 -4.28973466e-01 5.22566438e-01
-3.60359959e-02 9.37502265e-01 -7.46606052e-01 -9.16680515e-01
3.86201411e-01 9.46472645e-01 -4.70004469e-01 1.33334726e-01
1.56457767e-01 1.92215238e-02 -9.26105082e-01 8.70446563e-01
3.62232298e-01 -3.30037057e-01 1.03687532e-01 -5.18829107e-01
1.68107718e-01 -2.34022632e-01 -1.00552595e+00 2.16731429e+00
-3.74886930e-01 8.03699434e-01 1.16599917e-01 -1.12459803e+00
4.37543839e-01 5.07531583e-01 7.59421349e-01 -6.41880393e-01
1.15762822e-01 2.76755840e-01 -5.74284852e-01 -6.46183670e-01
8.19032848e-01 -1.29862204e-01 -1.50437921e-01 3.75306159e-01
2.14575320e-01 1.97759345e-01 1.61664590e-01 3.76741648e-01
8.63527119e-01 1.08926482e-01 2.24664122e-01 3.51767510e-01
5.40600061e-01 -4.07669067e-01 1.86068758e-01 6.71573043e-01
-2.74675339e-01 8.32638025e-01 4.06786241e-02 1.72228515e-01
-7.35906422e-01 -9.44664955e-01 -1.87116221e-01 1.26896966e+00
2.84163266e-01 -5.84820747e-01 -2.99819976e-01 -5.36078274e-01
1.07704550e-01 8.50602165e-02 -4.46031928e-01 -1.62195131e-01
-2.59784132e-01 -3.32103610e-01 4.83750433e-01 3.43308061e-01
4.44367588e-01 -5.87219715e-01 -1.56753838e-01 -1.53561169e-02
-5.14495432e-01 -1.34323716e+00 -8.54843974e-01 -4.64344680e-01
-7.53314257e-01 -8.12698543e-01 -1.20639801e+00 -8.64132643e-01
3.23514551e-01 1.07586884e+00 7.14405477e-01 -2.24849880e-02
-4.03520286e-01 1.40359235e+00 -6.85155988e-01 1.07608028e-01
7.82873183e-02 -1.73277304e-01 4.96694222e-02 2.84942091e-01
2.50804394e-01 -3.18008035e-01 -7.49346316e-01 2.90810645e-01
-1.46279323e+00 -3.02690536e-01 6.78396761e-01 9.13567483e-01
7.36456335e-01 -1.51100561e-01 5.52049518e-01 -3.84347618e-01
5.37136614e-01 -6.04944110e-01 -1.70584396e-01 5.74968278e-01
-2.52850264e-01 -1.70442536e-01 5.37872873e-02 -4.93857771e-01
-8.88526440e-01 -2.69558668e-01 2.02124864e-01 -9.58563447e-01
3.04662764e-01 9.81769979e-01 -9.01552960e-02 -1.94570467e-01
-3.46769020e-02 4.02535796e-01 7.80223683e-02 -3.72905642e-01
4.26910251e-01 7.25289345e-01 1.62786633e-01 -6.78029716e-01
8.76195729e-01 5.48462152e-01 -7.29643777e-02 -1.15954459e+00
-6.02273822e-01 -9.48706865e-01 -3.44442010e-01 -2.87382215e-01
9.08737659e-01 -1.31955552e+00 -3.59460145e-01 2.17722327e-01
-9.25729930e-01 3.88501525e-01 3.06284040e-01 8.04995120e-01
-5.47902465e-01 9.91684794e-01 -4.25629467e-01 -7.76222408e-01
-2.74956673e-01 -1.25668788e+00 1.49749947e+00 -2.33543247e-01
2.19734147e-01 -9.41583335e-01 1.08764812e-01 6.73397899e-01
3.33320320e-01 -1.18916994e-02 7.62994468e-01 -5.20413041e-01
-9.73107338e-01 -4.69742239e-01 -4.31513071e-01 3.69053692e-01
1.83363169e-01 1.01440758e-01 -7.28111446e-01 -5.82409739e-01
-2.86935598e-01 -6.53728306e-01 1.31813061e+00 3.68141145e-01
1.12150478e+00 -4.25198674e-03 -3.48890722e-01 2.55047083e-01
1.52877593e+00 4.46960405e-02 5.64640462e-01 5.49952872e-02
7.20872462e-01 4.41701680e-01 8.06374431e-01 6.15990043e-01
4.22163218e-01 9.96842921e-01 2.93987870e-01 2.62859195e-01
-2.63109468e-02 -1.78040117e-01 5.31424284e-01 1.05722928e+00
1.28587574e-01 -5.41881740e-01 -4.99790937e-01 5.92286527e-01
-2.14585400e+00 -1.11018610e+00 4.28429782e-01 2.13865876e+00
5.13996899e-01 -3.88914645e-01 -1.15467750e-01 8.52713361e-02
7.26069093e-01 5.02852380e-01 -3.44550610e-01 2.04288095e-01
-1.63495928e-01 -1.05803631e-01 8.09244290e-02 2.29029357e-01
-1.29543495e+00 7.69086540e-01 5.71656704e+00 1.26807702e+00
-1.00866091e+00 8.15913752e-02 6.92215040e-02 -1.38330340e-01
-4.57479626e-01 -1.10585742e-01 -4.94732559e-01 9.85219553e-02
4.26192105e-01 -1.89883456e-01 3.14226359e-01 4.36040789e-01
-4.89789322e-02 1.26990274e-01 -1.12993169e+00 1.47057474e+00
6.95731938e-01 -9.28032756e-01 6.19623244e-01 -8.98369104e-02
7.19348192e-01 -1.79446056e-01 5.01896083e-01 4.20197666e-01
-3.77004385e-01 -6.33648396e-01 6.32782161e-01 5.98852754e-01
7.87011743e-01 -6.34684741e-01 4.15032327e-01 -1.24344649e-02
-1.35970807e+00 1.26004908e-02 -2.76457727e-01 5.39997935e-01
2.90004194e-01 3.51693571e-01 -2.02653885e-01 1.05205953e+00
5.90225816e-01 1.13023567e+00 -5.26118875e-01 1.09298563e+00
3.16016883e-01 2.73472875e-01 -1.76017985e-01 6.46611899e-02
2.71076351e-01 -1.21023200e-01 7.16452658e-01 1.22272670e+00
5.42254686e-01 -4.77656610e-02 4.09033835e-01 3.29045475e-01
-3.40426415e-01 5.65346241e-01 -7.86000192e-01 -5.07418454e-01
3.90097111e-01 1.13616407e+00 -2.77838826e-01 -8.63197744e-02
-8.06932569e-01 1.06174648e+00 9.10369158e-02 7.42052794e-01
-6.93968654e-01 -2.86814272e-01 6.28616154e-01 -3.63115609e-01
5.60096085e-01 -2.60481268e-01 5.48497140e-01 -1.48788953e+00
2.69743770e-01 -8.80746961e-01 4.89109248e-01 -7.81412959e-01
-1.53812349e+00 2.02197909e-01 1.62550360e-01 -1.53317583e+00
-1.82635322e-01 -4.31538105e-01 -7.01889023e-02 5.93375087e-01
-1.78058410e+00 -1.52075148e+00 1.33661795e-02 8.84460926e-01
5.99074483e-01 -2.72901744e-01 7.13334620e-01 7.14090705e-01
-4.71444964e-01 7.06803918e-01 5.41938901e-01 1.25438645e-01
1.16492319e+00 -6.25043929e-01 -5.16354978e-01 5.18058777e-01
2.95435071e-01 8.12962472e-01 3.30009580e-01 -3.20468783e-01
-2.00847960e+00 -8.91251564e-01 5.42254686e-01 -6.95045143e-02
8.20097089e-01 -6.83443770e-02 -7.93546081e-01 6.95949078e-01
4.67977732e-01 -8.11663195e-02 1.01185083e+00 -9.20168459e-02
-7.38270521e-01 -2.31942341e-01 -7.58715510e-01 6.96900368e-01
5.69947541e-01 -1.02156484e+00 -4.50085104e-01 2.22478583e-01
6.27067387e-01 -3.43653969e-02 -9.99210298e-01 4.73278821e-01
9.29568410e-01 -6.64258599e-01 1.24077547e+00 -3.26655686e-01
3.98673594e-01 -1.48351222e-01 -8.67222726e-01 -9.29202437e-01
-4.60720509e-02 -4.61670935e-01 -9.87787470e-02 1.36816216e+00
-2.17159558e-02 -4.13104534e-01 2.45960444e-01 3.37394744e-01
5.90010639e-03 -6.39811337e-01 -1.19588029e+00 -5.50115883e-01
-1.23870604e-01 -3.85977566e-01 -1.84486695e-02 9.17664409e-01
-1.58496361e-04 3.20442110e-01 -7.81029463e-01 1.91263780e-02
7.37544417e-01 3.77609491e-01 6.73185408e-01 -1.03551888e+00
-2.31500521e-01 -5.29379904e-01 -6.80233061e-01 -1.32458615e+00
4.95013535e-01 -8.85830820e-01 -6.67031780e-02 -1.60361624e+00
8.34301412e-01 -1.13307266e-02 -5.71213901e-01 2.15394586e-01
-2.46607676e-01 3.60447705e-01 3.93862814e-01 4.70311910e-01
-1.22963941e+00 9.93010640e-01 1.16552138e+00 -4.18360353e-01
2.81378478e-01 -4.58629251e-01 -5.58895469e-01 4.03501153e-01
2.27536082e-01 -1.99705422e-01 -4.40598279e-01 -6.55789912e-01
2.78563708e-01 3.43452603e-01 3.29538763e-01 -7.27719963e-01
2.71753967e-01 5.88542670e-02 1.53787658e-01 -7.10521638e-01
9.94572639e-01 -9.50739741e-01 -1.06054299e-01 -1.54780790e-01
-5.06404281e-01 7.67198019e-03 6.53911158e-02 1.17590773e+00
-7.25627840e-01 -1.74757048e-01 3.86618525e-01 1.31392047e-01
-6.45254016e-01 6.03524089e-01 -3.80853534e-01 -8.45756158e-02
7.65701294e-01 -1.31739020e-01 -1.10143661e-01 -7.31810212e-01
-4.75581646e-01 4.71422642e-01 3.33380610e-01 7.27045476e-01
1.04559684e+00 -1.64922881e+00 -6.76201582e-01 -1.63194209e-01
5.98101974e-01 -5.30338109e-01 5.08849680e-01 9.01499629e-01
5.86615019e-02 6.43785834e-01 7.98191205e-02 -6.97865903e-01
-1.54710972e+00 4.83636498e-01 -6.89973086e-02 -2.63016105e-01
-2.93946981e-01 4.81811196e-01 3.65013272e-01 1.12536913e-02
4.22120184e-01 2.16543645e-01 -2.80677706e-01 3.92817289e-01
5.66891730e-01 3.18158686e-01 -1.24866344e-01 -9.61712956e-01
-1.70666814e-01 9.77689803e-01 -3.72531891e-01 -3.68198931e-01
1.18371642e+00 -5.72906315e-01 -5.32645658e-02 7.10502684e-01
1.74866343e+00 -1.70837581e-01 -7.42863774e-01 -6.69244528e-01
-2.75027454e-01 -7.45368898e-01 2.06451207e-01 -2.82744020e-01
-1.11768889e+00 9.72805500e-01 6.32492363e-01 -4.60320339e-03
1.26112854e+00 -8.43380988e-02 9.04828310e-01 6.52925670e-01
3.63861173e-01 -1.11114919e+00 6.44297957e-01 3.97096097e-01
8.86217177e-01 -1.50980031e+00 1.35758549e-01 -1.71956167e-01
-5.71120739e-01 1.01561999e+00 8.37703496e-02 4.84779030e-02
7.42266059e-01 -4.57027674e-01 -1.32475048e-01 -1.27812847e-01
-7.97541738e-01 -2.65190214e-01 5.95245838e-01 2.46965066e-01
5.38526833e-01 -1.96781486e-01 -3.58235151e-01 1.24028459e-01
6.93777025e-01 -1.58452749e-01 9.64713171e-02 1.18961096e+00
-4.36095387e-01 -1.19427705e+00 -3.68293345e-01 4.43197608e-01
-8.06702316e-01 -1.10815816e-01 -3.04509997e-01 6.94061875e-01
-5.87161720e-01 1.03250849e+00 -2.95211822e-01 -3.73318583e-01
1.24125347e-01 2.11291209e-01 5.96883595e-01 -3.86827171e-01
-1.44228145e-01 6.23607099e-01 -1.90059423e-01 -4.37411457e-01
-1.04157186e+00 -7.63551891e-01 -8.20254147e-01 2.42881067e-02
-4.92199600e-01 4.93471418e-03 6.33320451e-01 8.19080770e-01
2.63818890e-01 2.48428956e-01 7.60950446e-01 -1.10570323e+00
-5.36055088e-01 -8.05189431e-01 -7.11542904e-01 4.77880508e-01
4.34523731e-01 -9.73491430e-01 -4.52638090e-01 -2.67925560e-02] | [10.442986488342285, 0.9765657186508179] |
8a6da194-1831-4a7d-94e9-21403e963035 | your-local-gan-designing-two-dimensional | 1911.12287 | null | https://arxiv.org/abs/1911.12287v2 | https://arxiv.org/pdf/1911.12287v2.pdf | Your Local GAN: Designing Two Dimensional Local Attention Mechanisms for Generative Models | We introduce a new local sparse attention layer that preserves two-dimensional geometry and locality. We show that by just replacing the dense attention layer of SAGAN with our construction, we obtain very significant FID, Inception score and pure visual improvements. FID score is improved from $18.65$ to $15.94$ on ImageNet, keeping all other parameters the same. The sparse attention patterns that we propose for our new layer are designed using a novel information theoretic criterion that uses information flow graphs. We also present a novel way to invert Generative Adversarial Networks with attention. Our method extracts from the attention layer of the discriminator a saliency map, which we use to construct a new loss function for the inversion. This allows us to visualize the newly introduced attention heads and show that they indeed capture interesting aspects of two-dimensional geometry of real images. | ['Han Zhang', 'Augustus Odena', 'Giannis Daras', 'Alexandros G. Dimakis'] | 2019-11-27 | your-local-gan-designing-two-dimensional-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Daras_Your_Local_GAN_Designing_Two_Dimensional_Local_Attention_Mechanisms_for_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Daras_Your_Local_GAN_Designing_Two_Dimensional_Local_Attention_Mechanisms_for_CVPR_2020_paper.pdf | cvpr-2020-6 | ['deep-attention', 'deep-attention'] | ['computer-vision', 'natural-language-processing'] | [ 1.52348485e-02 4.10792470e-01 3.11512291e-01 -5.96717708e-02
-5.74096322e-01 -6.72802389e-01 6.21010005e-01 -3.42340916e-01
-2.46603608e-01 6.67820036e-01 2.65873224e-01 -1.74618527e-01
-1.00986809e-01 -9.73161340e-01 -1.15146637e+00 -6.49862111e-01
-2.29256898e-01 1.13172047e-01 2.00760499e-01 -3.95158887e-01
2.90916920e-01 9.05097008e-01 -1.12891424e+00 9.61699411e-02
7.70366251e-01 1.01247907e+00 -1.23394527e-01 8.09200346e-01
2.72704124e-01 9.46787000e-01 -5.98678827e-01 -5.82717001e-01
6.79110050e-01 -5.54962218e-01 -9.30224180e-01 -9.37677026e-02
9.01309192e-01 -3.19432497e-01 -6.24194860e-01 9.88939226e-01
4.74634469e-01 3.88400331e-02 6.62473798e-01 -1.12170517e+00
-1.03622484e+00 4.77270991e-01 -7.10419238e-01 3.14869285e-01
5.56279421e-02 2.49506235e-01 1.17497623e+00 -8.47038507e-01
9.26855922e-01 1.32352602e+00 8.54959309e-01 2.33589083e-01
-1.54972959e+00 -4.81970161e-01 3.03316452e-02 2.99110293e-01
-1.40455770e+00 -1.86852857e-01 9.88900304e-01 -2.92341769e-01
7.67487288e-01 1.33149743e-01 5.29021144e-01 9.10944879e-01
4.61209953e-01 6.09457016e-01 9.15085375e-01 -3.83385569e-01
1.54226292e-02 5.48492372e-02 -2.45534629e-01 1.11728823e+00
2.30873376e-01 2.14713737e-01 -1.23308532e-01 4.14844304e-01
1.17314482e+00 1.05733842e-01 -2.14621693e-01 -7.57848382e-01
-1.11910951e+00 1.06906831e+00 1.35006416e+00 3.84302437e-01
-2.89520800e-01 4.24373597e-01 -6.21630885e-02 2.08631977e-01
1.46534801e-01 9.80666757e-01 1.50856346e-01 3.61741871e-01
-6.99175715e-01 8.63547325e-02 5.09223878e-01 9.04383719e-01
1.01526761e+00 3.20772886e-01 -3.57921213e-01 4.63050187e-01
-1.30919009e-01 6.83602154e-01 9.91956741e-02 -1.25055075e+00
1.29269585e-01 4.09603000e-01 -1.20525241e-01 -1.52196074e+00
-4.55518097e-01 -7.91230500e-01 -1.14583910e+00 3.53799820e-01
2.68887967e-01 -1.77878991e-01 -9.86711800e-01 1.94665456e+00
-1.33972302e-01 2.12732509e-01 -1.42087176e-01 8.80018115e-01
4.45755005e-01 4.94183362e-01 -1.69930667e-01 4.71036971e-01
9.56686854e-01 -1.06985247e+00 -3.11562032e-01 -2.40489259e-01
4.76679653e-01 -3.68908137e-01 1.24014986e+00 1.69731975e-01
-1.42158878e+00 -5.20082772e-01 -1.39591873e+00 -3.16879272e-01
-4.11780208e-01 -1.37235507e-01 8.07864130e-01 4.92260247e-01
-1.35733128e+00 8.33209932e-01 -6.06043696e-01 -6.60192594e-02
7.20095098e-01 3.72216672e-01 -3.24182540e-01 -1.15708098e-01
-1.09644079e+00 7.27121711e-01 -5.42267151e-02 -2.76497066e-01
-1.16875136e+00 -8.41832280e-01 -1.00527799e+00 3.30072999e-01
2.59271681e-01 -8.18678677e-01 7.60128260e-01 -9.71391678e-01
-1.32138920e+00 7.85842538e-01 2.12581068e-01 -6.31986797e-01
3.62611145e-01 -5.68245649e-02 -1.96961332e-02 3.88546914e-01
2.10440114e-01 9.01622891e-01 8.96632671e-01 -1.40298498e+00
-1.81400225e-01 -2.28642032e-01 3.12687695e-01 -6.15273006e-02
-2.37246364e-01 -4.56406564e-01 -3.21806818e-01 -8.96777451e-01
-1.15280762e-01 -9.26121235e-01 -3.23687881e-01 5.74275367e-02
-7.11584747e-01 5.07300735e-01 8.15538347e-01 -6.28749907e-01
7.94030011e-01 -2.03621125e+00 3.33313853e-01 4.07873094e-01
6.85610533e-01 1.73341215e-01 -3.82850021e-01 9.84935462e-02
-3.47328335e-01 3.60642016e-01 -6.16590917e-01 -2.95489222e-01
-1.10315062e-01 -4.15133275e-02 -2.64742643e-01 4.30232078e-01
5.31316459e-01 1.28383017e+00 -8.43922257e-01 -8.60245675e-02
3.53950232e-01 6.06386662e-01 -1.00884163e+00 2.40663812e-01
4.01915722e-02 3.41847509e-01 -1.20340005e-01 1.75141320e-01
8.55773568e-01 -4.56164688e-01 -3.41775231e-02 -2.15659454e-01
1.64323762e-01 4.39714789e-02 -8.51926804e-01 1.99018562e+00
-6.87724292e-01 6.94484472e-01 1.06952693e-02 -1.14789748e+00
7.36675918e-01 -2.56454766e-01 3.37922007e-01 -9.78194296e-01
1.88514978e-01 -1.76097304e-01 -2.06783880e-02 -2.23387275e-02
2.36355022e-01 -2.71890014e-01 -2.10820690e-01 5.21446884e-01
3.78988028e-01 -2.86663622e-01 5.73446937e-02 6.38187349e-01
1.33308256e+00 -1.80911258e-01 -3.32077891e-02 -4.98245537e-01
5.56449652e-01 -2.53915012e-01 1.36622876e-01 8.08517277e-01
2.07041517e-01 9.10961211e-01 8.94171417e-01 -5.22293031e-01
-1.33100414e+00 -1.36479187e+00 1.04761280e-01 8.16755295e-01
1.56515270e-01 -1.65070325e-01 -9.17019486e-01 -9.06279087e-01
-2.96962727e-02 5.60074210e-01 -1.10089111e+00 -5.60435832e-01
-6.31567419e-01 -4.53207731e-01 5.01718223e-01 5.29737234e-01
7.60532081e-01 -9.40703213e-01 -3.35397959e-01 -1.10623658e-01
1.24443799e-01 -1.00019491e+00 -6.97802067e-01 1.27047867e-01
-5.49383640e-01 -9.40614998e-01 -9.60287869e-01 -7.45408058e-01
7.99649298e-01 1.32501364e-01 1.39517057e+00 8.29995945e-02
-5.71154475e-01 1.69475496e-01 -6.53861463e-02 -1.43863857e-01
-1.86160862e-01 3.91812116e-01 -3.43665659e-01 1.25681847e-01
-2.16567859e-01 -7.22617865e-01 -7.83357978e-01 1.75828129e-01
-8.96476567e-01 7.70730078e-02 7.94200838e-01 8.15773904e-01
3.38862151e-01 -2.74699777e-01 4.67805743e-01 -9.74452496e-01
3.97167951e-01 -4.73823279e-01 -6.12659574e-01 -1.11408584e-01
-3.92270416e-01 5.74128866e-01 1.01648164e+00 4.64016162e-02
-7.87805498e-01 7.14459782e-03 -2.30134457e-01 -6.46991253e-01
1.81011669e-02 -3.91516462e-02 -1.94529966e-01 -4.95227844e-01
8.15616310e-01 -3.76425683e-02 6.95770308e-02 -3.55163604e-01
6.63265646e-01 -2.54732762e-02 7.80434966e-01 -3.26986343e-01
1.08596718e+00 6.87729478e-01 2.86473006e-01 -4.91937250e-01
-9.19173479e-01 8.62970352e-02 -6.10075831e-01 1.76999018e-01
9.18717325e-01 -7.07351804e-01 -9.53048706e-01 3.02068442e-01
-1.00694060e+00 -3.19894940e-01 -7.36750126e-01 5.54112121e-02
-7.11352885e-01 1.39518127e-01 -4.84789103e-01 -1.81913957e-01
-2.17752039e-01 -1.10736299e+00 1.03339779e+00 -8.33884906e-03
1.01481222e-01 -1.07554018e+00 9.54586565e-02 -1.01038329e-01
7.38715768e-01 4.89074439e-01 9.28245366e-01 -2.05484495e-01
-9.79645908e-01 7.65435100e-02 -7.40716159e-01 4.57475692e-01
4.73140851e-02 -2.22596601e-01 -9.45229828e-01 -3.26549500e-01
-5.14871292e-02 -2.02572748e-01 1.20538902e+00 4.73767430e-01
1.53983331e+00 -5.00138640e-01 -6.10785224e-02 1.33305764e+00
1.63002646e+00 3.29012051e-02 1.00870669e+00 -8.42366461e-03
1.01449835e+00 2.64124304e-01 -1.63055241e-01 1.12228177e-01
1.31164417e-01 6.52271926e-01 6.73070848e-01 -6.33848548e-01
-4.24959630e-01 -3.94698322e-01 1.85683683e-01 4.52827841e-01
-2.06522316e-01 -1.06169075e-01 -7.26641059e-01 5.25563896e-01
-1.55873752e+00 -1.03328776e+00 3.91018540e-01 2.15505242e+00
6.80153728e-01 2.39100099e-01 1.24097914e-01 -5.52498177e-02
5.00491321e-01 3.18618476e-01 -5.34563422e-01 -5.16591966e-01
-2.98430115e-01 7.72655189e-01 7.03517616e-01 9.86597717e-01
-1.12689435e+00 9.41089034e-01 7.02872038e+00 6.59189999e-01
-1.18287063e+00 -4.18699160e-02 7.99263597e-01 -1.21264823e-01
-6.10294461e-01 -2.77415752e-01 -3.48869413e-01 2.88939685e-01
7.74003983e-01 -2.85271436e-01 5.07120609e-01 6.82043433e-01
-1.93949252e-01 2.54021913e-01 -1.01064420e+00 8.85658324e-01
2.16354191e-01 -1.89191270e+00 3.43235552e-01 1.54619128e-01
9.50084448e-01 -1.13647044e-01 4.47732508e-01 4.91751954e-02
5.11200309e-01 -1.51958048e+00 5.52554250e-01 6.03229940e-01
9.28123057e-01 -1.10607100e+00 6.84005499e-01 -2.63083130e-01
-8.95172656e-01 -2.04367880e-02 -3.14467847e-01 9.09796879e-02
-3.32305916e-02 5.07564306e-01 -7.43199289e-01 3.86954904e-01
4.25291270e-01 8.98324907e-01 -7.24027634e-01 8.87942195e-01
-3.31785262e-01 2.07346246e-01 -2.90105730e-01 2.35568255e-01
5.52307725e-01 -5.49280345e-02 5.88578582e-01 1.13518262e+00
3.34087074e-01 -3.39029491e-01 -2.96403944e-01 1.46175337e+00
-6.36563718e-01 -5.58011793e-02 -1.02878630e+00 2.80925721e-01
-6.82292953e-02 1.09922862e+00 -5.55385232e-01 -2.10652575e-01
-2.08640218e-01 1.28258359e+00 3.75176102e-01 3.89850527e-01
-1.00587058e+00 -8.47875595e-01 7.04528153e-01 3.02078545e-01
5.96665084e-01 -1.49472713e-01 -3.61678213e-01 -1.08051550e+00
-1.43542543e-01 -6.38097823e-01 4.25249115e-02 -7.16623127e-01
-1.09523261e+00 6.53234839e-01 -2.13117003e-01 -7.96206594e-01
-2.31132418e-01 -6.13097370e-01 -8.51486802e-01 9.20594335e-01
-1.36715102e+00 -9.30532634e-01 -4.56353158e-01 8.61753106e-01
1.18469678e-01 -1.77751020e-01 6.43409073e-01 3.19306612e-01
-4.16896999e-01 9.45353270e-01 -2.75910813e-02 3.30712050e-01
4.24463063e-01 -1.49443436e+00 8.92649710e-01 1.01651514e+00
2.92984217e-01 5.34184575e-01 3.92131329e-01 -1.70294121e-01
-1.20482838e+00 -1.07628918e+00 4.67282832e-01 -4.07068551e-01
5.12536883e-01 -6.08011723e-01 -7.62358546e-01 7.95290351e-01
4.32851374e-01 1.57925561e-01 2.32564136e-01 -1.74101561e-01
-5.91248035e-01 -2.14815289e-01 -1.34458375e+00 6.02673352e-01
1.22359180e+00 -4.31511283e-01 -2.67594099e-01 2.23571762e-01
1.09712005e+00 -1.08939707e-01 -6.53049409e-01 1.10798493e-01
2.67781585e-01 -1.30598748e+00 1.41667998e+00 -7.39352167e-01
7.40554452e-01 -2.15642434e-02 3.91890481e-02 -1.41582656e+00
-6.93016648e-01 -8.12774837e-01 3.65876108e-02 5.70210576e-01
5.06591558e-01 -7.09729910e-01 9.32933688e-01 2.04869419e-01
-2.35167891e-01 -6.36996806e-01 -6.65039480e-01 -7.86349356e-01
3.36193204e-01 -1.40018404e-01 5.29776096e-01 8.37279379e-01
-2.82035917e-01 5.51506937e-01 -4.40017492e-01 -8.03681761e-02
8.65491867e-01 8.49861093e-03 6.11829638e-01 -9.40941215e-01
-3.56359065e-01 -5.71462154e-01 -6.64051116e-01 -1.07981491e+00
-1.19429842e-01 -9.71020043e-01 -3.05942923e-01 -1.21463954e+00
1.47445455e-01 -3.11044991e-01 -4.55711365e-01 4.18632507e-01
9.98274833e-02 7.85253227e-01 4.02151436e-01 -1.56049326e-01
-3.64644051e-01 6.66925788e-01 1.51162648e+00 -2.43086457e-01
2.25861799e-02 -3.19771737e-01 -1.27596545e+00 6.66369021e-01
7.70043671e-01 -3.49826902e-01 -3.87692779e-01 -5.21897078e-01
1.42513290e-01 -4.25079435e-01 8.33513439e-01 -1.32926679e+00
-7.89986998e-02 6.10807911e-02 5.90770125e-01 -2.87524667e-02
2.87641913e-01 -6.99238479e-01 -6.06964119e-02 4.69844609e-01
-3.49017769e-01 -1.41558439e-01 3.54424447e-01 3.27128828e-01
-1.38217375e-01 1.65848076e-01 1.21093500e+00 -1.18204422e-01
-4.29279119e-01 4.99479890e-01 9.62845236e-02 4.04565483e-01
9.59677517e-01 9.46740359e-02 -3.27539235e-01 -6.65125251e-01
-9.18583691e-01 -1.03394628e-01 6.71640694e-01 1.90179437e-01
4.99801725e-01 -1.59588611e+00 -7.35312521e-01 6.22992933e-01
-2.18620956e-01 -6.67829812e-02 2.29077742e-01 6.08133852e-01
-9.55882788e-01 4.86594796e-01 -7.86828637e-01 -4.08066660e-01
-6.08869433e-01 7.45126903e-01 4.07138109e-01 -4.45278049e-01
-6.99106395e-01 9.23731148e-01 7.01970875e-01 -1.34296775e-01
8.25372785e-02 -2.80866534e-01 1.53638974e-01 -2.45338589e-01
3.05835873e-01 1.67668387e-01 9.05784294e-02 -3.98109406e-01
-3.32113564e-01 7.33780503e-01 7.94627219e-02 -3.69405141e-03
1.52849293e+00 -1.74280219e-02 1.46572858e-01 -1.73295587e-01
1.69054759e+00 1.79348096e-01 -1.49582016e+00 -1.05446100e-01
-4.97925997e-01 -6.13079488e-01 9.89244599e-03 -6.93057418e-01
-1.65293181e+00 1.27130926e+00 5.08709610e-01 3.20214957e-01
1.18095779e+00 3.71397547e-02 8.46661866e-01 3.74153852e-01
1.52185664e-01 -5.54871082e-01 3.71691555e-01 6.31063938e-01
1.30017006e+00 -9.32300627e-01 -1.87584311e-01 -2.37207457e-01
-3.70185107e-01 5.65223277e-01 4.53916520e-01 -8.10689628e-01
6.64741695e-01 4.23881650e-01 -2.94310421e-01 -2.91643173e-01
-3.53908390e-01 -2.78355867e-01 3.72377425e-01 8.69516015e-01
1.11402415e-01 -2.34499529e-01 1.82797641e-01 2.84384489e-01
-3.55589986e-01 -3.77810597e-01 5.00981927e-01 3.97336930e-01
-3.13975751e-01 -8.07779789e-01 -6.06297888e-02 2.66489148e-01
-4.16032821e-01 -2.98013359e-01 -4.77175623e-01 9.13546860e-01
-3.60001735e-02 2.27153420e-01 3.63524765e-01 -5.16786397e-01
4.50551271e-01 -1.75516114e-01 5.39271116e-01 -4.89607185e-01
-3.95507932e-01 -3.74126315e-01 -3.23715329e-01 -9.12864745e-01
1.88427418e-02 -1.45535782e-01 -1.01166070e+00 -6.76872611e-01
2.48029009e-01 -6.47474034e-03 4.45437551e-01 3.79018277e-01
7.53947556e-01 8.06777716e-01 8.93567860e-01 -1.15885878e+00
-2.47917101e-02 -6.19124234e-01 -5.58725297e-01 6.08254731e-01
5.72004437e-01 -7.28115141e-01 -5.86004019e-01 3.94574478e-02] | [11.454115867614746, -0.5203142762184143] |
d5f3d06f-00e3-4e86-a2e8-d64e46f88b5e | clanet-a-comprehensive-framework-for-cross | 2306.16538 | null | https://arxiv.org/abs/2306.16538v1 | https://arxiv.org/pdf/2306.16538v1.pdf | CLANet: A Comprehensive Framework for Cross-Batch Cell Line Identification Using Brightfield Images | Cell line authentication plays a crucial role in the biomedical field, ensuring researchers work with accurately identified cells. Supervised deep learning has made remarkable strides in cell line identification by studying cell morphological features through cell imaging. However, batch effects, a significant issue stemming from the different times at which data is generated, lead to substantial shifts in the underlying data distribution, thus complicating reliable differentiation between cell lines from distinct batch cultures. To address this challenge, we introduce CLANet, a pioneering framework for cross-batch cell line identification using brightfield images, specifically designed to tackle three distinct batch effects. We propose a cell cluster-level selection method to efficiently capture cell density variations, and a self-supervised learning strategy to manage image quality variations, thus producing reliable patch representations. Additionally, we adopt multiple instance learning(MIL) for effective aggregation of instance-level features for cell line identification. Our innovative time-series segment sampling module further enhances MIL's feature-learning capabilities, mitigating biases from varying incubation times across batches. We validate CLANet using data from 32 cell lines across 93 experimental batches from the AstraZeneca Global Cell Bank. Our results show that CLANet outperforms related approaches (e.g. domain adaptation, MIL), demonstrating its effectiveness in addressing batch effects in cell line identification. | ['Huiyu Zhou', 'Yinhai Wang', 'Jonathan Orme', 'Kerry Hallbrook', 'Navin Rathna Kumar', 'Adam Corrigan', 'Lei Tong'] | 2023-06-28 | null | null | null | null | ['self-supervised-learning', 'multiple-instance-learning'] | ['computer-vision', 'methodology'] | [-5.16761746e-03 -8.61294448e-01 -1.04343750e-01 3.39348055e-02
-1.06007552e+00 -5.58612585e-01 4.90429908e-01 8.04659545e-01
-5.24913311e-01 9.42086816e-01 -1.30281478e-01 -4.13282253e-02
-1.82966769e-01 -6.43852174e-01 -4.90668118e-01 -1.36856616e+00
3.11243329e-02 4.95081455e-01 -1.55503109e-01 -9.93449688e-02
4.61199373e-01 9.79523778e-01 -1.05045354e+00 3.36780965e-01
6.72663033e-01 9.27925825e-01 -6.87298924e-02 7.59172082e-01
-2.91691363e-01 6.44129634e-01 -9.03241575e-01 9.23283994e-02
-1.20227158e-01 -2.17991039e-01 -5.40367901e-01 3.47229958e-01
3.16009343e-01 3.15656401e-02 -6.81551769e-02 5.92789769e-01
9.03022468e-01 -3.55476975e-01 1.02260125e+00 -1.16391671e+00
-3.04488331e-01 6.10887222e-02 -6.05016828e-01 4.10007924e-01
1.30065948e-01 3.65621328e-01 5.47434390e-01 -8.33173096e-01
6.20659530e-01 6.40178263e-01 1.06230950e+00 3.78023118e-01
-1.68251848e+00 -6.02910578e-01 -1.31620571e-01 5.86825609e-02
-1.66958416e+00 -5.51684201e-01 4.61458683e-01 -6.35126889e-01
8.75592232e-01 1.43726617e-01 7.09218502e-01 1.10350358e+00
5.89619339e-01 6.02534056e-01 1.10717082e+00 -2.88838506e-01
4.69731271e-01 1.19350024e-01 -3.37984152e-02 2.21366271e-01
1.67101279e-01 -2.82319009e-01 -2.85790712e-01 -2.76656926e-01
9.36824501e-01 2.27455422e-01 -4.21730310e-01 -2.73829609e-01
-1.35388136e+00 5.14673829e-01 4.35809940e-02 7.38692403e-01
-2.27739185e-01 -1.10892482e-01 5.99595845e-01 2.62393236e-01
2.25970075e-01 6.32643700e-01 -6.85959160e-01 -3.61661310e-03
-1.06484222e+00 -5.01836911e-02 6.16538525e-01 5.99858940e-01
4.71816719e-01 -7.45099187e-02 -4.71842349e-01 1.05479145e+00
-9.37588289e-02 2.71519244e-01 9.12362397e-01 -5.37173450e-01
-2.98719853e-01 7.42048442e-01 4.53321822e-02 -8.60790253e-01
-8.56398463e-01 -7.13454306e-01 -1.18905008e+00 -5.34261204e-03
8.11724186e-01 1.74647570e-01 -7.85817981e-01 1.39176357e+00
2.79960781e-01 3.68755788e-01 -6.71830634e-03 5.06019115e-01
7.29132950e-01 2.80484974e-01 2.91855931e-01 -4.91119951e-01
1.44212449e+00 -3.46827090e-01 -5.91413021e-01 4.35591906e-01
1.13188851e+00 -6.13743663e-01 7.98879862e-01 5.70367992e-01
-6.31670356e-01 -3.78001571e-01 -8.54240835e-01 2.43052006e-01
-7.53211796e-01 1.08257703e-01 6.40882909e-01 5.26515126e-01
-9.37954068e-01 4.90859360e-01 -5.14308214e-01 -4.92207646e-01
1.03037608e+00 7.66433656e-01 -5.86710930e-01 1.13870958e-02
-7.73515940e-01 3.81377816e-01 8.77392665e-02 6.63298219e-02
-5.02620459e-01 -1.29964542e+00 -7.11870432e-01 -5.95826143e-03
-1.97858363e-01 -6.96694613e-01 6.54971063e-01 -7.43540466e-01
-1.55107486e+00 1.06341064e+00 -1.33646369e-01 -5.61291218e-01
2.73803771e-01 4.07984972e-01 -5.01982450e-01 9.61123854e-02
-2.08705086e-02 5.30590832e-01 4.74377424e-01 -1.31604052e+00
-6.49602473e-01 -3.27219397e-01 -6.87899590e-01 -1.99635237e-01
-3.93435001e-01 -1.30251318e-01 -2.74091959e-01 -6.79528534e-01
-2.02771500e-01 -6.61458850e-01 -3.37991744e-01 -3.23074788e-01
-2.32855678e-01 -3.75108235e-02 5.67243099e-01 -4.44692701e-01
1.08856976e+00 -2.09483576e+00 -2.67432451e-01 2.84312338e-01
4.07898635e-01 1.71177417e-01 -1.02497630e-01 3.42311800e-01
4.07025442e-02 6.34533837e-02 -9.35165435e-02 -4.42545444e-01
-3.79737914e-01 -1.15425438e-01 1.85839251e-01 9.08647120e-01
4.61373627e-01 9.93256211e-01 -7.77532101e-01 -5.80617607e-01
1.24760672e-01 6.59067452e-01 -1.82317778e-01 1.24167819e-02
3.93579416e-02 8.28359306e-01 -3.84647772e-02 9.71061587e-01
8.15883279e-01 -5.47462404e-01 -2.16355268e-02 -3.33299518e-01
-2.60765553e-02 -4.45744276e-01 -8.39126289e-01 1.42198896e+00
-4.21968520e-01 5.35028219e-01 -1.02668501e-01 -9.73308802e-01
9.27195966e-01 3.10416996e-01 1.21909213e+00 -8.05866301e-01
4.92971033e-01 3.87957633e-01 -2.06296146e-01 -2.77189046e-01
1.48867205e-01 -2.21981585e-01 6.91858828e-02 -1.48685463e-02
1.33237243e-01 1.61897987e-02 3.80913317e-01 -1.28419241e-02
9.85362947e-01 -3.32717896e-01 1.48418039e-01 -4.29597646e-01
7.84364879e-01 2.36952212e-02 6.36462629e-01 5.68556547e-01
-5.12906015e-01 9.04681087e-01 5.19211590e-01 -5.86589396e-01
-8.45655203e-01 -6.50729775e-01 -6.36537790e-01 6.32674992e-01
6.03278866e-03 4.50183526e-02 -4.55191553e-01 -5.09517014e-01
1.83892682e-01 1.09328628e-01 -8.87234092e-01 -8.67737979e-02
-4.34865505e-01 -1.26907432e+00 6.26560509e-01 5.07611394e-01
9.76474136e-02 -8.23326588e-01 -2.07578287e-01 5.32265365e-01
-2.58385967e-02 -1.10326540e+00 -5.16092479e-01 4.29558307e-01
-6.80699468e-01 -1.26895809e+00 -9.41020012e-01 -8.88932049e-01
8.03038418e-01 1.20402008e-01 1.12112665e+00 2.14996651e-01
-7.79383659e-01 2.27756515e-01 -1.35580271e-01 -4.76633400e-01
-4.32603717e-01 1.82036802e-01 6.00551926e-02 2.38385379e-01
6.50419176e-01 -3.29611987e-01 -7.33062625e-01 3.56873244e-01
-8.89562666e-01 -3.13989490e-01 5.44379771e-01 1.26333606e+00
1.13789439e+00 1.76925749e-01 9.82658803e-01 -9.99200881e-01
2.43889987e-01 -4.60852832e-01 -5.49168468e-01 2.48148933e-01
-5.12442112e-01 -3.31452668e-01 8.64512026e-01 -4.58252519e-01
-7.06064224e-01 4.28728946e-02 -1.92657635e-01 -2.51484156e-01
-5.30110359e-01 3.01320434e-01 -9.17831734e-02 -3.37422639e-01
6.68620646e-01 5.05648255e-01 4.60390747e-01 -6.38528913e-02
-3.47567201e-01 5.82691967e-01 5.51140249e-01 -3.41631681e-01
3.06779861e-01 8.02829981e-01 3.23679715e-01 -8.04777086e-01
-5.33099949e-01 -9.93940473e-01 -5.77280283e-01 -4.46101427e-02
6.15415394e-01 -9.79291439e-01 -9.02896881e-01 1.06070077e+00
-7.04715967e-01 -5.84170043e-01 -2.70190388e-01 3.09738100e-01
-5.71110725e-01 2.95509487e-01 -9.32884932e-01 -4.56920564e-01
-3.72393578e-01 -1.08422613e+00 1.26125348e+00 4.22500879e-01
-1.73776895e-01 -1.50589919e+00 4.10970360e-01 1.49554804e-01
4.65278536e-01 5.97948551e-01 7.63118625e-01 -9.25921261e-01
-1.30291611e-01 -4.18364257e-01 -1.58354715e-01 -8.55594873e-03
5.30056059e-01 2.91199654e-01 -1.17460537e+00 -6.41206682e-01
-1.95806846e-01 -1.86964199e-01 7.49437094e-01 9.24706876e-01
1.18897116e+00 1.24380596e-01 -7.85398901e-01 9.69678342e-01
1.62015820e+00 2.27375031e-01 7.68396914e-01 5.20325482e-01
7.68744886e-01 3.68226975e-01 3.14279199e-01 7.04704046e-01
1.28371596e-01 4.72404808e-01 1.01134658e-01 -7.48096645e-01
-3.08071617e-02 2.63949245e-01 -1.66248083e-01 3.99589270e-01
2.27448672e-01 -2.37818420e-01 -8.74782085e-01 5.95124185e-01
-1.46023870e+00 -8.20216715e-01 -1.62244841e-01 2.27574682e+00
9.50438082e-01 -1.10341743e-01 4.08042192e-01 3.33564728e-01
4.81017888e-01 -5.24788678e-01 -5.54327548e-01 4.89978381e-02
-7.65745103e-01 5.47844060e-02 6.19903684e-01 9.28534269e-02
-1.16183579e+00 4.82689291e-01 6.10310984e+00 1.06517696e+00
-1.57593930e+00 -3.48816484e-01 1.44918239e+00 1.70579813e-02
1.71246938e-02 -4.22928929e-01 -8.83409798e-01 5.66968322e-01
6.69392228e-01 -3.01299989e-02 -1.72398612e-01 2.73364991e-01
4.10814285e-01 -4.58486937e-02 -1.21029699e+00 1.09770012e+00
-1.12732068e-01 -1.79351842e+00 -1.39345005e-01 4.08892542e-01
8.35969090e-01 -1.92431599e-01 3.10602840e-02 1.73667893e-01
-1.01224259e-01 -1.34133089e+00 1.42294019e-01 5.82655549e-01
9.03491259e-01 -8.42510164e-01 1.11857116e+00 4.48157229e-02
-9.87923265e-01 -8.14972371e-02 -2.57968426e-01 3.59185517e-01
-1.13376148e-01 9.49931085e-01 -1.10523856e+00 6.13224685e-01
4.67635423e-01 8.14316213e-01 -6.95782959e-01 1.36048746e+00
9.67289388e-01 4.75680143e-01 -1.98038161e-01 2.09957197e-01
-1.41389832e-01 1.82682555e-02 7.84932524e-02 1.56588089e+00
3.79193455e-01 -2.54648119e-01 4.47507389e-02 5.64873517e-01
7.58352354e-02 4.05777514e-01 -3.21789086e-01 1.57263517e-01
5.51828802e-01 1.45139229e+00 -1.18608487e+00 -1.30493462e-01
-3.80975574e-01 6.76081061e-01 1.56615540e-01 3.48632067e-01
-7.09094405e-01 -2.56922096e-01 5.10812402e-01 2.73188919e-01
3.46775711e-01 9.93091911e-02 -3.99067402e-01 -7.27421582e-01
-5.89120507e-01 -8.37172329e-01 6.81009710e-01 -7.33251423e-02
-1.75139284e+00 3.61117989e-01 -4.88970846e-01 -1.40306294e+00
2.51992017e-01 -7.88580596e-01 -5.34850478e-01 6.70378566e-01
-1.63146317e+00 -1.21986651e+00 -3.64907444e-01 4.93657976e-01
4.20496434e-01 -1.23947434e-01 8.64920855e-01 4.16548997e-01
-9.71737444e-01 9.72696602e-01 4.26913112e-01 1.30711988e-01
9.87592220e-01 -1.31815970e+00 -1.90860644e-01 2.32672989e-01
1.08064599e-02 5.16367197e-01 5.52083313e-01 -2.97047555e-01
-1.49346209e+00 -1.16760528e+00 4.47976708e-01 -4.56251144e-01
4.96552140e-01 -2.02713698e-01 -1.12269592e+00 8.63708779e-02
-6.06481871e-03 2.63804376e-01 1.40825832e+00 -7.59278387e-02
-1.18023448e-01 -3.42834473e-01 -1.23596442e+00 5.55858910e-01
3.34793746e-01 -1.41130865e-01 3.62937480e-01 4.42969173e-01
7.23179430e-02 -5.19946277e-01 -1.47204614e+00 3.89584452e-01
4.72926080e-01 -8.93254519e-01 8.92345846e-01 -2.24552378e-01
8.65022838e-02 -3.75978291e-01 2.90564775e-01 -1.19832003e+00
-6.31616175e-01 -5.59541047e-01 1.49285803e-02 1.47489762e+00
3.82629663e-01 -7.78724253e-01 9.14004207e-01 3.45494598e-01
-5.26955537e-02 -1.07765639e+00 -8.99578571e-01 -7.88710415e-01
5.10438740e-01 1.11715965e-01 8.15385818e-01 1.06835008e+00
1.34684399e-01 -6.71514170e-03 1.39044330e-01 -9.76441503e-02
3.59841347e-01 -5.14806472e-02 8.83843243e-01 -1.25149620e+00
-2.85336792e-01 -9.99889314e-01 -5.85889339e-01 -5.34858644e-01
1.01317935e-01 -6.21853530e-01 9.36598144e-03 -1.17077327e+00
3.43288243e-01 -6.54921174e-01 -8.39347363e-01 2.54537016e-01
-2.93085009e-01 5.40698290e-01 -3.39472443e-01 2.08860993e-01
-6.28754199e-01 1.58814341e-01 1.33024812e+00 -2.69598722e-01
-2.38850117e-01 -1.02188006e-01 -6.63229108e-01 1.91463009e-01
8.24895799e-01 -1.49858952e-01 -1.01959482e-01 3.91571932e-02
-7.26628155e-02 -2.00381875e-01 2.93018818e-01 -1.40455580e+00
3.02258164e-01 -1.57885998e-01 8.71133208e-01 -5.18073857e-01
-4.96412963e-02 -7.28702784e-01 3.28022420e-01 5.08644044e-01
-1.69835195e-01 -2.17794880e-01 5.31443059e-01 5.04082620e-01
-3.50472391e-01 1.18275464e-01 1.00034273e+00 4.65971045e-02
-3.42945099e-01 6.29877627e-01 -5.71559429e-01 -9.81547534e-02
1.07407439e+00 -8.02719653e-01 -2.84219533e-01 1.14763148e-01
-5.57563901e-01 3.25528830e-01 9.13702250e-01 -1.61187440e-01
2.47564688e-01 -1.30947101e+00 -8.72280121e-01 3.72790694e-01
6.01775050e-01 1.97179243e-01 6.68712616e-01 1.49460542e+00
-6.14560723e-01 2.75569320e-01 -1.39324322e-01 -1.01037228e+00
-1.08106518e+00 5.27678072e-01 5.50680995e-01 -6.01290107e-01
-3.45604748e-01 1.00985765e+00 2.82670736e-01 -1.09796859e-01
7.42562115e-03 -2.17415079e-01 -4.38658744e-01 1.70028597e-01
4.80349869e-01 2.89042592e-01 4.79932487e-01 -7.09913611e-01
-4.06856418e-01 6.01730645e-01 -3.67459208e-01 5.20787895e-01
1.23838544e+00 -3.98817249e-02 1.47436466e-02 5.05567312e-01
1.36745489e+00 7.29035214e-02 -1.45729828e+00 2.90136226e-02
-7.67415836e-02 -1.96782306e-01 -4.38719131e-02 -5.91558576e-01
-1.14600837e+00 3.96409303e-01 7.22648144e-01 2.16595232e-01
1.24749565e+00 -2.69534379e-01 8.01141381e-01 1.48615073e-02
3.14731210e-01 -1.17698264e+00 -1.63993321e-03 3.82867038e-01
4.58636671e-01 -1.48920441e+00 5.05421013e-02 -2.17893943e-01
-4.28883076e-01 1.23709309e+00 4.52067614e-01 2.08977565e-01
5.60169876e-01 5.57907879e-01 3.88915718e-01 -3.23423833e-01
-6.32501364e-01 1.72180727e-01 -2.12451264e-01 8.76501143e-01
6.65824234e-01 -9.65030044e-02 6.60624206e-02 6.57198191e-01
2.12665781e-01 2.45775029e-01 3.32060844e-01 8.76447737e-01
-1.31129533e-01 -1.10337806e+00 -4.89531994e-01 6.53143167e-01
-5.34537077e-01 7.72168413e-02 -1.56800002e-01 6.94393516e-01
4.43526097e-02 6.69650078e-01 3.16955298e-01 -6.79030120e-02
6.46712258e-02 -1.28730342e-01 5.23991942e-01 -2.58835524e-01
-6.68329120e-01 3.04317623e-01 -3.75096709e-01 -2.01124474e-01
-3.47026795e-01 -7.78216481e-01 -1.27758217e+00 -2.43226230e-01
-4.55778241e-01 -2.82370541e-02 2.06755951e-01 8.73547316e-01
5.28309643e-01 6.07988536e-01 9.58294868e-01 -7.93575943e-01
-1.06704794e-01 -8.17787826e-01 -8.08001459e-01 6.87619984e-01
4.51270491e-01 -5.47386050e-01 -3.90482396e-01 4.60205644e-01] | [15.009220123291016, -3.0943586826324463] |
1a5e8215-c6e6-415b-93f3-83b45ee42dc0 | image-completion-via-inference-in-deep | 2102.12037 | null | https://arxiv.org/abs/2102.12037v3 | https://arxiv.org/pdf/2102.12037v3.pdf | Conditional Image Generation by Conditioning Variational Auto-Encoders | We present a conditional variational auto-encoder (VAE) which, to avoid the substantial cost of training from scratch, uses an architecture and training objective capable of leveraging a foundation model in the form of a pretrained unconditional VAE. To train the conditional VAE, we only need to train an artifact to perform amortized inference over the unconditional VAE's latent variables given a conditioning input. We demonstrate our approach on tasks including image inpainting, for which it outperforms state-of-the-art GAN-based approaches at faithfully representing the inherent uncertainty. We conclude by describing a possible application of our inpainting model, in which it is used to perform Bayesian experimental design for the purpose of guiding a sensor. | ['Frank Wood', 'Saeid Naderiparizi', 'William Harvey'] | 2021-02-24 | conditional-image-generation-by-conditioning | https://openreview.net/forum?id=7MV6uLzOChW | https://openreview.net/pdf?id=7MV6uLzOChW | iclr-2022-4 | ['conditional-image-generation'] | ['computer-vision'] | [ 5.33073664e-01 5.15589058e-01 1.60650790e-01 -4.08184022e-01
-1.11728144e+00 -1.61977783e-01 7.63576150e-01 -3.37799519e-01
-3.47165346e-01 8.96524191e-01 1.08870648e-01 -3.26210499e-01
1.45073012e-01 -9.11441028e-01 -1.33720434e+00 -6.91095293e-01
2.57562757e-01 6.38420165e-01 -1.17559522e-01 1.56366929e-01
-1.65334325e-02 3.41058999e-01 -1.29844570e+00 1.28664032e-01
6.22338831e-01 1.12953663e+00 2.32980266e-01 6.68545127e-01
3.26696634e-01 9.37889457e-01 -6.12748504e-01 -3.25884849e-01
-3.37144695e-02 -8.38183761e-01 -3.80344450e-01 1.67015433e-01
-8.23223125e-03 -7.02042103e-01 -2.13791445e-01 8.17734003e-01
1.10708982e-01 9.36922580e-02 9.95505512e-01 -9.09134209e-01
-4.85728890e-01 3.88482034e-01 -3.38045895e-01 -3.22834194e-01
1.36789441e-01 2.93197036e-01 9.05048609e-01 -7.00850606e-01
6.01149559e-01 1.07371616e+00 6.67230666e-01 7.59863734e-01
-1.72523391e+00 -2.35981673e-01 -1.49587646e-01 -3.23925227e-01
-1.07984114e+00 -5.10618210e-01 8.49816144e-01 -4.94279265e-01
6.78195179e-01 -7.93101117e-02 5.55510640e-01 1.55337048e+00
3.43834966e-01 8.54282796e-01 1.23552382e+00 -4.53849047e-01
8.71917427e-01 -7.48225674e-02 -4.34635997e-01 7.31742382e-01
-1.22129045e-01 5.23220241e-01 -5.36388218e-01 -1.30450919e-01
1.01636124e+00 -3.17392051e-01 -2.61345096e-02 -4.34474856e-01
-8.04821670e-01 9.49713111e-01 3.80631655e-01 -9.42771882e-02
-7.17656910e-01 8.61128449e-01 1.15831010e-01 1.13377161e-02
4.66321439e-01 3.01066846e-01 -2.95462310e-01 -2.07561612e-01
-1.19270122e+00 2.39594162e-01 5.88922858e-01 7.05073655e-01
7.99202502e-01 2.79197007e-01 -2.22548738e-01 4.05869782e-01
5.08921921e-01 5.49450576e-01 1.52208418e-01 -1.41296589e+00
-9.33668856e-03 -1.70813203e-01 2.77057141e-01 -2.27611393e-01
2.59536803e-01 -1.97901264e-01 -5.56337357e-01 6.33074701e-01
-2.18484580e-04 -3.15971076e-01 -1.33393955e+00 1.86507475e+00
1.91659667e-02 2.87644237e-01 7.36514153e-03 4.70068812e-01
5.28268469e-03 7.79271483e-01 2.08503269e-02 -2.00749397e-01
9.39675152e-01 -5.23817122e-01 -5.56813657e-01 -2.78189868e-01
-1.99733023e-02 -5.25225043e-01 1.06985927e+00 5.43552935e-01
-1.30166149e+00 -5.45313239e-01 -1.24675930e+00 -4.36750613e-03
-4.03513052e-02 3.38352844e-02 5.80977917e-01 6.07193410e-01
-9.24303353e-01 7.58062482e-01 -1.44356167e+00 -7.90907070e-02
5.54446816e-01 3.31288315e-02 -2.17909455e-01 3.23183239e-02
-7.89105654e-01 9.96964812e-01 1.93648085e-01 -3.81843038e-02
-1.83303535e+00 -7.44088888e-01 -1.03561401e+00 1.26460403e-01
2.83232778e-01 -1.07984388e+00 1.45453715e+00 -8.48567724e-01
-2.07914472e+00 5.07088006e-01 -9.35333446e-02 -6.05938852e-01
5.65251231e-01 -3.35672945e-01 1.38292089e-01 1.61406741e-01
4.49252836e-02 8.65315318e-01 1.25449109e+00 -1.47581494e+00
-1.36926901e-02 -1.59543946e-01 -1.05273072e-02 -2.78492093e-01
2.38887817e-01 -3.90576184e-01 -3.10032338e-01 -7.74018764e-01
-1.61365256e-01 -8.12294960e-01 -2.74832934e-01 1.31898031e-01
-4.22754019e-01 1.17794439e-01 7.11308837e-01 -7.23192453e-01
6.66447043e-01 -1.89985645e+00 5.01947463e-01 1.74515709e-01
-1.15130469e-01 -1.45209208e-01 4.38179150e-02 5.57918787e-01
1.50595918e-01 -9.90582630e-02 -7.89638460e-01 -9.72591162e-01
1.47747725e-01 5.55362225e-01 -5.98792970e-01 2.67497867e-01
5.85741460e-01 9.10554409e-01 -8.14715266e-01 -4.00110930e-01
5.53070903e-01 7.07210064e-01 -7.70203590e-01 7.06746340e-01
-9.41254795e-01 6.88672602e-01 -3.12161833e-01 3.73851180e-01
3.47868502e-01 -5.12022451e-02 2.13957369e-01 -2.05710337e-01
1.17129460e-01 2.63950437e-01 -6.97692811e-01 2.13352561e+00
-7.80356228e-01 5.41199446e-01 1.75265700e-01 -9.54407036e-01
6.67952955e-01 1.46348387e-01 3.59416038e-01 -2.77708381e-01
3.36917639e-01 1.17853507e-01 -4.57748830e-01 -2.55482733e-01
2.28734553e-01 -4.58845973e-01 -1.06730618e-01 5.29510975e-01
5.03284097e-01 -6.42227948e-01 -2.11179212e-01 3.78787756e-01
1.12197554e+00 8.55006754e-01 2.89223120e-02 -2.09053740e-01
9.58337113e-02 -4.53661889e-01 3.75415564e-01 6.96979403e-01
3.08317780e-01 7.16504514e-01 5.03804505e-01 -1.20789386e-01
-1.33257759e+00 -1.40177238e+00 -1.61347717e-01 3.83773893e-01
-5.00242352e-01 -3.93773198e-01 -8.77076626e-01 -6.08879745e-01
-3.75984125e-02 1.28441453e+00 -1.05762148e+00 -4.92451824e-02
-2.33777374e-01 -4.87400800e-01 2.55038500e-01 7.48524725e-01
4.37269717e-01 -8.16714644e-01 -9.09729123e-01 2.35489100e-01
-5.38985021e-02 -9.24276173e-01 -9.42566618e-02 4.67411786e-01
-8.24031830e-01 -7.26822853e-01 -5.71128488e-01 -1.09633364e-01
6.21854603e-01 -4.69905317e-01 1.29444337e+00 -4.17394310e-01
-2.09698081e-01 8.63745272e-01 -6.43931776e-02 -6.29763305e-01
-7.85717368e-01 -3.76446515e-01 -2.89921045e-01 -9.99434292e-02
8.20245966e-02 -9.83781040e-01 -5.51843464e-01 -1.30960479e-01
-1.10146546e+00 -9.49908886e-03 5.36293209e-01 1.14258182e+00
8.09692979e-01 -1.61408812e-01 3.27666521e-01 -7.77308464e-01
3.75031084e-01 -4.27769452e-01 -9.77265835e-01 1.41268566e-01
-5.03820956e-01 5.51484644e-01 3.92217636e-01 -1.79894626e-01
-1.24746478e+00 1.19211674e-01 -4.23076540e-01 -7.50769138e-01
3.22432555e-02 5.69544435e-01 -1.02679729e-01 1.29107028e-01
4.74995077e-01 1.27282366e-01 2.30439276e-01 -4.87192005e-01
5.98863006e-01 4.14364427e-01 7.43723154e-01 -9.97785747e-01
5.77570200e-01 5.48976898e-01 3.57547671e-01 -6.16655946e-01
-7.50943840e-01 3.84976208e-01 -3.73276770e-01 -1.86465934e-01
1.11461031e+00 -9.40959811e-01 -5.60513854e-01 2.66712189e-01
-1.23644638e+00 -6.54867113e-01 -9.01523232e-01 4.03128475e-01
-1.12579083e+00 1.24498934e-01 -5.16524076e-01 -1.00181484e+00
3.95787992e-02 -1.06253958e+00 1.35936570e+00 -1.13346554e-01
-1.21537074e-01 -1.04385924e+00 3.28241855e-01 7.37787113e-02
4.49454963e-01 5.44514179e-01 8.69343519e-01 7.60385208e-03
-8.75003457e-01 -4.03328724e-02 2.04573855e-01 7.23192155e-01
-6.55750707e-02 1.27524793e-01 -1.28414714e+00 -2.30113402e-01
3.18285853e-01 -7.59178996e-01 1.08112347e+00 6.86036110e-01
1.26548374e+00 -2.48892568e-02 -1.59274265e-01 6.35968745e-01
1.55038726e+00 2.63687279e-02 8.54037583e-01 -1.11461982e-01
3.26884896e-01 3.83669108e-01 3.01002264e-01 5.79116642e-01
7.64575601e-02 4.22064215e-01 7.57318735e-01 6.51871562e-02
-7.41983056e-02 -7.39207208e-01 3.34011257e-01 3.81461352e-01
1.06318546e-02 -2.27695033e-01 -3.54434073e-01 5.49172521e-01
-1.74778736e+00 -7.95455158e-01 6.38584793e-01 2.10782719e+00
8.13124895e-01 1.25276357e-01 -1.36441484e-01 -1.74931943e-01
1.32584482e-01 1.75916404e-01 -7.83653617e-01 -3.27830732e-01
3.30703497e-01 7.32683778e-01 3.38114738e-01 8.16634953e-01
-8.18804502e-01 6.77115023e-01 7.53034449e+00 6.48487210e-01
-7.90910840e-01 1.48101509e-01 6.45752907e-01 1.38294905e-01
-7.03985274e-01 2.79899836e-01 -3.30568075e-01 4.30411369e-01
1.11668551e+00 2.86017090e-01 7.61396170e-01 8.01435173e-01
6.54903278e-02 -3.94686103e-01 -1.40638375e+00 6.32402599e-01
1.78315174e-02 -1.35050535e+00 -1.48066813e-02 5.38357310e-02
7.14166045e-01 -9.30711743e-04 2.12983847e-01 2.19910026e-01
7.92769909e-01 -1.07384932e+00 7.62441635e-01 7.18162298e-01
8.44433486e-01 -4.67255950e-01 3.34404290e-01 3.39345157e-01
-5.36751866e-01 1.74446300e-01 -2.71295369e-01 9.20357332e-02
3.98275673e-01 7.22316027e-01 -5.44972122e-01 4.87057775e-01
4.59192038e-01 3.55149895e-01 3.43441293e-02 2.68450558e-01
-5.53942144e-01 7.87158728e-01 -5.27789712e-01 4.21317905e-01
8.96543190e-02 -3.49147290e-01 4.26070452e-01 8.18974435e-01
4.83398736e-01 2.79178564e-02 -6.94378391e-02 1.54865587e+00
-1.22725219e-01 -5.93754470e-01 -6.62851512e-01 -1.26496539e-01
3.95784169e-01 9.15918291e-01 -3.23730171e-01 -3.45227480e-01
-2.53536075e-01 1.08211148e+00 2.50387430e-01 3.76957208e-01
-9.22993660e-01 -2.75079906e-02 4.49661702e-01 -1.91415504e-01
7.54916131e-01 -2.95870155e-01 -2.00600028e-01 -1.25381386e+00
-4.63613644e-02 -6.52719796e-01 4.76716608e-02 -1.18367422e+00
-1.21822464e+00 4.41533387e-01 1.85675293e-01 -7.95161784e-01
-8.86881769e-01 -6.13652825e-01 -7.22833395e-01 9.64128613e-01
-1.28537095e+00 -1.39799881e+00 -9.59778875e-02 5.40252805e-01
3.50288868e-01 -6.39766231e-02 1.05985725e+00 -6.44813552e-02
-2.99618959e-01 2.31277093e-01 4.59815599e-02 -3.12844396e-01
4.76659924e-01 -1.36550570e+00 3.28083217e-01 9.94648933e-01
1.21924445e-01 5.93535185e-01 1.23525536e+00 -5.86999834e-01
-1.35996091e+00 -7.95686185e-01 2.45632961e-01 -6.02436721e-01
6.52985454e-01 -4.06069726e-01 -4.26411420e-01 1.19568789e+00
5.35503030e-01 -5.41482083e-02 4.25663322e-01 1.07848374e-02
-2.09275737e-01 -9.09056142e-02 -1.26342607e+00 2.61439085e-01
5.51096499e-01 -8.59418452e-01 -5.68977296e-01 1.20619945e-01
7.04536974e-01 -3.13737482e-01 -7.69376814e-01 3.86209518e-01
5.67635953e-01 -1.09962046e+00 8.23328495e-01 -4.01744962e-01
7.43843257e-01 -2.49668553e-01 -3.90833467e-01 -1.37361777e+00
-5.13101481e-02 -7.46530831e-01 -4.03672010e-01 1.21172464e+00
1.68348283e-01 -3.67391616e-01 9.66579497e-01 6.17629409e-01
-1.97151393e-01 -6.32181585e-01 -9.93535399e-01 -5.59606969e-01
1.65860996e-01 -6.16994679e-01 4.20744449e-01 2.71113932e-01
-5.08678675e-01 2.82942057e-01 -6.80413663e-01 4.24330123e-02
8.53832424e-01 -1.09230474e-01 8.08929026e-01 -8.89776886e-01
-9.93481159e-01 1.51453942e-01 -1.03253521e-01 -1.11532784e+00
1.78745046e-01 -3.36272836e-01 5.41055739e-01 -1.20206130e+00
1.15578219e-01 -1.54203728e-01 -2.48083308e-01 2.37838194e-01
1.30160481e-01 1.18030891e-01 2.23895535e-02 -4.08978574e-02
-1.57445446e-01 9.89998043e-01 8.83697510e-01 -2.43889719e-01
9.07107815e-02 -1.42498221e-02 -4.74936992e-01 7.26318479e-01
4.64361429e-01 -5.89626610e-01 -7.00141430e-01 -4.40720260e-01
2.88480192e-01 3.02965999e-01 7.58183479e-01 -9.96404409e-01
2.75508687e-02 -7.59271532e-02 4.45249140e-01 -4.22311604e-01
6.45900846e-01 -9.81683969e-01 4.87320393e-01 3.41261148e-01
-3.96052241e-01 -2.26127088e-01 2.06495821e-01 7.49535084e-01
-1.82465717e-01 -3.77448201e-01 8.55747223e-01 -1.52976021e-01
-1.33481860e-01 1.62037835e-01 -5.57897687e-01 -1.09755635e-01
8.25581610e-01 3.14923108e-01 9.46541056e-02 -6.03162408e-01
-7.63309240e-01 -1.44557074e-01 6.40135586e-01 -1.74982920e-01
6.56061411e-01 -1.17937350e+00 -4.61566538e-01 3.33606482e-01
-1.68357596e-01 8.23725909e-02 1.21099964e-01 2.34828725e-01
-3.89933765e-01 1.25522628e-01 -2.31978357e-01 -5.70402443e-01
-5.47734380e-01 7.19007254e-01 4.17452693e-01 -3.75519961e-01
-4.46327358e-01 8.28025341e-01 2.13087931e-01 -8.18392336e-02
7.45410323e-02 -3.83240104e-01 6.33288562e-01 -4.27426577e-01
2.10497364e-01 3.70257981e-02 -1.79861426e-01 -1.45449769e-02
-7.23489746e-02 3.37177038e-01 2.50254065e-01 -8.17423224e-01
1.36618841e+00 5.81607642e-03 2.57683592e-03 5.93857765e-01
9.48051989e-01 2.40153484e-02 -1.99810004e+00 4.36042622e-02
-4.30637389e-01 -3.05299431e-01 3.47398162e-01 -9.27329481e-01
-8.56234729e-01 1.12038410e+00 4.94144469e-01 -1.66959912e-01
1.13567483e+00 -3.98004465e-02 5.98722160e-01 3.84700149e-01
5.74252784e-01 -9.38348234e-01 2.06125215e-01 8.09374452e-02
9.32461560e-01 -1.09297681e+00 2.29878023e-01 -9.99665074e-03
-5.23253441e-01 1.04932094e+00 1.46784246e-01 -4.99695480e-01
8.93013775e-01 5.96518159e-01 -3.27742904e-01 -2.23689660e-01
-7.16690958e-01 -4.04888056e-02 7.35784844e-02 6.87412977e-01
1.64002910e-01 3.21607217e-02 1.01629652e-01 2.69519120e-01
3.00146453e-02 3.22570741e-01 3.37504089e-01 1.04218280e+00
-1.30380109e-01 -1.36753225e+00 -4.19165641e-02 2.85678685e-01
-4.86053526e-01 -3.10105458e-02 -1.21440962e-01 7.05058932e-01
1.84753370e-02 5.56523919e-01 7.23492950e-02 -2.72276580e-01
4.31868397e-02 1.47480801e-01 8.63464057e-01 -4.05408204e-01
-8.65599290e-02 2.08376050e-01 9.78480652e-02 -8.24974537e-01
-5.92723668e-01 -7.41551101e-01 -7.30655432e-01 -6.00199923e-02
-1.22565411e-01 2.78785024e-02 8.37345481e-01 9.15805042e-01
2.57596642e-01 8.21339548e-01 3.48633289e-01 -1.07065344e+00
-7.57984281e-01 -1.02216601e+00 -6.15667999e-01 1.95908114e-01
3.65618587e-01 -6.59447491e-01 -2.86656350e-01 2.26832464e-01] | [11.544386863708496, -0.32839682698249817] |
3e80003e-0d73-4b36-a589-7781cf0d1eab | drone-as-a-service-composition-under | 2103.06513 | null | https://arxiv.org/abs/2103.06513v1 | https://arxiv.org/pdf/2103.06513v1.pdf | Drone-as-a-Service Composition Under Uncertainty | We propose an uncertainty-aware service approach to provide drone-based delivery services called Drone-as-a-Service (DaaS) effectively. Specifically, we propose a service model of DaaS based on the dynamic spatiotemporal features of drones and their in-flight contexts. The proposed DaaS service approach consists of three components: scheduling, route-planning, and composition. First, we develop a DaaS scheduling model to generate DaaS itineraries through a Skyway network. Second, we propose an uncertainty-aware DaaS route-planning algorithm that selects the optimal Skyways under weather uncertainties. Third, we develop two DaaS composition techniques to select an optimal DaaS composition at each station of the planned route. A spatiotemporal DaaS composer first selects the optimal DaaSs based on their spatiotemporal availability and drone capabilities. A predictive DaaS composer then utilises the outcome of the first composer to enable fast and accurate DaaS composition using several Machine Learning classification methods. We train the classifiers using a new set of spatiotemporal features which are in addition to other DaaS QoS properties. Our experiments results show the effectiveness and efficiency of the proposed approach. | ['Athman Bouguettaya', 'Azadeh Ghari Neiat', 'Du Yong Kim', 'Flora D. Salim', 'Ali Hamdi'] | 2021-03-11 | null | null | null | null | ['service-composition'] | ['miscellaneous'] | [-3.78063053e-01 -5.26588738e-01 -1.93726018e-01 -6.52519107e-01
-5.37720263e-01 -8.18229139e-01 7.46590018e-01 1.26135047e-03
-2.64243595e-02 6.89354241e-01 -1.08283207e-01 -6.30669743e-02
-1.04830849e+00 -1.20462930e+00 -5.03279924e-01 -7.81621516e-01
-5.46574950e-01 9.50126827e-01 4.59554166e-01 -6.41847789e-01
-5.09888865e-03 1.30534315e+00 -2.11266470e+00 3.26396406e-01
8.30184996e-01 1.67619777e+00 -5.54341972e-02 8.01363528e-01
-7.21616298e-03 3.68668109e-01 -6.64624214e-01 -1.68488935e-01
7.00913429e-01 -4.55560982e-02 -2.40066275e-01 9.56811234e-02
-7.08069131e-02 -1.70104310e-01 -2.29240596e-01 3.69749099e-01
2.00527236e-01 7.05259442e-01 3.91304463e-01 -2.25908685e+00
2.92560488e-01 5.13927460e-01 4.45849270e-01 3.83938372e-01
3.21167886e-01 1.62200570e-01 1.02786613e+00 -5.78842163e-01
2.73278713e-01 7.28973329e-01 2.19827026e-01 3.76754642e-01
-5.66748083e-01 -6.23573720e-01 4.84341919e-01 1.76022187e-01
-1.59980524e+00 -4.73187029e-01 6.11985505e-01 -2.03987107e-01
9.43933010e-01 6.06163979e-01 1.06979144e+00 2.88290739e-01
3.38366598e-01 5.38317978e-01 5.08464634e-01 6.25027269e-02
4.94180739e-01 -5.09791411e-02 -4.17775899e-01 6.96708083e-01
-1.61853567e-01 2.31117979e-01 -5.16653955e-01 -4.94698852e-01
1.92564756e-01 4.31366146e-01 -8.26910883e-02 -2.78764576e-01
-1.23468494e+00 5.06235719e-01 2.69618690e-01 -9.17883664e-02
-9.29607928e-01 -4.57374454e-02 1.83878720e-01 2.16302738e-01
1.83536649e-01 4.68102813e-01 -5.69059074e-01 -1.16615795e-01
-1.33786774e+00 8.53967428e-01 8.86899769e-01 1.38931799e+00
6.06188416e-01 2.82894611e-01 -2.79192976e-03 4.68888819e-01
-6.47160877e-03 5.08563697e-01 2.89397836e-02 -8.25106084e-01
2.32480496e-01 5.24561465e-01 6.52859330e-01 -9.82092917e-01
-3.49128366e-01 -2.60465622e-01 -2.78830439e-01 -2.81467102e-02
-2.90763408e-01 -2.25342557e-01 -6.99308693e-01 1.33816063e+00
6.18390322e-01 6.16966844e-01 -1.84862874e-03 1.13398433e+00
9.25902501e-02 1.02697730e+00 -1.68418050e-01 -4.43015337e-01
6.89470232e-01 -8.70701611e-01 -5.91776848e-01 4.00148422e-01
1.66085258e-01 -3.47290397e-01 4.41642731e-01 8.64045858e-01
-1.10697889e+00 -2.91798264e-01 -1.06876743e+00 7.80140817e-01
-5.96041918e-01 1.68921240e-02 7.00999677e-01 4.09405410e-01
-7.46016204e-01 3.86106908e-01 -4.97012377e-01 -2.49467522e-01
-9.78952460e-03 6.02181971e-01 -1.47990420e-01 2.09183954e-02
-1.16394639e+00 6.86373353e-01 4.49295700e-01 -6.48195595e-02
-1.64451313e+00 -7.75054276e-01 -6.70814335e-01 8.16633254e-02
7.72332788e-01 -6.65192425e-01 1.01258647e+00 -7.02230453e-01
-1.34526980e+00 1.17430791e-01 9.87434685e-02 -5.83862543e-01
1.48485079e-01 1.82977200e-01 -1.25169444e+00 1.72142118e-01
-1.85982689e-01 1.55887082e-01 8.30568969e-01 -1.29025197e+00
-1.81361830e+00 -2.08197460e-01 4.04003471e-01 3.27417493e-01
1.35223612e-01 -4.01237793e-02 -9.75129679e-02 -3.88768733e-01
7.12263882e-02 -1.02900875e+00 -2.06854448e-01 -2.42265493e-01
1.19024940e-01 -2.48860806e-01 5.60557425e-01 -2.40385577e-01
1.69448400e+00 -1.81382239e+00 2.76457518e-01 7.82673299e-01
-8.72032717e-02 4.68124188e-02 6.13503270e-02 8.97051871e-01
1.78988457e-01 -7.28855729e-02 -1.11528397e-01 -1.87262278e-02
1.16248399e-01 5.10416687e-01 -5.73315978e-01 2.04067498e-01
9.06106979e-02 5.74271157e-02 -8.58322740e-01 -2.66957842e-03
1.45005509e-01 2.44863108e-02 -4.78831023e-01 6.74581707e-01
-7.28428602e-01 1.92489624e-01 -6.37825429e-01 1.13997293e+00
5.17506421e-01 5.63215017e-01 -8.43054056e-02 -1.52814627e-01
-4.63592649e-01 3.22016597e-01 -1.20880461e+00 1.57862043e+00
-7.92912185e-01 -7.30581135e-02 1.95968226e-01 -7.81843603e-01
1.09153008e+00 4.19013262e-01 1.18176162e+00 -2.86238473e-02
5.62988147e-02 4.12503868e-01 -1.81504682e-01 -4.79892105e-01
5.47900021e-01 3.29862721e-02 -3.66946608e-01 5.24438918e-01
-6.84920549e-02 -5.03593624e-01 2.11722091e-01 1.17617466e-01
9.37916219e-01 2.98116863e-01 1.72603369e-01 -1.23630807e-01
7.16045797e-01 2.67386496e-01 8.22523296e-01 2.43945390e-01
-3.60119581e-01 1.83229342e-01 1.13222443e-01 -1.05137694e+00
-8.69793773e-01 -1.00318897e+00 3.96045357e-01 1.00131321e+00
2.79767066e-01 -3.85732651e-01 -4.76329297e-01 -7.54204512e-01
2.90244371e-01 1.09739316e+00 -2.60479122e-01 -1.53360814e-01
-3.35630476e-01 -3.63564432e-01 3.44609052e-01 -5.97308949e-03
2.97902524e-01 -4.74480480e-01 -5.79745710e-01 2.67572522e-01
-3.04513928e-02 -1.18442047e+00 -5.93550146e-01 -2.57541060e-01
-3.84363413e-01 -8.35620880e-01 2.13002086e-01 -4.74128872e-02
3.11172932e-01 3.41135085e-01 8.16669583e-01 2.56166786e-01
6.98228255e-02 6.01038992e-01 -3.77865493e-01 -3.38025510e-01
-8.46800953e-02 -1.63131997e-01 7.96100557e-01 5.93177855e-01
3.68590266e-01 -6.89448059e-01 -5.86521566e-01 7.09728360e-01
-9.89469886e-01 -2.61001498e-01 7.48261884e-02 2.52359569e-01
8.46016526e-01 8.86979401e-01 7.59492755e-01 -3.06094587e-01
5.28900087e-01 -1.07248783e+00 -9.43552792e-01 3.51797432e-01
-7.64897287e-01 -3.47098053e-01 7.61018455e-01 -1.05906010e-01
-7.82984316e-01 7.79064447e-02 1.51061222e-01 -6.98435724e-01
-3.47068310e-01 6.00962043e-01 -4.29648459e-01 2.94791684e-02
2.55157024e-01 3.48701715e-01 -2.17027456e-01 -1.71365350e-01
2.33620271e-01 7.49044836e-01 3.69278133e-01 -8.19825232e-01
8.76358747e-01 5.56721032e-01 6.34145588e-02 -4.22391474e-01
-5.77244639e-01 -5.47354221e-01 -4.42989022e-01 -5.38151801e-01
4.41812128e-01 -7.60914028e-01 -8.64842594e-01 -4.09000106e-02
-6.07570887e-01 9.33857858e-02 -1.18846767e-01 5.00269651e-01
-8.55692565e-01 -4.42750394e-01 1.75749362e-01 -1.11243987e+00
-9.59628448e-02 -1.27670813e+00 8.81215811e-01 1.33763015e-01
3.05631220e-01 -5.61815739e-01 -7.08448812e-02 3.06101322e-01
3.10968280e-01 4.18059796e-01 6.43707812e-01 -1.00837076e+00
-1.04954278e+00 -5.17971337e-01 2.84112096e-01 1.82531569e-02
2.02936620e-01 3.95309120e-01 -4.85324532e-01 -3.62070829e-01
-4.45306540e-01 2.45770752e-01 7.93424174e-02 1.32938474e-01
1.13780928e+00 -7.32261717e-01 -4.52635169e-01 7.32254028e-01
1.48868477e+00 7.56614566e-01 1.23029865e-01 6.38885275e-02
2.17128679e-01 9.00576055e-01 1.54958642e+00 9.25162792e-01
6.73961639e-01 1.08177578e+00 1.04489481e+00 4.80065137e-01
4.15070295e-01 -2.66544908e-01 3.56947064e-01 5.99354088e-01
-1.72452092e-01 -6.59174323e-01 -8.94011497e-01 8.28354061e-01
-1.96325505e+00 -8.70720565e-01 3.81879210e-01 2.56447530e+00
2.28151113e-01 7.09534809e-02 6.42716527e-01 2.93074846e-02
3.50080818e-01 7.33760744e-02 -3.87050509e-01 -6.66914403e-01
2.43617162e-01 -1.74979717e-01 6.12352967e-01 3.52700919e-01
-9.00461435e-01 1.01801443e+00 4.83591986e+00 7.60824084e-01
-8.59859169e-01 -5.56025058e-02 3.89062762e-02 -4.50688630e-01
-5.48510969e-01 3.97976898e-02 -9.90181506e-01 6.29847646e-01
1.29570377e+00 -4.21200693e-01 1.10698998e+00 8.66905749e-01
7.43682742e-01 1.21492296e-01 -9.85640407e-01 6.97596669e-01
1.86490670e-01 -1.56211245e+00 3.73078048e-01 -1.24846064e-01
6.69217885e-01 -3.92338783e-02 -2.45851308e-01 -1.01939827e-01
2.53136754e-01 -6.85372651e-01 7.51270771e-01 1.06055784e+00
4.62970704e-01 -1.56829059e+00 4.75432962e-01 4.98203635e-01
-1.35634482e+00 -7.29465127e-01 9.03392732e-02 2.94196844e-01
3.82404774e-01 4.38218355e-01 -9.34951961e-01 1.08741391e+00
7.62146950e-01 3.17136884e-01 1.47661105e-01 1.15582883e+00
2.19817162e-02 3.26701194e-01 -4.91474897e-01 6.92134947e-02
4.29138362e-01 -5.06915390e-01 1.01180220e+00 6.90113604e-01
7.67858565e-01 5.14211535e-01 6.89055860e-01 5.56547523e-01
4.24900502e-01 -1.37357771e-01 -7.78009832e-01 -2.18710914e-01
9.88239646e-01 1.27839577e+00 -3.14694107e-01 -1.88663557e-01
-7.30034709e-02 5.79901159e-01 -4.65170026e-01 3.59507620e-01
-8.88704479e-01 -4.05716360e-01 1.47086406e+00 4.06921953e-01
2.95985863e-02 -5.20640969e-01 3.50123465e-01 -8.87005091e-01
-3.33575785e-01 -9.79422390e-01 5.75372934e-01 -9.39892471e-01
-9.76841033e-01 1.02410901e+00 2.39543349e-01 -1.65377641e+00
-4.04562920e-01 -1.53861940e-01 -5.11523843e-01 5.65206587e-01
-1.61305416e+00 -1.53007686e+00 -3.80080223e-01 8.62025499e-01
8.68813038e-01 -7.46115208e-01 7.87188411e-01 4.60188866e-01
-4.64888424e-01 3.87695711e-03 -4.56891984e-01 -6.63060963e-01
4.46459055e-01 -8.64642739e-01 1.58199668e-01 1.08349502e+00
-1.11962035e-02 3.07989269e-01 7.59236991e-01 -8.19562852e-01
-1.55503130e+00 -1.16979182e+00 8.55034471e-01 -3.01245004e-01
6.42870665e-01 -1.37997404e-01 -3.68397087e-01 7.60079443e-01
-1.21632636e-01 1.70092918e-02 7.38101423e-01 -3.88698727e-01
2.02866733e-01 -8.92793834e-01 -1.39723468e+00 4.27591980e-01
1.01673126e+00 -4.56822179e-02 -1.63731575e-01 5.83147466e-01
8.65736365e-01 -1.70367122e-01 -1.02251101e+00 5.78504086e-01
4.82088119e-01 -7.80490816e-01 7.39958704e-01 -9.51357663e-01
-2.94216394e-01 -8.06998074e-01 -5.46182752e-01 -1.50361657e+00
-1.80192828e-01 -9.69355762e-01 -5.30411117e-03 1.00855112e+00
4.31072801e-01 -6.24779344e-01 3.44232440e-01 7.74789155e-01
-5.02687037e-01 -7.43383467e-01 -9.70103443e-01 -7.32682765e-01
-8.42843890e-01 -5.39291322e-01 1.71219087e+00 7.62125731e-01
-3.92425090e-01 -5.43870926e-01 -4.70920920e-01 9.11731958e-01
4.95445669e-01 4.33724523e-01 7.81251431e-01 -1.11944282e+00
1.12084821e-02 -1.69532865e-01 -3.00485373e-01 -4.28988755e-01
6.02834821e-02 -6.60465539e-01 1.18751690e-01 -1.17961407e+00
-9.43450391e-01 -1.10087192e+00 -5.68478703e-01 3.04296494e-01
5.88231206e-01 -4.79219317e-01 4.98604849e-02 9.82471630e-02
-6.03171945e-01 5.48590600e-01 6.78705215e-01 8.36111754e-02
-5.08758724e-01 9.65064049e-01 2.96093021e-02 3.46650362e-01
7.89429605e-01 -4.57243085e-01 -9.32270288e-01 -1.88504249e-01
2.81890810e-01 4.81742799e-01 -8.14208202e-03 -1.12396538e+00
4.99568075e-01 -9.44727659e-01 -2.60090798e-01 -8.25735986e-01
6.33490503e-01 -1.40868008e+00 6.34632885e-01 1.14706807e-01
-1.61529303e-01 6.02549374e-01 1.39205262e-01 6.65181458e-01
-2.57894099e-01 7.78598040e-02 4.48174030e-01 8.14916417e-02
-8.39334369e-01 1.06368828e+00 -6.31436825e-01 -6.48374915e-01
1.41118789e+00 -9.03828666e-02 7.78999701e-02 -5.04565299e-01
-6.43686891e-01 6.75452769e-01 3.32229435e-01 7.16541409e-01
9.49323416e-01 -1.22149122e+00 -5.12288690e-01 3.78851146e-01
2.60857582e-01 -2.99293578e-01 1.87717125e-01 5.18219650e-01
-6.31475031e-01 4.08677250e-01 -4.62112308e-01 -3.11311692e-01
-9.09997582e-01 7.58488595e-01 4.51792717e-01 2.35883415e-01
-8.06673989e-02 6.64361179e-01 -6.14807487e-01 -4.69387889e-01
1.80317715e-01 -2.43293375e-01 -2.74175048e-01 7.52310529e-02
5.79636395e-01 5.44253707e-01 3.28548491e-01 -8.26563060e-01
-5.97513020e-01 2.06653252e-01 6.23405159e-01 -3.25489789e-01
1.35507166e+00 -4.21336055e-01 -1.77453220e-01 1.64593413e-01
3.27472836e-01 2.30715513e-01 -1.16934907e+00 9.43039730e-02
1.55391386e-02 -9.00429249e-01 3.06537360e-01 -1.00489676e+00
-1.17060721e+00 3.52145433e-01 3.20000499e-01 5.19143939e-01
1.51869643e+00 -5.75726092e-01 1.13193882e+00 3.42518203e-02
8.20056260e-01 -1.08509326e+00 -6.30115092e-01 4.76566330e-02
8.70522559e-01 -9.50376511e-01 -1.09701671e-01 -1.60550043e-01
-1.03595185e+00 1.00546288e+00 4.67351496e-01 -1.12184040e-01
8.51522863e-01 1.10893264e-01 -2.15402514e-01 -2.89229631e-01
-1.38177025e+00 -3.90862823e-01 3.60110819e-01 6.01411879e-01
-6.14557445e-01 5.05891562e-01 -2.15228438e-01 8.38900030e-01
-1.76728845e-01 1.70361251e-01 5.09482265e-01 6.00475788e-01
-5.56805730e-01 -1.00324178e+00 -5.15148640e-02 2.17184231e-01
-7.30994716e-03 2.78444171e-01 6.01285137e-03 5.17577589e-01
5.86431265e-01 1.05174112e+00 2.07905352e-01 -9.54069138e-01
5.19637465e-01 -5.23121655e-02 -1.87543798e-02 -7.32304275e-01
-6.63966119e-01 -6.30968288e-02 4.50542718e-01 -9.50541973e-01
-1.57111928e-01 -6.56680822e-01 -1.31284332e+00 -3.35369587e-01
-1.20250545e-01 5.94525814e-01 1.23436463e+00 9.07811821e-01
6.62423849e-01 4.35201854e-01 1.33932865e+00 -6.33912563e-01
-2.15685010e-01 -1.34080768e-01 -6.73305094e-01 -2.43635893e-01
4.35248584e-01 -9.52367842e-01 -3.72087300e-01 -3.61997187e-01] | [5.526534557342529, 1.9600681066513062] |
d8a218a0-5147-41e1-9991-d501dbfababc | mask-guided-feature-extraction-and | 2109.07755 | null | https://arxiv.org/abs/2109.07755v1 | https://arxiv.org/pdf/2109.07755v1.pdf | Mask-Guided Feature Extraction and Augmentation for Ultra-Fine-Grained Visual Categorization | While the fine-grained visual categorization (FGVC) problems have been greatly developed in the past years, the Ultra-fine-grained visual categorization (Ultra-FGVC) problems have been understudied. FGVC aims at classifying objects from the same species (very similar categories), while the Ultra-FGVC targets at more challenging problems of classifying images at an ultra-fine granularity where even human experts may fail to identify the visual difference. The challenges for Ultra-FGVC mainly comes from two aspects: one is that the Ultra-FGVC often arises overfitting problems due to the lack of training samples; and another lies in that the inter-class variance among images is much smaller than normal FGVC tasks, which makes it difficult to learn discriminative features for each class. To solve these challenges, a mask-guided feature extraction and feature augmentation method is proposed in this paper to extract discriminative and informative regions of images which are then used to augment the original feature map. The advantage of the proposed method is that the feature detection and extraction model only requires a small amount of target region samples with bounding boxes for training, then it can automatically locate the target area for a large number of images in the dataset at a high detection accuracy. Experimental results on two public datasets and ten state-of-the-art benchmark methods consistently demonstrate the effectiveness of the proposed method both visually and quantitatively. | ['Yongsheng Gao', 'Miaohua Zhang', 'Xiaohan Yu', 'Zicheng Pan'] | 2021-09-16 | null | null | null | null | ['fine-grained-visual-categorization'] | ['computer-vision'] | [ 2.74196208e-01 -4.36179399e-01 -1.05389453e-01 -8.69005993e-02
-4.76641238e-01 -5.60574412e-01 6.69933259e-01 3.13528657e-01
-2.94957578e-01 5.97534299e-01 -7.90420324e-02 -2.64078397e-02
-1.73299655e-01 -7.23590195e-01 -3.47906172e-01 -8.20903480e-01
7.65384212e-02 7.69810285e-03 6.00438654e-01 7.05399215e-02
4.86425400e-01 5.40267169e-01 -2.00809765e+00 2.09766924e-01
9.41468596e-01 1.52297747e+00 3.67294490e-01 3.06335360e-01
-4.27464932e-01 4.48476493e-01 -5.46676695e-01 -9.87474173e-02
4.05931503e-01 -1.13664843e-01 -6.96366847e-01 3.55699539e-01
5.59550822e-01 -2.65687287e-01 -2.28032116e-02 1.30400026e+00
1.54441610e-01 9.74075943e-02 1.03047359e+00 -1.34122241e+00
-6.96042001e-01 -5.17362095e-02 -1.10538256e+00 4.30215746e-01
-3.40390056e-02 6.91573396e-02 9.31999922e-01 -1.00663447e+00
3.04191917e-01 1.33089399e+00 6.09334111e-01 2.84724146e-01
-1.13257301e+00 -8.08390200e-01 4.04720247e-01 3.05960417e-01
-1.81791782e+00 -9.91945416e-02 6.36224747e-01 -6.99702621e-01
4.77985471e-01 3.72656584e-01 5.24783075e-01 5.22484303e-01
1.37319982e-01 5.14325500e-01 1.34284079e+00 -2.94602096e-01
1.17713496e-01 2.80485392e-01 2.77154028e-01 6.63931012e-01
3.05586278e-01 -3.01488279e-03 -8.81463215e-02 -8.71772394e-02
7.62768984e-01 4.77064431e-01 -3.07909817e-01 -4.44265574e-01
-1.02830803e+00 8.45750332e-01 9.06051159e-01 4.54013258e-01
-3.41848314e-01 -2.43230134e-01 5.04055679e-01 7.25637600e-02
2.94078439e-01 2.73333371e-01 -1.36187285e-01 2.12189019e-01
-7.97175646e-01 4.44942079e-02 1.85872763e-01 8.03056479e-01
1.03465629e+00 -1.04478069e-01 -3.19980919e-01 9.66045320e-01
1.39458766e-02 3.02881241e-01 7.38384128e-01 -3.14275622e-01
4.68434989e-01 1.01383376e+00 8.03838819e-02 -1.50685799e+00
-1.95401594e-01 -6.65863931e-01 -1.22246563e+00 1.82538554e-01
2.94729203e-01 3.48479301e-01 -1.02323532e+00 1.37909341e+00
3.97693276e-01 7.41749927e-02 -2.97670871e-01 9.34206486e-01
9.18632209e-01 7.48690486e-01 1.09866560e-01 -1.80964440e-01
1.48958588e+00 -8.43792737e-01 -3.52842867e-01 -4.63444263e-01
2.00904310e-01 -6.27870262e-01 1.18830884e+00 7.77803212e-02
-2.51447976e-01 -7.87494540e-01 -1.16511452e+00 2.86161602e-01
-6.96506262e-01 2.45380968e-01 7.02125251e-01 4.37931091e-01
-5.95887005e-01 1.44480392e-01 -2.63161749e-01 -2.36640334e-01
7.07489133e-01 1.88701406e-01 -5.75550258e-01 -2.29138955e-01
-7.52586544e-01 5.05888700e-01 6.76062644e-01 1.12746522e-01
-7.78830826e-01 -5.06693959e-01 -6.95223272e-01 2.09615484e-01
3.87943417e-01 -8.56558308e-02 6.76185906e-01 -1.10110390e+00
-8.23977411e-01 7.59046614e-01 -4.52637039e-02 -1.56412765e-01
5.07952332e-01 2.22915500e-01 -4.26041991e-01 2.18494043e-01
4.15815085e-01 5.75975597e-01 9.38026488e-01 -1.31932509e+00
-1.10241616e+00 -5.72281003e-01 -7.62087628e-02 8.32816511e-02
-4.98318285e-01 -1.20168872e-01 -5.36851764e-01 -8.07047427e-01
1.54282391e-01 -7.26893842e-01 -1.03853934e-01 2.17786461e-01
-3.78739566e-01 -4.68346775e-01 1.22146738e+00 -2.91801572e-01
1.12225604e+00 -2.20507598e+00 -6.42122254e-02 1.72687784e-01
3.25590730e-01 3.91011715e-01 -8.65916684e-02 2.21462101e-01
8.45638737e-02 1.93236470e-01 -1.49897665e-01 1.63550824e-01
-3.75298202e-01 -7.42404982e-02 -3.55916470e-01 5.26830137e-01
2.23684192e-01 6.70291245e-01 -6.84546173e-01 -8.00102174e-01
4.51206863e-01 1.88720211e-01 -1.41829401e-01 2.61470258e-01
1.19124472e-01 2.05465987e-01 -6.54676914e-01 8.53846371e-01
9.75794554e-01 -3.70064318e-01 -1.83311388e-01 -4.44762498e-01
-2.23455921e-01 -5.88775754e-01 -1.41834402e+00 1.17734802e+00
-2.01657832e-01 3.77511978e-01 -1.21568710e-01 -1.11082518e+00
8.77103031e-01 -1.33303210e-01 2.03981191e-01 -6.45424902e-01
2.93853462e-01 1.12311289e-01 -3.18327360e-02 -2.96417147e-01
3.25125575e-01 -1.42135903e-01 -1.92902416e-01 2.56253123e-01
-1.70374252e-02 5.80378175e-02 2.17773765e-01 9.18108374e-02
5.19543290e-01 -1.79882362e-01 7.23026931e-01 -3.68270427e-01
7.10024774e-01 1.21758588e-01 8.55006099e-01 7.60403156e-01
-4.97099817e-01 3.77200782e-01 1.06456123e-01 -4.60709363e-01
-6.99618042e-01 -9.74347353e-01 -4.23288047e-01 9.82241929e-01
7.93175399e-01 -2.96586752e-01 -6.00477636e-01 -8.79825592e-01
1.93043411e-01 2.06908509e-01 -9.98113990e-01 -1.87627450e-01
-1.10414855e-01 -6.93461239e-01 2.96141058e-01 6.65665090e-01
8.45125079e-01 -8.39890301e-01 -6.41137600e-01 -1.16234966e-01
-6.47551790e-02 -9.85165656e-01 -5.07072926e-01 6.47765025e-02
-6.03047967e-01 -1.18500865e+00 -6.51993692e-01 -9.81191218e-01
9.29211974e-01 9.40345705e-01 5.91086924e-01 1.68203816e-01
-8.61198246e-01 -1.48339078e-01 -4.77917075e-01 -4.21999335e-01
7.50622302e-02 -3.29581887e-01 -2.16968078e-02 5.03284514e-01
4.45186555e-01 -1.46026686e-01 -7.90022373e-01 6.45605981e-01
-7.46963739e-01 5.69712035e-02 8.42863023e-01 1.09065127e+00
9.04946744e-01 3.92020017e-01 7.50507414e-01 -8.35332870e-01
3.16729933e-01 -3.98694694e-01 -6.29161239e-01 2.33881816e-01
-5.61023057e-01 -1.92001775e-01 9.29014802e-01 -5.20780444e-01
-8.23996007e-01 -2.37274840e-02 2.14844838e-01 -4.39709604e-01
-2.31783062e-01 1.61317065e-01 -1.15883522e-01 -4.05461133e-01
5.15262008e-01 4.38694507e-01 -1.73754856e-01 -5.16517341e-01
3.15553188e-01 1.00417709e+00 5.72698772e-01 -2.49397576e-01
9.50594068e-01 4.32007074e-01 -1.26882559e-02 -8.21474969e-01
-7.48023629e-01 -7.54063606e-01 -7.47193277e-01 -2.24024758e-01
7.88638175e-01 -8.30481768e-01 -6.32577360e-01 3.34107995e-01
-6.24136269e-01 2.98050910e-01 -3.20713334e-02 2.72085607e-01
-2.01028019e-01 6.28090262e-01 -1.97249800e-01 -8.01688313e-01
-4.37835485e-01 -1.15803218e+00 1.06468976e+00 6.06390774e-01
2.15240926e-01 -6.22675598e-01 -5.27991354e-01 2.53601938e-01
3.62285852e-01 3.86956066e-01 1.10201478e+00 -3.11954200e-01
-5.90605259e-01 -5.22370160e-01 -8.81963849e-01 4.03323382e-01
4.73033816e-01 -1.37655050e-01 -9.10227537e-01 -5.33085406e-01
-3.27357620e-01 -6.25396430e-01 8.39769840e-01 1.54689714e-01
1.64955199e+00 -2.13273853e-01 -6.39867425e-01 5.29010117e-01
1.69357336e+00 2.23945394e-01 1.58286691e-01 3.18410009e-01
7.57887244e-01 4.56062376e-01 1.09675837e+00 3.76384586e-01
1.13677979e-01 7.83491552e-01 6.22767985e-01 -1.30691454e-01
-2.96675116e-01 -2.21770838e-01 -2.74275303e-01 3.70058656e-01
-6.16996840e-04 1.02684163e-01 -6.94793463e-01 6.30866647e-01
-1.68144155e+00 -8.22208166e-01 -3.13501097e-02 2.18320417e+00
6.02173805e-01 -5.06524853e-02 1.40889391e-01 3.42068344e-01
9.92563426e-01 1.01750843e-01 -6.65367186e-01 -6.39850348e-02
-4.75660451e-02 -1.84217870e-01 3.74284327e-01 -9.25958678e-02
-1.43185484e+00 7.66929567e-01 5.36677647e+00 1.37171769e+00
-1.36698377e+00 4.50509651e-05 7.48047769e-01 3.35015446e-01
3.05562973e-01 -1.58751562e-01 -8.76036406e-01 6.38968289e-01
1.77443236e-01 -2.95927469e-02 2.14159921e-01 1.05144703e+00
-8.17197636e-02 -1.40194491e-01 -8.10596228e-01 1.39743400e+00
6.22698888e-02 -1.23783338e+00 2.83863544e-01 4.95757684e-02
6.04613423e-01 -3.44734788e-01 1.72511139e-03 3.33577842e-01
-9.52496231e-02 -1.00266933e+00 7.87811160e-01 1.12636976e-01
1.17151642e+00 -8.95234704e-01 8.02481711e-01 4.90948945e-01
-1.68893039e+00 -4.55161572e-01 -9.33117807e-01 5.24408445e-02
-3.67862344e-01 5.08939326e-01 -7.01472521e-01 6.44578278e-01
8.85317624e-01 5.54555237e-01 -1.10900688e+00 1.31321573e+00
2.16632262e-01 3.41387004e-01 -4.33770083e-02 -2.22394854e-01
4.38473940e-01 -6.70099258e-02 2.41711959e-01 1.17596090e+00
2.62063473e-01 -7.75066167e-02 5.37988424e-01 7.27990329e-01
2.21223310e-02 3.60387057e-01 -4.75394756e-01 7.12544005e-03
5.70715606e-01 1.67192638e+00 -1.02925813e+00 -1.33575603e-01
-4.04726326e-01 1.00245607e+00 4.07405555e-01 1.76988348e-01
-6.85583889e-01 -7.67778575e-01 3.94133389e-01 5.38504086e-02
5.25644124e-01 1.01855785e-01 -1.61141545e-01 -1.10492444e+00
6.19286075e-02 -7.07373023e-01 5.49826086e-01 -4.43194300e-01
-1.40468717e+00 7.68874228e-01 -2.14220703e-01 -1.51983082e+00
-3.64000089e-02 -6.63976014e-01 -5.48439622e-01 8.38452935e-01
-1.33179069e+00 -1.38081300e+00 -9.02258813e-01 7.24741280e-01
6.28819883e-01 -1.33552849e-01 7.44929552e-01 2.30529055e-01
-7.13321865e-01 6.78270459e-01 2.81945258e-01 2.16826558e-01
4.94046569e-01 -1.12223220e+00 -1.34993255e-01 8.73007298e-01
-1.21837996e-01 4.79789257e-01 4.41992790e-01 -4.05601591e-01
-1.36840951e+00 -1.49980736e+00 3.49493414e-01 -9.10044611e-02
4.63606179e-01 -4.65243936e-01 -8.65649223e-01 1.85354725e-01
-2.05439523e-01 5.46350956e-01 4.43479449e-01 -1.87513992e-01
-6.03085399e-01 -4.40766245e-01 -1.29939497e+00 2.52771109e-01
8.34808409e-01 -4.31357354e-01 -6.18796706e-01 1.87190339e-01
3.55576575e-01 1.58170685e-01 -5.82911611e-01 4.34819460e-01
5.32642365e-01 -8.01656783e-01 9.21391368e-01 -3.52811784e-01
1.30705088e-01 -7.70339549e-01 -3.77961129e-01 -1.36458623e+00
-5.96037447e-01 2.27079272e-01 2.64992505e-01 1.60578740e+00
-1.54719874e-01 -6.41040564e-01 5.53006709e-01 -6.34867921e-02
4.24609371e-02 -8.64368439e-01 -8.83233249e-01 -9.06351209e-01
-1.58500910e-01 8.22170451e-02 5.59505343e-01 8.93990755e-01
-4.22853410e-01 2.83010095e-01 -2.45070085e-01 9.62145403e-02
8.06471825e-01 8.26425433e-01 7.00748146e-01 -1.53521645e+00
-1.39699101e-01 -4.75407213e-01 -8.35686147e-01 -8.23801339e-01
-1.22669205e-01 -7.88376749e-01 2.37301037e-01 -1.58458304e+00
6.56656146e-01 -7.03620136e-01 -3.01014990e-01 4.63514388e-01
-4.33637857e-01 6.68322504e-01 2.25264609e-01 3.81902516e-01
-5.18454134e-01 5.45909166e-01 1.14996517e+00 -3.70245636e-01
7.85057694e-02 -5.69211668e-04 -1.02611899e+00 5.84739387e-01
4.11377072e-01 -1.40547574e-01 -4.54709053e-01 2.29829196e-02
-5.35716295e-01 -2.74213135e-01 4.62161988e-01 -1.13651800e+00
1.96691807e-02 -3.55759114e-01 8.69226456e-01 -7.46141553e-01
1.70978650e-01 -8.38084519e-01 -4.07099351e-02 4.51837093e-01
-1.46906509e-03 -1.87188148e-01 2.29429990e-01 9.68180120e-01
-4.86539245e-01 -1.88126639e-01 1.26843190e+00 -6.81763962e-02
-1.26108325e+00 4.11191583e-01 -1.63200051e-01 3.12825926e-02
1.42250192e+00 -4.32810128e-01 -5.11207759e-01 5.31551912e-02
-9.60142389e-02 1.38017967e-01 4.06772107e-01 5.72230041e-01
6.21299684e-01 -1.42127454e+00 -5.61746180e-01 2.44659856e-01
7.35103130e-01 -6.06850833e-02 5.39849341e-01 6.61683857e-01
-3.17738146e-01 5.09625316e-01 -4.80932832e-01 -8.55317056e-01
-1.37204647e+00 8.98894310e-01 1.58161744e-01 -7.04601780e-02
-6.35201633e-01 8.18782866e-01 8.93571377e-01 -1.43233895e-01
2.07451001e-01 -3.70904580e-02 -4.56330210e-01 2.14950547e-01
9.05070901e-01 1.95625007e-01 5.48164174e-02 -9.77363944e-01
-5.68305850e-01 8.85218501e-01 -3.21443498e-01 5.90209663e-01
1.14221060e+00 -2.05310956e-01 3.92139331e-02 2.43046045e-01
1.24272346e+00 -1.27166837e-01 -1.18966913e+00 -3.52877527e-01
-3.08846503e-01 -9.54863846e-01 2.10684165e-01 -8.34039032e-01
-9.80665684e-01 1.08545518e+00 9.81563568e-01 3.30555052e-01
1.24225378e+00 -6.80271070e-03 4.78265762e-01 8.38351250e-02
5.20322025e-01 -9.79844332e-01 1.41540229e-01 1.65909827e-01
8.87686789e-01 -1.24997747e+00 2.92201750e-02 -7.03583062e-01
-5.68175673e-01 1.01389313e+00 8.39997768e-01 -6.33940324e-02
5.85121632e-01 -1.03574052e-01 -6.72889128e-02 3.61481286e-03
-2.70984739e-01 -3.48111689e-01 5.04724681e-01 6.51736200e-01
5.26346900e-02 1.72360048e-01 -2.65842527e-01 7.74351180e-01
1.20805010e-01 -1.59774363e-01 2.25662395e-01 7.67520189e-01
-8.28178585e-01 -7.41257250e-01 -4.65786755e-01 9.51316893e-01
-3.70751083e-01 1.40935808e-01 -4.46805388e-01 7.46314108e-01
4.33917761e-01 1.04552746e+00 1.59723595e-01 -5.86919010e-01
1.97408065e-01 -3.04648459e-01 3.10857147e-01 -5.57973087e-01
-3.37360144e-01 7.46406689e-02 -2.38230437e-01 -4.32384968e-01
-7.25573674e-02 -3.87725741e-01 -1.14406371e+00 -1.27827615e-01
-5.02792776e-01 2.76689410e-01 4.39424515e-01 8.14423800e-01
3.23328376e-01 3.95316064e-01 8.78385365e-01 -8.66668999e-01
-6.44256175e-01 -9.93390918e-01 -8.66331577e-01 6.67688489e-01
3.29686821e-01 -1.21515524e+00 -3.99571210e-01 -2.53733963e-01] | [9.633298873901367, 1.9892250299453735] |
745473d2-b541-49ba-ba9a-287a008ce0dc | customizing-an-affective-tutoring-system | 2111.14262 | null | https://arxiv.org/abs/2111.14262v1 | https://arxiv.org/pdf/2111.14262v1.pdf | Customizing an Affective Tutoring System Based on Facial Expression and Head Pose Estimation | In recent years, the main problem in e-learning has shifted from analyzing content to personalization of learning environment by Intelligence Tutoring Systems (ITSs). Therefore, by designing personalized teaching models, learners are able to have a successful and satisfying experience in achieving their learning goals. Affective Tutoring Systems (ATSs) are some kinds of ITS that can recognize and respond to affective states of learner. In this study, we designed, implemented, and evaluated a system to personalize the learning environment based on the facial emotions recognition, head pose estimation, and cognitive style of learners. First, a unit called Intelligent Analyzer (AI) created which was responsible for recognizing facial expression and head angles of learners. Next, the ATS was built which mainly made of two units: ITS, IA. Results indicated that with the ATS, participants needed less efforts to pass the tests. In other words, we observed when the IA unit was activated, learners could pass the final tests in fewer attempts than those for whom the IA unit was deactivated. Additionally, they showed an improvement in terms of the mean passing score and academic satisfaction. | ['Ebrahim Mousavi', 'Gholam Ali Montazer', 'Mahdi Pourmirzaei'] | 2021-11-21 | null | null | null | null | ['head-pose-estimation'] | ['computer-vision'] | [-3.53166044e-01 4.95932877e-01 1.59714252e-01 -5.58241487e-01
-3.22826542e-02 -3.94410551e-01 1.29960239e-01 3.30702513e-01
-2.49688402e-01 4.96319503e-01 2.03140676e-02 -5.03771789e-02
-5.45472726e-02 -8.48678768e-01 -3.75160426e-01 -5.53696990e-01
4.07827049e-01 1.39297798e-01 2.91314691e-01 -3.45322728e-01
3.38625073e-01 6.96542084e-01 -2.19458461e+00 8.95547643e-02
1.19296682e+00 6.65784478e-01 2.02024743e-01 3.76835048e-01
-4.91050869e-01 5.31931758e-01 -6.73292696e-01 -4.38705226e-03
-3.31282437e-01 -4.16059822e-01 -5.41933477e-01 1.80980414e-02
1.71811789e-01 -3.24824095e-01 -7.06338361e-02 8.78694117e-01
6.81606054e-01 5.65520644e-01 4.02081907e-01 -1.31827414e+00
-2.34423280e-01 3.02577585e-01 -1.35495931e-01 -3.31298500e-01
6.59303546e-01 -1.35599613e-01 2.43281901e-01 -5.86276591e-01
1.55322433e-01 8.76887083e-01 2.68921584e-01 9.65223134e-01
-5.39284289e-01 -7.86491454e-01 1.78547740e-01 5.55729926e-01
-9.06118691e-01 -2.12790087e-01 6.71861827e-01 -3.97723109e-01
4.29394931e-01 1.79888651e-01 1.09400439e+00 7.87110567e-01
1.00799739e-01 7.97449768e-01 1.09236765e+00 -7.02845931e-01
3.47783685e-01 9.67849553e-01 4.83949572e-01 5.90792120e-01
2.11896330e-01 -2.91586906e-01 -6.23713434e-01 3.87631029e-01
4.04972613e-01 -1.80105224e-01 -3.12563747e-01 -4.66320097e-01
-5.11338830e-01 5.50801516e-01 1.24345580e-02 3.87593091e-01
-5.14041722e-01 -5.84895134e-01 3.07194829e-01 1.60931051e-01
-7.58780837e-02 4.41541076e-01 -3.35393339e-01 -5.65841496e-01
-2.02044576e-01 -2.97664940e-01 8.73950422e-01 8.59408259e-01
3.79761517e-01 8.18483680e-02 -2.24439517e-01 6.00248218e-01
3.67558211e-01 4.36055183e-01 9.12188172e-01 -5.70070505e-01
-3.01249683e-01 1.16380572e+00 -2.29977041e-01 -5.36662996e-01
-4.93862003e-01 -2.12990209e-01 -2.27943420e-01 3.17988902e-01
1.38195381e-02 -5.81863761e-01 -5.47278225e-01 1.72052824e+00
8.16460192e-01 -1.27216965e-01 1.81808621e-01 6.01426661e-01
1.25209844e+00 4.90907997e-01 3.98000628e-01 -5.17310143e-01
1.52222157e+00 -5.99843502e-01 -1.23733675e+00 1.20053306e-01
6.97844386e-01 -7.76035726e-01 1.21358466e+00 6.21661663e-01
-1.01957250e+00 -6.84800327e-01 -8.51437986e-01 3.86140555e-01
-4.29214329e-01 -3.09922285e-02 4.26145822e-01 9.48945582e-01
-1.03157234e+00 2.56919175e-01 -4.37426150e-01 -5.96252561e-01
-2.65748054e-01 4.69657689e-01 -4.05481607e-01 4.64488119e-01
-1.02813387e+00 9.83409286e-01 3.95401604e-02 -2.49227300e-01
-5.17978847e-01 -4.48336780e-01 -6.51239157e-01 3.33497256e-01
4.94467802e-02 -3.73586774e-01 1.41539681e+00 -1.14569771e+00
-2.32888842e+00 7.52851963e-01 -1.46465719e-01 3.48068267e-01
1.03727452e-01 -4.11991984e-01 -2.27392256e-01 1.09073713e-01
-3.33932936e-01 4.02644843e-01 2.95173645e-01 -1.01636028e+00
-6.50215209e-01 -8.46013188e-01 -2.00196564e-01 7.86982000e-01
-8.04954469e-01 -1.24197043e-01 -3.43601674e-01 3.56790684e-02
-2.95785181e-02 -8.84701431e-01 3.28225642e-01 -4.27022159e-01
9.79178101e-02 -5.46238244e-01 1.09398890e+00 -4.89703357e-01
1.28809929e+00 -2.12745094e+00 -8.12525377e-02 5.60988903e-01
6.87994361e-02 8.47948909e-01 3.31969261e-01 1.47582039e-01
-2.42311984e-01 -2.14719668e-01 7.37456918e-01 1.10277027e-01
2.63846785e-01 2.40486398e-01 3.86183947e-01 -1.75519809e-02
-5.79353571e-01 2.00677902e-01 -5.26094079e-01 -5.42190492e-01
5.74509263e-01 2.64282256e-01 -5.09321749e-01 8.62525165e-01
1.08061574e-01 3.99218410e-01 -5.34819484e-01 3.77402693e-01
4.34359968e-01 3.70081455e-01 1.25223622e-01 1.66534632e-01
-4.16216642e-01 2.05317941e-02 -1.12080848e+00 9.06465292e-01
-5.29457271e-01 5.30813038e-01 2.90904671e-01 -6.99863017e-01
1.31556153e+00 7.70164788e-01 5.47148228e-01 -1.07897854e+00
4.10140038e-01 -1.00008726e-01 -9.60546955e-02 -1.20128059e+00
4.26882148e-01 2.80188501e-01 3.31587613e-01 4.12818968e-01
-1.08647734e-01 -6.43130615e-02 -2.50976264e-01 -2.80170534e-02
6.96766078e-01 2.60838687e-01 1.46651715e-01 -1.04357805e-02
7.52882242e-01 -4.55945402e-01 3.20379883e-01 1.20214455e-01
-4.90210325e-01 -2.99486876e-01 2.30918482e-01 -1.74597666e-01
-3.79266143e-01 -6.49373412e-01 4.22966741e-02 1.40332401e+00
-3.84449475e-02 -2.34267846e-01 -1.25737166e+00 -3.80316406e-01
-3.74894828e-01 9.59964335e-01 -2.83286273e-01 -5.74231505e-01
-1.84431419e-01 -1.06645375e-01 6.89125881e-02 1.96967810e-01
7.78487802e-01 -1.35425591e+00 -8.31427574e-01 3.74272442e-03
-1.79353476e-01 -6.03473306e-01 -3.27221990e-01 8.31248611e-02
-6.51010215e-01 -8.10582697e-01 -3.72309238e-01 -1.02418303e+00
8.78289461e-01 2.98490465e-01 5.52019238e-01 1.28828019e-01
-9.67271347e-03 9.10048008e-01 -5.58982670e-01 -8.46701741e-01
-9.42586958e-02 6.84425980e-02 2.02158540e-01 -9.82149914e-02
7.10602343e-01 -5.26493490e-01 -5.88462293e-01 2.94064164e-01
-6.73132300e-01 1.01283044e-01 6.09060466e-01 2.60626554e-01
3.11767250e-01 -1.70030117e-01 7.05169141e-01 -6.86533153e-01
8.97229791e-01 -2.64838189e-01 -3.87532055e-01 4.31252807e-01
-8.42768967e-01 -2.35851202e-02 4.19990838e-01 -3.18302006e-01
-1.43755484e+00 6.93201274e-02 -3.35209966e-01 -1.04472771e-01
-7.26307392e-01 3.09490144e-01 -5.46516955e-01 -2.34429896e-01
4.54036385e-01 2.50056498e-02 3.28977019e-01 -1.49116740e-01
-3.13891351e-01 1.16294920e+00 4.75328177e-01 -6.17282808e-01
3.46644640e-01 -5.42208076e-01 -1.13155931e-01 -9.29329813e-01
-6.21834636e-01 -3.85790944e-01 -5.48742592e-01 -1.18077886e+00
8.97987068e-01 -5.85527360e-01 -1.47566748e+00 5.47779739e-01
-5.44624150e-01 -3.02721143e-01 3.57098468e-02 9.54018712e-01
-2.24706218e-01 -3.64773832e-02 -2.84483641e-01 -9.94096458e-01
-4.26256657e-01 -9.57315803e-01 3.13702494e-01 1.20445228e+00
-4.87469822e-01 -9.48777020e-01 3.57323214e-02 6.73965096e-01
4.70516413e-01 -5.42916283e-02 8.87716830e-01 -6.65525913e-01
1.16067603e-01 -1.96289510e-01 4.06913042e-01 1.98114529e-01
1.19662561e-01 3.58171493e-01 -1.06010401e+00 7.50864521e-02
1.41005099e-01 -1.97123155e-01 -2.34063864e-01 2.03437164e-01
1.14304793e+00 -1.87947050e-01 -5.89254573e-02 2.67200559e-01
8.32898855e-01 1.01346350e+00 7.09332407e-01 6.72596693e-01
2.40162492e-01 9.78857994e-01 8.72527242e-01 3.89312476e-01
5.32634914e-01 7.16747820e-01 2.48273894e-01 -2.89136972e-02
2.22080722e-01 -1.38104737e-01 7.84917951e-01 1.11324656e+00
-1.10240608e-01 1.87304273e-01 -7.46718585e-01 2.11441949e-01
-1.48118901e+00 -6.90811276e-01 -2.56596088e-01 2.33721685e+00
8.42395842e-01 -1.56331450e-01 7.53820539e-02 9.08380821e-02
7.30423391e-01 -5.96618235e-01 -5.18928111e-01 -9.92199421e-01
6.01095676e-01 3.72438759e-01 -3.37018371e-02 4.88233358e-01
-5.52326977e-01 8.15697014e-01 5.63066387e+00 1.85310274e-01
-1.30044639e+00 -4.16093737e-01 3.10213864e-01 3.03807240e-02
-2.36306071e-01 -4.10043180e-01 -8.04197192e-01 1.80253968e-01
1.07713377e+00 -2.98979908e-01 1.67950794e-01 1.00746083e+00
3.30794126e-01 -3.40655386e-01 -6.95349097e-01 6.16757929e-01
3.40070613e-02 -3.67505759e-01 -1.98326528e-01 -2.17642143e-01
3.15306544e-01 -7.70722508e-01 4.21055593e-03 6.57144248e-01
-7.13777766e-02 -6.30414069e-01 3.53007466e-01 8.17054152e-01
3.45672876e-01 -1.01134765e+00 1.00556874e+00 4.37502235e-01
-7.97059059e-01 1.11201443e-01 6.27012029e-02 -6.94469875e-03
-4.25314963e-01 -1.10791571e-01 -1.14009547e+00 3.17132026e-01
7.24636078e-01 -1.36066288e-01 -4.54629868e-01 9.55152154e-01
-5.26760161e-01 5.71038365e-01 -1.38688549e-01 -6.04036987e-01
-2.29619414e-01 -4.10328627e-01 2.48018399e-01 6.79027915e-01
5.16463161e-01 6.38240993e-01 -9.01072621e-02 1.34929538e-01
2.57288575e-01 4.60237771e-01 -3.68824124e-01 2.33587831e-01
6.67772472e-01 1.43822682e+00 -3.89676750e-01 -1.83635831e-01
-3.39636832e-01 5.01720309e-01 2.00490151e-02 2.17637494e-01
-8.62197816e-01 -6.39508367e-01 4.17228550e-01 2.03986853e-01
-1.98386952e-01 3.05653542e-01 -2.56770223e-01 -6.00514472e-01
-2.42657959e-01 -8.78302634e-01 2.58501649e-01 -1.08866918e+00
-5.50368369e-01 4.48533118e-01 -9.71494615e-02 -7.89474964e-01
-1.71235994e-01 -5.53996325e-01 -9.72678483e-01 6.76604211e-01
-8.56083393e-01 -5.31011343e-01 -1.08893049e+00 8.09167683e-01
2.85385549e-01 -1.41068429e-01 1.09550345e+00 3.04458141e-01
-1.17866552e+00 8.14807117e-01 -2.08336651e-01 -2.31414035e-01
1.04045177e+00 -1.29171598e+00 -9.35600042e-01 3.60628128e-01
-4.90218550e-01 3.55696440e-01 8.37445617e-01 -4.01178330e-01
-1.45278382e+00 -5.51546633e-01 8.19584131e-01 3.17895859e-01
-9.20765623e-02 2.38289326e-01 -1.01344478e+00 6.29096031e-01
2.84009606e-01 -7.89874732e-01 1.08400726e+00 1.78363875e-01
3.59811664e-01 -3.86371285e-01 -1.24817383e+00 6.63065255e-01
3.50399315e-01 -1.41062453e-01 -5.66989660e-01 -6.85321614e-02
1.19927973e-01 -5.18967032e-01 -1.04937196e+00 2.42561311e-01
6.36898339e-01 -9.73562598e-01 3.88869226e-01 -3.04608434e-01
1.80928156e-01 2.44342685e-02 3.82144451e-01 -1.39820337e+00
-1.17595442e-01 -4.50362116e-01 1.75771683e-01 1.40328276e+00
1.70300558e-01 -6.64188147e-01 8.93950939e-01 1.50997424e+00
-5.01520932e-01 -8.30062270e-01 -4.11971301e-01 -7.47450590e-02
-1.72373265e-01 1.53685242e-01 6.54435396e-01 8.05742621e-01
8.38208616e-01 4.17887986e-01 2.06190899e-01 1.27993762e-01
1.49203852e-01 -2.18626544e-01 1.08463633e+00 -1.23847044e+00
7.21416324e-02 -5.32145441e-01 -2.51021653e-01 -6.03509665e-01
5.14621846e-02 -3.58005315e-01 4.06367779e-02 -1.44652128e+00
-9.37734991e-02 -5.78554869e-02 -7.79056773e-02 6.43482924e-01
-4.37330425e-01 -5.60322046e-01 -1.14497833e-01 -3.46261948e-01
-5.65588236e-01 5.98730206e-01 1.36333418e+00 5.91821432e-01
-5.73247135e-01 3.76609027e-01 -6.79378927e-01 8.49761605e-01
1.15535867e+00 -1.13291601e-02 -5.31120777e-01 5.99648803e-02
-7.00530931e-02 3.66354465e-01 -2.73093581e-01 -1.02080202e+00
5.20534039e-01 -3.98681164e-01 5.82105517e-01 -5.02727330e-01
1.47304416e-01 -9.88321006e-01 6.65587708e-02 3.67310703e-01
-2.27406263e-01 -7.88360164e-02 5.28140664e-01 -4.20152932e-01
-1.86003864e-01 -5.75189054e-01 7.69468188e-01 2.49715149e-01
-5.34459114e-01 -1.09013729e-01 -1.05099070e+00 -6.09346151e-01
1.65625381e+00 -3.62968028e-01 -2.20732823e-01 -7.31383324e-01
-7.51600802e-01 4.43192899e-01 1.42963558e-01 3.04010719e-01
6.71075583e-01 -1.02575397e+00 -1.42226905e-01 4.11048472e-01
-6.20250143e-02 6.77358359e-02 5.32787323e-01 8.25800896e-01
-4.51797962e-01 5.37150621e-01 -7.10144341e-01 -2.28214189e-01
-2.05115509e+00 1.09368272e-01 3.82687658e-01 1.06636470e-03
-2.54947960e-01 5.77223361e-01 7.83669651e-02 -4.92730439e-01
7.79755473e-01 6.25981838e-02 -1.10068655e+00 3.54467839e-01
6.07895672e-01 5.30885816e-01 1.42612666e-01 -4.26457167e-01
1.44847622e-02 3.64639342e-01 3.88343297e-02 -2.56865937e-02
1.25533438e+00 -2.35567898e-01 1.15987584e-01 4.34535831e-01
5.23491085e-01 2.26392761e-01 -8.52336168e-01 1.54833570e-01
-1.31103218e-01 -1.37800515e-01 -1.38544450e-02 -1.00868320e+00
-1.06635606e+00 7.53037810e-01 8.59704018e-01 5.12263589e-02
1.39808917e+00 -3.66506934e-01 5.34242153e-01 3.73818964e-01
1.23745136e-01 -1.45242572e+00 1.53369293e-01 6.36095941e-01
5.39659917e-01 -8.74974370e-01 -4.44117904e-01 -1.80277452e-01
-7.18321621e-01 1.30159473e+00 1.52908850e+00 5.08977234e-01
5.39771736e-01 1.52477575e-02 2.11457863e-01 -2.92768419e-01
-7.43786812e-01 5.93123287e-02 4.43710148e-01 4.80882883e-01
8.12128365e-01 3.01683903e-01 -3.88369322e-01 1.07021201e+00
-5.09924710e-01 4.36459519e-02 5.72053611e-01 9.04614508e-01
-1.14870369e+00 -1.02429497e+00 -7.58531511e-01 2.18342856e-01
-1.32146850e-01 4.44895118e-01 -6.33867323e-01 8.04558754e-01
2.19177514e-01 9.90363300e-01 -2.27214191e-02 -5.57801723e-01
8.42436254e-01 4.52501416e-01 4.03286994e-01 -5.46515822e-01
-8.16233516e-01 -2.32744530e-01 -7.09619373e-02 -3.68231803e-01
-1.46137223e-01 -6.31762207e-01 -1.69560707e+00 -5.34834802e-01
-3.45859379e-01 7.35798120e-01 9.75093067e-01 7.76277542e-01
2.07248405e-01 9.44772959e-01 8.41229439e-01 -3.69552284e-01
-1.72110200e-01 -1.15467393e+00 -6.32042646e-01 1.58412188e-01
-3.68595958e-01 -5.41145623e-01 -2.73824513e-01 -2.47656003e-01] | [13.436083793640137, 2.9597575664520264] |
76b479aa-e4b4-46b4-8777-4426868d2103 | make-a-story-visual-memory-conditioned | 2211.13319 | null | https://arxiv.org/abs/2211.13319v3 | https://arxiv.org/pdf/2211.13319v3.pdf | Make-A-Story: Visual Memory Conditioned Consistent Story Generation | There has been a recent explosion of impressive generative models that can produce high quality images (or videos) conditioned on text descriptions. However, all such approaches rely on conditional sentences that contain unambiguous descriptions of scenes and main actors in them. Therefore employing such models for more complex task of story visualization, where naturally references and co-references exist, and one requires to reason about when to maintain consistency of actors and backgrounds across frames/scenes, and when not to, based on story progression, remains a challenge. In this work, we address the aforementioned challenges and propose a novel autoregressive diffusion-based framework with a visual memory module that implicitly captures the actor and background context across the generated frames. Sentence-conditioned soft attention over the memories enables effective reference resolution and learns to maintain scene and actor consistency when needed. To validate the effectiveness of our approach, we extend the MUGEN dataset and introduce additional characters, backgrounds and referencing in multi-sentence storylines. Our experiments for story generation on the MUGEN, the PororoSV and the FlintstonesSV dataset show that our method not only outperforms prior state-of-the-art in generating frames with high visual quality, which are consistent with the story, but also models appropriate correspondences between the characters and the background. | ['Leonid Sigal', 'Shweta Mahajan', 'Sergey Tulyakov', 'Jian Ren', 'Hsin-Ying Lee', 'Tanzila Rahman'] | 2022-11-23 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Rahman_Make-a-Story_Visual_Memory_Conditioned_Consistent_Story_Generation_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Rahman_Make-a-Story_Visual_Memory_Conditioned_Consistent_Story_Generation_CVPR_2023_paper.pdf | cvpr-2023-1 | ['story-visualization', 'story-generation'] | ['computer-vision', 'natural-language-processing'] | [ 5.98568246e-02 -2.81172931e-01 1.43575564e-01 -1.30553916e-01
-4.59963739e-01 -4.20669138e-01 1.13139272e+00 -1.21834010e-01
7.88340420e-02 7.65924990e-01 6.25741363e-01 2.73305357e-01
1.58594415e-01 -8.38751793e-01 -9.06901956e-01 -5.70370138e-01
3.64785284e-01 3.03629458e-01 6.51815414e-01 -3.26069057e-01
1.99802577e-01 9.78711024e-02 -1.67674160e+00 7.70397305e-01
5.84288836e-01 6.43746197e-01 5.60590029e-01 9.90858316e-01
-1.87366232e-01 1.47904265e+00 -6.86455607e-01 -4.82333928e-01
-5.50489202e-02 -8.58566403e-01 -5.32979906e-01 5.02068222e-01
8.50915611e-01 -4.99045193e-01 -6.52937293e-01 7.58955121e-01
5.57155728e-01 3.13040972e-01 7.57875621e-01 -1.12013888e+00
-9.92279887e-01 7.38564193e-01 -7.20803380e-01 3.14000666e-01
5.66323578e-01 4.22507882e-01 7.16918409e-01 -7.34136701e-01
1.05342245e+00 1.41264439e+00 3.17516983e-01 5.49975634e-01
-1.29652643e+00 -5.65855086e-01 4.97323602e-01 4.84568417e-01
-1.22748661e+00 -7.05212951e-01 8.38967204e-01 -7.34252930e-01
8.03344727e-01 3.91656280e-01 8.76679778e-01 1.63149393e+00
4.29580733e-02 6.75445497e-01 8.09922218e-01 -4.02252913e-01
1.76791996e-01 7.09050000e-02 -1.53067037e-01 5.63475549e-01
-1.92993492e-01 -1.34770513e-01 -1.02892613e+00 3.42264920e-01
1.03586733e+00 -1.41052514e-01 -4.80489969e-01 -3.85390878e-01
-1.40974760e+00 6.21295214e-01 6.26331568e-02 3.45241904e-01
-4.23678279e-01 3.14639986e-01 2.59741843e-01 -1.63746864e-01
5.16199529e-01 9.97671112e-02 3.10699672e-01 -1.23325393e-01
-1.32827830e+00 4.46723074e-01 5.02529621e-01 1.25021458e+00
2.44289249e-01 4.02214766e-01 -8.03730249e-01 6.58325493e-01
3.96616980e-02 5.13885856e-01 1.49886072e-01 -1.00696838e+00
5.12109756e-01 3.77158254e-01 2.47707412e-01 -1.46069229e+00
3.31076868e-02 -3.02124619e-01 -1.07517457e+00 8.36709142e-02
2.82038063e-01 1.90925404e-01 -7.68080473e-01 1.76622069e+00
3.72980297e-01 5.94252169e-01 4.06367667e-02 1.00313413e+00
1.08256030e+00 9.74580646e-01 8.82658884e-02 -3.04847777e-01
1.22673845e+00 -1.05783319e+00 -1.02908278e+00 -2.57871628e-01
7.95473903e-02 -8.83017182e-01 1.19544458e+00 2.06350774e-01
-1.46120119e+00 -8.22093844e-01 -9.51146185e-01 -3.11027050e-01
-1.88920889e-02 1.68605939e-01 3.44665080e-01 1.13904566e-01
-1.11864710e+00 4.15172935e-01 -6.68337822e-01 -3.68456572e-01
3.65698874e-01 -3.23877305e-01 -1.83447257e-01 -1.01073220e-01
-1.07654953e+00 8.84646237e-01 4.53263938e-01 -2.35938802e-02
-1.21364212e+00 -7.43750095e-01 -9.01554823e-01 1.04370907e-01
2.68943310e-01 -1.02025521e+00 9.55235600e-01 -1.02691388e+00
-1.47935927e+00 7.38212109e-01 -1.06085159e-01 -5.51851630e-01
8.44353735e-01 -2.61809140e-01 -3.43788087e-01 3.11668515e-01
8.23215097e-02 8.40360403e-01 9.67196047e-01 -1.44973278e+00
-5.04987657e-01 1.38397187e-01 3.03270429e-01 2.26206318e-01
-1.70374006e-01 1.78281367e-01 -5.88550925e-01 -7.92035103e-01
-2.45382473e-01 -5.58748841e-01 1.24177650e-01 7.75568634e-02
-5.48026085e-01 1.71672672e-01 9.40984547e-01 -8.84051204e-01
1.37040401e+00 -2.18999505e+00 4.66281295e-01 -4.15125281e-01
6.48834407e-02 -2.63693407e-02 -9.75672454e-02 3.24222177e-01
-3.87796089e-02 -1.32001847e-01 -1.45399526e-01 -7.20033646e-01
-1.15899667e-01 1.42859444e-01 -6.00543439e-01 3.78800452e-01
2.31789231e-01 7.89736807e-01 -9.56682563e-01 -8.55921209e-01
5.42644203e-01 8.68143022e-01 -4.27793622e-01 3.61274958e-01
-5.82344532e-01 7.39347279e-01 -4.53595445e-03 2.63696134e-01
4.42803979e-01 -4.12293851e-01 -1.12979054e-01 -3.42917800e-01
-1.43200323e-01 -3.10994554e-02 -1.39784384e+00 1.90119147e+00
-2.60893106e-01 9.49695408e-01 -1.27652898e-01 -5.52530050e-01
7.27885127e-01 2.60518134e-01 1.22998908e-01 -5.84245443e-01
-2.72866059e-03 -2.93991089e-01 -3.73530924e-01 -7.15872765e-01
8.25309217e-01 1.17924973e-01 1.65210918e-01 2.82706350e-01
1.53808855e-03 -3.14583212e-01 6.84986889e-01 6.09266102e-01
6.87402964e-01 5.06512582e-01 2.83162370e-02 -4.19851393e-02
4.36373413e-01 -6.59898967e-02 3.01845491e-01 7.43384421e-01
3.10103059e-01 1.20419788e+00 4.72460538e-01 -3.60818446e-01
-1.27691495e+00 -9.86799896e-01 2.78180599e-01 9.35039580e-01
3.49635303e-01 -5.86959958e-01 -6.89941645e-01 -2.97257960e-01
-5.52758157e-01 1.23563528e+00 -7.18750596e-01 1.18572675e-01
-6.72686398e-01 -4.43749636e-01 1.24593481e-01 5.56850553e-01
6.74321890e-01 -1.07763243e+00 -8.62645507e-01 1.67765245e-01
-5.05585611e-01 -1.27206647e+00 -6.70194149e-01 -5.03349066e-01
-3.83476377e-01 -1.01830459e+00 -8.88415933e-01 -4.80015814e-01
6.14782691e-01 2.82626271e-01 1.36868417e+00 -5.19390255e-02
-1.34648800e-01 4.35535431e-01 -3.79734397e-01 -6.46485984e-02
-6.72669947e-01 -3.79788637e-01 -3.52325797e-01 2.60629833e-01
-4.76101518e-01 -5.72398484e-01 -6.32969916e-01 1.53740034e-01
-1.08037448e+00 1.02525365e+00 9.00848955e-02 7.22640216e-01
6.44770503e-01 -1.69829935e-01 1.51028126e-01 -7.67415166e-01
2.83447385e-01 -3.85883391e-01 -3.98549795e-01 5.06824732e-01
1.99366920e-02 -1.49338497e-02 4.63937521e-01 -6.74557805e-01
-1.38337970e+00 -1.10595018e-01 1.91686198e-01 -7.73476362e-01
-2.02918481e-02 4.08788100e-02 -2.30701730e-01 5.82773805e-01
5.31155229e-01 4.08823073e-01 -4.76475358e-01 -1.70927003e-01
7.16895580e-01 2.28894100e-01 8.10786486e-01 -5.37860513e-01
6.82740152e-01 6.39847755e-01 8.21030140e-03 -7.20757365e-01
-7.92469919e-01 1.77979097e-02 -6.81109190e-01 -6.45106137e-01
1.18306470e+00 -9.83820140e-01 -3.05880725e-01 3.84886086e-01
-1.67246652e+00 -4.63431060e-01 -4.94010359e-01 2.02666864e-01
-8.10999632e-01 3.22069705e-01 -6.22747481e-01 -8.63493264e-01
-1.30155668e-01 -1.07623076e+00 1.09528244e+00 3.22119296e-01
-4.79734123e-01 -9.09318566e-01 -2.61373296e-02 2.52371907e-01
4.53778684e-01 6.78760529e-01 7.47566402e-01 -2.49150787e-02
-9.42367733e-01 2.12944210e-01 -2.00653419e-01 -5.23271486e-02
2.45550349e-02 3.92880589e-01 -8.85319114e-01 -9.64948237e-02
-1.55871153e-01 -1.14711314e-01 8.81192684e-01 4.52825397e-01
1.01245928e+00 -4.48153973e-01 -1.35879904e-01 5.46675682e-01
1.13305235e+00 8.62827748e-02 8.28513563e-01 2.35684857e-01
8.78126025e-01 6.24150395e-01 4.77054745e-01 6.15689874e-01
5.60213685e-01 8.04040492e-01 3.85954887e-01 -1.90986723e-01
-6.72582150e-01 -5.85367620e-01 3.83166522e-01 7.20093548e-01
-2.74269402e-01 -6.46860421e-01 -5.13072729e-01 5.77746153e-01
-2.08669233e+00 -1.56123650e+00 -3.48235607e-01 1.90733004e+00
9.00516093e-01 -1.05204672e-01 5.80629557e-02 2.05462612e-02
8.70789945e-01 5.19650817e-01 -4.33939040e-01 -3.23958881e-02
-6.36336982e-01 -1.63488925e-01 -1.19613357e-01 4.88318831e-01
-1.00944328e+00 9.97116685e-01 5.97252846e+00 9.67507124e-01
-9.97683406e-01 2.25243449e-01 9.30290341e-01 -4.75761533e-01
-3.88381720e-01 3.63365468e-03 -6.76195502e-01 6.08532906e-01
5.59907615e-01 -1.58334345e-01 4.11562860e-01 5.73721170e-01
4.41375643e-01 -2.75369048e-01 -1.24211144e+00 1.16021657e+00
5.45496106e-01 -1.81652999e+00 3.54324281e-01 -5.24881840e-01
1.05639637e+00 -8.08027506e-01 1.17626846e-01 7.39983767e-02
3.06008548e-01 -1.01720202e+00 1.46537685e+00 9.52734113e-01
7.95220256e-01 -5.84988117e-01 3.12842786e-01 3.16360176e-01
-1.18776703e+00 2.69959092e-01 -1.88963845e-01 -2.34285071e-02
5.96515119e-01 2.92906642e-01 -4.47182715e-01 4.72239196e-01
7.04654336e-01 9.97670889e-01 -6.73284054e-01 8.08911562e-01
-4.08167660e-01 2.88699120e-01 8.62396657e-02 2.18013108e-01
-7.76931345e-02 -8.12791884e-02 6.41425848e-01 1.45451188e+00
5.51645160e-01 2.41400436e-01 9.58385225e-03 1.28091049e+00
1.53135866e-01 5.45376614e-02 -4.75107133e-01 6.30119517e-02
2.68632889e-01 1.02524543e+00 -7.73770869e-01 -6.28461003e-01
-2.88547814e-01 1.13570559e+00 1.32725134e-01 5.88670731e-01
-1.13480353e+00 2.89566934e-01 3.23453516e-01 5.40909648e-01
2.05905885e-01 -3.16198677e-01 -1.66924506e-01 -1.28848958e+00
7.35874660e-03 -8.29823792e-01 3.07739049e-01 -1.36513698e+00
-1.02185535e+00 7.37230182e-01 2.50952870e-01 -1.23891044e+00
-3.58315885e-01 1.26690743e-02 -6.90258622e-01 6.04819298e-01
-1.16389847e+00 -1.48861003e+00 -5.24315000e-01 6.80068016e-01
1.03726006e+00 -7.51806470e-03 4.70218748e-01 2.14703739e-01
-6.41782641e-01 2.09993646e-01 -2.22994834e-01 -6.81924075e-03
7.29756594e-01 -9.10043895e-01 1.40941590e-01 1.21676362e+00
4.70123649e-01 2.98047245e-01 1.07779467e+00 -8.34760904e-01
-1.23161626e+00 -1.06438553e+00 6.20736599e-01 -3.96598458e-01
5.41883469e-01 -5.14533222e-01 -9.25439715e-01 6.40109777e-01
7.96408296e-01 -2.28401601e-01 1.31614536e-01 -4.45963383e-01
-1.54534832e-01 5.03602065e-02 -7.80615449e-01 8.58220458e-01
1.19637871e+00 -2.70690322e-01 -4.59920913e-01 3.81576359e-01
7.53640711e-01 -7.75011837e-01 -3.75004530e-01 5.42239621e-02
2.36718148e-01 -1.33731163e+00 1.08290362e+00 -4.18846548e-01
1.06008005e+00 -5.59287548e-01 -2.02061996e-01 -1.17651844e+00
-2.54112870e-01 -6.87028408e-01 -3.16216648e-01 1.76725066e+00
-3.37784700e-02 1.23076379e-01 3.84807408e-01 7.73787975e-01
-6.81324080e-02 -3.89408708e-01 -8.42135489e-01 -3.49904239e-01
-3.55213791e-01 -3.50206494e-01 5.01142025e-01 9.30156350e-01
-3.69566441e-01 3.30281436e-01 -9.39533591e-01 2.12561805e-04
3.95206332e-01 2.72056401e-01 1.00590277e+00 -6.72365308e-01
-5.11745214e-01 -4.33124244e-01 -3.07350010e-01 -1.05524719e+00
2.00298950e-01 -4.94125307e-01 -4.33521755e-02 -1.84726739e+00
5.49597263e-01 -1.32145546e-03 2.39414349e-01 2.28252143e-01
-2.65468389e-01 1.34882063e-01 5.48439503e-01 1.76863089e-01
-9.19004679e-01 7.43911982e-01 1.41832519e+00 -2.80042946e-01
-1.34456962e-01 -4.75379586e-01 -3.15373987e-01 7.22081244e-01
2.89540380e-01 -1.70520604e-01 -5.52613020e-01 -6.80074215e-01
2.41857246e-01 3.24233472e-01 7.46713161e-01 -1.10504007e+00
3.68574768e-01 -3.79795015e-01 7.48866916e-01 -7.89180100e-01
6.32258534e-01 -4.92592096e-01 7.88435340e-01 8.17044526e-02
-4.83608127e-01 1.85684219e-01 1.83768749e-01 6.71799958e-01
-1.67491719e-01 5.56405634e-02 8.62299323e-01 -2.22854644e-01
-7.75933504e-01 1.36799380e-01 -1.97918534e-01 1.20724894e-01
1.22059190e+00 -3.09767067e-01 -4.87924218e-01 -6.92708254e-01
-6.78342581e-01 2.11022541e-01 6.11471176e-01 5.47406197e-01
8.51485968e-01 -1.42486155e+00 -1.05424213e+00 1.29686631e-02
-8.73434693e-02 8.69877860e-02 7.79792964e-01 6.14103198e-01
-5.71328521e-01 2.57674102e-02 -2.09800005e-01 -7.56306291e-01
-1.34897566e+00 7.47186244e-01 1.73701674e-01 -5.60928695e-02
-8.40213120e-01 8.18738878e-01 6.03897929e-01 4.93220866e-01
2.17398912e-01 -1.83333948e-01 -3.90853226e-01 1.81879029e-01
7.99426019e-01 2.85409838e-01 -4.79776263e-01 -9.99893665e-01
1.35821849e-03 5.10420442e-01 1.05659261e-01 -2.35662133e-01
1.38186669e+00 -4.06739295e-01 1.78721398e-01 6.46608591e-01
4.97477710e-01 -7.97120333e-02 -1.76112068e+00 -5.33608198e-02
-4.17656183e-01 -8.48914444e-01 -1.27314195e-01 -6.29836500e-01
-1.14045298e+00 9.56548750e-01 4.00661945e-01 6.75362349e-02
1.06141484e+00 1.00510940e-01 5.59017122e-01 -3.08915854e-01
2.91156828e-01 -8.32796812e-01 4.82094020e-01 3.01193953e-01
1.44595098e+00 -9.46117938e-01 5.52328639e-02 -3.70267898e-01
-1.01814294e+00 1.13573074e+00 7.38244057e-01 -1.48161352e-01
2.93365642e-02 3.33463371e-01 -1.12739712e-01 -4.54907343e-02
-9.46913004e-01 1.99054517e-02 2.83473909e-01 6.12228692e-01
4.28895772e-01 -2.04917222e-01 -1.22059081e-02 5.15730262e-01
-1.92520171e-01 -1.56073004e-01 5.76851070e-01 5.99411488e-01
-2.04134732e-01 -7.34401464e-01 -5.59273720e-01 1.62740238e-03
-1.79702267e-01 -2.21745789e-01 -2.72323698e-01 7.06344247e-01
2.21398056e-01 1.03459048e+00 2.02481419e-01 -6.90491572e-02
3.20807189e-01 -8.23108405e-02 6.59504592e-01 -4.15479749e-01
-5.50912857e-01 3.13784629e-01 1.07352957e-01 -4.91738319e-01
-7.14737415e-01 -6.38543904e-01 -9.91625249e-01 -3.70081127e-01
-1.23433419e-01 -2.52217770e-01 3.05595815e-01 7.67940223e-01
4.11889911e-01 9.34862077e-01 1.89338431e-01 -1.10008669e+00
2.66285744e-02 -9.28918839e-01 -4.05536532e-01 7.12009430e-01
1.55577630e-01 -7.06865191e-01 -1.30601496e-01 7.92877793e-01] | [11.143046379089355, 0.5451802611351013] |
45f772ea-b347-45e4-a62d-9f4ae146c268 | damage-accumulation-during-high-temperature | 2102.13575 | null | https://arxiv.org/abs/2102.13575v1 | https://arxiv.org/pdf/2102.13575v1.pdf | Damage accumulation during high temperature fatigue of Ti/SiC$_f$ metal matrix composites under different stress amplitudes | The damage mechanisms and load redistribution of high strength TC17 titanium alloy/unidirectional SiC fibre composite (fibre diameter = 100 $\mu$m) under high temperature (350 {\deg}C) fatigue cycling have been investigated in situ using synchrotron X-ray computed tomography (CT) and X-ray diffraction (XRD) for high cycle fatigue (HCF) under different stress amplitudes. The three-dimensional morphology of the crack and fibre fractures has been mapped by CT. During stable growth, matrix cracking dominates with the crack deflecting (by 50-100 $\mu$m in height) when bypassing bridging fibres. A small number of bridging fibres have fractured close to the matrix crack plane especially under relatively high stress amplitude cycling. Loading to the peak stress led to rapid crack growth accompanied by a burst of fibre fractures. Many of the fibre fractures occurred 50-300 $\mu$m from the matrix crack plane during rapid growth, in contrast to that in the stable growth stage, leading to extensive fibre pull-out on the fracture surface. The changes in fibre loading, interfacial stress, and the extent of fibre-matrix debonding in the vicinity of the crack have been mapped for the fatigue cycle and after the rapid growth by high spatial resolution XRD. The fibre/matrix interfacial sliding extends up to 600 $\mu$m (in the stable growth zone) or 700 $\mu$m (in the rapid growth zone) either side of the crack plane. The direction of interfacial shear stress reverses with the loading cycle, with the maximum frictional sliding stress reaching ~55 MPa in both the stable growth and rapid growth regimes. | ['Timothy L. Burnett', 'Xiaorong Zhou', 'Philip J. Withers', 'Yingwei Fan', 'Stefan Michalik', 'Katherine J. Dobson', 'Michael Atkinson', 'Yuan Chai', 'Samuel A. McDonald', 'Nan Li', 'Wenxia Zhao', 'Xu Xu', 'Ying Wang'] | 2021-02-26 | null | null | null | null | ['x-ray-diffraction'] | ['miscellaneous'] | [ 3.60136926e-01 2.14816496e-01 2.21827462e-01 6.44607723e-01
-2.59967029e-01 -8.88951868e-02 2.46668369e-01 2.38617696e-02
-4.38083798e-01 4.29801792e-01 1.33148625e-01 3.77867632e-02
-4.05816674e-01 -7.21032560e-01 -4.17090476e-01 -1.12141740e+00
-1.79560795e-01 7.54267573e-01 8.42514336e-01 -4.58548695e-01
4.54328805e-01 3.82043928e-01 -1.56219041e+00 1.86702207e-01
2.29835480e-01 6.62775517e-01 5.37037909e-01 3.79266679e-01
4.77740198e-01 1.87530547e-01 -1.75040901e-01 9.90340561e-02
-2.03312740e-01 -2.36860588e-01 -8.48809183e-01 1.60663813e-01
-5.87918341e-01 -2.47393832e-01 1.37161806e-01 6.12115383e-01
1.19963758e-01 1.21656312e-02 8.06286335e-01 -5.83754778e-01
-2.67906748e-02 4.72805649e-01 -1.09028625e+00 1.98314071e-01
5.64855278e-01 1.13893010e-01 7.70847142e-01 -1.10349417e+00
9.88541543e-01 8.29829991e-01 2.92737901e-01 6.24037623e-01
-9.76566315e-01 -3.40028346e-01 -6.16133213e-01 7.93420970e-02
-1.15106928e+00 -1.68180242e-01 7.64884055e-01 -9.59823430e-01
1.20910919e+00 7.12566972e-02 4.59784836e-01 6.43443584e-01
1.03924084e+00 -1.39698282e-01 1.16169810e+00 -5.73749542e-01
2.92617202e-01 -5.99299252e-01 2.74255909e-02 8.70922655e-02
4.15490568e-01 5.79246819e-01 3.09891496e-02 1.85043886e-01
9.77641165e-01 -2.67945737e-01 -5.69194138e-01 3.42113346e-01
-8.99411261e-01 4.70683604e-01 2.43542224e-01 5.62962234e-01
-3.26258600e-01 1.78083375e-01 5.98128736e-01 3.07508349e-01
4.26982164e-01 6.72377571e-02 -3.41894865e-01 -6.79498196e-01
-7.99373448e-01 5.43788552e-01 6.77553713e-01 3.77933651e-01
4.50400949e-01 -2.98624843e-01 7.52494574e-01 9.76172686e-01
4.19820964e-01 4.55059320e-01 5.14443778e-02 -7.82581866e-01
3.50691259e-01 5.05109608e-01 -1.77666262e-01 -4.63099897e-01
-1.09026264e-02 -8.30937698e-02 -7.47773767e-01 7.15855479e-01
6.14170372e-01 -1.69285163e-01 -7.01059759e-01 1.17173278e+00
2.33949661e-01 -4.59701777e-01 -2.05765545e-01 1.10226595e+00
4.03141230e-02 4.18798715e-01 9.87587124e-03 -7.16489494e-01
1.55693960e+00 -3.85686070e-01 -4.83414412e-01 -1.90519974e-01
5.32884121e-01 -8.54954302e-01 8.30483198e-01 -7.02191237e-03
-1.55063570e+00 -1.49429440e-01 -1.09378469e+00 5.10476708e-01
3.76275957e-01 -6.27429008e-01 -5.60416281e-01 3.48035455e-01
-4.91283685e-01 9.33933794e-01 -9.72264171e-01 -3.66464525e-01
7.29113668e-02 4.12627548e-01 -6.21091008e-01 -4.97550428e-01
-1.00259995e+00 6.52729928e-01 1.66522544e-02 2.64291316e-01
-5.61107337e-01 -4.73691225e-01 -4.70063359e-01 -1.69140637e-01
3.58383983e-01 -5.53894222e-01 1.09880960e+00 -8.51443410e-02
-1.53780019e+00 4.16384935e-01 -1.58145234e-01 2.09809706e-01
2.81206518e-01 -3.19391370e-01 -2.96568274e-01 6.70326352e-01
4.92633134e-01 -6.09497018e-02 4.49887037e-01 -1.39196396e+00
-1.62885919e-01 -5.38965225e-01 -3.03250432e-01 6.47061095e-02
3.05589348e-01 4.21199709e-01 1.72257438e-01 -4.92706388e-01
6.41227722e-01 -1.03873956e+00 -2.05462813e-01 -9.41035509e-01
-2.88998961e-01 -2.61360794e-01 1.24975038e+00 -1.72760040e-01
1.14945495e+00 -2.17506170e+00 1.90335974e-01 5.54949880e-01
2.48739108e-01 -4.24792200e-01 6.61129355e-01 1.31789744e+00
-8.74861851e-02 7.96193480e-02 -7.40269601e-01 2.75380313e-01
-5.64488888e-01 3.67242515e-01 3.97561848e-01 5.79035103e-01
-3.30240518e-01 6.28483236e-01 -7.81560600e-01 -1.46875024e-01
-3.89235139e-01 1.88650545e-02 -2.27906108e-01 -3.56240988e-01
4.60900627e-02 5.04408740e-02 -2.09738731e-01 9.47945535e-01
7.17627823e-01 1.11773215e-01 2.57789284e-01 -4.01954986e-02
-5.99047303e-01 -3.24717283e-01 -9.28847611e-01 1.06388223e+00
1.76931322e-02 -5.32341702e-03 6.65130496e-01 -5.63776910e-01
9.18447196e-01 8.86079431e-01 6.65705383e-01 -7.07161129e-01
-2.52034292e-02 8.76887023e-01 2.93600261e-01 -8.96493793e-01
6.97448626e-02 -1.20913720e+00 -2.15898827e-01 9.98030722e-01
-4.83598381e-01 -2.76033551e-01 3.70006800e-01 1.90281898e-01
1.01427805e+00 4.33002748e-02 -3.88532966e-01 -8.53493631e-01
5.46192169e-01 -3.30155678e-02 5.07671356e-01 9.77507085e-02
3.05169642e-01 6.00229800e-01 7.80562460e-01 -1.84728876e-01
-1.47283030e+00 -9.41976905e-01 -4.14076537e-01 7.34344363e-01
3.18816781e-01 -1.75633833e-01 -9.85356987e-01 2.45819002e-01
-1.48170695e-01 -9.84920748e-03 -5.12249172e-01 -2.20612839e-01
-9.73003030e-01 -6.56559229e-01 -1.34063466e-02 8.38551819e-01
3.78008395e-01 -1.53744614e+00 -6.85384572e-01 5.66202164e-01
2.32792627e-02 -8.55125427e-01 -2.16191635e-01 4.35610674e-02
-1.39868164e+00 -1.16920531e+00 -3.50770950e-01 -4.97276992e-01
6.00405991e-01 -1.93645984e-01 6.60426080e-01 6.25122607e-01
-2.84107447e-01 2.21559018e-01 -4.25333112e-01 1.23906717e-01
-6.86481833e-01 -5.05695678e-02 1.76124692e-01 -7.05628216e-01
-9.50209145e-03 -5.32914996e-01 -8.58595192e-01 9.65268254e-01
-9.73601103e-01 -3.76507044e-01 3.91702831e-01 8.59252930e-01
6.37862861e-01 4.07933831e-01 9.08775628e-01 -8.59244645e-01
4.43219870e-01 -7.33469486e-01 1.04804128e-01 -6.12268746e-01
-8.71939719e-01 -5.06630301e-01 3.34984869e-01 7.61301583e-03
-1.25760806e+00 -8.70492458e-01 -5.35886645e-01 -8.92172195e-03
-3.61054569e-01 1.01633847e+00 -1.21289514e-01 3.41585070e-01
6.69901729e-01 -2.82179683e-01 5.20481765e-01 -6.22530103e-01
-2.57657111e-01 3.69717956e-01 6.24838471e-01 -8.55640948e-01
7.77654171e-01 4.82245713e-01 -3.01341731e-02 -9.31300998e-01
-1.49803042e-01 -3.02526623e-01 -6.71087921e-01 -9.10943985e-01
1.01328945e+00 -2.71670282e-01 -8.94879282e-01 6.78060710e-01
-3.32212806e-01 -4.70147401e-01 -3.12989593e-01 7.33262718e-01
-4.07903135e-01 6.22303367e-01 -1.47941995e+00 -5.42876363e-01
-4.04676855e-01 -1.33314955e+00 5.78453243e-01 5.51390424e-02
-5.10230362e-01 -9.76036429e-01 1.29127279e-01 4.79003102e-01
3.56394947e-01 7.09623635e-01 1.28931975e+00 5.44877887e-01
-3.11085939e-01 -3.86726320e-01 2.76520163e-01 1.60501167e-01
-8.29560161e-02 2.13191718e-01 -2.15546608e-01 -5.58427930e-01
4.59760934e-01 6.67934865e-02 7.60536134e-01 5.26286542e-01
-1.12351283e-01 1.88891426e-01 -8.30436707e-01 -1.01377860e-01
1.98346090e+00 7.31186569e-01 1.12488544e+00 6.69830501e-01
4.28410262e-01 6.66913986e-01 7.33022273e-01 -1.15872204e-01
-4.43812966e-01 4.25043821e-01 5.52617252e-01 1.95082456e-01
3.73606123e-02 4.71762300e-01 5.40121913e-01 9.32493031e-01
-1.10843372e+00 2.04896659e-01 -9.86783862e-01 6.72773898e-01
-1.07092035e+00 -9.81529236e-01 -8.62876356e-01 2.21702361e+00
8.13901067e-01 7.73785889e-01 1.51208118e-01 6.69999778e-01
6.36290371e-01 -3.45145643e-01 -1.05873823e-01 -8.52519274e-01
1.72242537e-01 4.51426268e-01 3.42789143e-01 7.65443623e-01
1.01715997e-01 3.69455665e-01 5.97143316e+00 4.54733521e-01
-1.18107879e+00 -3.25490348e-02 -7.43548647e-02 6.73522800e-02
-4.48615491e-01 4.60298181e-01 -4.30881858e-01 8.89338255e-01
1.13028133e+00 2.27883846e-01 -4.59905677e-02 3.09445202e-01
2.31055573e-01 -9.71275151e-01 -4.85645533e-01 7.37915337e-02
-7.35375941e-01 -1.45955122e+00 -8.78968537e-01 5.85561395e-01
3.05258125e-01 1.83696568e-01 -4.85008150e-01 -2.62967020e-01
-3.54020476e-01 -6.44796073e-01 9.02119040e-01 3.94939870e-01
1.55980575e+00 -7.98820138e-01 9.53496277e-01 1.96964830e-01
-1.45698094e+00 -2.56309956e-01 1.16183730e-02 -5.64669669e-01
8.61373723e-01 1.03357410e+00 -5.32749295e-01 8.39908421e-01
1.05460608e+00 3.20883065e-01 5.43799810e-02 1.86439484e-01
5.15362769e-02 6.85831845e-01 -5.03013611e-01 5.74736536e-01
1.13549165e-01 -5.68944037e-01 8.75410199e-01 4.79401350e-01
1.80812962e-02 3.23465019e-01 -3.61714512e-01 4.80026752e-01
3.01635057e-01 -3.22301924e-01 -3.17870766e-01 4.60279226e-01
7.09601939e-01 7.87944794e-01 -1.30791914e+00 1.07777961e-01
-1.99303120e-01 2.46436432e-01 -3.94077480e-01 2.81863630e-01
-2.38594532e-01 -3.23935270e-01 3.63883287e-01 1.11124492e+00
6.90999180e-02 -6.09589279e-01 -6.16631091e-01 -2.17588112e-01
1.92610994e-01 -7.55118057e-02 1.17672287e-01 -2.64711916e-01
-1.15323651e+00 3.80405605e-01 4.31742072e-01 -6.18070602e-01
-2.32945621e-01 -3.24326783e-01 -1.17216897e+00 9.17210996e-01
-4.91173118e-01 -6.75156772e-01 -4.69523249e-03 2.85249472e-01
2.37319902e-01 1.35018215e-01 4.80708361e-01 4.85040471e-02
-6.55060112e-01 -5.83282411e-02 2.72390455e-01 -1.88223347e-01
2.44179457e-01 -1.03409231e+00 5.49482033e-02 3.94910544e-01
-9.76004183e-01 5.82569957e-01 9.83399510e-01 -1.28858805e+00
-1.29772604e+00 -4.05322313e-01 7.87375867e-01 2.43963599e-01
6.16180122e-01 -3.35276276e-01 -1.03759694e+00 2.78862357e-01
3.54612201e-01 -3.65241915e-01 8.41442823e-01 -1.27170309e-01
5.18938899e-01 1.88990712e-01 -1.04860330e+00 4.35827523e-01
5.50673425e-01 -3.70043814e-01 -4.52137440e-01 -5.35196960e-02
5.35483174e-02 -1.59374163e-01 -1.56190240e+00 7.70185947e-01
8.44865143e-01 -8.25338364e-01 5.98454356e-01 2.09763736e-01
8.90466809e-01 -2.99697042e-01 1.24971233e-01 -5.14590025e-01
-6.91661596e-01 -2.98532218e-01 3.90406430e-01 1.06419504e+00
3.99119139e-01 -5.44945598e-01 7.91506886e-01 1.75857604e-01
-9.30105031e-01 -1.42544162e+00 -1.11357749e+00 -3.26465070e-01
6.01981878e-01 -1.28474027e-01 -1.23271100e-01 4.19750482e-01
4.15711999e-01 1.18332371e-01 4.35554415e-01 -2.02569783e-01
5.35319626e-01 -1.57063738e-01 -2.26017505e-01 -1.21808910e+00
-1.00086533e-01 4.67903018e-02 -9.10987630e-02 -2.63644248e-01
-4.60061163e-01 -5.68664670e-01 -1.02623053e-01 -1.59246087e+00
-9.30686966e-02 -6.32693768e-01 3.63394499e-01 3.20720263e-02
2.76400447e-01 3.45318645e-01 -2.59570688e-01 1.01454747e+00
5.66786170e-01 5.44205680e-02 1.46270621e+00 2.51435339e-01
-1.87623560e-01 -2.20658347e-01 -1.70750931e-01 5.80954492e-01
6.48748755e-01 -3.36607128e-01 -2.15056062e-01 -5.98535240e-02
7.02208102e-01 5.77460110e-01 1.53394774e-01 -9.67120111e-01
1.34126738e-01 -4.48594391e-02 4.38554138e-02 -7.31409490e-01
-7.76625052e-02 -8.59585524e-01 9.42809343e-01 8.25476706e-01
3.69132489e-01 1.14323243e-01 -2.90709108e-01 5.06049097e-01
-2.53681600e-01 -5.12958288e-01 1.07286811e+00 -3.75660628e-01
-1.05537754e-03 -1.29257992e-01 -1.24824142e+00 -3.02022666e-01
1.38899016e+00 -1.07158542e+00 -2.10517406e-01 4.17223871e-01
-1.50024009e+00 -1.37839079e-01 8.88586760e-01 9.66104958e-03
5.53281486e-01 -1.19762969e+00 -4.96110290e-01 2.81574577e-01
-2.96992987e-01 3.15893918e-01 8.34530234e-01 1.34138775e+00
-9.83384252e-01 -5.97213805e-02 -2.66253322e-01 -9.16586459e-01
-7.99637556e-01 2.66926169e-01 5.12870610e-01 -2.39525288e-01
-8.94632638e-01 9.79447484e-01 -1.57607108e-01 7.09842443e-01
-7.52173364e-01 4.20389250e-02 -2.66628325e-01 2.62837503e-02
4.14118469e-02 9.60892081e-01 3.71373057e-01 -6.15285218e-01
-1.85543403e-01 1.00216496e+00 -2.62462527e-01 -1.89125225e-01
1.31283259e+00 -4.45898563e-01 -6.43535852e-01 5.12054026e-01
1.03193390e+00 7.30990544e-02 -1.38119173e+00 2.74454713e-01
1.34250730e-01 -3.21303904e-01 -5.27266264e-01 -3.88019562e-01
-1.12597811e+00 3.26118201e-01 9.38348398e-02 2.86869049e-01
8.33316922e-01 7.83571064e-01 1.08241475e+00 -6.94343030e-01
5.17489314e-01 -1.53748631e+00 2.85446316e-01 2.99494088e-01
8.31347167e-01 -1.62781402e-01 -1.32213747e-02 -8.23968887e-01
-1.29861191e-01 1.42045486e+00 5.08429885e-01 -4.08218831e-01
9.29472685e-01 7.44297981e-01 -9.57325771e-02 -1.05932677e+00
-7.04645514e-01 4.91991550e-01 -3.34204704e-01 2.52197921e-01
6.82433009e-01 -3.56947817e-02 -8.69093955e-01 2.50400007e-01
-3.20178270e-01 -1.93542257e-01 8.72585118e-01 1.52940845e+00
-7.60989070e-01 -1.21633422e+00 -6.60939991e-01 5.84033668e-01
-6.46839321e-01 3.01728100e-01 3.62493366e-01 8.73857141e-01
1.14676498e-01 8.58433604e-01 3.08628678e-01 -2.73602515e-01
4.16215092e-01 2.42585734e-01 3.09589416e-01 -5.75794339e-01
-3.77947181e-01 7.79603124e-01 1.27995089e-01 -2.46498033e-01
-5.71526110e-01 -1.04318798e+00 -1.81180644e+00 -3.05309296e-01
-6.21097565e-01 5.28202176e-01 5.17216265e-01 1.30505311e+00
-4.29153949e-01 3.14699382e-01 6.85047925e-01 -7.98842669e-01
3.18657767e-05 -1.27707672e+00 -1.37111747e+00 3.36548746e-01
-1.99545085e-01 -1.09495640e+00 -5.95667362e-01 2.87060171e-01] | [6.14755916595459, 3.2689406871795654] |
72a9eedf-636e-4c17-9911-e31faa050320 | supervised-hyperalignment-for-multi-subject | 2001.02894 | null | https://arxiv.org/abs/2001.02894v1 | https://arxiv.org/pdf/2001.02894v1.pdf | Supervised Hyperalignment for multi-subject fMRI data alignment | Hyperalignment has been widely employed in Multivariate Pattern (MVP) analysis to discover the cognitive states in the human brains based on multi-subject functional Magnetic Resonance Imaging (fMRI) datasets. Most of the existing HA methods utilized unsupervised approaches, where they only maximized the correlation between the voxels with the same position in the time series. However, these unsupervised solutions may not be optimum for handling the functional alignment in the supervised MVP problems. This paper proposes a Supervised Hyperalignment (SHA) method to ensure better functional alignment for MVP analysis, where the proposed method provides a supervised shared space that can maximize the correlation among the stimuli belonging to the same category and minimize the correlation between distinct categories of stimuli. Further, SHA employs a generalized optimization solution, which generates the shared space and calculates the mapped features in a single iteration, hence with optimum time and space complexities for large datasets. Experiments on multi-subject datasets demonstrate that SHA method achieves up to 19% better performance for multi-class problems over the state-of-the-art HA algorithms. | ['Liangxiu Han', 'Alessandro Selvitella', 'Daoqiang Zhang', 'Muhammad Yousefnezhad'] | 2020-01-09 | null | null | null | null | ['multi-subject-fmri-data-alignment'] | ['medical'] | [ 4.69991356e-01 -3.05749923e-01 1.48701116e-01 -5.23359537e-01
-3.97950768e-01 -5.48150949e-02 3.16532254e-01 1.93178020e-02
-4.81741250e-01 6.54441476e-01 9.79419276e-02 2.88341165e-01
-7.83434868e-01 -4.29648280e-01 -3.13715905e-01 -1.10877264e+00
-4.73647565e-01 2.70722926e-01 2.25215912e-01 1.54489443e-01
5.49645841e-01 2.94024974e-01 -1.54403460e+00 3.24114531e-01
9.54199612e-01 8.12485456e-01 5.75939119e-01 1.08333984e-02
1.57300442e-01 3.26345414e-01 -2.82716513e-01 4.71522957e-02
3.35371941e-01 -5.91329575e-01 -5.93571365e-01 1.22429403e-02
2.78103262e-01 3.35001171e-01 -2.29386333e-02 1.22724545e+00
6.32342339e-01 4.19349521e-01 8.02389622e-01 -1.31625903e+00
-2.32930988e-01 4.95428979e-01 -8.92650306e-01 7.16295362e-01
3.72469351e-02 -6.48183748e-02 8.61822188e-01 -9.69093323e-01
4.18463469e-01 1.03292000e+00 3.35730076e-01 9.35246870e-02
-1.42401814e+00 -8.41592312e-01 -1.69984907e-01 7.02262163e-01
-1.54802978e+00 -2.91031748e-01 9.24612999e-01 -6.64563477e-01
8.70901883e-01 2.52807915e-01 7.18967497e-01 7.45839417e-01
6.34866238e-01 3.90392065e-01 1.66577637e+00 -2.39761442e-01
2.90991962e-01 -2.30701104e-01 2.74921387e-01 4.60497588e-01
1.49202913e-01 -1.33559406e-01 -7.82479942e-01 -2.18941629e-01
7.05947459e-01 -7.57325888e-02 -1.93991423e-01 -4.98338073e-01
-1.75462282e+00 7.45306492e-01 3.34450603e-01 8.01789045e-01
-7.45972037e-01 -5.25926411e-01 2.15152800e-01 3.28246020e-02
7.22185001e-02 6.03959143e-01 -2.70620406e-01 2.34745771e-01
-1.18843627e+00 3.20782363e-01 2.59303659e-01 4.77077544e-01
7.45227039e-01 5.15126530e-03 -2.06700742e-01 8.93152356e-01
2.60873675e-01 3.73898476e-01 9.57942903e-01 -7.30364084e-01
4.59122211e-01 5.98940134e-01 -3.53021234e-01 -1.45162463e+00
-8.68676960e-01 -6.62180364e-01 -9.88643289e-01 7.42135867e-02
2.46680096e-01 -6.91325963e-02 -6.56644642e-01 1.91160727e+00
2.53983915e-01 1.65122822e-01 -1.21241055e-01 9.17380989e-01
3.00422877e-01 4.60268557e-01 8.54082629e-02 -8.20420146e-01
1.36516595e+00 -6.92616343e-01 -7.65158236e-01 -3.61123115e-01
3.55430305e-01 -5.56308985e-01 8.15673709e-01 4.10309583e-01
-9.40571249e-01 -5.77958286e-01 -1.13352227e+00 5.20592511e-01
-1.10816903e-01 -6.34012371e-02 6.34316504e-01 2.61428773e-01
-7.65913844e-01 6.49020314e-01 -8.58633578e-01 -2.83952117e-01
4.10976887e-01 5.35679638e-01 -6.78448081e-01 2.57189810e-01
-1.01468754e+00 9.27856743e-01 4.95208591e-01 3.61652792e-01
-9.10471082e-01 -6.63040400e-01 -5.45720160e-01 5.44674918e-02
2.29238316e-01 -1.91549286e-01 5.22217453e-01 -9.70000863e-01
-1.08118403e+00 6.47101939e-01 -2.78443992e-01 -1.19583257e-01
9.66643542e-02 1.54893965e-01 -4.94895279e-01 1.18200466e-01
2.83974677e-01 5.88597953e-01 8.95761609e-01 -8.78331602e-01
-1.69270679e-01 -7.30588973e-01 -6.05295837e-01 4.37730908e-01
-4.27109271e-01 4.77883816e-02 1.99259385e-01 -6.00883126e-01
7.64420867e-01 -8.56075048e-01 -2.47618780e-01 -4.02980387e-01
-2.61584669e-01 -1.77255541e-01 5.20947456e-01 -8.09764147e-01
1.16852558e+00 -2.18059349e+00 6.07302606e-01 6.32501006e-01
2.26127088e-01 -1.06650844e-01 4.53112908e-02 1.00846395e-01
-4.35671061e-01 -3.80817294e-01 -5.80418110e-01 1.13845356e-01
-2.82727182e-01 1.09484904e-01 2.84555584e-01 8.05669487e-01
-1.42387465e-01 5.24951458e-01 -6.31602883e-01 -7.43477643e-01
-1.25943124e-01 1.22517407e-01 -4.45223033e-01 2.79181600e-01
3.77344817e-01 7.27755547e-01 -2.48479813e-01 3.77052099e-01
6.83435559e-01 -1.88831583e-01 5.59117317e-01 -5.62764227e-01
-2.01345518e-01 -2.76595861e-01 -1.26836467e+00 1.74119759e+00
1.31053641e-01 3.22868764e-01 -8.08287412e-02 -1.39512348e+00
1.04628944e+00 4.49970365e-01 1.04057658e+00 -9.35192704e-01
2.04223707e-01 2.85148829e-01 6.51953042e-01 -6.01216197e-01
-7.58194402e-02 -1.89792305e-01 2.22132280e-01 4.67069864e-01
2.33715847e-01 4.25998658e-01 2.31396928e-01 -2.12802887e-01
9.71968293e-01 -2.24199921e-01 5.40010154e-01 -9.01412368e-01
7.88830042e-01 -4.16777641e-01 7.82395244e-01 5.30446470e-01
-4.42235202e-01 2.53445357e-01 4.07729894e-01 -2.27012202e-01
-8.92236412e-01 -1.16561639e+00 -2.98197031e-01 6.15694463e-01
3.37499045e-02 -9.94923040e-02 -6.24739468e-01 -3.59622031e-01
-2.74324328e-01 2.70874143e-01 -6.07519448e-01 -2.13189926e-02
-5.94425619e-01 -1.27886593e+00 4.49016504e-02 3.45105439e-01
6.59691930e-01 -1.23253882e+00 -8.95404816e-01 1.35980338e-01
-2.09246427e-01 -7.06169903e-01 -1.98590875e-01 2.51542538e-01
-1.06086242e+00 -9.85720932e-01 -6.13207102e-01 -1.00305986e+00
1.04418719e+00 1.51897281e-01 2.94188231e-01 -2.50561923e-01
-3.39813650e-01 -8.08139369e-02 -1.99143335e-01 6.55188262e-02
2.98674256e-01 -1.38618961e-01 3.25693309e-01 4.19136703e-01
2.51919836e-01 -1.07771134e+00 -6.91734672e-01 3.77536625e-01
-7.55427301e-01 2.33982876e-01 8.13243389e-01 1.03067732e+00
7.84681797e-01 2.60291845e-01 7.95678973e-01 -4.93824720e-01
5.22533298e-01 -7.18188584e-01 -2.43292570e-01 2.90328056e-01
-8.02840233e-01 1.29925579e-01 4.90725368e-01 -6.51954055e-01
-9.41016674e-01 6.81624487e-02 3.04005414e-01 -2.11569637e-01
4.79928590e-02 6.78512990e-01 -2.89389193e-01 -2.26874664e-01
6.05170071e-01 5.45367002e-01 1.83119372e-01 -4.92609262e-01
-1.18808359e-01 3.37553084e-01 5.77540100e-01 -4.50393140e-01
4.78558600e-01 3.01274389e-01 3.72196257e-01 -6.06925666e-01
-4.05231178e-01 -4.13952351e-01 -1.08998501e+00 -3.32511038e-01
1.00396419e+00 -5.05396068e-01 -6.11969650e-01 3.95713449e-01
-7.51666069e-01 7.79615790e-02 3.08935970e-01 9.39751983e-01
-4.74117339e-01 5.64449251e-01 -1.93784744e-01 -6.28772140e-01
-4.66741741e-01 -1.24848020e+00 5.05602777e-01 2.09798321e-01
-2.76563466e-01 -5.80309570e-01 1.87599361e-01 2.29439214e-01
2.89572001e-01 3.20124149e-01 1.30574524e+00 -7.42650867e-01
-9.87123549e-02 1.42886668e-01 1.06151477e-01 1.01862587e-01
1.46322176e-01 -5.19749343e-01 -7.06437528e-01 -3.34622502e-01
4.05965626e-01 1.05402246e-01 3.61906618e-01 4.17963922e-01
1.24013734e+00 -3.21980447e-01 -1.96974635e-01 3.89188230e-01
1.35380983e+00 7.26450562e-01 6.47868276e-01 3.01975638e-01
6.16998911e-01 8.71502161e-01 5.43464482e-01 3.95275593e-01
7.48106465e-03 6.90940559e-01 5.67627847e-02 6.58108145e-02
2.72424400e-01 2.28321940e-01 2.50196695e-01 1.24260652e+00
-4.54169139e-02 2.54440099e-01 -9.54358876e-01 6.28460884e-01
-1.96787405e+00 -1.14990592e+00 -1.25817508e-01 2.32424355e+00
8.02999079e-01 -3.71001139e-02 2.71928869e-02 3.53057563e-01
8.58650327e-01 1.32952049e-01 -4.82542068e-01 -7.93460608e-02
-1.91573694e-01 3.31258953e-01 2.05125019e-01 1.74207255e-01
-9.02583420e-01 4.00369465e-01 6.59661245e+00 6.89843833e-01
-9.09670830e-01 4.36104238e-01 3.45520586e-01 -1.94714218e-01
-4.75795902e-02 1.23921095e-03 -3.07389408e-01 5.71913600e-01
8.40092778e-01 -4.20557737e-01 6.34247541e-01 6.62497222e-01
4.02579188e-01 -2.97316462e-01 -8.85876775e-01 1.21091211e+00
2.60884494e-01 -7.71928310e-01 -1.33599475e-01 5.40459119e-02
7.61736333e-01 -1.87693149e-01 -6.31403401e-02 -1.27526879e-01
-7.20988870e-01 -9.34331656e-01 4.89040226e-01 8.16599488e-01
2.56261170e-01 -7.30767906e-01 6.04551494e-01 4.40604627e-01
-1.07018304e+00 -1.14163600e-01 -4.15366918e-01 -4.86573018e-02
1.38203338e-01 6.55377090e-01 -6.29238427e-01 5.08279681e-01
5.91765523e-01 6.47794843e-01 -7.65227675e-01 1.03319466e+00
1.14412755e-01 3.30838323e-01 -8.55234638e-02 4.00804542e-02
-1.14553541e-01 -3.08847100e-01 6.72701836e-01 7.31879473e-01
3.89435977e-01 8.33447576e-02 3.18318635e-01 7.39678383e-01
4.98083621e-01 6.29850566e-01 -3.23679656e-01 8.26894026e-03
3.52563620e-01 1.43440950e+00 -9.12033796e-01 -1.36097804e-01
-2.46817872e-01 6.82343125e-01 3.41276199e-01 1.03950322e-01
-7.09850073e-01 -2.99481899e-01 2.78592050e-01 4.01914082e-02
3.63093987e-02 -4.89535600e-01 -4.33166713e-01 -9.83125925e-01
7.22613260e-02 -9.07411993e-01 5.97273409e-01 -6.77252710e-01
-1.36576152e+00 6.71164036e-01 4.02810425e-01 -1.13347173e+00
-1.18612358e-02 -4.70053047e-01 -4.92540479e-01 9.17153239e-01
-1.03529775e+00 -7.46228755e-01 -2.57416874e-01 9.43939805e-01
3.92720848e-01 -2.83879995e-01 8.30205500e-01 4.42626834e-01
-6.75069392e-01 2.49822333e-01 -7.87920654e-02 -3.23456198e-01
6.62809193e-01 -8.89013052e-01 -4.83568132e-01 1.14606905e+00
-2.00979803e-02 8.02657068e-01 5.31289756e-01 -8.32639277e-01
-1.22645724e+00 -5.24397969e-01 8.01098526e-01 2.00219586e-01
5.78202546e-01 -1.41613171e-01 -1.01288557e+00 3.77194136e-01
1.86506256e-01 -1.86726734e-01 9.66082275e-01 -7.33550787e-02
5.63417748e-02 -2.78905839e-01 -1.18251991e+00 4.66401935e-01
1.01271427e+00 -3.65765870e-01 -7.84335017e-01 4.85962421e-01
9.07149017e-02 3.31645012e-02 -1.25854862e+00 5.22714674e-01
5.75054467e-01 -8.24798703e-01 7.15542614e-01 -4.35419410e-01
2.39988223e-01 -5.96876383e-01 -1.34125829e-01 -1.32587922e+00
-8.88008773e-01 -1.70828328e-01 -1.88935473e-06 9.74076152e-01
2.94438899e-01 -6.26027465e-01 3.06754500e-01 5.35751998e-01
-7.73330033e-02 -7.59930909e-01 -1.16742206e+00 -6.56224787e-01
-3.10828149e-01 -8.69160220e-02 3.89429569e-01 1.05913174e+00
3.24559987e-01 1.96755692e-01 -4.14117277e-01 1.23395391e-01
9.12463903e-01 1.82131425e-01 9.34192985e-02 -1.27655089e+00
-2.96804070e-01 -3.69738489e-01 -6.59446001e-01 -1.31423667e-01
1.74324647e-01 -1.19534075e+00 1.49364531e-01 -1.25048256e+00
6.64475441e-01 -2.52624840e-01 -7.51508534e-01 6.16129458e-01
-1.82811707e-01 3.91823538e-02 -3.33488509e-02 4.75408047e-01
-2.99634486e-01 4.90812480e-01 1.21524823e+00 5.11272363e-02
-2.51302183e-01 -3.72132093e-01 -4.85517889e-01 3.81981820e-01
7.96589911e-01 -5.38970709e-01 -6.68764174e-01 -2.18094736e-01
-3.94130275e-02 -6.64031133e-02 1.82452038e-01 -1.37012291e+00
2.39946708e-01 -3.14949334e-01 6.69259429e-01 -4.55270529e-01
1.11279480e-01 -8.14013600e-01 5.78872263e-01 5.08409142e-01
-2.43811414e-01 4.31715339e-01 -1.17399991e-01 3.73544574e-01
-9.86698642e-02 -2.19369486e-01 9.29541826e-01 -1.45702913e-01
-6.88389301e-01 2.75191754e-01 -2.15019152e-01 -3.00803185e-01
1.20901608e+00 -8.73918235e-02 -2.18917638e-01 2.00504839e-01
-7.93439984e-01 6.09336980e-02 -8.80604237e-02 2.17872381e-01
6.78579390e-01 -1.47294700e+00 -5.26809454e-01 3.80574495e-01
-2.49850094e-01 -4.69008178e-01 6.50606811e-01 1.53434479e+00
-1.34521425e-01 3.68200094e-01 -1.03596485e+00 -6.88427269e-01
-1.23939836e+00 4.80266511e-01 2.35743031e-01 -2.94313818e-01
-6.36435747e-01 4.77765858e-01 2.33760044e-01 1.93270966e-02
-2.16512784e-01 1.48407683e-01 -6.76061690e-01 3.50787282e-01
5.39834261e-01 4.51997072e-01 1.54217020e-01 -8.70610297e-01
-6.63069367e-01 4.44655508e-01 3.23131122e-02 -3.53012294e-01
1.54952407e+00 -1.39485151e-02 -4.75449026e-01 5.99598646e-01
1.28400147e+00 -2.62727976e-01 -1.00063729e+00 -2.24134654e-01
8.85786191e-02 -5.69662273e-01 1.02408014e-01 -5.95877767e-01
-1.15717375e+00 5.90965927e-01 9.04341519e-01 -2.00848132e-01
1.27517068e+00 -3.08659256e-01 4.41049039e-01 2.28895769e-01
6.25064015e-01 -1.17472529e+00 1.21130243e-01 5.16928919e-02
9.06242490e-01 -8.44848514e-01 2.66130120e-01 -2.33388901e-01
-8.51543069e-01 1.11503661e+00 8.05684626e-01 -1.86607391e-01
7.61604667e-01 -9.41980630e-02 -4.39624906e-01 -4.96341050e-01
-3.66618961e-01 1.57447740e-01 4.95639145e-01 3.39672953e-01
2.70092636e-01 1.88250780e-01 -1.14388573e+00 8.21790814e-01
-1.72376007e-01 -3.45952123e-01 1.08391516e-01 9.95604813e-01
-4.50657815e-01 -9.58198965e-01 -3.52273047e-01 7.88084805e-01
-2.17482403e-01 3.84395719e-02 1.59281999e-01 3.94270480e-01
2.47850016e-01 7.30423808e-01 -1.55712729e-02 -4.41586375e-01
1.81227699e-01 4.16126043e-01 6.38074040e-01 -3.94826770e-01
-5.02307594e-01 2.31216058e-01 -3.02205145e-01 -6.77430511e-01
-7.19561875e-01 -1.01346397e+00 -1.37707686e+00 9.85852107e-02
-9.10062417e-02 1.30124375e-01 5.67494154e-01 1.18538404e+00
2.59574085e-01 4.81112719e-01 7.92364776e-01 -7.61818349e-01
-3.37313235e-01 -1.20102596e+00 -9.10686433e-01 6.06173098e-01
-2.50395536e-01 -1.12279809e+00 -3.30282152e-01 -2.49440153e-03] | [12.671313285827637, 3.3923285007476807] |
63c96483-ff80-475b-8032-f8d1d4a0d0d9 | ofa-2-a-multi-objective-perspective-for-the | 2303.13683 | null | https://arxiv.org/abs/2303.13683v1 | https://arxiv.org/pdf/2303.13683v1.pdf | OFA$^2$: A Multi-Objective Perspective for the Once-for-All Neural Architecture Search | Once-for-All (OFA) is a Neural Architecture Search (NAS) framework designed to address the problem of searching efficient architectures for devices with different resources constraints by decoupling the training and the searching stages. The computationally expensive process of training the OFA neural network is done only once, and then it is possible to perform multiple searches for subnetworks extracted from this trained network according to each deployment scenario. In this work we aim to give one step further in the search for efficiency by explicitly conceiving the search stage as a multi-objective optimization problem. A Pareto frontier is then populated with efficient, and already trained, neural architectures exhibiting distinct trade-offs among the conflicting objectives. This could be achieved by using any multi-objective evolutionary algorithm during the search stage, such as NSGA-II and SMS-EMOA. In other words, the neural network is trained once, the searching for subnetworks considering different hardware constraints is also done one single time, and then the user can choose a suitable neural network according to each deployment scenario. The conjugation of OFA and an explicit algorithm for multi-objective optimization opens the possibility of a posteriori decision-making in NAS, after sampling efficient subnetworks which are a very good approximation of the Pareto frontier, given that those subnetworks are already trained and ready to use. The source code and the final search algorithm will be released at https://github.com/ito-rafael/once-for-all-2 | ['Fernando J. Von Zuben', 'Rafael C. Ito'] | 2023-03-23 | null | null | null | null | ['architecture-search'] | ['methodology'] | [ 1.43529847e-01 1.01894245e-01 -1.43638209e-01 -7.86133483e-02
-2.53660679e-01 -5.20142257e-01 -7.52666406e-03 2.70239003e-02
-4.67429459e-01 8.13283503e-01 -4.84481812e-01 -4.54223216e-01
-8.24046016e-01 -1.01958656e+00 -5.75872421e-01 -8.30142856e-01
-1.06491975e-01 8.56633902e-01 -8.29193965e-02 -1.87170714e-01
9.84075069e-02 8.93231332e-01 -2.06460905e+00 -1.00629240e-01
7.74851203e-01 1.07776046e+00 4.91780639e-01 6.07316971e-01
-1.92361191e-01 -9.94319911e-04 -7.56255209e-01 -2.14174122e-01
3.05791974e-01 -4.11644846e-01 -7.35960424e-01 -1.98751807e-01
-4.36507165e-01 1.39647186e-01 3.45000327e-01 9.47552383e-01
7.39370346e-01 8.15788805e-02 4.52189595e-01 -1.16729271e+00
2.79315889e-01 6.41305983e-01 -9.50137302e-02 -1.76667094e-01
-1.50518492e-01 8.15406814e-02 7.98959255e-01 -5.63826084e-01
4.59627777e-01 8.46392691e-01 3.97968888e-01 3.68395746e-01
-1.04124105e+00 -5.80773592e-01 -7.58701488e-02 4.82563907e-03
-1.62401509e+00 -3.56193274e-01 8.38402987e-01 -2.10566461e-01
9.50442076e-01 7.04329133e-01 1.04928434e+00 6.94957972e-01
-1.55115694e-01 5.26971817e-01 5.82797050e-01 -6.61341071e-01
7.77519047e-01 1.28476664e-01 -5.62686101e-02 5.49024463e-01
6.46951914e-01 2.28334174e-01 -1.41270325e-01 -8.20686594e-02
4.69508171e-01 -3.37642938e-01 -1.59344062e-01 -4.13747668e-01
-7.63357878e-01 6.67553723e-01 2.02377662e-01 6.40818596e-01
-7.47844160e-01 -4.25584428e-02 2.69538701e-01 2.32625738e-01
-7.40140378e-02 1.10669005e+00 -6.44000828e-01 -1.00966558e-01
-1.23593605e+00 2.57910118e-02 8.74876201e-01 5.29326200e-01
7.83683479e-01 1.49118781e-01 2.73473226e-02 6.57643378e-01
1.13957584e-01 1.07126303e-01 3.85909170e-01 -8.01938355e-01
-2.80970819e-02 9.25421774e-01 2.11912990e-02 -6.64411128e-01
-6.49481416e-01 -9.46167171e-01 -7.43961453e-01 6.29180074e-01
4.00808930e-01 -5.72166562e-01 -5.49832582e-01 1.52026367e+00
4.78127122e-01 -1.88070416e-01 1.03773743e-01 8.06143999e-01
3.16604018e-01 7.07929075e-01 -3.79984975e-01 -4.99974102e-01
1.41887450e+00 -8.93353701e-01 -4.36799973e-01 -7.19001666e-02
5.09693503e-01 -4.93359685e-01 6.11926138e-01 6.46000564e-01
-1.24589646e+00 -3.79580140e-01 -1.35101104e+00 5.73453009e-01
-6.40265107e-01 6.72392905e-01 5.41759431e-01 1.06673503e+00
-1.12042522e+00 6.75709903e-01 -6.63735390e-01 -2.75075585e-01
2.75810331e-01 1.07411468e+00 2.65893996e-01 2.17253730e-01
-9.59753096e-01 7.54173815e-01 8.64008367e-01 3.37043613e-01
-6.85002923e-01 -3.15226644e-01 -3.19628596e-01 5.49752474e-01
6.89415991e-01 -7.54190445e-01 7.36303985e-01 -1.36463881e+00
-1.89601481e+00 3.61534834e-01 -6.12513162e-02 -3.62470955e-01
2.93982834e-01 5.81883848e-01 -3.00829440e-01 -1.67208195e-01
-4.52350169e-01 4.84893292e-01 5.81891596e-01 -1.16344273e+00
-4.16002274e-01 -1.17919236e-01 3.73734951e-01 1.07617274e-01
-5.82221687e-01 1.48559555e-01 -5.14742613e-01 -2.95768917e-01
-4.00677562e-01 -8.89990568e-01 -3.70818794e-01 -2.89951980e-01
-4.46626067e-01 -5.38545176e-02 3.23842615e-01 -3.29064190e-01
1.52558923e+00 -1.80914760e+00 5.20816922e-01 7.37957656e-01
-1.62097633e-01 4.11107153e-01 -2.31941044e-01 2.37765044e-01
-1.71073690e-01 1.17914125e-01 -1.14563992e-02 -2.52677858e-01
2.48897020e-02 6.39775693e-02 5.20512402e-01 1.32034257e-01
2.26455182e-01 5.47015071e-01 -6.16635323e-01 -3.16549033e-01
1.24619380e-01 4.80351567e-01 -5.97561061e-01 3.68361697e-02
-3.30018044e-01 3.13012630e-01 -6.93721354e-01 6.72932982e-01
6.00812674e-01 -3.74677032e-01 6.53795898e-01 -2.70760238e-01
-4.99368489e-01 -5.03694899e-02 -1.45753741e+00 1.53719950e+00
-7.96469986e-01 5.28680801e-01 3.05714548e-01 -1.22640669e+00
9.28417981e-01 3.05318058e-01 5.78471243e-01 -7.14549959e-01
8.74184489e-01 6.92755163e-01 2.90789992e-01 -4.84992340e-02
3.70208442e-01 3.04561406e-01 3.87822650e-02 4.08905357e-01
1.98846698e-01 2.67971128e-01 5.62733948e-01 -6.34623647e-01
9.26456511e-01 4.06746864e-02 1.36038825e-01 -6.63884461e-01
6.97994411e-01 1.09199241e-01 4.95172262e-01 4.46696460e-01
4.06051338e-01 2.80810207e-01 4.82536405e-01 -4.66890007e-01
-1.06610906e+00 -4.99235928e-01 -1.10543668e-01 7.03302383e-01
-2.99661160e-02 -1.71426386e-01 -8.77365947e-01 -4.89602357e-01
-5.43548644e-01 8.15692484e-01 -4.76523906e-01 -1.19270504e-01
-5.70835054e-01 -7.03018069e-01 2.62455463e-01 -1.24754459e-01
2.80947208e-01 -1.04891860e+00 -1.44112647e+00 2.33141705e-01
8.07455629e-02 -5.64172387e-01 1.93226188e-01 7.98963606e-01
-8.47128630e-01 -9.15897608e-01 -7.47687876e-01 -5.83065569e-01
9.12752807e-01 -3.70333970e-01 1.13336682e+00 4.65097249e-01
-4.07099605e-01 -1.30787864e-01 -4.65304628e-02 -2.68668890e-01
-4.60204810e-01 3.88885975e-01 -1.44902855e-01 -9.50903445e-03
-2.14749560e-01 -7.59670615e-01 -4.01139468e-01 4.52635080e-01
-8.01250935e-01 1.05595015e-01 8.32403898e-01 6.80232048e-01
7.56994605e-01 7.35081255e-01 4.03248787e-01 -1.77060723e-01
6.86019361e-01 -4.29426670e-01 -1.31791270e+00 7.15186000e-01
-5.70709288e-01 5.25116742e-01 9.19229925e-01 -4.67056930e-01
-6.06634498e-01 3.53819013e-01 -2.55209804e-01 -6.92771375e-01
-2.26602852e-01 5.74923575e-01 -5.85304141e-01 -2.15945229e-01
3.87857109e-01 -2.07120664e-02 -1.40571162e-01 -4.16030616e-01
-7.57162347e-02 3.31412256e-01 -5.91623113e-02 -5.69165945e-01
5.65680802e-01 -1.02001727e-02 4.29951310e-01 -6.78645015e-01
-1.17850408e-01 -1.34048104e-01 -3.95327777e-01 -6.29630744e-01
8.03782463e-01 -1.72354311e-01 -1.11638916e+00 2.03658611e-01
-1.19885278e+00 -2.54905015e-01 -5.32380342e-01 3.42415929e-01
-1.93822846e-01 -3.13532174e-01 3.38322043e-01 -1.14692318e+00
-3.97002190e-01 -1.55881202e+00 5.25726914e-01 6.40437007e-01
-2.12323159e-01 -8.16574931e-01 -1.96922839e-01 -2.75014713e-02
5.71089804e-01 2.51392871e-01 1.04079366e+00 -7.88810313e-01
-5.31443715e-01 -2.41486385e-01 2.55943656e-01 7.89551064e-02
-1.30333051e-01 2.16795534e-01 -6.37145460e-01 -4.37404066e-01
-3.78711000e-02 -1.10527677e-02 2.56647199e-01 6.64281607e-01
1.19224143e+00 -4.16411549e-01 -3.33114535e-01 6.53622329e-01
1.72662640e+00 7.14159250e-01 6.69554472e-01 5.36184072e-01
1.01608656e-01 6.53561711e-01 4.35126305e-01 5.05229414e-01
-9.51160192e-02 8.03891003e-01 6.99620485e-01 -2.95721218e-02
1.97431058e-01 2.63983607e-01 -5.02772555e-02 3.69558930e-01
-2.20256194e-01 -6.31574214e-01 -9.10719931e-01 4.88482863e-01
-1.62405431e+00 -6.29709244e-01 1.76816031e-01 2.46810222e+00
2.55832464e-01 1.30911529e-01 4.40945596e-01 3.35952908e-01
8.31627905e-01 -2.38894656e-01 -6.60511672e-01 -7.14451015e-01
3.09780370e-02 4.10377264e-01 6.54033840e-01 3.03264141e-01
-6.01754665e-01 2.67898172e-01 4.98598003e+00 1.10372472e+00
-1.47495377e+00 -2.19954908e-01 7.86051035e-01 -5.94668031e-01
-3.54133368e-01 4.48693149e-02 -6.42920613e-01 5.72553456e-01
1.21082544e+00 -5.94646335e-02 9.12836730e-01 7.44391501e-01
2.11872399e-01 -1.79857552e-01 -7.58594751e-01 8.41370642e-01
-4.05326486e-01 -1.55742979e+00 -2.35859424e-01 4.38098252e-01
6.77455008e-01 -1.13922320e-01 -1.20400973e-01 7.88484737e-02
4.02557440e-02 -9.92025018e-01 9.07347441e-01 5.35505652e-01
5.16177952e-01 -1.19509363e+00 7.19678402e-01 4.26627845e-01
-1.10513580e+00 -4.87611949e-01 -1.86627954e-01 2.46630505e-01
9.65435803e-02 7.53359675e-01 -3.89558583e-01 1.01890039e+00
7.18365669e-01 -2.42299035e-01 -2.38615647e-01 1.36192465e+00
1.88865736e-01 2.55020291e-01 -7.04943717e-01 -6.41682923e-01
2.39815161e-01 -2.98112571e-01 8.00614715e-01 7.96701789e-01
9.17527080e-01 -2.00628757e-01 -2.54118741e-01 1.15309966e+00
1.04174025e-01 1.54970825e-01 -1.33631155e-01 -1.61780477e-01
5.21466970e-01 1.24746740e+00 -1.14880717e+00 4.76254635e-02
2.01988444e-01 5.84707439e-01 -3.27519365e-02 2.06912354e-01
-9.87150252e-01 -6.13094866e-01 3.38032514e-01 -8.33925977e-02
4.94725496e-01 1.03643322e-02 -4.43543911e-01 -6.48729682e-01
6.12721108e-02 -7.70511687e-01 1.86002165e-01 -5.99922776e-01
-5.06567061e-01 9.17288005e-01 5.83536066e-02 -1.12793672e+00
-4.00071830e-01 -6.75989866e-01 -7.32336700e-01 8.64984035e-01
-1.31215537e+00 -7.02266514e-01 -4.18594964e-02 2.91439056e-01
3.88682455e-01 -3.18748027e-01 7.76150465e-01 4.07027245e-01
-8.87776494e-01 5.50536692e-01 2.26302952e-01 -6.43358350e-01
-1.75799236e-01 -7.54130602e-01 -2.11122289e-01 8.28140974e-01
-1.89827576e-01 4.38190460e-01 7.14209795e-01 -2.77632535e-01
-1.52832615e+00 -5.47524393e-01 7.59597182e-01 4.42241937e-01
3.78079772e-01 -2.67513394e-01 -5.06037235e-01 -2.22277835e-01
3.31170142e-01 -5.73440790e-01 4.72494721e-01 1.22802109e-01
6.09910667e-01 -2.45313361e-01 -1.07278621e+00 5.24789929e-01
7.68356562e-01 1.59321740e-01 3.04735929e-01 2.47941330e-01
2.38221213e-01 -2.16395766e-01 -7.03262210e-01 4.67530429e-01
5.20846963e-01 -9.47712421e-01 1.01631343e+00 -2.94791609e-02
1.34759828e-01 -3.91246676e-01 1.64506715e-02 -1.17886603e+00
-1.39873251e-02 -8.06841671e-01 -4.96573746e-02 1.22478306e+00
8.26719761e-01 -7.59307504e-01 8.78193080e-01 3.52887988e-01
-1.73464864e-01 -1.23593092e+00 -1.18186545e+00 -1.00767076e+00
-1.86354890e-01 -7.11702555e-02 1.15999210e+00 7.28285849e-01
-3.98310125e-01 8.14956799e-02 -1.40882567e-01 3.33096564e-01
3.54842365e-01 2.66581714e-01 3.13621610e-01 -1.42838371e+00
-7.70628333e-01 -9.28587139e-01 -8.61766934e-02 -6.48322403e-01
-6.99259639e-02 -6.59265459e-01 -9.73129421e-02 -1.38250363e+00
-3.94719303e-01 -6.48346484e-01 -9.66437832e-02 4.43322659e-01
4.64686483e-01 -2.11048126e-01 2.36750513e-01 6.91576302e-03
-3.50964755e-01 3.39228392e-01 9.48468029e-01 -6.56099021e-02
-4.35363859e-01 2.96954334e-01 -4.71440822e-01 5.22708178e-01
1.12876260e+00 -4.81523424e-01 -4.73227680e-01 -4.05882001e-01
5.18733621e-01 2.32054114e-01 7.38946274e-02 -1.36568618e+00
3.62335771e-01 -9.41329226e-02 1.83430761e-01 -5.13865769e-01
4.54049677e-01 -1.28531682e+00 7.78392136e-01 7.21893966e-01
1.54248655e-01 1.33150131e-01 4.67740893e-01 -2.43474785e-02
-6.33738190e-02 -9.18568909e-01 6.61447883e-01 2.84815785e-02
-4.69364733e-01 -1.44893691e-01 -5.75885952e-01 -4.13747042e-01
1.09951746e+00 -7.29139686e-01 -3.48251336e-03 5.51000610e-02
-8.38380873e-01 3.47381532e-01 3.66906017e-01 -1.39881927e-03
1.93409979e-01 -8.31619620e-01 -3.50826979e-01 1.24960862e-01
-3.00341427e-01 3.04753873e-02 3.69898796e-01 7.21300960e-01
-5.70554674e-01 3.77181590e-01 -4.76620078e-01 -3.80706996e-01
-1.21517909e+00 5.47254741e-01 7.60854840e-01 -6.47593796e-01
1.86589271e-01 6.87213480e-01 -4.42410380e-01 -3.91385317e-01
1.97630957e-01 -1.38060421e-01 -2.75027364e-01 1.73218235e-01
1.34743050e-01 6.35102451e-01 3.20421666e-01 -3.29220265e-01
-5.83281755e-01 5.79042912e-01 5.84395826e-01 5.28854020e-02
1.55015874e+00 5.00184074e-02 -2.30250135e-01 -8.79187807e-02
1.17929447e+00 -7.26462230e-02 -8.00723493e-01 5.66551387e-01
-1.22167841e-01 -1.83524862e-02 2.38153607e-01 -1.16162252e+00
-1.43164027e+00 6.23933613e-01 7.84562171e-01 3.73709798e-01
1.88579869e+00 -2.74775386e-01 3.62350017e-01 4.10050005e-01
5.75364053e-01 -1.06823123e+00 -3.08998853e-01 3.86938930e-01
7.10256875e-01 -4.53012198e-01 -7.67160207e-02 -1.12216465e-01
-9.06828195e-02 1.42869508e+00 4.88225162e-01 1.13785565e-02
4.49870199e-01 4.65614617e-01 -3.64594489e-01 -3.10403913e-01
-6.32731199e-01 -2.00280905e-01 3.53928775e-01 2.34065548e-01
-5.68978451e-02 4.34839837e-02 -5.00821650e-01 6.57062888e-01
-1.66004106e-01 4.15010452e-02 3.46242972e-02 8.87597144e-01
-2.59599924e-01 -1.43436527e+00 -3.49233508e-01 3.07700038e-01
-2.27711514e-01 1.23617932e-01 -4.05969501e-01 6.79935396e-01
4.72190082e-01 7.98020959e-01 -4.32066694e-02 -4.58679795e-01
3.26083958e-01 4.28789444e-02 3.97307217e-01 -2.87650526e-02
-1.03518701e+00 6.45541251e-02 4.88836676e-01 -5.15715003e-01
-8.61358419e-02 -5.33929527e-01 -1.21213746e+00 -2.35561401e-01
-6.09162271e-01 4.66700912e-01 1.15987551e+00 7.68085420e-01
5.90777457e-01 1.01801074e+00 6.44097626e-01 -1.09345126e+00
-7.72042647e-02 -1.88931629e-01 -2.29004905e-01 -5.67346931e-01
-9.05491114e-02 -6.55724287e-01 -1.70022681e-01 -6.13317013e-01] | [8.157873153686523, 3.207883834838867] |
d655bdb2-0471-4ee1-910f-d1385b35e4de | wikichat-a-few-shot-llm-based-chatbot | 2305.14292 | null | https://arxiv.org/abs/2305.14292v1 | https://arxiv.org/pdf/2305.14292v1.pdf | WikiChat: A Few-Shot LLM-Based Chatbot Grounded with Wikipedia | Despite recent advances in Large Language Models (LLMs), users still cannot trust the information provided in their responses. LLMs cannot speak accurately about events that occurred after their training, which are often topics of great interest to users, and, as we show in this paper, they are highly prone to hallucination when talking about less popular (tail) topics. This paper presents WikiChat, a few-shot LLM-based chatbot that is grounded with live information from Wikipedia. Through many iterations of experimentation, we have crafte a pipeline based on information retrieval that (1) uses LLMs to suggest interesting and relevant facts that are individually verified against Wikipedia, (2) retrieves additional up-to-date information, and (3) composes coherent and engaging time-aware responses. We propose a novel hybrid human-and-LLM evaluation methodology to analyze the factuality and conversationality of LLM-based chatbots. We focus on evaluating important but previously neglected issues such as conversing about recent and tail topics. We evaluate WikiChat against strong fine-tuned and LLM-based baselines across a diverse set of conversation topics. We find that WikiChat outperforms all baselines in terms of the factual accuracy of its claims, by up to 12.1%, 28.3% and 32.7% on head, recent and tail topics, while matching GPT-3.5 in terms of providing natural, relevant, non-repetitive and informational responses. | ['Monica S. Lam', 'Heidi C. Zhang', 'Violet Z. Yao', 'Sina J. Semnani'] | 2023-05-23 | null | null | null | null | ['chatbot', 'chatbot'] | ['methodology', 'natural-language-processing'] | [-4.15340275e-01 4.14294094e-01 -1.29496962e-01 -2.10158408e-01
-1.43852890e+00 -7.38912165e-01 1.04627359e+00 4.76726174e-01
-3.47977877e-01 8.99421811e-01 8.14964414e-01 -1.01585448e-01
-5.07497750e-02 -4.25739557e-01 -3.66948515e-01 -1.38256282e-01
1.47621185e-01 6.79416001e-01 3.06106269e-01 -5.71038485e-01
4.63166237e-01 -2.29913116e-01 -9.72114682e-01 1.00028074e+00
7.17228770e-01 4.86143917e-01 3.13804328e-01 7.68623292e-01
-4.46952581e-01 1.59958589e+00 -1.09726858e+00 -9.63727236e-01
-5.22255182e-01 -4.12869871e-01 -1.56462622e+00 -3.67306709e-01
3.95446002e-01 -1.44956186e-01 -9.67188105e-02 5.35333395e-01
5.31861186e-01 3.65142554e-01 5.44217587e-01 -1.07491839e+00
-3.38136435e-01 1.01468325e+00 -1.25189319e-01 1.11701421e-01
9.01926816e-01 2.11025625e-01 1.07345009e+00 -7.70155013e-01
9.20403600e-01 1.53906631e+00 9.17526424e-01 7.20678389e-01
-9.64159369e-01 -2.85197765e-01 -6.68367445e-02 4.01011407e-02
-9.62245703e-01 -4.92659092e-01 5.35086393e-01 -4.63450819e-01
1.24542737e+00 6.23683393e-01 4.89505939e-02 1.75198972e+00
1.72971915e-02 8.09919894e-01 8.16466510e-01 -2.73745269e-01
-3.11026052e-02 6.53555572e-01 2.60400295e-01 2.81579792e-01
-3.21253747e-01 -7.60416687e-01 -9.61842716e-01 -5.59140801e-01
5.32488190e-02 -3.51416409e-01 -2.85762072e-01 6.86998725e-01
-1.18415654e+00 9.25269485e-01 1.89349100e-01 5.22569597e-01
-7.27615833e-01 -7.94728547e-02 6.50174141e-01 4.24376160e-01
8.03167403e-01 8.34549785e-01 -3.92223150e-01 -9.91138995e-01
-6.99304044e-01 4.64688241e-01 1.74792802e+00 8.98308396e-01
6.95977449e-01 -5.57518721e-01 -6.65014863e-01 1.19262397e+00
1.03765577e-01 2.52598703e-01 5.02833486e-01 -1.22339904e+00
8.98878634e-01 6.52258277e-01 4.79709476e-01 -1.18870604e+00
-3.98504198e-01 -1.74311638e-01 -5.14268517e-01 -6.35912776e-01
3.33933443e-01 -3.76711130e-01 1.79044008e-01 1.54718840e+00
1.63987622e-01 -1.17031746e-01 1.02736309e-01 6.79673016e-01
1.38187766e+00 9.46111023e-01 2.12074980e-01 -3.46922517e-01
1.35284793e+00 -1.11370575e+00 -8.09659600e-01 -2.99357921e-01
7.59681225e-01 -1.09953082e+00 1.47902966e+00 2.73574829e-01
-1.21848261e+00 -2.32184350e-01 -4.48625892e-01 -2.69971460e-01
-1.23492599e-01 -8.31394643e-02 3.96345019e-01 1.80905089e-01
-8.70492160e-01 6.02179408e-01 -2.03570768e-01 -7.62109995e-01
5.74181788e-02 -2.16002181e-01 -2.84657717e-01 8.62531513e-02
-1.40363252e+00 1.17119908e+00 -8.43950137e-02 -2.48274267e-01
-9.16431665e-01 -7.04316616e-01 -5.45541167e-01 4.06550691e-02
4.81159121e-01 -3.74446422e-01 1.97676861e+00 -3.58202606e-01
-1.65117776e+00 9.78211462e-01 -3.27099711e-01 -5.06059051e-01
5.70337296e-01 -5.34061670e-01 -1.72539726e-01 3.33200425e-01
4.41475749e-01 3.07005823e-01 3.96396101e-01 -9.57076252e-01
-4.48667347e-01 2.50129372e-01 4.74024564e-01 2.02455357e-01
-3.93215001e-01 4.02645648e-01 -2.16862231e-01 -2.46756732e-01
-4.45071638e-01 -7.09303737e-01 9.22682695e-03 -5.26983500e-01
-5.58008730e-01 -7.59919822e-01 7.29266346e-01 -7.48607934e-01
1.38641584e+00 -1.85477924e+00 -2.76481777e-01 -2.59057105e-01
3.36547762e-01 4.83502030e-01 -1.05908275e-01 1.38348722e+00
7.11069882e-01 2.66356051e-01 2.36516654e-01 -6.66478515e-01
7.40795061e-02 -2.86450842e-03 -4.92067128e-01 -1.60212200e-02
-3.76057066e-02 1.01950443e+00 -1.20749962e+00 -6.06446922e-01
8.43615159e-02 3.35187823e-01 -4.18420464e-01 6.12763524e-01
-5.57320893e-01 3.79560679e-01 -5.06136596e-01 2.21348315e-01
-3.08058634e-02 -4.89525110e-01 -5.82713773e-03 1.94547579e-01
-1.30649656e-01 1.21060979e+00 -4.20741320e-01 1.65396452e+00
-8.53071988e-01 9.81561422e-01 1.38388453e-02 -4.67983186e-01
8.51488054e-01 9.16035235e-01 1.46271333e-01 -4.87908095e-01
8.03498104e-02 1.15625188e-01 -3.58785003e-01 -9.36958849e-01
5.99596560e-01 5.86390309e-02 -3.16887826e-01 9.16532755e-01
2.11819455e-01 -2.17542872e-01 1.49275169e-01 8.00400555e-01
1.26071250e+00 -3.25609922e-01 3.95235777e-01 -8.71432796e-02
5.81250489e-01 9.93762463e-02 4.13822150e-03 1.09724629e+00
-9.29441303e-02 3.63585144e-01 7.71703601e-01 -2.44932070e-01
-5.47032833e-01 -6.20919108e-01 3.07874054e-01 1.39053929e+00
-9.02888924e-02 -8.34808648e-01 -6.77514613e-01 -6.28344119e-01
-3.03940952e-01 1.19147754e+00 -4.73704278e-01 5.28570414e-02
-5.19428551e-01 -2.77955741e-01 7.04747021e-01 -1.02868661e-01
5.15520513e-01 -1.40635931e+00 -4.89831895e-01 6.35177672e-01
-1.24562728e+00 -1.22501302e+00 -5.06656945e-01 -1.80724084e-01
-4.41329598e-01 -1.10047662e+00 -4.66764331e-01 -4.91419435e-01
-6.53838739e-02 5.30990362e-01 1.52116871e+00 3.05360466e-01
1.65979013e-01 5.10894358e-01 -6.75481558e-01 -4.98110682e-01
-8.95721138e-01 1.88312784e-01 -2.53708929e-01 -1.30463660e-01
4.58834678e-01 -7.32170284e-01 -3.15976739e-01 3.60882849e-01
-2.34806940e-01 -4.51783538e-02 3.11851501e-01 6.93986595e-01
-3.85878593e-01 -5.35746694e-01 8.17171395e-01 -1.19744170e+00
1.32878387e+00 -9.28080201e-01 2.06944332e-01 2.35446885e-01
-3.24495554e-01 -2.32187703e-01 4.78050113e-01 -6.43774748e-01
-1.30855334e+00 -7.49029696e-01 -2.32840300e-01 1.72869772e-01
-3.04591991e-02 6.02704585e-01 2.71196038e-01 2.28523999e-01
1.17175865e+00 -7.10347220e-02 -1.89466119e-01 -6.00868046e-01
4.49128985e-01 1.22321439e+00 4.93810147e-01 -8.49168003e-01
3.44270885e-01 2.11203068e-01 -8.57201576e-01 -1.00729835e+00
-1.31384242e+00 -1.00941801e+00 -4.34703790e-02 -5.93747258e-01
3.41451198e-01 -7.87492812e-01 -1.33493865e+00 2.39082754e-01
-1.77789998e+00 -5.49618125e-01 3.89458076e-03 2.74713218e-01
-5.72654009e-01 4.48236316e-01 -1.19522429e+00 -1.12811494e+00
-8.05950463e-01 -5.50014853e-01 8.49210680e-01 2.04770751e-02
-1.30732155e+00 -1.10729122e+00 1.95494965e-01 7.45858669e-01
6.73222542e-01 1.80096123e-02 6.95630312e-01 -1.16086578e+00
-9.12878066e-02 -4.41567928e-01 -2.25183945e-02 -8.02838802e-03
-9.39633623e-02 -1.56199768e-01 -1.20194113e+00 8.45802352e-02
2.02636093e-01 -9.07307506e-01 5.03516078e-01 -6.62949681e-02
4.84794587e-01 -1.11446357e+00 -2.23049894e-01 -3.16462457e-01
8.15351427e-01 -1.54253915e-01 5.46367049e-01 3.70698392e-01
1.23345897e-01 1.13600636e+00 5.87222695e-01 6.81629121e-01
7.47434437e-01 5.67654490e-01 3.70767266e-01 4.28850561e-01
5.91303371e-02 -5.74030340e-01 5.83820045e-01 9.29245532e-01
6.03812262e-02 -4.67397988e-01 -6.27187133e-01 8.64661217e-01
-2.07713246e+00 -1.32334793e+00 -3.57699662e-01 1.76173222e+00
1.34449279e+00 1.37729555e-01 1.68687448e-01 -3.66022378e-01
4.79903042e-01 3.17846000e-01 -7.68480450e-03 -6.16966784e-01
1.85038857e-02 4.29405905e-02 -2.62233585e-01 7.78918564e-01
-7.14466691e-01 9.51931417e-01 5.62715006e+00 8.43796849e-01
-7.52307713e-01 5.98496318e-01 4.07702804e-01 -1.01151019e-01
-1.25351623e-01 4.75357883e-02 -6.67696357e-01 4.42733228e-01
1.19591510e+00 -4.70935822e-01 2.08321750e-01 9.59029078e-01
5.05095363e-01 -2.60071516e-01 -1.03610623e+00 7.04441905e-01
1.59910277e-01 -1.51612616e+00 -1.81281239e-01 -4.58609819e-01
6.09967232e-01 1.14040986e-01 -3.70514303e-01 7.41700768e-01
5.26291370e-01 -8.70322764e-01 6.23887241e-01 4.31166500e-01
1.67867303e-01 -2.74062037e-01 1.01556587e+00 1.04946458e+00
-6.59874618e-01 1.96262166e-01 -1.42763928e-01 -4.36644316e-01
3.71318519e-01 4.52644676e-01 -1.27976227e+00 2.42046624e-01
6.93342745e-01 6.31859779e-01 -2.51500219e-01 8.03260803e-01
-6.23072267e-01 9.17963386e-01 -2.88654000e-01 -7.70030081e-01
4.59244907e-01 3.09063435e-01 6.20966792e-01 1.70294631e+00
4.18794900e-03 2.55899101e-01 1.12150475e-01 1.06695616e+00
-3.57808024e-01 2.07636654e-01 -4.66178268e-01 -6.16628528e-02
9.35412765e-01 1.49423182e+00 -1.24545492e-01 -4.33601707e-01
-3.27704728e-01 9.07067537e-01 6.05876803e-01 2.88208395e-01
-4.36527252e-01 -2.45684400e-01 4.47215527e-01 1.08061070e-02
-2.82609910e-01 1.65597796e-01 1.74121648e-01 -9.56216335e-01
1.25048757e-01 -9.69526827e-01 4.35032338e-01 -9.00367260e-01
-1.60965216e+00 7.85876572e-01 -1.35979712e-01 -9.21906590e-01
-6.72575414e-01 -6.83895424e-02 -1.09679294e+00 8.93252313e-01
-1.12838709e+00 -1.08064079e+00 -3.02636713e-01 4.71985459e-01
7.88103163e-01 1.53785467e-01 1.03419018e+00 1.93191260e-01
-1.66244373e-01 3.70788962e-01 -3.62446338e-01 1.01117335e-01
1.10639274e+00 -1.00924349e+00 3.15449208e-01 3.22549880e-01
2.22780854e-01 9.11488712e-01 1.09445024e+00 -6.20487750e-01
-9.43502724e-01 -8.92314494e-01 1.77872193e+00 -1.07033932e+00
8.37214649e-01 -3.55188340e-01 -1.15593910e+00 7.09897816e-01
5.45600474e-01 -8.41558874e-01 6.62611246e-01 4.72839147e-01
-4.07561958e-01 3.00440013e-01 -1.01275063e+00 5.84268332e-01
5.66843510e-01 -8.37947071e-01 -1.00863373e+00 9.89351273e-01
1.01470435e+00 -4.20723438e-01 -8.18808615e-01 -2.10649624e-01
4.58454847e-01 -9.38424468e-01 7.04569340e-01 -8.93908560e-01
5.76647699e-01 4.26645547e-01 5.87412417e-02 -1.21226168e+00
2.13490855e-02 -1.35371041e+00 -1.18220426e-01 1.59559083e+00
5.42168856e-01 -3.11672896e-01 4.76598769e-01 7.78654575e-01
-9.93206501e-02 -4.59111363e-01 -6.85463846e-01 -5.68594575e-01
-6.68088645e-02 -6.55701399e-01 5.50237969e-02 1.09644353e+00
9.18103516e-01 8.61457765e-01 -7.36552060e-01 -2.36349717e-01
1.35415450e-01 1.91506166e-02 1.07415378e+00 -1.18687630e+00
-1.39922529e-01 -5.37840605e-01 2.43669197e-01 -1.05765831e+00
1.10930890e-01 -6.41756713e-01 2.77343124e-01 -1.74934232e+00
4.07710075e-01 3.65984440e-02 2.74778694e-01 4.80385244e-01
-1.71585187e-01 8.80528521e-03 7.16941431e-02 4.60366279e-01
-1.05115020e+00 3.29834819e-01 8.48735809e-01 9.23966020e-02
-3.05895090e-01 2.53515542e-01 -8.54397058e-01 7.40803421e-01
6.84441209e-01 -5.22081971e-01 -1.07107103e-01 -1.10647548e-02
4.63688254e-01 3.77460301e-01 3.80927503e-01 -6.39495075e-01
4.73776251e-01 -1.30879492e-01 -6.09475851e-01 -4.82816309e-01
4.41036910e-01 -2.70639122e-01 -2.79579520e-01 1.00017481e-01
-8.71852219e-01 -2.38261566e-01 1.10623538e-01 3.93360406e-01
-3.73186201e-01 -4.55695719e-01 2.88066149e-01 -4.90449995e-01
-4.92625982e-01 -1.52628899e-01 -7.76872575e-01 6.34500682e-01
4.38745856e-01 3.25285226e-01 -7.10710168e-01 -1.34912789e+00
-3.71269107e-01 2.98883080e-01 -5.13629988e-02 6.34181917e-01
4.15835232e-01 -9.07873690e-01 -1.05350983e+00 -6.98920727e-01
2.83636630e-01 -3.35065007e-01 3.32669228e-01 1.08386588e+00
-1.56425133e-01 6.42364800e-01 2.78231800e-01 -3.58035475e-01
-1.23204279e+00 1.93823233e-01 -4.43206243e-02 -4.72819418e-01
-6.39548659e-01 1.08306813e+00 -1.16217375e-01 -5.18337965e-01
5.47736228e-01 -4.75166738e-02 -2.80220121e-01 2.03464642e-01
8.11191678e-01 4.40070540e-01 4.24977094e-02 -4.17614520e-01
-2.44752124e-01 -1.49276620e-02 -9.00182351e-02 -3.48930836e-01
1.20986176e+00 -2.99709469e-01 -3.59524280e-01 8.00446391e-01
1.22728324e+00 1.80054769e-01 -7.40338564e-01 -6.39843524e-01
4.70777124e-01 -2.03855485e-01 -5.77681839e-01 -1.18605638e+00
-2.44325966e-01 1.05917132e+00 -4.87826765e-01 5.24828911e-01
4.76763010e-01 3.44839066e-01 1.07584465e+00 7.89259493e-01
3.36289436e-01 -9.39303398e-01 5.20635188e-01 8.31794620e-01
1.26700759e+00 -1.30191314e+00 -2.75342077e-01 -2.10468501e-01
-9.75753725e-01 1.14092505e+00 6.33739352e-01 2.81894803e-01
2.00958222e-01 3.21333744e-02 3.02902907e-01 -4.81753767e-01
-1.49872065e+00 1.22615851e-01 2.65347689e-01 1.86737269e-01
7.56826580e-01 -1.75804850e-02 -4.00558919e-01 9.17365432e-01
-3.61053556e-01 -3.06868523e-01 7.01526225e-01 6.24237716e-01
-6.13504171e-01 -5.76760173e-01 -8.61377642e-02 1.47855669e-01
-6.06873870e-01 -1.69056654e-01 -8.58039200e-01 7.43724167e-01
-5.45746326e-01 1.55954170e+00 -4.37611520e-01 -4.20007974e-01
3.36799741e-01 2.12053522e-01 -1.45246446e-01 -9.17622805e-01
-1.21420825e+00 -2.43046463e-01 9.21226084e-01 -6.14618301e-01
-4.62574720e-01 -4.54019397e-01 -9.60582912e-01 -5.96429169e-01
-3.65766257e-01 7.23643064e-01 4.32535768e-01 1.16449845e+00
3.77479672e-01 5.74771687e-02 5.03109217e-01 -5.65072119e-01
-6.36088312e-01 -1.66299713e+00 5.31748161e-02 6.59237266e-01
1.25780851e-01 -2.90262133e-01 -4.74033356e-01 -1.75547749e-01] | [12.293328285217285, 8.129841804504395] |
ea6c5496-0be8-470f-99f5-fd8d176a8393 | modal-regression-based-atomic-representation | 1711.05861 | null | http://arxiv.org/abs/1711.05861v1 | http://arxiv.org/pdf/1711.05861v1.pdf | Modal Regression based Atomic Representation for Robust Face Recognition | Representation based classification (RC) methods such as sparse RC (SRC) have
shown great potential in face recognition in recent years. Most previous RC
methods are based on the conventional regression models, such as lasso
regression, ridge regression or group lasso regression. These regression models
essentially impose a predefined assumption on the distribution of the noise
variable in the query sample, such as the Gaussian or Laplacian distribution.
However, the complicated noises in practice may violate the assumptions and
impede the performance of these RC methods. In this paper, we propose a modal
regression based atomic representation and classification (MRARC) framework to
alleviate such limitation. Unlike previous RC methods, the MRARC framework does
not require the noise variable to follow any specific predefined distributions.
This gives rise to the capability of MRARC in handling various complex noises
in reality. Using MRARC as a general platform, we also develop four novel RC
methods for unimodal and multimodal face recognition, respectively. In
addition, we devise a general optimization algorithm for the unified MRARC
framework based on the alternating direction method of multipliers (ADMM) and
half-quadratic theory. The experiments on real-world data validate the efficacy
of MRARC for robust face recognition. | ['Luoqing Li', 'Hong Chen', 'Yulong Wang', 'Yuan Yan Tang'] | 2017-11-05 | null | null | null | null | ['robust-face-recognition'] | ['computer-vision'] | [ 2.44369477e-01 -4.50932980e-01 -2.29712531e-01 -5.24267435e-01
-7.70437181e-01 5.43611385e-02 3.71694475e-01 -6.59778297e-01
-5.16190492e-02 6.41736269e-01 6.98885471e-02 -4.34013084e-02
-4.43118513e-01 -5.57083726e-01 -5.08078098e-01 -1.02971494e+00
3.67637068e-01 1.55967742e-01 -6.28673732e-01 -2.34863475e-01
1.72734857e-01 5.85167766e-01 -1.29007185e+00 -1.56997647e-02
1.07908058e+00 9.39532936e-01 -1.12612583e-01 -2.69064754e-02
-2.20751017e-01 6.65113330e-01 -3.70707303e-01 -2.37637267e-01
1.83764026e-01 -3.71558636e-01 -1.40857145e-01 2.27468997e-01
3.40861946e-01 8.11029077e-02 -5.62110543e-01 1.15212893e+00
7.09729671e-01 3.30730468e-01 7.34811842e-01 -1.37942529e+00
-9.69085813e-01 6.40251100e-01 -1.07778513e+00 -1.65850185e-02
3.18746269e-01 -2.97058553e-01 6.24384880e-01 -1.29074490e+00
2.78755099e-01 1.54773808e+00 7.63429224e-01 5.48946321e-01
-1.27851677e+00 -8.84256423e-01 4.15544629e-01 1.09832533e-01
-1.81030047e+00 -7.55794048e-01 1.06435108e+00 -4.03084576e-01
3.08309078e-01 4.29352820e-01 -4.13538236e-03 1.06558800e+00
-1.52332142e-01 7.33677030e-01 1.18897700e+00 -2.70909756e-01
7.67003819e-02 1.07275173e-01 1.49226800e-01 5.87572038e-01
2.93937057e-01 -1.70595810e-01 -3.38368535e-01 -4.06760871e-01
7.99869001e-01 2.69073427e-01 -3.82172018e-01 -2.12532103e-01
-1.07910991e+00 9.46323514e-01 2.05651686e-01 4.21474546e-01
-4.01262850e-01 1.06577739e-01 3.33628684e-01 4.65443023e-02
4.79485393e-01 -2.32846618e-01 1.22038396e-02 4.11389738e-01
-1.03963864e+00 -1.42480746e-01 4.79914725e-01 9.57894564e-01
5.92094362e-01 6.20848119e-01 -2.09834427e-01 1.16746426e+00
6.98296189e-01 8.25725734e-01 5.92685819e-01 -8.03192496e-01
4.89828378e-01 4.35003936e-01 -1.37199342e-01 -1.68336570e+00
-1.98203459e-01 -6.22450590e-01 -1.38314164e+00 -1.17208250e-01
-2.59504430e-02 -4.39043064e-03 -8.13335061e-01 1.80856526e+00
2.76562452e-01 3.58542293e-01 4.78720963e-02 1.10819149e+00
9.50550258e-01 6.82773769e-01 -5.12150340e-02 -6.73703969e-01
8.16496968e-01 -6.54750228e-01 -1.10106146e+00 1.29915893e-01
2.99351901e-01 -6.58341587e-01 7.19900608e-01 5.06408155e-01
-6.95471883e-01 -4.98647332e-01 -7.92587340e-01 2.53478944e-01
-7.16718584e-02 6.26304805e-01 6.76122367e-01 7.78089583e-01
-7.61296511e-01 2.24380240e-01 -6.85269415e-01 -4.55566533e-02
3.55828404e-01 5.67908645e-01 -6.42547369e-01 -2.69192159e-01
-8.99441719e-01 5.57764709e-01 -1.11078501e-01 9.92379963e-01
-6.87412977e-01 -3.90695959e-01 -9.51419771e-01 -1.28330827e-01
2.87861437e-01 -3.35601628e-01 5.42351961e-01 -1.19152474e+00
-1.61143434e+00 5.50624669e-01 -5.25880635e-01 2.09570661e-01
4.12197471e-01 -1.41770810e-01 -5.25070369e-01 -1.83234513e-01
-7.65856430e-02 -2.97380309e-03 1.14777029e+00 -1.27886355e+00
1.24667808e-02 -6.78969085e-01 -3.08162838e-01 -7.91814029e-02
-2.09115878e-01 2.77367383e-01 -3.99866521e-01 -8.44652772e-01
4.90582287e-01 -7.93411434e-01 -3.26199412e-01 -1.41060531e-01
-5.92546105e-01 -3.53259683e-01 8.10385942e-01 -6.22835815e-01
1.23371136e+00 -2.40726900e+00 4.86306489e-01 5.24848938e-01
-2.30219625e-02 1.02663018e-01 -2.50816882e-01 2.34020427e-01
-3.67939025e-01 4.13325727e-02 -5.26229978e-01 -4.16260153e-01
-6.86377883e-02 1.84731469e-01 -2.58549780e-01 1.03035843e+00
1.72146298e-02 5.04202545e-01 -6.58807516e-01 -5.12947261e-01
-9.26247090e-02 8.22105467e-01 -4.08823758e-01 1.95974872e-01
2.02684641e-01 7.18011498e-01 -7.29109168e-01 8.97681117e-01
1.12402868e+00 7.09112957e-02 5.47257811e-02 -5.90583622e-01
-1.21023528e-01 -4.14718121e-01 -1.63682151e+00 1.41768992e+00
-2.48847261e-01 2.82068074e-01 5.02161086e-01 -1.48224485e+00
1.30212426e+00 4.49019194e-01 7.24486589e-01 -3.09290439e-01
6.36583343e-02 2.62572378e-01 -3.05837035e-01 -5.03508568e-01
-3.43120284e-02 -2.47861385e-01 2.32835054e-01 1.51792228e-01
-1.93336546e-01 3.12755167e-01 -4.63057980e-02 -5.92053160e-02
4.81426448e-01 1.48288682e-01 -5.15847877e-02 -1.82279974e-01
1.12520099e+00 -3.36837411e-01 1.08766949e+00 4.57901597e-01
-4.91300039e-03 7.62751043e-01 2.61594832e-01 -1.91998079e-01
-6.44100249e-01 -8.27659428e-01 -4.34372723e-01 8.74410987e-01
-4.08769399e-02 -2.15344921e-01 -2.82860458e-01 -5.41713417e-01
2.23798528e-02 3.40116978e-01 -5.26132524e-01 -1.28159702e-01
-6.41892493e-01 -1.30528569e+00 4.93791759e-01 3.64831507e-01
5.27513325e-01 -6.57237649e-01 4.72657323e-01 9.67301279e-02
-2.59988993e-01 -1.04818082e+00 -6.07395768e-01 -2.57911831e-01
-6.86078131e-01 -9.28576112e-01 -7.08185971e-01 -7.75523186e-01
1.13770866e+00 4.74724263e-01 5.22882581e-01 -8.17035362e-02
-4.04898405e-01 4.65432405e-01 -2.86258817e-01 4.37083915e-02
-5.49421236e-02 -4.48581010e-01 4.90424156e-01 8.46733272e-01
2.22971961e-01 -5.60722530e-01 -5.09064257e-01 5.59788465e-01
-8.59848082e-01 -3.25602502e-01 6.05888963e-01 9.61232185e-01
8.27905476e-01 2.80096438e-02 8.77376735e-01 -9.20252502e-01
6.48423493e-01 -8.41236234e-01 -5.98304927e-01 3.45817685e-01
-5.58801174e-01 -1.79197833e-01 7.14671493e-01 -7.62198687e-01
-1.11710286e+00 1.69996187e-01 1.68863293e-02 -8.83366585e-01
3.26656789e-01 7.82680154e-01 -4.52686459e-01 -3.07244033e-01
3.43918860e-01 5.87980032e-01 1.68930694e-01 -6.05587780e-01
2.62831151e-01 8.97620201e-01 2.64055610e-01 -6.74785674e-01
1.03729844e+00 5.47990680e-01 1.40239075e-01 -9.85714912e-01
-5.96954584e-01 -4.32505161e-01 -4.50220257e-01 -3.57153080e-02
7.14344800e-01 -1.01000237e+00 -7.94653893e-01 2.87783056e-01
-8.63819361e-01 1.91623375e-01 1.18377917e-01 7.15348661e-01
-4.07269686e-01 6.71802461e-01 -3.41116130e-01 -1.19701636e+00
-3.98293436e-01 -1.28790414e+00 8.13076377e-01 2.72974819e-01
3.30489367e-01 -7.56531119e-01 -8.72584730e-02 5.42116821e-01
6.23736441e-01 4.00987357e-01 6.44385159e-01 -3.92807752e-01
-2.91358382e-01 -2.46812433e-01 -2.27688715e-01 5.78267336e-01
3.49215865e-01 1.90107912e-01 -8.04171085e-01 -5.21986783e-01
2.46502742e-01 -1.56016141e-01 7.08624721e-01 3.57702106e-01
1.32106221e+00 -3.35526526e-01 -3.16726506e-01 9.36421692e-01
1.44829237e+00 7.31295794e-02 6.86871231e-01 -8.06737244e-02
8.86174381e-01 5.46890199e-01 6.04879558e-01 5.64280570e-01
1.11325681e-01 7.66193449e-01 1.65733367e-01 -1.45096615e-01
3.78746420e-01 -7.32173324e-02 6.38708353e-01 1.04271376e+00
-3.23058337e-01 1.92150697e-01 -4.54309374e-01 1.84399396e-01
-2.03898668e+00 -1.05840933e+00 -3.09651345e-01 2.36350226e+00
6.64644837e-01 -5.80388606e-01 -1.86322987e-01 5.23685031e-02
9.46100771e-01 1.45553589e-01 -4.51327890e-01 -3.35314423e-01
-3.77899677e-01 -4.65235189e-02 2.68629402e-01 2.40005270e-01
-1.04268205e+00 7.03807890e-01 6.07301807e+00 1.04332483e+00
-1.19612777e+00 1.38479054e-01 5.48473120e-01 2.66531497e-01
-3.18197221e-01 -1.15393125e-01 -8.28375697e-01 4.25605208e-01
5.28165281e-01 -1.90630052e-02 6.88742518e-01 8.03226411e-01
5.10155916e-01 3.28221142e-01 -8.21271598e-01 1.39481592e+00
3.49146307e-01 -7.25911558e-01 2.06314340e-01 2.49424344e-03
8.01650107e-01 -3.13156009e-01 2.81170875e-01 3.94189030e-01
-1.44662648e-01 -1.43239069e+00 1.02506995e-01 8.53222072e-01
9.30814743e-01 -6.24540269e-01 7.01512039e-01 2.24554688e-01
-1.03675747e+00 -4.10059318e-02 -8.44540417e-01 4.50274646e-01
-1.05530068e-01 6.29518867e-01 -6.57007471e-02 8.28574717e-01
3.46103638e-01 8.90196621e-01 -3.04090142e-01 9.04513776e-01
-6.43116310e-02 6.46672189e-01 -4.10575539e-01 3.24701220e-01
-4.77132648e-02 -8.29617381e-01 8.01211655e-01 1.14188647e+00
2.96155572e-01 1.59844965e-01 2.36750364e-01 9.31139648e-01
-4.14380282e-02 5.98956883e-01 -5.60674250e-01 -4.54873331e-02
3.75061333e-01 1.30127251e+00 -2.19003990e-01 6.14782572e-02
-5.75959027e-01 7.11508512e-01 2.39745006e-01 7.31717706e-01
-8.84789228e-01 -2.18517795e-01 4.82583553e-01 -6.98016956e-02
2.59723574e-01 -3.39293212e-01 -9.33008119e-02 -1.50839615e+00
1.39706388e-01 -1.22320759e+00 3.48431736e-01 -3.16118807e-01
-1.59295309e+00 5.60159504e-01 -8.54807645e-02 -1.23631001e+00
3.28631699e-01 -4.80734080e-01 -7.55856574e-01 9.00814593e-01
-1.56949842e+00 -1.24205661e+00 -3.66029650e-01 9.32013631e-01
3.70111078e-01 -4.61743265e-01 6.40173137e-01 6.31792009e-01
-1.25611567e+00 7.63422370e-01 4.50543046e-01 1.59029201e-01
7.62236416e-01 -7.74249971e-01 -5.84393442e-01 6.56340718e-01
-7.29790097e-03 1.00703311e+00 5.37742198e-01 -4.95814502e-01
-1.97934365e+00 -1.16102362e+00 4.23011959e-01 -4.07282636e-03
4.11226749e-01 -2.03483254e-01 -1.08618140e+00 7.81457841e-01
-3.63000572e-01 3.69073987e-01 7.68652022e-01 1.23191401e-01
-4.52465624e-01 -5.81303537e-01 -1.05001783e+00 4.16949630e-01
8.33054245e-01 -3.70514393e-01 -3.92733753e-01 5.69869876e-01
2.53363401e-01 7.70635903e-02 -1.04296196e+00 7.53935397e-01
4.91656721e-01 -5.07400334e-01 1.11468017e+00 -7.07457483e-01
7.61777386e-02 -6.13146782e-01 -4.85068679e-01 -1.02111697e+00
-4.69288796e-01 -6.13099933e-01 -3.99443209e-02 1.61470544e+00
2.65208155e-01 -8.39969575e-01 5.64614475e-01 5.51455140e-01
1.09992884e-01 -8.05726171e-01 -1.25278747e+00 -7.79289544e-01
-6.73107803e-02 -2.53192484e-01 3.05830777e-01 1.09065187e+00
-1.41634583e-01 1.10873401e-01 -6.80981100e-01 3.24205995e-01
9.10333812e-01 2.03667760e-01 7.41590738e-01 -1.06861639e+00
-1.32977411e-01 -3.32035810e-01 -4.62630481e-01 -7.66031861e-01
5.89829385e-01 -9.60869908e-01 -6.15613200e-02 -1.32915080e+00
3.77728492e-01 -8.08133066e-01 -5.50691068e-01 3.21824253e-01
-2.84210414e-01 2.97406912e-01 -5.64242266e-02 5.61990380e-01
-4.54858899e-01 1.14449286e+00 1.06980920e+00 -2.63046861e-01
-2.29299024e-01 2.09933534e-01 -6.29123926e-01 5.71000576e-01
6.58936024e-01 -3.91297609e-01 -3.17608207e-01 -2.74021715e-01
-4.05682437e-03 1.57897233e-03 4.50942554e-02 -6.23515964e-01
2.74828434e-01 -3.52777153e-01 3.99481237e-01 -2.18978614e-01
4.64455962e-01 -9.56031501e-01 3.12015861e-01 1.25986263e-01
-1.67908132e-01 -5.12459055e-02 -1.28657818e-01 7.76767671e-01
-4.62223947e-01 -7.75451139e-02 7.97367215e-01 1.32244840e-01
-2.88930714e-01 4.97356206e-01 -1.55818045e-01 -1.85551301e-01
9.17102933e-01 -2.18557909e-01 -1.22123033e-01 -3.59496951e-01
-7.94045568e-01 2.68430978e-01 -5.88264391e-02 2.79426783e-01
9.43972766e-01 -1.50519753e+00 -1.15633595e+00 3.47079515e-01
-9.49899480e-02 -2.49323532e-01 2.68137097e-01 1.42172325e+00
-1.29160672e-01 3.53861719e-01 9.34145451e-02 -5.38894176e-01
-1.21357310e+00 7.18286037e-01 4.98206049e-01 1.03258453e-01
-4.20129120e-01 5.31725287e-01 2.85852909e-01 -6.70513034e-01
2.87021250e-01 2.63388276e-01 -5.24761140e-01 -2.21731290e-02
4.84959126e-01 4.67334330e-01 -2.59404212e-01 -1.05871356e+00
-5.03484368e-01 9.95225906e-01 -3.63646797e-03 1.85000643e-01
1.34171212e+00 -1.14793368e-01 -6.39976203e-01 3.71234596e-01
1.32458210e+00 2.91733772e-01 -7.72243381e-01 -3.61207008e-01
-1.70702741e-01 -5.68397224e-01 2.31214732e-01 -2.78474420e-01
-1.37261212e+00 5.71287572e-01 4.55555022e-01 -1.93470061e-01
1.13890886e+00 -3.48055005e-01 3.77453834e-01 2.52487570e-01
3.61347616e-01 -7.82435775e-01 -3.10653687e-01 1.26590341e-01
1.23037446e+00 -1.35521948e+00 2.16339812e-01 -7.22371042e-01
-5.10770679e-01 1.19034445e+00 6.13339663e-01 -1.51059836e-01
6.57396674e-01 -1.02318443e-01 -6.08526357e-02 9.13640484e-02
-4.14576441e-01 -1.01422509e-02 3.89029324e-01 4.55051750e-01
6.03752196e-01 -1.30138144e-01 -7.10555673e-01 6.45790756e-01
3.24872613e-01 -3.90594117e-02 2.33996496e-01 6.46147549e-01
-1.02980524e-01 -9.91412580e-01 -7.55236924e-01 2.92475611e-01
-6.08038127e-01 1.04366004e-01 -8.52495953e-02 4.83761489e-01
-2.43888889e-02 1.20543802e+00 -3.22511017e-01 -1.84298277e-01
3.09260577e-01 -1.18797667e-01 3.36995900e-01 -4.36681449e-01
-2.49370202e-01 1.80471703e-01 -2.06414536e-01 -4.63263273e-01
-5.12419224e-01 -6.46808565e-01 -1.08061504e+00 -1.66325778e-01
-6.11651182e-01 4.02075857e-01 8.11661899e-01 8.08271289e-01
3.08043778e-01 1.52392641e-01 1.06111979e+00 -5.61221242e-01
-8.63796592e-01 -1.04362929e+00 -6.27968252e-01 2.92513728e-01
2.43118331e-01 -7.26649761e-01 -5.65722585e-01 -1.20253004e-01] | [12.545652389526367, 0.3993406295776367] |
82ba0357-ed4e-4592-9f78-80bd700470fe | adaptive-threshold-selective-self-attention | null | null | https://aclanthology.org/2022.coling-1.157 | https://aclanthology.org/2022.coling-1.157.pdf | Adaptive Threshold Selective Self-Attention for Chinese NER | Recently, Transformer has achieved great success in Chinese named entity recognition (NER) owing to its good parallelism and ability to model long-range dependencies, which utilizes self-attention to encode context. However, the fully connected way of self-attention may scatter the attention distribution and allow some irrelevant character information to be integrated, leading to entity boundaries being misidentified. In this paper, we propose a data-driven Adaptive Threshold Selective Self-Attention (ATSSA) mechanism that aims to dynamically select the most relevant characters to enhance the Transformer architecture for Chinese NER. In ATSSA, the attention score threshold of each query is automatically generated, and characters with attention score higher than the threshold are selected by the query while others are discarded, so as to address irrelevant attention integration. Experiments on four benchmark Chinese NER datasets show that the proposed ATSSA brings 1.68 average F1 score improvements to the baseline model and achieves state-of-the-art performance. | ['Yong Dou', 'Ziwen Zhang', 'Minghao Hu', 'Zhen Huang', 'Biao Hu'] | null | null | null | null | coling-2022-10 | ['chinese-named-entity-recognition'] | ['natural-language-processing'] | [ 4.90954984e-03 -1.84513494e-01 3.21201459e-02 -3.04816425e-01
-7.84913301e-01 -3.91577035e-01 1.12807579e-01 2.71068662e-01
-8.85972679e-01 6.15946949e-01 5.32599151e-01 -2.65125901e-01
2.52583236e-01 -7.20866263e-01 -3.65308523e-01 -7.95152783e-01
2.67754167e-01 2.96171010e-01 5.77142119e-01 -2.06189141e-01
2.80572027e-01 3.30196798e-01 -9.84000623e-01 2.69154161e-01
1.46337187e+00 7.06832826e-01 6.25102639e-01 3.61570507e-01
-7.52087474e-01 7.19480693e-01 -8.72947991e-01 -5.34685135e-01
-3.81990373e-01 -4.45377856e-01 -1.02102053e+00 -4.73321974e-01
-4.38832976e-02 -1.29917478e-02 -3.24822307e-01 1.19216597e+00
7.90681183e-01 1.66206002e-01 2.00904980e-01 -6.30029321e-01
-9.50624943e-01 1.04947567e+00 -3.35043103e-01 6.59439981e-01
-6.80209100e-02 -2.92565465e-01 1.17471445e+00 -1.04564583e+00
2.14562938e-01 9.05351758e-01 4.13641840e-01 7.30162442e-01
-8.95833194e-01 -6.84321046e-01 5.27268589e-01 3.12442809e-01
-1.44531488e+00 -1.55496135e-01 5.56317210e-01 1.89582020e-01
1.23794937e+00 2.89368600e-01 3.41092765e-01 6.65779173e-01
1.03742167e-01 1.07526433e+00 4.67846602e-01 -4.13945377e-01
2.22147301e-01 -1.96501002e-01 4.10560101e-01 4.09250617e-01
1.27565891e-01 -4.61758077e-01 -2.99713343e-01 -5.36872000e-02
5.71749091e-01 -1.29919246e-01 -1.46697864e-01 2.79295623e-01
-9.65881407e-01 5.69365799e-01 6.78566098e-01 7.42776692e-01
-7.50892520e-01 -1.11543298e-01 3.70400488e-01 -3.65754142e-02
3.82787347e-01 3.98845583e-01 -8.20868313e-01 -2.41771996e-01
-5.66122472e-01 -2.96258897e-01 3.76778364e-01 9.92156327e-01
7.02502131e-01 1.48511425e-01 -6.86262310e-01 9.71631825e-01
-3.84933576e-02 3.78962338e-01 6.65001452e-01 -3.55475307e-01
7.56497204e-01 8.98651659e-01 -1.46564851e-02 -6.00068331e-01
-1.97248295e-01 -6.07695997e-01 -9.60535169e-01 -5.75126290e-01
-1.24688474e-02 -4.85533386e-01 -1.17625928e+00 1.63526368e+00
2.49524470e-02 6.06297664e-02 2.25168511e-01 8.58597815e-01
8.03624153e-01 8.88358414e-01 6.49909616e-01 -2.27470994e-01
1.44315970e+00 -9.95824754e-01 -8.49962711e-01 -4.27293301e-01
7.23166287e-01 -6.50579512e-01 1.17714834e+00 -1.58798229e-02
-9.66266394e-01 -5.16990066e-01 -7.00253248e-01 -7.86263794e-02
-3.35914105e-01 2.87359744e-01 3.32971752e-01 4.77632999e-01
-7.93577909e-01 3.25842649e-01 -9.73954916e-01 -1.96985230e-01
3.83602202e-01 4.76315469e-01 -6.94757001e-03 1.41782507e-01
-1.48624396e+00 6.39302433e-01 6.30398333e-01 4.57713187e-01
-4.10995632e-01 -4.14265901e-01 -6.83953583e-01 6.67685032e-01
3.14494342e-01 -2.52641439e-01 1.08493054e+00 -8.25148642e-01
-1.35119629e+00 3.30133021e-01 -5.38914919e-01 -2.88907290e-01
1.09031565e-01 -5.03961384e-01 -6.48420632e-01 -8.74048769e-02
2.33792275e-01 6.28557742e-01 1.49162039e-01 -7.07837105e-01
-7.60530710e-01 -4.08365458e-01 -2.22600132e-01 3.48674506e-01
-8.33095849e-01 3.38968366e-01 -1.00077975e+00 -6.86329305e-01
1.24905817e-01 -7.27431059e-01 -5.04315794e-01 -1.03467798e+00
-4.81850922e-01 -5.55805206e-01 8.66215706e-01 -5.81594527e-01
1.98326457e+00 -2.20092154e+00 1.23307090e-02 -8.61810818e-02
-1.82473227e-01 7.52913475e-01 -3.07111651e-01 2.71075904e-01
2.80630477e-02 5.15583277e-01 -2.04844609e-01 -1.22499496e-01
-3.19066018e-01 1.11385562e-01 -1.23925038e-01 -1.77688763e-01
5.35431445e-01 9.34948027e-01 -9.30368125e-01 -6.91999316e-01
-1.09913595e-01 5.31734467e-01 -3.88804168e-01 3.15609187e-01
-1.25709876e-01 2.96914011e-01 -9.59518373e-01 4.69604254e-01
7.18525052e-01 -2.12542430e-01 1.64891735e-01 -1.40938699e-01
-3.67783159e-01 6.58754349e-01 -8.88458252e-01 1.30097902e+00
-3.18019599e-01 3.32315087e-01 -2.41956308e-01 -4.98926729e-01
1.08528411e+00 3.49770337e-01 8.22305903e-02 -9.92927253e-01
1.27652913e-01 7.02925175e-02 9.60283726e-02 -2.81074107e-01
6.67087972e-01 2.84758687e-01 -2.41538346e-01 1.07632935e-01
-2.50199556e-01 6.31364942e-01 2.53684223e-01 2.22674727e-01
1.11724520e+00 -3.31092551e-02 2.22110990e-02 -2.22866490e-01
9.15231287e-01 5.28288893e-02 1.16257548e+00 6.21725738e-01
-1.69765994e-01 4.20193970e-01 3.44410598e-01 -3.99680406e-01
-8.34939659e-01 -7.59864748e-01 1.71585009e-01 1.39262569e+00
1.30040184e-01 -4.17512119e-01 -1.04748452e+00 -1.11032546e+00
-5.72985590e-01 8.13804746e-01 -4.98304188e-01 -2.59019107e-01
-9.93111730e-01 -8.15812647e-01 6.94699109e-01 9.06841278e-01
7.53710508e-01 -1.68411386e+00 -4.61112142e-01 6.54339314e-01
-5.73970616e-01 -8.97950649e-01 -9.74302173e-01 3.83201271e-01
-8.57429862e-01 -5.78590274e-01 -9.48081791e-01 -1.07151520e+00
8.31555486e-01 1.52835473e-01 8.96291196e-01 1.61936730e-01
1.26544744e-01 -2.94851601e-01 -6.36750758e-01 -2.33553782e-01
-2.52989177e-02 6.76782727e-01 -4.72398371e-01 -8.13787896e-03
8.47855330e-01 -1.50776142e-02 -5.91911256e-01 4.30458635e-01
-8.10695291e-01 -3.06880206e-01 8.40322554e-01 1.03175235e+00
6.63320363e-01 -4.66413284e-03 7.25467205e-01 -8.32853556e-01
6.05877995e-01 -3.32371682e-01 -4.59681690e-01 5.54228604e-01
-5.18921196e-01 3.65136445e-01 9.90146399e-01 -5.21356285e-01
-1.50342369e+00 -4.06510457e-02 -3.05115134e-01 -8.13953951e-02
-5.24961203e-02 6.06622159e-01 -6.40840292e-01 4.02292013e-01
1.65181071e-01 4.22364146e-01 -7.89493918e-01 -7.83488989e-01
-3.50718312e-02 8.23586822e-01 4.37418818e-01 -3.53422046e-01
1.49485767e-01 2.88520800e-03 -6.75887346e-01 -7.01820195e-01
-9.22697723e-01 -4.80641723e-01 -6.01620734e-01 1.89759821e-01
1.10074425e+00 -7.43075311e-01 -3.89905274e-01 5.29601097e-01
-1.21471310e+00 -2.05803037e-01 9.99451280e-02 4.96141165e-01
2.19743684e-01 2.69969493e-01 -9.30693686e-01 -9.57841992e-01
-7.95743465e-01 -1.03473985e+00 7.54898131e-01 8.79806578e-01
7.15916976e-02 -6.53903127e-01 -1.40871614e-01 3.90269645e-02
5.40537417e-01 -5.30719876e-01 9.56755936e-01 -1.00240934e+00
-4.66124773e-01 -5.22707812e-02 -3.11010569e-01 1.23413153e-01
5.27611859e-02 -2.13810340e-01 -8.10311198e-01 -1.14478305e-01
-3.48338157e-01 8.84358436e-02 9.61825252e-01 2.03891650e-01
1.23062122e+00 -1.55317977e-01 -6.05841517e-01 4.85293716e-01
1.30192161e+00 6.22191608e-01 8.79603624e-01 4.85275984e-01
8.18531394e-01 2.58335471e-01 6.83527470e-01 2.28334710e-01
4.45666641e-01 4.16504055e-01 2.99114496e-01 -3.27505887e-01
2.18384773e-01 -3.78940850e-01 3.10944319e-01 9.64521885e-01
2.13807635e-02 -5.78533411e-01 -9.21147525e-01 8.03313255e-01
-1.62800026e+00 -8.43885660e-01 -2.25868732e-01 2.20775914e+00
8.11537266e-01 3.95921946e-01 -1.93025127e-01 -1.93298772e-01
1.23213327e+00 1.01082884e-01 -6.28303409e-01 -4.05587971e-01
-3.40060294e-01 2.07565367e-01 5.53295374e-01 1.99648365e-01
-1.05636096e+00 1.42919934e+00 5.40106678e+00 8.10403407e-01
-1.08095753e+00 6.31002188e-02 7.93465316e-01 3.95818353e-01
-4.16051418e-01 9.00196880e-02 -1.17097843e+00 5.45949578e-01
1.01780784e+00 -1.78160682e-01 1.93851050e-02 7.18184233e-01
1.19374789e-01 1.68060556e-01 -3.66531521e-01 3.87585700e-01
-5.83531745e-02 -9.10500109e-01 2.00087264e-01 -9.50264651e-03
6.54597759e-01 2.66981900e-01 -1.62591591e-01 4.80698675e-01
4.41477269e-01 -5.46053827e-01 3.49419147e-01 2.47029662e-01
6.41354144e-01 -1.06493044e+00 1.05636084e+00 3.24476719e-01
-1.49554849e+00 -1.96984470e-01 -6.07334971e-01 3.58325064e-01
3.02704245e-01 4.10049409e-01 -5.48184693e-01 5.40647924e-01
9.08375740e-01 2.66045988e-01 -5.05311728e-01 1.11981308e+00
-3.93893450e-01 8.58785391e-01 -2.79366255e-01 -4.96878952e-01
4.55719113e-01 -1.04572594e-01 3.97379398e-01 1.46181178e+00
3.21160614e-01 4.00146127e-01 -6.10400513e-02 5.82684457e-01
-1.45081490e-01 4.51563805e-01 1.41692713e-01 -3.00158449e-02
7.90563762e-01 1.07715881e+00 -8.10744822e-01 -4.38280165e-01
-4.52836037e-01 1.29751039e+00 7.15867102e-01 4.20528829e-01
-9.14047301e-01 -9.93955612e-01 3.85173708e-01 -3.01049590e-01
9.93518114e-01 9.31956917e-02 -2.92622298e-01 -1.09575808e+00
-1.50547192e-01 -5.07045090e-01 8.01694691e-01 -6.05265021e-01
-1.14580631e+00 1.07648134e+00 -6.09997213e-01 -8.27561200e-01
2.58692712e-01 -1.80973470e-01 -9.85831082e-01 8.98939550e-01
-1.39919853e+00 -7.63726056e-01 -1.95406169e-01 3.82668167e-01
6.72855794e-01 1.19838253e-01 7.90966868e-01 5.38461447e-01
-1.02135992e+00 9.25209403e-01 -1.52441356e-02 6.76981509e-01
5.64333797e-01 -1.32999277e+00 7.38002300e-01 1.12553585e+00
8.29906240e-02 7.75861621e-01 2.49471799e-01 -8.67288888e-01
-9.73633468e-01 -1.15592647e+00 1.54959345e+00 -3.98611799e-02
2.84066737e-01 -2.14352578e-01 -1.26528370e+00 5.96696079e-01
2.69566774e-01 -2.81343218e-02 5.97337425e-01 4.97502275e-02
-2.01877892e-01 -1.68835551e-01 -8.49055827e-01 7.23652005e-01
9.45511281e-01 -2.77623087e-01 -7.65718162e-01 -3.62935185e-01
8.01467001e-01 -1.63893402e-01 -7.39605069e-01 2.88477093e-01
1.81341916e-01 -7.62956977e-01 6.40085042e-01 -5.21767914e-01
1.24867046e-02 -3.99084121e-01 1.71966806e-01 -1.24449778e+00
-8.29840243e-01 -5.77275753e-01 1.15076192e-01 1.70518649e+00
8.69861066e-01 -4.24936354e-01 7.79345691e-01 5.11539817e-01
-4.15748149e-01 -6.47942245e-01 -8.32476795e-01 -5.64490139e-01
8.34196731e-02 4.38540503e-02 9.36409831e-01 8.41042578e-01
-2.33248388e-03 6.95573926e-01 -1.94801316e-01 3.99556279e-01
2.80991554e-01 -1.18511923e-01 6.92163259e-02 -1.02978730e+00
2.30125993e-01 -4.45517421e-01 1.44243361e-02 -1.38445151e+00
9.63851884e-02 -5.55946290e-01 3.28400940e-01 -1.64007711e+00
2.64850110e-01 -6.01198554e-01 -9.01664376e-01 6.25123501e-01
-8.22340429e-01 6.69731293e-03 6.73624426e-02 3.91222611e-02
-9.32740688e-01 6.80040836e-01 9.72373903e-01 6.29108995e-02
-5.30921638e-01 -5.67876510e-02 -7.57751942e-01 3.99208069e-01
8.74102712e-01 -6.21653318e-01 -2.46493407e-02 -9.04215336e-01
-6.99357241e-02 -4.61410098e-02 -2.87238568e-01 -8.41061234e-01
5.61839640e-01 -8.54981691e-02 5.90537488e-01 -1.02089953e+00
-1.23308972e-01 -6.12477064e-01 -2.30492353e-01 4.76798654e-01
-3.78255486e-01 4.24726009e-01 2.04616085e-01 3.82686824e-01
-2.49270618e-01 -3.04662138e-01 4.72845435e-01 -9.39016119e-02
-9.18065786e-01 3.95211995e-01 -4.32321727e-01 4.18043613e-01
6.70696020e-01 -1.59273315e-02 -2.91149050e-01 6.24683835e-02
-5.87394834e-01 5.37127197e-01 1.69465274e-01 5.48443675e-01
4.76608366e-01 -1.25191462e+00 -7.90124714e-01 2.39161387e-01
1.64635688e-01 -1.72443036e-02 5.48862100e-01 3.92210603e-01
-1.39719382e-01 5.05973458e-01 1.41682193e-01 -2.60616213e-01
-1.14214993e+00 5.32122970e-01 2.19313920e-01 -5.01268744e-01
-6.11085296e-01 9.57678616e-01 2.10189015e-01 -2.19725311e-01
1.35529175e-01 -2.08787650e-01 -5.91999650e-01 -1.66295096e-01
7.84251571e-01 2.85619527e-01 1.66241467e-01 -6.22758210e-01
-6.41863644e-01 5.00864148e-01 -4.08837169e-01 9.55429599e-02
1.21597862e+00 -2.65046686e-01 -2.12009512e-02 -4.99589033e-02
9.23782349e-01 3.25164467e-01 -1.07315147e+00 -3.46193999e-01
3.65296692e-01 -1.19762577e-01 1.89279154e-01 -9.23149943e-01
-1.16980088e+00 8.19552422e-01 3.13906968e-01 2.73589253e-01
1.31311274e+00 1.86773366e-03 1.12592363e+00 4.09274668e-01
1.05379432e-01 -1.17595232e+00 -2.37599075e-01 9.82182443e-01
2.90958017e-01 -8.42092216e-01 -5.25596321e-01 -2.71097481e-01
-9.69539046e-01 8.89191389e-01 1.10566306e+00 1.95843309e-01
2.81207353e-01 4.07458335e-01 1.75499767e-01 1.22853175e-01
-7.66569793e-01 -3.56278747e-01 1.21664196e-01 3.49968195e-01
6.84660137e-01 -1.68278605e-01 -5.35994291e-01 9.29824531e-01
1.73040941e-01 -2.11887613e-01 1.63501412e-01 8.18599582e-01
-5.79304874e-01 -1.21722519e+00 -1.52862728e-01 2.86254138e-01
-9.29044068e-01 -5.24295270e-01 -3.55503410e-01 3.79472017e-01
-3.09949135e-03 9.66678023e-01 3.55886400e-01 -1.48923337e-01
6.09083772e-01 1.41336828e-01 -1.11587368e-01 -5.06837428e-01
-9.94549990e-01 4.00046945e-01 4.93219867e-02 -1.71868846e-01
-4.18511666e-02 -4.88753110e-01 -1.79867101e+00 1.32637486e-01
-7.85393119e-01 8.77147675e-01 2.85709113e-01 7.94451475e-01
7.13974714e-01 6.53921723e-01 6.77638173e-01 -6.81682825e-02
-3.41942221e-01 -1.06249535e+00 -1.43395811e-01 1.61775321e-01
2.69348342e-02 -2.24620193e-01 -2.42208555e-01 -2.85007983e-01] | [9.791481018066406, 9.824719429016113] |
a9906084-a50e-472b-9bed-5de8e3d2d7fd | multi-person-absolute-3d-human-pose | 2004.03989 | null | https://arxiv.org/abs/2004.03989v1 | https://arxiv.org/pdf/2004.03989v1.pdf | Multi-Person Absolute 3D Human Pose Estimation with Weak Depth Supervision | In 3D human pose estimation one of the biggest problems is the lack of large, diverse datasets. This is especially true for multi-person 3D pose estimation, where, to our knowledge, there are only machine generated annotations available for training. To mitigate this issue, we introduce a network that can be trained with additional RGB-D images in a weakly supervised fashion. Due to the existence of cheap sensors, videos with depth maps are widely available, and our method can exploit a large, unannotated dataset. Our algorithm is a monocular, multi-person, absolute pose estimator. We evaluate the algorithm on several benchmarks, showing a consistent improvement in error rates. Also, our model achieves state-of-the-art results on the MuPoTS-3D dataset by a considerable margin. | ['Andras Lorincz', 'Marton Veges'] | 2020-04-08 | null | null | null | null | ['3d-multi-person-pose-estimation-absolute', '3d-multi-person-pose-estimation-root-relative'] | ['computer-vision', 'computer-vision'] | [-2.15673238e-01 6.55541196e-03 -7.06946552e-02 -2.75538713e-01
-7.41311729e-01 -5.41989684e-01 2.94222534e-01 -1.18644647e-01
-7.23109245e-01 8.38027239e-01 4.60225910e-01 4.08580303e-01
3.01765859e-01 -3.92495424e-01 -7.08461821e-01 -2.15660706e-01
5.98821454e-02 8.13256800e-01 3.64442140e-01 -2.41554305e-01
-2.99534589e-01 4.82624143e-01 -1.40851676e+00 -9.85213518e-02
3.60017926e-01 1.06757534e+00 -3.05049717e-01 6.36805952e-01
6.52899981e-01 3.10086608e-01 -5.59176862e-01 -6.28098965e-01
5.79591513e-01 -1.44641042e-01 -6.10983491e-01 3.03090960e-01
8.84612501e-01 -7.71906018e-01 -6.30079746e-01 5.84620416e-01
8.97400260e-01 -3.34803644e-03 4.68991131e-01 -1.28540432e+00
-2.76551545e-02 1.65780485e-02 -6.38973892e-01 -1.81345791e-01
1.09025335e+00 2.57244021e-01 6.16621733e-01 -8.83964062e-01
7.33983457e-01 1.34990525e+00 8.92332971e-01 5.54534137e-01
-1.05949128e+00 -3.76603127e-01 1.13075510e-01 -2.14674458e-01
-1.48928344e+00 -2.66461581e-01 5.91952264e-01 -3.39484543e-01
9.30148423e-01 6.62614331e-02 1.00519073e+00 1.55783224e+00
1.63990520e-02 9.16098595e-01 1.13713980e+00 -2.93639243e-01
4.47961204e-02 8.57224315e-02 -3.81585956e-01 7.34892964e-01
4.55185860e-01 1.15010329e-01 -8.22392881e-01 4.92774919e-02
8.48883271e-01 -1.36250979e-03 -1.88009590e-01 -9.89542782e-01
-1.32812369e+00 4.63959843e-01 6.03265345e-01 -1.91184089e-01
-1.37824342e-01 1.80739239e-01 4.32145566e-01 2.01569721e-01
5.90385973e-01 3.06534886e-01 -6.32749259e-01 -5.12807846e-01
-6.65135562e-01 7.28843808e-01 7.54363835e-01 8.99413347e-01
3.75010610e-01 -5.19182324e-01 1.03541546e-01 4.11595970e-01
2.22697973e-01 7.12104082e-01 1.08048066e-01 -1.03230751e+00
7.49981523e-01 6.57643855e-01 4.93473738e-01 -1.02213168e+00
-7.91019619e-01 -3.62419605e-01 -6.67288899e-01 3.19088668e-01
1.00581086e+00 -2.29519174e-01 -6.43426359e-01 1.70031369e+00
5.29247999e-01 -2.79244363e-01 -2.79322684e-01 1.32683420e+00
6.30275249e-01 2.69124703e-03 -1.48818284e-01 2.13830575e-01
1.19445920e+00 -9.75918293e-01 -3.14143449e-01 -6.92613184e-01
4.27998692e-01 -4.28644270e-01 8.90403032e-01 5.59847474e-01
-7.54643500e-01 -4.90324616e-01 -8.10719848e-01 -1.37160808e-01
-3.62260640e-01 3.23338896e-01 7.64832079e-01 8.40952873e-01
-6.44656301e-01 5.40620744e-01 -9.64364767e-01 -7.24536896e-01
3.86354715e-01 3.24700654e-01 -9.38390791e-01 -2.48601258e-01
-1.07842147e+00 1.23321724e+00 3.46483648e-01 2.55567819e-01
-6.27066612e-01 -3.78818989e-01 -1.05140388e+00 -4.46932346e-01
6.45472944e-01 -8.67167413e-01 1.00624144e+00 -4.41292465e-01
-1.32022929e+00 1.11250663e+00 8.52904469e-02 -4.19715881e-01
1.37450981e+00 -7.47584999e-01 1.50404908e-02 7.80866072e-02
1.12959050e-01 9.02497470e-01 5.88262141e-01 -1.13003051e+00
-3.97683203e-01 -7.94570625e-01 2.32439086e-01 4.10435140e-01
-3.08425307e-01 -2.50327796e-01 -7.03530192e-01 -3.33739161e-01
8.00112858e-02 -1.26229203e+00 -3.68828714e-01 5.04340827e-01
-5.12746274e-01 -2.24238206e-02 5.74501514e-01 -6.80940032e-01
6.80011868e-01 -1.82553339e+00 4.55963045e-01 4.43568965e-03
3.14243942e-01 6.22514561e-02 2.49898493e-01 2.30219066e-01
3.00018102e-01 -3.13726366e-01 -3.39752771e-02 -8.32444191e-01
1.78338066e-01 6.17225580e-02 2.73600310e-01 7.81434298e-01
3.07442644e-03 9.33066785e-01 -8.90314221e-01 -4.72977906e-01
4.17849481e-01 5.57676375e-01 -4.96730983e-01 3.08885157e-01
2.46994998e-02 7.64365911e-01 -2.18939751e-01 8.47071946e-01
5.18696547e-01 -2.81198353e-01 9.44932178e-02 -1.94176629e-01
1.56911403e-01 -4.96361665e-02 -1.55699992e+00 2.35365915e+00
-2.44520634e-01 5.02667367e-01 -2.92590857e-01 -5.10056555e-01
5.67866206e-01 2.69259721e-01 5.35602570e-01 -2.49626070e-01
2.91709721e-01 3.53675187e-01 -4.44632053e-01 -2.03314662e-01
5.87055027e-01 1.58137739e-01 -3.56908649e-01 2.92493075e-01
3.93821895e-02 -7.37698004e-02 1.14744313e-01 3.63419279e-02
1.24486196e+00 6.04982138e-01 4.01944607e-01 1.83283642e-01
1.33069277e-01 -7.69510940e-02 5.41236162e-01 6.48470461e-01
-5.69154203e-01 1.02826762e+00 3.50074440e-01 -7.46931612e-01
-1.12610829e+00 -1.27667594e+00 1.14737796e-02 6.67128503e-01
1.43590689e-01 -5.92896819e-01 -6.91001713e-01 -7.38247097e-01
3.92425179e-01 -1.95215777e-01 -6.73253834e-01 1.09762795e-01
-5.75426221e-01 -4.56458867e-01 5.85643530e-01 7.98043132e-01
6.63797140e-01 -6.05812073e-01 -1.05196559e+00 4.87115309e-02
-2.75479913e-01 -1.71929789e+00 -2.65456289e-01 -7.10925907e-02
-7.18611658e-01 -1.15336967e+00 -1.00068831e+00 -3.27561647e-01
7.16074049e-01 7.56177679e-02 1.24905503e+00 -3.08652282e-01
-3.90019357e-01 4.36277241e-01 -2.30473816e-01 -3.48798752e-01
2.31486991e-01 2.81466275e-01 4.13298905e-01 -3.89694035e-01
4.21832770e-01 -3.08386743e-01 -6.77977979e-01 4.63733435e-01
-2.04877079e-01 -1.06214106e-01 5.51963449e-01 6.57910764e-01
4.59330261e-01 -2.55944669e-01 1.37996048e-01 -6.46933377e-01
1.55521184e-01 -6.67644106e-03 -4.83189940e-01 -1.39595062e-01
-1.30211800e-01 -2.41989136e-01 2.36052349e-01 -3.09956193e-01
-7.99867690e-01 6.19800389e-01 -1.73849557e-02 -3.28715533e-01
-5.37149310e-01 5.72498254e-02 -2.00445890e-01 -2.49030367e-01
9.22484815e-01 -1.88129500e-01 1.07598573e-01 -5.54313362e-01
3.37447822e-01 6.22429430e-01 7.03477919e-01 -4.98092979e-01
9.67909098e-01 7.16023445e-01 1.87718198e-01 -6.75945759e-01
-1.12192786e+00 -6.42444670e-01 -1.13317144e+00 -3.65896672e-01
8.89846444e-01 -1.43739557e+00 -8.72781992e-01 8.87840569e-01
-1.07809508e+00 -3.08446407e-01 -1.00683503e-01 5.84405839e-01
-7.41399527e-01 3.74592662e-01 -6.17686331e-01 -8.22922111e-01
-5.04824184e-02 -9.34649408e-01 1.46989167e+00 3.79648507e-02
-6.12779021e-01 -5.09219408e-01 7.48787597e-02 5.81426322e-01
4.36190031e-02 9.19891000e-01 -5.49115427e-02 -2.37162054e-01
-5.29313743e-01 -6.82013154e-01 -3.42278555e-02 1.45118773e-01
8.58799443e-02 -6.21030390e-01 -9.30300534e-01 -5.41846812e-01
-4.23088729e-01 -8.81010115e-01 6.45453215e-01 1.88980132e-01
9.11494911e-01 2.65219241e-01 -4.47685987e-01 4.44428593e-01
9.91010547e-01 -7.39446402e-01 3.64011914e-01 4.09772366e-01
9.01742339e-01 5.35659909e-01 7.22723246e-01 6.88643396e-01
7.03281045e-01 1.09169793e+00 3.93628240e-01 -9.26492363e-02
-5.76068126e-02 -4.15380120e-01 3.72339040e-02 2.01389134e-01
-4.80079830e-01 -1.59260660e-01 -9.79245424e-01 3.95154148e-01
-1.93442333e+00 -7.42584586e-01 -1.80694889e-02 2.34626603e+00
6.29545152e-01 3.85521740e-01 5.89957654e-01 3.13382179e-01
3.98998260e-01 1.34997502e-01 -5.25990367e-01 4.52583879e-01
-1.18275732e-01 -1.35314520e-02 7.12245405e-01 1.18507080e-01
-1.42086208e+00 7.25164711e-01 6.24744415e+00 1.88129455e-01
-7.15289533e-01 9.73335188e-03 3.31527561e-01 -6.27665818e-01
4.39027458e-01 -3.10042918e-01 -9.15209711e-01 4.19687033e-01
5.46980858e-01 3.81647050e-01 1.23123847e-01 9.78574634e-01
9.61345434e-02 -3.74533534e-01 -1.41030586e+00 1.47655129e+00
2.54695654e-01 -7.58529186e-01 -5.86134911e-01 1.73969939e-01
6.53801262e-01 -5.35117127e-02 -2.49579355e-01 2.10317567e-01
-4.56988737e-02 -1.05193889e+00 8.02719116e-01 3.90313029e-01
7.01224804e-01 -9.30705905e-01 9.06266928e-01 5.27002633e-01
-1.18655157e+00 1.29936352e-01 -3.64242196e-01 -3.23179841e-01
3.63690019e-01 6.70588076e-01 -3.73238951e-01 4.24731463e-01
1.00015342e+00 7.72616267e-01 -7.49902308e-01 1.05370772e+00
-5.87281764e-01 -1.36794910e-01 -7.51658201e-01 -4.48674746e-02
-1.87694132e-01 2.70496309e-01 4.01256740e-01 9.13866162e-01
1.76830024e-01 -1.46594876e-02 6.11766815e-01 4.31288660e-01
-1.15308248e-01 -2.01268658e-01 -5.99182248e-01 3.79313856e-01
3.64755064e-01 1.14742553e+00 -6.17445052e-01 6.90681264e-02
-3.50689262e-01 1.33487356e+00 5.12893319e-01 -1.21396165e-02
-7.79624283e-01 -4.46546637e-02 6.44203663e-01 3.08520824e-01
1.38328478e-01 -6.08382463e-01 -1.59178182e-01 -1.35570753e+00
5.27786136e-01 -9.57588375e-01 3.52372795e-01 -6.36291504e-01
-1.27667058e+00 2.52778500e-01 1.85836926e-01 -1.30934644e+00
-4.73760366e-01 -8.86802554e-01 1.48085043e-01 6.41851723e-01
-1.28050745e+00 -1.31902766e+00 -7.44539618e-01 6.03258669e-01
1.02559403e-01 -8.41204971e-02 8.86613131e-01 3.60488206e-01
-2.46167064e-01 6.68952107e-01 -4.44597572e-01 4.45440710e-01
1.09768176e+00 -1.33500791e+00 6.04335964e-01 7.27785408e-01
8.51464197e-02 4.04708207e-01 7.92127192e-01 -5.45835197e-01
-1.66585219e+00 -7.38593578e-01 8.98712695e-01 -1.15011680e+00
1.58839762e-01 -8.11015606e-01 -3.33305985e-01 8.51386309e-01
-3.89442444e-01 3.65362763e-01 5.21502495e-01 3.47943455e-01
-3.48461568e-01 3.64742093e-02 -1.19942355e+00 4.28041041e-01
1.55228066e+00 -3.69932443e-01 -5.22803068e-01 5.40658832e-01
4.54032063e-01 -1.30977225e+00 -9.44322824e-01 2.72671759e-01
9.17487562e-01 -1.10128820e+00 1.27309096e+00 -2.72766471e-01
3.77313048e-01 -3.03077757e-01 -2.27611497e-01 -1.23822868e+00
4.55014072e-02 -3.03280532e-01 -3.42058659e-01 8.13512266e-01
1.82329059e-01 -4.09413606e-01 1.34359884e+00 8.09373319e-01
3.40109646e-01 -6.00279152e-01 -1.02891636e+00 -1.04309905e+00
-1.88198388e-01 -3.46788198e-01 4.45925921e-01 4.59086448e-01
-1.41985938e-01 1.38737425e-01 -9.13887024e-01 -2.80140005e-02
9.45891619e-01 -9.41377431e-02 1.46160865e+00 -1.28002763e+00
-5.35797119e-01 5.88288642e-02 -8.72521281e-01 -1.20703876e+00
1.29134312e-01 -3.19740772e-01 1.31215602e-01 -1.38819742e+00
1.68671146e-01 -1.68027595e-01 8.17634761e-02 4.02312309e-01
-2.06642553e-01 6.81925476e-01 3.50693017e-01 4.90932874e-02
-8.49002302e-01 5.30792892e-01 1.12044215e+00 9.83038843e-02
5.53879235e-03 3.35759148e-02 -3.98039371e-01 8.66811216e-01
5.00228345e-01 -4.18545783e-01 -1.15014620e-01 -4.43241060e-01
1.84871912e-01 -1.93741560e-01 7.94306517e-01 -1.44621706e+00
1.75601721e-01 1.52160987e-01 1.01833284e+00 -7.00704694e-01
9.00629580e-01 -7.57704198e-01 1.60310581e-01 5.11316419e-01
-6.14796020e-02 3.78354751e-02 -8.55969545e-03 5.56200445e-01
-3.04581076e-02 4.12533641e-01 6.60104036e-01 -5.31383097e-01
-7.52997339e-01 4.83821869e-01 2.13208228e-01 3.75952572e-01
8.81134570e-01 -3.74976128e-01 -7.38086626e-02 -4.60818678e-01
-3.45634729e-01 1.98690578e-01 6.84413791e-01 5.88014185e-01
4.83793497e-01 -1.58455563e+00 -6.78930998e-01 3.87775600e-02
3.90984327e-01 3.68741930e-01 3.21678296e-02 6.84887588e-01
-4.64247912e-01 4.22096610e-01 -4.15998936e-01 -8.10532749e-01
-1.17872572e+00 1.77665994e-01 3.22686404e-01 -2.80579865e-01
-6.78147972e-01 6.82259321e-01 -3.12675327e-01 -8.50605667e-01
3.17383170e-01 -1.34785380e-02 2.86802799e-01 -1.13337904e-01
5.27215838e-01 4.70588773e-01 2.08947267e-02 -8.47108960e-01
-6.91042781e-01 8.03062916e-01 2.43352845e-01 -1.15684837e-01
1.31915641e+00 -1.89992204e-01 2.17179880e-01 3.62885445e-01
1.06716776e+00 -6.07102327e-02 -1.60516918e+00 -2.19903648e-01
-3.12934548e-01 -8.56086612e-01 -4.07247305e-01 -7.21035540e-01
-8.38517606e-01 7.14072049e-01 6.49345279e-01 -3.73294026e-01
7.69119143e-01 3.56414951e-02 8.38873625e-01 6.23057485e-01
9.25450027e-01 -1.32241917e+00 3.30508858e-01 4.81475145e-01
7.51211107e-01 -1.73602426e+00 4.47791696e-01 -5.21449029e-01
-4.53376591e-01 8.96579921e-01 8.02894473e-01 -1.93567649e-01
3.62853974e-01 1.72919229e-01 1.71872914e-01 -4.66031507e-02
-2.41299838e-01 -2.50216246e-01 2.57256657e-01 7.90099919e-01
4.19864982e-01 6.27475157e-02 2.05524801e-03 2.97411233e-01
-4.71272260e-01 8.62718523e-02 2.53914684e-01 1.08503091e+00
-2.55704880e-01 -1.01843667e+00 -5.31223655e-01 8.45708027e-02
-4.01030570e-01 5.49763918e-01 -5.60654461e-01 1.07718897e+00
5.67283444e-02 9.04043913e-01 -2.81763077e-01 -2.99413085e-01
7.66226590e-01 -1.56346843e-01 9.72955942e-01 -4.60282117e-01
-3.53150070e-01 -1.69659615e-01 1.93275452e-01 -1.01853240e+00
-5.00849366e-01 -8.13470602e-01 -8.34018707e-01 -5.74832380e-01
7.13110249e-03 -4.05274749e-01 5.65141559e-01 1.12302256e+00
1.68931589e-01 1.94212675e-01 2.10185051e-01 -1.33356977e+00
-6.41891420e-01 -7.72675872e-01 -5.10858655e-01 6.72081113e-01
1.72754452e-01 -9.60150838e-01 -1.24962375e-01 -3.35026801e-01] | [7.00563383102417, -0.9182350039482117] |
a1370600-71d0-4e5d-b13e-091332a99d89 | spatio-temporal-deformable-convolution-for | null | null | https://www.aiide.org/ojs/index.php/AAAI/article/view/6697 | https://www.aiide.org/ojs/index.php/AAAI/article/download/6697/6551 | Spatio-temporal deformable convolution for compressed video quality enhancement | Recent years have witnessed remarkable success of deep learning methods in quality enhancement for compressed video. To better explore temporal information, existing methods usually estimate optical flow for temporal motion compensation. However, since compressed video could be seriously distorted by various compression artifacts, the estimated optical flow tends to be inaccurate and unreliable, thereby resulting in ineffective quality enhancement. In addition, optical flow estimation for consecutive frames is generally conducted in a pairwise manner, which is computational expensive and inefficient. In this paper, we propose a fast yet effective method for compressed video quality enhancement by incorporating a novel Spatio-Temporal Deformable Fusion (STDF) scheme to aggregate temporal information. Specifically, the proposed STDF takes a target frame along with its neighboring reference frames as input to jointly predict an offset field to deform the spatio-temporal sampling positions of convolution. As a result, complementary information from both target and reference frames can be fused within a single Spatio-Temporal Deformable Convolution (STDC) operation. Extensive experiments show that our method achieves the state-of-the-art performance of compressed video quality enhancement in terms of both accuracy and efficiency. | ['Jianing Deng'] | 2020-04-03 | null | null | null | null | ['video-enhancement', 'video-restoration'] | ['computer-vision', 'computer-vision'] | [ 2.85139054e-01 -7.55433977e-01 5.64143285e-02 -3.48871112e-01
-6.32151246e-01 -3.02076280e-01 2.23664671e-01 -9.59217623e-02
-4.77755010e-01 6.39634967e-01 3.06417972e-01 3.46431062e-02
-1.55282095e-01 -6.99442565e-01 -5.47823012e-01 -7.43253410e-01
1.74923055e-02 -5.39770603e-01 3.01521659e-01 3.18741538e-02
2.28230730e-01 3.57374847e-01 -1.20634270e+00 2.66838789e-01
1.02304423e+00 1.18816984e+00 4.83497322e-01 6.18709266e-01
2.53492624e-01 9.37257349e-01 -3.44912767e-01 -3.14109653e-01
3.35786968e-01 -5.53488851e-01 -6.91963971e-01 2.76020885e-01
4.13461298e-01 -1.03240609e+00 -9.97179449e-01 1.31350076e+00
4.84202981e-01 4.84649867e-01 9.97602567e-02 -9.76063550e-01
-6.96157634e-01 2.70370990e-01 -6.53385043e-01 5.38169026e-01
9.78707820e-02 3.11658978e-01 5.79086483e-01 -7.52554834e-01
4.63158131e-01 1.13805962e+00 3.67652714e-01 4.35492039e-01
-9.11163747e-01 -4.70283955e-01 1.44235581e-01 4.12005126e-01
-1.17429793e+00 -5.71064949e-01 8.64782810e-01 -1.85113370e-01
5.94696343e-01 1.18202038e-01 8.10129642e-01 4.26713139e-01
2.77182877e-01 7.96475947e-01 6.78082526e-01 2.98031941e-02
3.55891176e-02 -4.71531421e-01 -4.82505649e-01 6.04670465e-01
-2.84458939e-02 2.82421619e-01 -3.67419004e-01 2.89145470e-01
1.20583367e+00 3.08167756e-01 -7.50808120e-01 -8.61908495e-02
-1.45260513e+00 3.96840096e-01 5.91952920e-01 3.74588430e-01
-4.72138882e-01 2.61726856e-01 3.81160796e-01 1.82167023e-01
4.41019624e-01 -1.48702143e-02 -1.59061983e-01 -2.89868563e-01
-1.30431736e+00 3.55191201e-01 8.44037086e-02 5.33014894e-01
7.06847131e-01 1.97417393e-01 -3.02447766e-01 7.40618944e-01
3.37144077e-01 3.18282276e-01 3.44281226e-01 -1.51993275e+00
7.02762544e-01 1.33288041e-01 5.21693170e-01 -1.41845441e+00
-3.36027183e-02 -3.75699818e-01 -1.15807903e+00 1.08919151e-01
4.31066364e-01 -4.67303433e-02 -6.20741367e-01 1.47928786e+00
2.21757233e-01 6.60291374e-01 -1.51333332e-01 1.41084027e+00
4.36813980e-01 9.46443617e-01 6.60250187e-02 -6.51067793e-01
9.27599192e-01 -1.03045309e+00 -1.01175272e+00 -1.08341803e-03
3.89506698e-01 -1.01077735e+00 5.02373338e-01 3.19391757e-01
-1.52276170e+00 -8.35216820e-01 -1.09926474e+00 -2.28770003e-01
3.10517937e-01 5.02936468e-02 3.62785816e-01 1.24454074e-01
-9.53546226e-01 1.02284110e+00 -1.10545444e+00 3.43236953e-01
4.95447338e-01 2.88264990e-01 -1.85146630e-01 -3.36481005e-01
-1.25354147e+00 4.45677280e-01 4.74568844e-01 2.37639323e-01
-9.13462579e-01 -7.06779540e-01 -8.43096614e-01 1.83528036e-01
2.64726341e-01 -6.90191984e-01 1.17408156e+00 -9.81543481e-01
-1.38774371e+00 2.70078689e-01 -3.81055325e-01 -3.45722586e-01
6.31712079e-01 -3.09595406e-01 -5.00194192e-01 5.20075917e-01
-4.47444394e-02 6.50820971e-01 9.57062364e-01 -1.06752002e+00
-7.97490478e-01 -1.39636919e-01 1.01506568e-01 3.10082227e-01
-3.50452602e-01 7.65165612e-02 -6.26890600e-01 -1.05709052e+00
2.41058871e-01 -4.05373245e-01 -9.28334221e-02 4.78899747e-01
1.42327547e-01 6.13979213e-02 1.14472234e+00 -1.05391312e+00
1.71173537e+00 -2.08501434e+00 2.11109951e-01 -4.16620612e-01
4.07102555e-01 7.36540556e-01 -1.41082734e-01 -7.26788938e-02
1.38914600e-01 7.73717090e-02 -4.36431348e-01 -2.82465875e-01
-4.34467554e-01 8.09835047e-02 -1.29792199e-01 5.43737054e-01
2.30376065e-01 8.18140984e-01 -1.24117136e+00 -6.59591258e-01
6.60110056e-01 9.87146616e-01 -7.08890259e-01 3.12913150e-01
1.07950397e-01 7.14852691e-01 -5.42923272e-01 5.16516566e-01
9.91603434e-01 -2.11525857e-01 -1.68387577e-01 -6.00055993e-01
-3.84875655e-01 1.58076048e-01 -1.11651146e+00 1.87959743e+00
-4.12595332e-01 6.94550395e-01 7.18727857e-02 -9.61191952e-01
6.47736669e-01 4.92595166e-01 8.75384986e-01 -6.88600838e-01
2.34607771e-01 1.78543672e-01 -8.64883810e-02 -7.26863563e-01
6.70579076e-01 -9.51208249e-02 5.75113714e-01 2.63774931e-01
-2.22450182e-01 -1.81580056e-02 1.51987299e-01 -2.29964256e-02
9.05636609e-01 3.32394838e-01 1.24495946e-01 2.39086628e-01
8.08656037e-01 -5.98100066e-01 9.38850939e-01 3.06666940e-01
-7.12167203e-01 8.36000204e-01 2.48802677e-02 -4.95519578e-01
-1.30060804e+00 -9.26274061e-01 1.55391926e-02 4.20828789e-01
6.19137168e-01 -2.89156348e-01 -7.32953787e-01 -3.79671872e-01
-3.65793109e-01 1.55091435e-01 -1.39764801e-01 -1.67133197e-01
-9.80895579e-01 -4.18184787e-01 2.46573687e-01 4.46457386e-01
1.07564735e+00 -8.81217003e-01 -6.11548901e-01 6.64856315e-01
-6.16145492e-01 -1.23124409e+00 -9.43375051e-01 -6.57953918e-01
-1.10860097e+00 -8.04510772e-01 -1.20482433e+00 -7.59814262e-01
3.64347786e-01 7.76977539e-01 7.12171793e-01 3.36267292e-01
1.67038161e-02 -3.04500133e-01 -3.34135115e-01 4.66700524e-01
-1.72264338e-01 -4.52800512e-01 -1.63005367e-01 3.50560814e-01
-8.46665204e-02 -5.82875252e-01 -1.23786759e+00 3.75990182e-01
-1.26629007e+00 1.37927681e-01 3.19507778e-01 8.78270686e-01
5.60989261e-01 3.48948091e-01 2.95032114e-01 -1.02732830e-01
3.40096116e-01 -2.65439600e-01 -6.20320499e-01 1.68627352e-01
-3.10932606e-01 -1.68888807e-01 7.40322053e-01 -3.55970860e-01
-1.30196261e+00 -1.75146237e-01 -6.83451891e-02 -8.58979702e-01
6.59670830e-02 4.63260263e-01 -1.74270734e-01 -2.08322629e-01
1.42150939e-01 4.97441739e-01 -4.15194929e-02 -4.43388373e-01
-3.29988524e-02 5.70903480e-01 7.24174619e-01 -2.09311098e-01
6.71127856e-01 5.80462217e-01 -8.82076798e-04 -4.83453423e-01
-6.72342241e-01 -3.02611589e-01 -5.08423269e-01 -4.26328748e-01
9.44027126e-01 -8.72866392e-01 -7.92981923e-01 9.14324760e-01
-1.31994116e+00 -1.53134540e-01 1.62146553e-01 8.89691055e-01
-4.02384907e-01 8.75324428e-01 -1.02281356e+00 -5.21769941e-01
-2.47876137e-01 -1.51342833e+00 9.64277327e-01 4.57046360e-01
3.02195102e-01 -1.09854579e+00 -2.10877910e-01 3.98952663e-01
6.61686480e-01 1.29678115e-01 5.15349448e-01 3.78450423e-01
-9.20765162e-01 4.82443683e-02 -6.02904558e-01 6.61191106e-01
4.63520706e-01 -8.46745260e-03 -5.99907398e-01 -3.88092577e-01
2.42054015e-01 2.49821898e-02 8.46558690e-01 6.42997563e-01
1.43724096e+00 -2.44259298e-01 -1.62622277e-02 9.69581902e-01
1.51256061e+00 4.65514213e-01 9.72714961e-01 1.99397862e-01
8.29268813e-01 1.87798098e-01 8.92017782e-01 5.77873111e-01
2.04929218e-01 7.76787996e-01 4.04514521e-01 -6.32937849e-02
-2.03326881e-01 -1.81493163e-02 2.88710922e-01 7.74884045e-01
-3.95207167e-01 -4.21513855e-01 -4.26180631e-01 5.28556764e-01
-1.79640794e+00 -1.33242714e+00 1.25336424e-01 2.27874637e+00
8.70820105e-01 -9.04979780e-02 -2.36599684e-01 3.95657748e-01
9.15932596e-01 4.70225692e-01 -6.02247238e-01 1.02445379e-01
-3.12970392e-02 4.40749265e-02 4.35020298e-01 6.76671922e-01
-1.20120931e+00 6.07748747e-01 5.35498095e+00 8.74903023e-01
-1.33276808e+00 6.42693788e-02 7.20076561e-01 -1.79621398e-01
-3.35662246e-01 -3.47483642e-02 -2.87039101e-01 7.89027870e-01
5.09444118e-01 -2.09397346e-01 6.62018001e-01 2.69886166e-01
8.31037283e-01 -6.18497878e-02 -6.22587025e-01 1.29012179e+00
-1.65675640e-01 -1.46154988e+00 7.94296414e-02 2.78719477e-02
8.97083044e-01 -3.21054935e-01 1.36786178e-01 -1.45572186e-01
-3.49053055e-01 -7.40805387e-01 6.68847859e-01 6.94137037e-01
8.14054310e-01 -8.59779656e-01 6.88773215e-01 1.79045126e-01
-1.37118149e+00 -6.93829805e-02 -3.66970778e-01 -8.31472054e-02
5.81077278e-01 6.39132023e-01 1.01126663e-01 6.33561015e-01
7.58813560e-01 1.36141872e+00 -1.21248506e-01 1.29889441e+00
4.22433801e-02 4.02045369e-01 1.19082630e-01 5.51731348e-01
3.94330680e-01 -2.89718866e-01 5.58340311e-01 8.16010475e-01
6.51997745e-01 5.10577202e-01 7.40588158e-02 6.65387869e-01
-1.60720512e-01 -1.31293446e-01 -8.04295987e-02 4.85610217e-03
2.83679157e-01 1.12882710e+00 -3.40916187e-01 -5.39419293e-01
-6.09843791e-01 1.18174899e+00 -1.67179406e-02 4.89860952e-01
-1.00072312e+00 -4.12747741e-01 7.51840651e-01 1.30486026e-01
3.32331032e-01 -2.85323232e-01 8.94881561e-02 -1.57076371e+00
1.83829159e-01 -6.93181336e-01 6.67408109e-02 -8.32674205e-01
-9.39142048e-01 4.29924726e-01 -2.90525943e-01 -1.78954363e+00
8.13450515e-02 -2.05209762e-01 -4.23726827e-01 9.22446430e-01
-1.73083746e+00 -6.58686578e-01 -5.32589197e-01 7.42574334e-01
7.05718637e-01 1.51503712e-01 1.54621989e-01 8.49914134e-01
-4.78833050e-01 4.61034834e-01 1.32816583e-01 2.24344373e-01
8.84489894e-01 -7.54247606e-01 2.14241445e-01 1.25469494e+00
-4.33981538e-01 3.62997025e-01 5.32577872e-01 -5.89299083e-01
-1.14014864e+00 -1.27634382e+00 7.72318125e-01 4.21879441e-01
4.80898231e-01 4.61927116e-01 -1.19986665e+00 3.39618593e-01
1.68218523e-01 5.01004338e-01 1.50974423e-01 -8.68915319e-01
3.37582156e-02 -3.78500730e-01 -1.16940868e+00 4.53781664e-01
9.90474939e-01 -4.91143644e-01 -2.53512949e-01 4.66897450e-02
7.89358735e-01 -5.17756641e-01 -1.08623433e+00 5.77810943e-01
5.06372929e-01 -1.24715471e+00 8.98124337e-01 -8.46348852e-02
9.15477216e-01 -6.96503758e-01 -1.90055475e-01 -1.01826513e+00
-4.16095555e-01 -7.27493823e-01 -4.05625641e-01 1.02507615e+00
-3.02941889e-01 -4.36784893e-01 5.81950963e-01 5.02412438e-01
-3.27694297e-01 -8.52110326e-01 -8.00186694e-01 -4.83129442e-01
-2.88253367e-01 -3.27414215e-01 5.30629933e-01 9.12310243e-01
-3.45218956e-01 -1.95171461e-01 -7.79732585e-01 2.28174955e-01
7.17246413e-01 -5.74105456e-02 3.06520671e-01 -6.55794144e-01
-4.41018641e-01 -4.10727590e-01 -5.22592604e-01 -1.59529650e+00
-1.13169089e-01 -5.22528052e-01 3.21197324e-02 -1.42590773e+00
1.32938087e-01 -3.30231756e-01 -5.72557867e-01 9.83823091e-02
-5.51894844e-01 4.15554434e-01 2.49772578e-01 3.55023980e-01
-5.13302028e-01 7.47305393e-01 1.93304777e+00 -1.53520703e-01
-1.87708244e-01 -1.28141999e-01 -1.92694977e-01 5.64448357e-01
6.65468037e-01 -1.37580648e-01 -4.03452486e-01 -8.35253656e-01
-1.22725874e-01 7.57892370e-01 5.21979868e-01 -1.10387671e+00
1.46472305e-01 -3.19720536e-01 5.45132875e-01 -5.93850791e-01
3.16950679e-01 -6.19415164e-01 4.23346609e-02 4.51742977e-01
-2.47332290e-01 6.08883500e-02 1.81093905e-02 5.98209977e-01
-7.54696965e-01 -6.36625960e-02 9.54141021e-01 -1.28701270e-01
-7.09448159e-01 7.51410544e-01 -1.35642126e-01 -1.68296501e-01
7.84998238e-01 -2.30906650e-01 -1.72436237e-01 -5.13413787e-01
-5.50354064e-01 7.23031685e-02 5.79367697e-01 2.65850335e-01
9.29267406e-01 -1.41568792e+00 -7.07812071e-01 1.71372727e-01
-4.64462638e-01 -5.62398285e-02 8.53652835e-01 9.86580014e-01
-8.06952417e-01 2.08375409e-01 -2.43161142e-01 -5.76939285e-01
-1.11461830e+00 7.74314940e-01 6.02314293e-01 -2.54508734e-01
-5.21692157e-01 7.31533289e-01 1.04118384e-01 4.14785594e-01
4.46384363e-02 -2.80222207e-01 -6.40573502e-02 -2.49244645e-01
8.85930777e-01 5.42114198e-01 -1.44007236e-01 -7.15435684e-01
-7.65181929e-02 7.09661782e-01 -5.20358421e-02 -4.32456806e-02
1.12405276e+00 -5.75963140e-01 -8.98820534e-02 -1.49761876e-02
1.54670107e+00 -1.49196446e-01 -1.88629782e+00 -3.08102071e-01
-5.34478545e-01 -1.10746801e+00 3.96870583e-01 -4.13500249e-01
-1.73503649e+00 1.03771722e+00 6.84375703e-01 2.04804000e-02
1.62344980e+00 -5.81571579e-01 1.46683669e+00 -2.08736002e-01
2.02895328e-01 -8.62051904e-01 9.52054113e-02 1.69450775e-01
5.86793244e-01 -1.34441805e+00 1.44463062e-01 -4.43658561e-01
-3.43078554e-01 1.28894842e+00 4.15945441e-01 -1.50796875e-01
5.59270263e-01 4.17838842e-02 -6.66219667e-02 3.41389298e-01
-6.15310669e-01 9.96341482e-02 3.07202190e-01 2.86025286e-01
5.50048470e-01 -1.97434127e-01 -5.04658401e-01 -4.51029278e-02
3.70018899e-01 3.35871220e-01 3.75201315e-01 8.96465898e-01
-3.55310261e-01 -1.05109298e+00 -2.69226044e-01 1.99774086e-01
-6.60003364e-01 -1.49828851e-01 4.68733579e-01 3.93278986e-01
2.20737532e-01 1.03436470e+00 6.88072890e-02 -3.56743038e-01
-5.77712059e-03 -5.20313323e-01 5.93942702e-01 2.96027493e-03
-3.51256639e-01 5.42844772e-01 -3.93442512e-01 -7.72190034e-01
-7.73888648e-01 -6.41237855e-01 -1.35206676e+00 -4.58985716e-01
9.64572001e-03 -1.43913463e-01 2.32616529e-01 9.71700847e-01
2.26667389e-01 6.60984874e-01 8.92769873e-01 -1.06642532e+00
-3.01312923e-01 -7.12841213e-01 -4.98915344e-01 5.99112272e-01
7.28335798e-01 -5.48412204e-01 -4.28736925e-01 3.86442542e-01] | [11.124772071838379, -1.771977424621582] |
8b443e8a-b1e4-4db8-b9ed-b7402ae9ff8a | nnrepair-constraint-based-repair-of-neural | 2103.12535 | null | https://arxiv.org/abs/2103.12535v2 | https://arxiv.org/pdf/2103.12535v2.pdf | NNrepair: Constraint-based Repair of Neural Network Classifiers | We present NNrepair, a constraint-based technique for repairing neural network classifiers. The technique aims to fix the logic of the network at an intermediate layer or at the last layer. NNrepair first uses fault localization to find potentially faulty network parameters (such as the weights) and then performs repair using constraint solving to apply small modifications to the parameters to remedy the defects. We present novel strategies to enable precise yet efficient repair such as inferring correctness specifications to act as oracles for intermediate layer repair, and generation of experts for each class. We demonstrate the technique in the context of three different scenarios: (1) Improving the overall accuracy of a model, (2) Fixing security vulnerabilities caused by poisoning of training data and (3) Improving the robustness of the network against adversarial attacks. Our evaluation on MNIST and CIFAR-10 models shows that NNrepair can improve the accuracy by 45.56 percentage points on poisoned data and 10.40 percentage points on adversarial data. NNrepair also provides small improvement in the overall accuracy of models, without requiring new data or re-training. | ['Corina Pasareanu', 'Yannic Noller', 'Youcheng Sun', 'Divya Gopinath', 'Muhammad Usman'] | 2021-03-23 | null | null | null | null | ['fault-localization'] | ['computer-code'] | [ 6.40173778e-02 4.80373204e-01 -2.14434434e-02 -2.89644539e-01
-6.35456979e-01 -1.15235579e+00 1.10820509e-01 1.18034072e-01
-2.99258471e-01 7.73447692e-01 -2.59660482e-01 -8.01358283e-01
-1.79701731e-01 -6.68045163e-01 -1.31091428e+00 -6.23596847e-01
-1.45244211e-01 6.32819161e-02 4.22518790e-01 7.11937100e-02
3.39910150e-01 8.13755453e-01 -1.08146405e+00 3.51870954e-01
6.32049561e-01 1.00714040e+00 -7.39851356e-01 8.27633202e-01
4.69126999e-01 9.19495463e-01 -1.19658875e+00 -7.95890868e-01
4.82192785e-01 4.22307774e-02 -7.64501154e-01 -3.59642118e-01
5.70752680e-01 -3.39174777e-01 -3.82207304e-01 1.38519943e+00
2.35933408e-01 -9.35259685e-02 2.75143057e-01 -1.64862335e+00
-4.79348063e-01 1.04168642e+00 -5.10088839e-02 3.51561278e-01
-7.80395195e-02 3.31937015e-01 8.43602836e-01 -4.25379574e-01
1.63276911e-01 1.09244430e+00 8.86268377e-01 9.15041387e-01
-1.22687531e+00 -8.74341488e-01 1.50763869e-01 -3.88727821e-02
-1.45829856e+00 -6.52242541e-01 5.43161273e-01 -1.94728523e-01
1.25084925e+00 3.77889663e-01 -5.83040267e-02 1.10224605e+00
4.78551716e-01 1.44933239e-01 5.28353930e-01 -2.92577595e-01
6.54998124e-01 2.23236382e-01 4.08474535e-01 1.01842356e+00
5.01683235e-01 2.87695676e-01 -4.34760004e-01 -7.76295006e-01
4.96079385e-01 -1.32098094e-01 -2.78916389e-01 2.93118954e-01
-5.97127020e-01 7.23629296e-01 6.40393198e-01 -1.03994273e-01
-2.15562105e-01 5.39803803e-01 5.22681475e-01 5.15982389e-01
2.37268344e-01 9.29748535e-01 -9.60038364e-01 4.49073404e-01
-4.99916375e-01 2.13638887e-01 9.91227984e-01 6.01711154e-01
3.31250489e-01 5.50776005e-01 -6.82796687e-02 4.24613655e-01
1.54296219e-01 2.02219576e-01 6.29831254e-02 -1.06503618e+00
7.60887086e-01 4.56670314e-01 -5.19827008e-02 -8.10634613e-01
-1.68915480e-01 -5.89655519e-01 -5.12026012e-01 4.10738945e-01
5.04658043e-01 -7.49965489e-01 -1.12144899e+00 1.94154835e+00
1.89495459e-02 7.02774405e-01 2.23533839e-01 4.24356192e-01
2.74167478e-01 5.16344726e-01 1.05052747e-01 1.94952413e-01
9.69095528e-01 -6.77362323e-01 -1.28801882e-01 -2.76205331e-01
7.51318753e-01 -7.85239562e-02 6.87654197e-01 7.10214615e-01
-1.08502007e+00 -8.63002762e-02 -1.34317577e+00 4.77322519e-01
-5.27140021e-01 -3.78270894e-01 4.35647786e-01 1.10840285e+00
-1.00215995e+00 7.40995407e-01 -7.05339491e-01 3.76741201e-01
6.51365280e-01 6.82978630e-01 -5.24986863e-01 -1.83289424e-01
-1.29514408e+00 9.10852134e-01 4.29461062e-01 3.40509228e-02
-1.42058945e+00 -1.17464387e+00 -7.76695669e-01 4.58801448e-01
2.83223629e-01 -5.51031828e-01 1.07225597e+00 -7.17840731e-01
-9.14907515e-01 1.97290644e-01 2.53570616e-01 -1.01102984e+00
2.40836799e-01 -7.47174695e-02 -4.42350805e-01 5.78749068e-02
-4.94879276e-01 5.09002745e-01 8.01092148e-01 -1.20496213e+00
-4.94134724e-01 -1.13112725e-01 3.71609598e-01 -2.79093236e-01
-4.82663095e-01 7.45273530e-02 5.21516576e-02 -6.02997243e-01
4.30134758e-02 -8.16346049e-01 -1.98002398e-01 -1.17451707e-02
-9.49480116e-01 3.59597430e-02 6.32398546e-01 -7.42667735e-01
9.84173357e-01 -2.08420825e+00 -2.36307140e-02 7.01780319e-01
2.06564680e-01 6.15834832e-01 -1.14258081e-01 -9.84134227e-02
-5.44106901e-01 6.76259220e-01 -3.20569038e-01 -2.15481400e-01
-1.54732496e-01 3.97236586e-01 -7.18576670e-01 3.81156206e-01
3.33450854e-01 6.78447247e-01 -4.40411508e-01 1.85973525e-01
-2.79005289e-01 3.63061488e-01 -8.55504751e-01 9.68792960e-02
-4.41340059e-01 -8.52364749e-02 -1.59731254e-01 7.93223381e-01
4.76191014e-01 2.22533882e-01 3.02119888e-02 -2.84966696e-02
5.43710947e-01 2.68380761e-01 -1.07087481e+00 8.02636504e-01
-2.49476120e-01 3.86343539e-01 1.26649603e-01 -7.63348520e-01
7.63725460e-01 5.27808487e-01 -1.48741454e-01 -2.88125724e-01
3.06129962e-01 -1.00973598e-03 1.65687352e-02 -2.64151245e-01
-5.23693003e-02 7.74876550e-02 -3.03677052e-01 5.22886992e-01
6.06675297e-02 2.72970706e-01 -3.36958766e-01 2.26363078e-01
1.68395638e+00 -8.49434137e-01 -2.55177736e-01 -1.69812232e-01
3.09128612e-01 -2.18639016e-01 8.32272112e-01 8.68086994e-01
-1.96778998e-02 2.07765400e-01 1.01887274e+00 -4.71879601e-01
-9.27584887e-01 -1.06978285e+00 3.07084322e-02 6.40941679e-01
-4.77918655e-01 -1.36910871e-01 -1.12858081e+00 -1.27084744e+00
3.63711476e-01 1.00458848e+00 -7.85919309e-01 -9.35995936e-01
-2.41495863e-01 -8.10835004e-01 1.37238228e+00 7.48716235e-01
5.69745123e-01 -7.71913528e-01 -1.35971218e-01 -1.53830960e-01
2.05439448e-01 -8.53438020e-01 -3.22925299e-01 6.62859321e-01
-8.10573459e-01 -1.17766821e+00 1.61980525e-01 -5.68681836e-01
1.20030951e+00 -2.72526324e-01 6.87038958e-01 8.89486194e-01
-3.52087617e-02 -7.82504305e-02 -2.71421694e-03 -3.77515674e-01
-5.42198241e-01 -5.17628677e-02 2.99817771e-01 -1.85398087e-01
-1.02377377e-01 -5.69995403e-01 7.88673013e-02 3.45957249e-01
-9.45005357e-01 -7.40858436e-01 1.85087785e-01 7.26222754e-01
2.25196123e-01 7.83254385e-01 8.12151015e-01 -1.35529137e+00
6.83255792e-01 -7.64128983e-01 -8.04021239e-01 5.15248001e-01
-8.02331924e-01 1.77867368e-01 8.93213689e-01 -7.59697258e-01
-6.00629151e-01 4.04244736e-02 -3.04382801e-01 -7.71288514e-01
-2.31372099e-03 4.92191941e-01 -4.09958690e-01 -5.76967776e-01
1.20098257e+00 -3.86970222e-01 -4.25939828e-01 -4.13645834e-01
1.34678371e-02 2.50896037e-01 9.49653029e-01 -7.49675512e-01
1.07243991e+00 -1.63073137e-01 1.94355454e-02 -1.47179309e-02
-8.18156719e-01 3.26534420e-01 -3.03343415e-01 1.25517413e-01
2.78122246e-01 -6.60446644e-01 -7.88152099e-01 7.51301885e-01
-1.08661842e+00 -6.24542594e-01 -1.14440039e-01 -9.97393951e-02
1.90024137e-01 -1.02637075e-01 -8.04874539e-01 -5.76310277e-01
-4.53091919e-01 -1.03261518e+00 1.92252815e-01 2.46786967e-01
5.56406640e-02 -1.06393194e+00 -3.69396508e-01 2.62084901e-01
3.39367211e-01 3.80575120e-01 1.34904754e+00 -1.25185740e+00
-3.91821653e-01 -6.74749911e-01 -1.54692016e-03 8.08203757e-01
-2.15583459e-01 6.78976029e-02 -1.28743565e+00 -2.99811929e-01
2.53329761e-02 -3.85308892e-01 6.13203108e-01 -3.67415771e-02
1.41664743e+00 -8.42408717e-01 -2.13730812e-01 6.89192951e-01
1.33447373e+00 3.41561943e-01 8.78449500e-01 -2.05378719e-02
7.53334463e-01 2.67398298e-01 7.73157105e-02 1.43531606e-01
-3.87726501e-02 2.13828459e-01 8.71613681e-01 2.30944961e-01
1.88838884e-01 -2.80655921e-01 4.36837465e-01 -5.13185300e-02
5.41989863e-01 -3.98042679e-01 -9.50527906e-01 3.83089453e-01
-1.57157981e+00 -7.61974216e-01 2.04317451e-01 2.12705040e+00
9.38238263e-01 6.90959930e-01 -1.59925774e-01 5.04971027e-01
5.81125081e-01 -3.46763223e-01 -9.26080644e-01 -6.86025918e-01
1.67839155e-01 2.24032179e-01 8.70220900e-01 8.08587670e-01
-8.03082466e-01 9.57293034e-01 7.12067604e+00 3.90866935e-01
-9.48248684e-01 1.73338458e-01 7.49412119e-01 -2.86823153e-01
-1.50984868e-01 1.42082930e-01 -8.94748807e-01 4.35625345e-01
1.28518665e+00 7.85193220e-02 8.36965859e-01 8.59129786e-01
-3.49079281e-01 1.28798798e-01 -1.11268580e+00 3.74792486e-01
2.47001454e-01 -1.50464833e+00 -1.19354114e-01 -1.20065793e-01
7.22261488e-01 -8.27916712e-02 -2.97084060e-02 3.23332936e-01
6.62097991e-01 -1.13861406e+00 8.33742380e-01 7.08003342e-01
5.08813262e-01 -1.44928217e+00 9.04639602e-01 4.62913513e-01
-4.48307633e-01 -5.65742671e-01 -3.68264169e-01 2.58058533e-02
-3.06080520e-01 6.48656964e-01 -1.03726804e+00 -2.69936416e-02
6.57092392e-01 2.59514283e-02 -9.14325058e-01 9.58306909e-01
-6.29802644e-01 9.52485561e-01 -3.76668096e-01 3.67009670e-01
-1.51382908e-01 7.16751695e-01 3.29274356e-01 8.82404923e-01
-1.94624603e-01 -2.64213495e-02 2.53873579e-02 8.59603047e-01
-6.12354100e-01 -8.97986531e-01 -4.02422100e-01 1.16917945e-01
1.10509133e+00 1.04902339e+00 -5.49584270e-01 -2.43929133e-01
2.58013219e-01 5.77703595e-01 4.61721539e-01 4.71258402e-01
-1.04537308e+00 -7.04446912e-01 7.98796833e-01 -1.57637373e-01
1.32657722e-01 -2.03619730e-02 -6.59587681e-01 -7.95778751e-01
-1.36465504e-04 -1.07527447e+00 4.69134480e-01 -7.88405299e-01
-1.06356263e+00 5.42455077e-01 -5.15682176e-02 -3.01428825e-01
-1.66962385e-01 -5.66008151e-01 -1.04203868e+00 9.34556603e-01
-9.57822263e-01 -7.01491714e-01 1.25834182e-01 6.29573345e-01
6.63695857e-02 -4.18478340e-01 9.02259231e-01 2.41309404e-01
-1.26923156e+00 1.38026667e+00 -4.42861766e-01 3.12859148e-01
3.56180459e-01 -1.03970969e+00 7.30272651e-01 1.47199690e+00
2.78558093e-03 8.94048452e-01 6.93234444e-01 -7.74145305e-01
-1.09660840e+00 -1.31178105e+00 7.97779143e-01 -5.88177621e-01
6.40572906e-01 -6.34554029e-01 -1.23729420e+00 9.06255305e-01
-2.67613500e-01 1.92854777e-01 4.02981102e-01 -8.54704082e-02
-9.99862790e-01 -1.79514572e-01 -1.99865830e+00 5.77748775e-01
6.96708143e-01 -6.63938642e-01 -2.35679463e-01 4.43134010e-01
1.13660192e+00 -5.73915362e-01 -9.31711316e-01 2.36311778e-01
1.80263907e-01 -5.34086883e-01 9.32046056e-01 -9.13395226e-01
4.22147885e-02 -2.88686872e-01 -3.76553684e-01 -1.33005941e+00
-1.10304542e-01 -5.80302775e-01 -4.69544679e-01 1.40711296e+00
8.99242878e-01 -9.88184094e-01 7.71337569e-01 1.25384283e+00
-2.77172118e-01 -7.74487436e-01 -8.41310978e-01 -5.83434939e-01
1.23310223e-01 -4.74623859e-01 9.47917461e-01 1.07002723e+00
-9.37532037e-02 -1.09695904e-01 -2.02876791e-01 9.43638802e-01
5.64615250e-01 -8.97290230e-01 3.26823086e-01 -9.98950601e-01
-5.09823799e-01 -1.19485646e-01 -2.72281736e-01 -7.62949660e-02
5.81694603e-01 -8.23626995e-01 8.14529955e-02 -8.78305793e-01
-2.74473071e-01 -6.78867400e-01 -4.26734090e-01 1.50092375e+00
-2.03271210e-02 3.45099956e-01 -1.25559783e-02 -1.46075562e-01
-1.99036255e-01 -2.40892112e-01 2.55577654e-01 -9.38298702e-02
1.53809711e-01 1.83974102e-01 -1.18455815e+00 5.28604805e-01
1.12743092e+00 -1.00006104e+00 -5.74897826e-01 -5.48312008e-01
4.69789356e-01 -5.37478030e-02 8.94705653e-01 -1.20473063e+00
6.13905966e-01 3.51489000e-02 5.30623734e-01 4.65954877e-02
2.57023815e-02 -8.82674754e-01 1.12661004e-01 7.92280555e-01
-5.60402811e-01 3.14189374e-01 4.98169661e-01 5.69701672e-01
3.67320299e-01 -7.29097068e-01 9.13335681e-01 2.22265795e-01
-1.92074686e-01 9.00214389e-02 -3.84239793e-01 -8.44335929e-02
1.01435232e+00 3.05002987e-01 -9.97134984e-01 8.11013207e-02
-7.99546421e-01 3.39916706e-01 3.72799277e-01 3.31723839e-01
6.80901170e-01 -8.72177899e-01 -3.16903830e-01 6.47843003e-01
-3.71661305e-01 1.87195372e-02 -5.44090476e-03 1.91596821e-01
-3.63637775e-01 1.62196413e-01 -9.84117985e-02 2.95870262e-03
-1.37392247e+00 6.42326534e-01 8.79453540e-01 -1.48106301e-02
-2.84717411e-01 1.31449950e+00 -5.12632668e-01 -3.97870034e-01
8.54065478e-01 -3.68934333e-01 2.78624475e-01 -5.00511944e-01
6.06647968e-01 2.92355508e-01 5.12094617e-01 2.62531769e-02
-4.61853564e-01 -1.47526324e-01 -3.56275231e-01 -7.09459186e-02
1.10288239e+00 5.17145932e-01 -1.73565879e-01 -9.30575356e-02
9.53374386e-01 -1.18942656e-01 -1.41371560e+00 -4.99574915e-02
2.90093701e-02 -1.11177653e-01 3.64222944e-01 -1.47911239e+00
-1.57521725e+00 5.90723336e-01 2.30639577e-01 1.64792314e-01
1.10467243e+00 -3.13007087e-01 6.38009727e-01 7.26729691e-01
4.67711717e-01 -5.60565054e-01 -3.86227816e-02 5.03176630e-01
6.23686969e-01 -4.72386986e-01 -7.98821822e-02 -2.27593705e-01
-3.79638821e-01 7.54852593e-01 8.02745640e-01 -3.79870594e-01
7.63542712e-01 6.17064059e-01 -1.77559271e-01 -1.46269053e-01
-1.23590696e+00 8.36602211e-01 3.76820713e-02 6.09303772e-01
-3.17521840e-01 -1.26994818e-01 6.08089864e-01 8.83434236e-01
-1.66492984e-01 -4.62628514e-01 6.54285491e-01 9.67371285e-01
-4.32895631e-01 -8.25621963e-01 -6.05831981e-01 5.98609686e-01
-7.44130790e-01 -1.11055337e-01 -6.74108207e-01 4.23295110e-01
2.56019115e-01 1.17710149e+00 -1.75921455e-01 -8.85643780e-01
5.84630787e-01 4.13508922e-01 3.65753978e-01 -5.77744424e-01
-1.17853880e+00 -6.54428244e-01 3.73422176e-01 -5.05026698e-01
6.28194988e-01 -4.64009196e-01 -1.36694717e+00 -6.99757338e-01
-5.01257658e-01 2.12417021e-01 8.60436082e-01 1.05212879e+00
4.75553662e-01 7.75207520e-01 6.85750484e-01 -5.41599095e-01
-8.29150915e-01 -5.66689372e-01 -4.67849702e-01 -1.18787624e-01
3.76296580e-01 -5.22909939e-01 -8.60750437e-01 -1.69826925e-01] | [5.975515365600586, 7.706060886383057] |
57767812-63bb-417b-a117-fc21de173e6a | long-tail-prediction-uncertainty-aware | 2207.00788 | null | https://arxiv.org/abs/2207.00788v2 | https://arxiv.org/pdf/2207.00788v2.pdf | Long-Tail Prediction Uncertainty Aware Trajectory Planning for Self-driving Vehicles | A typical trajectory planner of autonomous driving commonly relies on predicting the future behavior of surrounding obstacles. Recently, deep learning technology has been widely adopted to design prediction models due to their impressive performance. However, such models may fail in the "long-tail" driving cases where the training data is sparse or unavailable, leading to planner failures. To this end, this work proposes a trajectory planner to consider the prediction model uncertainty arising from insufficient data for safer performance. Firstly, an ensemble network structure estimates the prediction model's uncertainty due to insufficient training data. Then a trajectory planner is designed to consider the worst-case arising from prediction uncertainty. The results show that the proposed method can improve the safety of trajectory planning under the prediction uncertainty caused by insufficient data. At the same time, with sufficient data, the framework will not lead to overly conservative results. This technology helps to improve the safety and reliability of autonomous vehicles under the long-tail data distribution of the real world. | ['Kun Jiang', 'Yunkang Xu', 'Diange Yang', 'Xiaoyu Liu', 'Nanshan Deng', 'Zhong Cao', 'Weitao Zhou'] | 2022-07-02 | null | null | null | null | ['trajectory-planning'] | ['robots'] | [-3.43636036e-01 2.26798967e-01 -2.71685570e-01 -4.30758864e-01
-4.54347372e-01 -2.63647661e-02 5.28694868e-01 -2.04069912e-01
-2.08865508e-01 7.95310616e-01 1.54543787e-01 -6.47510946e-01
-2.28096500e-01 -1.05774164e+00 -7.71144271e-01 -7.22782254e-01
-7.27878660e-02 2.07908005e-01 3.05227876e-01 -2.65140831e-01
1.71446681e-01 3.28820765e-01 -1.81763804e+00 -1.57492802e-01
1.28519523e+00 1.11631989e+00 3.65953833e-01 6.88558221e-02
-2.94087768e-01 4.86646354e-01 -1.48444593e-01 2.06655059e-02
2.75437981e-01 1.72617242e-01 -2.49858156e-01 -2.78998554e-01
-4.20449585e-01 -6.67412877e-01 -7.43949771e-01 1.01887870e+00
2.57652193e-01 2.47803345e-01 6.16232753e-01 -1.78860307e+00
7.42214024e-02 3.83090824e-01 -2.77256429e-01 -7.38224164e-02
-2.46491730e-01 4.30171192e-01 3.25647920e-01 -5.31962931e-01
1.88233986e-01 1.17142689e+00 7.07646906e-01 4.96447563e-01
-6.09443069e-01 -7.67280936e-01 4.04809415e-01 5.96535027e-01
-1.28606522e+00 -3.06590080e-01 7.30302870e-01 -4.69813496e-01
6.19717479e-01 -9.94651243e-02 3.72975349e-01 1.08572650e+00
8.03111553e-01 7.00914681e-01 5.51814556e-01 3.67472827e-01
3.45275521e-01 9.43399295e-02 -3.11925616e-02 1.70412928e-01
2.14711115e-01 8.32619071e-01 4.59450781e-02 2.35676199e-01
-1.50854066e-01 1.01913981e-01 1.11735135e-01 -6.05289638e-02
-7.10051179e-01 6.56132400e-01 4.67070997e-01 -1.38600186e-01
-4.85754311e-01 2.45033158e-03 4.61609364e-01 1.33777976e-01
1.90036714e-01 -4.12376272e-03 -2.54993081e-01 -2.75974333e-01
-7.98339188e-01 5.54095984e-01 5.38570046e-01 1.27279580e+00
6.56592727e-01 4.32651669e-01 -1.09919712e-01 3.37099642e-01
4.35983330e-01 6.44330919e-01 9.04159918e-02 -1.02626395e+00
7.71484554e-01 4.78353620e-01 6.73483133e-01 -1.11199701e+00
-8.38614106e-01 -4.61004317e-01 -9.30772543e-01 4.76947337e-01
3.15533817e-01 -5.37132680e-01 -8.07152390e-01 1.67423439e+00
2.53061622e-01 3.07366818e-01 2.39650905e-01 8.50496113e-01
4.22255874e-01 9.05920267e-01 1.15708143e-01 -1.47987023e-01
8.40682924e-01 -6.33045077e-01 -9.54554915e-01 -3.32387835e-01
7.43316412e-01 -3.80744308e-01 6.41673803e-01 3.74741137e-01
-5.25264859e-01 -7.40677536e-01 -1.11984503e+00 4.73368406e-01
-3.92377883e-01 -3.34853679e-02 2.12327674e-01 5.09004951e-01
-5.21061599e-01 5.67669690e-01 -7.09217727e-01 -1.28913373e-01
5.44737339e-01 2.46419191e-01 -6.35195598e-02 -3.21502894e-01
-1.47254109e+00 1.21966004e+00 6.95561647e-01 4.38991368e-01
-1.07610250e+00 -7.91646302e-01 -5.60452998e-01 -2.81270687e-02
5.58224559e-01 -3.08130562e-01 1.15610957e+00 -4.58517075e-01
-1.23435652e+00 -3.25145632e-01 -7.54433051e-02 -5.16353130e-01
8.39274228e-01 -1.96888700e-01 -7.16440618e-01 -5.90547562e-01
6.19126484e-03 3.75479281e-01 4.86287713e-01 -1.25621974e+00
-1.13290060e+00 -2.03379542e-01 -8.68891701e-02 1.60794407e-01
3.77129987e-02 -5.85058987e-01 -5.27430996e-02 -3.65130268e-02
-6.44144788e-02 -1.12784529e+00 -5.60246110e-01 -3.80338311e-01
-3.02696109e-01 -3.96445125e-01 1.12144482e+00 -5.42654157e-01
1.26616180e+00 -2.23066449e+00 -4.61846679e-01 2.08339587e-01
-1.63974926e-01 2.08901748e-01 1.05633633e-02 6.25563025e-01
2.64649540e-01 4.58117248e-03 -1.38983324e-01 1.60639193e-02
1.96184963e-02 5.79824090e-01 -5.47124326e-01 3.67746741e-01
2.02836186e-01 4.84536439e-01 -7.88569033e-01 -2.80721754e-01
3.54418933e-01 1.50799438e-01 -2.28510767e-01 3.10166299e-01
-4.24594849e-01 5.73097289e-01 -9.51071560e-01 3.09786558e-01
1.03024781e+00 4.15068179e-01 -2.57875502e-01 -1.11898854e-02
-6.21203780e-01 -7.76099786e-02 -1.09916830e+00 1.24383509e+00
-5.30043840e-01 5.33268631e-01 5.31585291e-02 -9.46668983e-01
1.18692958e+00 1.99448496e-01 5.81472397e-01 -8.52328956e-01
3.23199809e-01 4.40665781e-01 1.83038756e-01 -8.21940005e-01
5.59287727e-01 -6.64460883e-02 -6.89013228e-02 4.73408028e-02
-6.79219723e-01 -7.19689718e-03 -2.12849766e-01 -8.57921243e-02
1.02032447e+00 2.46620968e-01 -1.58653259e-01 -2.01299965e-01
5.93677640e-01 1.68802187e-01 1.10695982e+00 3.97432387e-01
-5.00133753e-01 8.61158893e-02 4.05586898e-01 -6.40104651e-01
-1.09878683e+00 -6.62981212e-01 -2.25166991e-01 4.13166523e-01
5.24464607e-01 2.91056484e-02 -4.15447831e-01 -5.53982317e-01
2.41107956e-01 1.17222655e+00 -3.39196265e-01 -5.92198074e-01
-5.02432942e-01 -6.47840381e-01 4.26415771e-01 5.41159153e-01
5.88077247e-01 -7.96261966e-01 -5.68342447e-01 4.45644706e-01
-1.37638062e-01 -9.33111012e-01 -1.02971926e-01 -1.90115765e-01
-5.27687371e-01 -9.98042822e-01 -2.73909211e-01 -2.96201795e-01
4.68996346e-01 3.07174146e-01 3.72710854e-01 1.15569323e-01
2.87128985e-01 -1.51199788e-01 -1.87019274e-01 -7.66974211e-01
-4.49904025e-01 -7.02984035e-02 2.74962902e-01 -5.20031154e-03
3.75838846e-01 -4.35939759e-01 -5.94562888e-01 6.19920254e-01
-7.38198221e-01 -4.74990048e-02 3.31620991e-01 7.01021314e-01
4.01270151e-01 8.40341747e-01 1.28922117e+00 -3.16654295e-01
5.80719113e-01 -1.04005241e+00 -7.54807770e-01 6.61557689e-02
-9.49566960e-01 8.60817805e-02 1.07349122e+00 -1.72075361e-01
-1.34238112e+00 1.30458757e-01 -1.91976309e-01 -3.43802541e-01
-2.29125157e-01 6.88403487e-01 -4.13125366e-01 1.93070397e-01
3.03725183e-01 1.54341415e-01 2.21514642e-01 -2.76864052e-01
5.03085703e-02 8.74048769e-01 3.19503248e-01 -4.63527769e-01
7.88010657e-01 3.85128230e-01 4.35816616e-01 -5.18168092e-01
-4.10276413e-01 -2.00484529e-01 -2.65077233e-01 -6.45352185e-01
6.33416176e-01 -9.71573412e-01 -7.68315852e-01 4.70142454e-01
-1.09407270e+00 -2.05725014e-01 1.75157577e-01 7.31063545e-01
-6.95598364e-01 2.42287308e-01 3.69034335e-02 -1.31920195e+00
-2.63690893e-02 -1.40148640e+00 4.40912873e-01 4.00463909e-01
2.38804474e-01 -6.40245438e-01 -2.11045876e-01 -3.70695107e-02
5.45189440e-01 1.84335470e-01 7.80606151e-01 -4.36131477e-01
-7.39722610e-01 -4.98877525e-01 -2.41801843e-01 1.90547884e-01
-1.26643017e-01 9.92808416e-02 -7.70365894e-01 -9.09567997e-02
-1.11555181e-01 7.32931122e-02 5.85110188e-01 3.06987524e-01
1.33166468e+00 -2.63875723e-01 -6.08075202e-01 3.31122398e-01
1.26216459e+00 7.19444215e-01 7.04618871e-01 4.22319710e-01
3.82287025e-01 1.12023139e+00 1.33942449e+00 5.37941098e-01
7.04332232e-01 5.63873053e-01 1.08461380e+00 4.21509027e-01
3.19029868e-01 -5.46234369e-01 3.18074793e-01 3.38304073e-01
2.79128581e-01 -3.14315170e-01 -1.20056903e+00 6.91066086e-01
-2.30175257e+00 -1.04171145e+00 -3.85757118e-01 2.31557274e+00
1.33351386e-01 3.22620958e-01 1.31348416e-01 -1.15916207e-02
6.07066870e-01 -3.86779346e-02 -8.93967211e-01 -4.70047325e-01
1.77792981e-01 -8.35965335e-01 8.57746601e-01 3.00456107e-01
-9.59099531e-01 6.68254852e-01 5.36775732e+00 1.07748115e+00
-1.06104243e+00 -1.66810021e-01 4.60881561e-01 1.41147316e-01
-5.15850067e-01 8.49882290e-02 -9.81923997e-01 1.03920376e+00
1.23665380e+00 -3.93028051e-01 1.43256336e-01 9.15249527e-01
8.69518101e-01 -2.25772753e-01 -7.65750289e-01 6.24849975e-01
-5.22476435e-01 -1.16932893e+00 -4.15005028e-01 1.60627320e-01
6.23399377e-01 2.24668339e-01 1.39802516e-01 6.52531147e-01
3.13573867e-01 -1.07145834e+00 6.14542425e-01 9.06757534e-01
4.40208942e-01 -1.17768097e+00 1.07458866e+00 1.09314752e+00
-9.83799994e-01 -5.86890101e-01 -4.73540634e-01 -1.51200712e-01
6.17923737e-01 6.22723401e-01 -9.03651774e-01 7.30030477e-01
4.77368802e-01 7.27759719e-01 -1.80910125e-01 1.38561821e+00
-1.18033877e-02 4.51178461e-01 -4.59825397e-01 -1.43899962e-01
5.68164945e-01 -2.93571979e-01 6.68705404e-01 6.80963814e-01
7.93628693e-01 3.36385854e-02 3.38866442e-01 6.12075388e-01
4.10091072e-01 -8.27890337e-02 -1.13047409e+00 2.56953716e-01
6.63346350e-01 7.77475357e-01 -1.31009892e-01 -5.87739684e-02
-2.99173474e-01 4.38538045e-02 4.30076532e-02 5.42274654e-01
-9.29834664e-01 -3.45102161e-01 6.74707353e-01 1.48884058e-01
-3.92446369e-02 -2.05980033e-01 -6.41481340e-01 -5.40322721e-01
7.74122700e-02 -3.08863878e-01 3.75013240e-02 -6.38064265e-01
-1.06088448e+00 5.81295907e-01 5.33430651e-02 -1.72447944e+00
-3.19968671e-01 -4.56242323e-01 -1.06865013e+00 8.84815633e-01
-1.67039132e+00 -9.92729306e-01 -9.61931050e-02 2.14724824e-01
4.99866664e-01 -3.83367598e-01 4.10456151e-01 4.26531821e-01
-8.60943437e-01 2.52380908e-01 4.11817580e-01 -4.61842686e-01
5.23150802e-01 -6.90693557e-01 2.00078309e-01 8.24852407e-01
-8.66863370e-01 9.30979252e-02 1.14398503e+00 -9.01315093e-01
-1.31396866e+00 -1.51311290e+00 7.63026297e-01 -1.48081183e-01
6.41748190e-01 1.74226925e-01 -1.04692149e+00 2.92297512e-01
-1.71570554e-01 6.70444071e-02 1.31109878e-01 6.93015456e-02
4.33785953e-02 -4.85200614e-01 -1.28804386e+00 7.14776874e-01
7.02440143e-01 -1.46442935e-01 -2.55119652e-01 1.11799367e-01
6.80985808e-01 -2.10465610e-01 -8.04472089e-01 7.81557620e-01
6.83824956e-01 -7.09564388e-01 5.02791166e-01 -6.63895607e-01
3.68730605e-01 -4.50320899e-01 -1.18055277e-01 -1.53465545e+00
-2.75966942e-01 -2.90549874e-01 -3.61645371e-02 1.02195847e+00
6.44385099e-01 -7.17333615e-01 7.34930158e-01 1.09290338e+00
-6.33680582e-01 -8.04579198e-01 -1.52674806e+00 -9.61432219e-01
2.76055604e-01 -9.14401472e-01 1.18434322e+00 2.89022148e-01
6.08562073e-03 -2.48459339e-01 -7.14339972e-01 4.65936989e-01
5.80709994e-01 -7.07626566e-02 7.95594811e-01 -1.03296065e+00
4.21190768e-01 -3.50233406e-01 -2.97164977e-01 -8.81359398e-01
4.50882196e-01 -4.37063992e-01 6.36179209e-01 -1.53124392e+00
-3.37356478e-01 -9.67766523e-01 -2.49430671e-01 7.63562843e-02
-1.41674086e-01 -6.08204842e-01 7.60751292e-02 3.04760244e-02
-4.07340258e-01 9.94053006e-01 1.22650874e+00 -2.20658779e-01
-1.37027904e-01 7.52124012e-01 -2.93860316e-01 6.10685825e-01
1.30203748e+00 -4.35693502e-01 -7.94361115e-01 -2.65104473e-01
1.87842652e-01 4.78513598e-01 2.07046032e-01 -1.15212190e+00
4.20310378e-01 -7.59439647e-01 -8.76461342e-02 -1.20817053e+00
1.28666922e-01 -1.33704901e+00 4.97230262e-01 5.31733096e-01
9.68074799e-03 -1.15270920e-01 4.06007111e-01 1.09989715e+00
-1.86130896e-01 -1.91582829e-01 5.37766933e-01 2.78366804e-01
-9.67666924e-01 7.48268247e-01 -6.37251616e-01 -3.27794284e-01
1.41894472e+00 -2.01935023e-01 -3.15265924e-01 -5.28896093e-01
-3.41295302e-01 1.12684476e+00 1.93490177e-01 7.66302586e-01
5.87789536e-01 -1.46201408e+00 -5.83793044e-01 2.65802085e-01
2.35902295e-01 1.04000874e-01 7.96400011e-01 8.77958655e-01
-6.00874498e-02 6.01965249e-01 -2.11136922e-01 -3.76048952e-01
-6.22445285e-01 7.97390342e-01 2.86264271e-01 -2.76946388e-02
-6.36009514e-01 2.51441449e-01 1.62821725e-01 -4.51654911e-01
2.65121460e-01 -7.31632859e-02 -4.84156907e-01 -1.04414560e-01
4.72012639e-01 7.52543271e-01 2.13909224e-02 -8.08009446e-01
-1.89195409e-01 1.67730689e-01 2.62046665e-01 1.89185902e-01
9.24510598e-01 -5.40380120e-01 5.07914782e-01 4.23676908e-01
8.20329785e-01 -4.03921187e-01 -1.73113060e+00 2.95294542e-02
8.11341964e-03 -4.47897911e-01 3.18819970e-01 -7.48853326e-01
-9.72440660e-01 1.02021706e+00 5.05198717e-01 6.47028089e-02
7.94475079e-01 -6.95961177e-01 1.13679481e+00 2.89963841e-01
6.65033579e-01 -1.40763819e+00 -6.80329084e-01 6.27081692e-01
8.37427080e-01 -1.47526836e+00 -2.45751172e-01 -1.38687804e-01
-8.64631593e-01 1.09149754e+00 9.39969659e-01 3.87930162e-02
1.06146646e+00 1.33380279e-01 -1.60864487e-01 1.90543756e-01
-9.35026348e-01 -8.16383138e-02 1.43875316e-01 8.11861157e-01
-2.72045046e-01 3.91213834e-01 -3.58701319e-01 8.80041718e-01
-1.52930975e-01 1.86252415e-01 5.52124441e-01 5.05074382e-01
-6.37548566e-01 -9.23914373e-01 -2.36486822e-01 6.99684978e-01
2.19622022e-03 4.34305996e-01 4.52054620e-01 7.21293986e-01
3.95981997e-01 1.29595184e+00 -1.15028992e-02 -7.08178818e-01
4.62589055e-01 -1.18098214e-01 -3.47999543e-01 -7.89891779e-02
-1.56366900e-02 -3.19461972e-01 2.77605504e-01 -6.21798754e-01
1.81408022e-02 -6.53665841e-01 -1.51989162e+00 -5.83220661e-01
-9.46260244e-02 1.78788334e-01 7.12709308e-01 1.15494764e+00
5.42901695e-01 6.72546089e-01 8.09927464e-01 -6.20498896e-01
-9.89874661e-01 -7.59788454e-01 -5.69263220e-01 -6.93841800e-02
4.58283573e-01 -1.12716043e+00 -2.56611049e-01 -6.47731483e-01] | [5.692999362945557, 1.2393370866775513] |
0c97e716-6013-4d00-9bb9-6af9edab095a | multiverse-transformer-1st-place-solution-for | 2306.11868 | null | https://arxiv.org/abs/2306.11868v1 | https://arxiv.org/pdf/2306.11868v1.pdf | Multiverse Transformer: 1st Place Solution for Waymo Open Sim Agents Challenge 2023 | This technical report presents our 1st place solution for the Waymo Open Sim Agents Challenge (WOSAC) 2023. Our proposed MultiVerse Transformer for Agent simulation (MVTA) effectively leverages transformer-based motion prediction approaches, and is tailored for closed-loop simulation of agents. In order to produce simulations with a high degree of realism, we design novel training and sampling methods, and implement a receding horizon prediction mechanism. In addition, we introduce a variable-length history aggregation method to mitigate the compounding error that can arise during closed-loop autoregressive execution. On the WOSAC, our MVTA and its enhanced version MVTE reach a realism meta-metric of 0.5091 and 0.5168, respectively, outperforming all the other methods on the leaderboard. | ['Fan Yi', 'Tiebiao Zhao', 'Yu Wang'] | 2023-06-20 | null | null | null | null | ['motion-prediction'] | ['computer-vision'] | [-5.44724166e-01 -1.36030361e-01 -8.05373862e-02 3.30370575e-01
-6.31352484e-01 -6.22511148e-01 1.07393777e+00 -3.10625136e-02
-4.01964873e-01 9.60642755e-01 1.99843988e-01 -6.06179655e-01
-6.66856617e-02 -7.08305478e-01 -6.95042074e-01 -3.75858694e-01
-7.50494123e-01 7.26382494e-01 4.66176748e-01 -6.66130126e-01
-1.43687829e-01 1.86959192e-01 -1.51201653e+00 -9.74371284e-02
8.93868506e-01 8.17257762e-01 2.35879526e-01 1.04095399e+00
4.58337724e-01 1.26194417e+00 -6.35367632e-01 -1.21665746e-01
4.08171862e-01 -2.56839097e-01 -4.66008723e-01 -2.15309769e-01
-2.12185442e-01 -5.21921813e-01 -3.68611366e-01 4.10221785e-01
5.30122221e-01 3.62100333e-01 3.91796380e-01 -1.78384840e+00
3.89103621e-01 7.18506396e-01 -3.59934777e-01 1.53949231e-01
6.72656596e-01 9.26025748e-01 5.09584367e-01 -3.83702129e-01
8.51048350e-01 1.22487605e+00 7.20861197e-01 3.85967195e-01
-9.58660245e-01 -5.13903081e-01 3.62154365e-01 9.44587290e-02
-1.21332276e+00 -2.92594403e-01 2.01468706e-01 -4.71578002e-01
1.22077119e+00 4.23202366e-01 1.08071899e+00 1.26425970e+00
6.71890438e-01 6.51836753e-01 1.03698826e+00 9.09503400e-02
5.99194407e-01 -3.45155120e-01 -2.00530678e-01 7.39817500e-01
1.79197639e-01 6.81281328e-01 -3.57043058e-01 -5.45833409e-01
6.92548156e-01 -4.03055519e-01 -8.10157787e-03 -3.33314538e-01
-1.59251738e+00 6.65796161e-01 1.62271172e-01 -4.31099921e-01
-8.18974376e-01 6.07506514e-01 5.83836555e-01 3.20439905e-01
9.09687579e-02 3.91274810e-01 -3.64608437e-01 -6.53382421e-01
-6.68342948e-01 1.04228961e+00 1.07176709e+00 1.13159513e+00
2.84504503e-01 4.35787201e-01 -3.21211934e-01 -1.45216882e-01
5.15009880e-01 7.02591658e-01 2.06377864e-01 -1.59479523e+00
3.41213167e-01 3.25251877e-01 6.89320087e-01 -6.54634893e-01
-6.59269869e-01 -4.67961669e-01 -5.45568109e-01 7.35747457e-01
1.37179613e-01 -6.59509480e-01 -5.77155530e-01 1.61709297e+00
5.77298641e-01 7.21343815e-01 4.52892065e-01 1.06494415e+00
4.64319170e-01 8.71339858e-01 7.56414086e-02 -4.10885006e-01
1.18046224e+00 -1.35731292e+00 -6.94148421e-01 -1.19248271e-01
7.37349212e-01 -5.56705713e-01 8.93883705e-01 4.28422242e-01
-1.27530205e+00 -7.36221224e-02 -1.32856584e+00 5.63984215e-01
-1.07460469e-02 -4.62014735e-01 6.00219011e-01 5.07404029e-01
-1.07246840e+00 5.71832120e-01 -1.34118843e+00 -5.17004803e-02
-3.74422312e-01 2.84537196e-01 9.60858539e-02 4.13979083e-01
-9.41909492e-01 1.18289816e+00 2.02279583e-01 -4.33849424e-01
-1.57716131e+00 -9.79058504e-01 -7.92941928e-01 -1.25522435e-01
6.30534172e-01 -8.52019310e-01 1.73104858e+00 -2.88025826e-01
-2.24901915e+00 -3.24430168e-02 1.75496399e-01 -1.04017806e+00
8.78670216e-01 -2.28718221e-01 -3.31609935e-01 -3.27322781e-02
8.31649154e-02 4.79898125e-01 4.79507923e-01 -1.41650963e+00
-7.54396319e-01 2.00815618e-01 4.83855397e-01 4.15406495e-01
3.24319303e-01 -1.01297870e-01 -7.59446397e-02 -6.15537822e-01
-5.50882101e-01 -1.22973502e+00 -9.29695427e-01 -4.84267265e-01
7.82261565e-02 5.20611517e-02 6.68835402e-01 -4.41491336e-01
1.20773578e+00 -1.65336120e+00 3.48445207e-01 1.63206130e-01
2.13747814e-01 9.66173410e-02 -2.22397059e-01 8.22883606e-01
6.89518750e-01 -3.01306784e-01 9.49719325e-02 -4.74646419e-01
3.51348460e-01 2.15251088e-01 -4.76896971e-01 3.38479638e-01
-2.86074758e-01 8.26447010e-01 -1.13919544e+00 -3.30626637e-01
3.78434390e-01 1.47208899e-01 -6.16119742e-01 1.26920834e-01
-8.20643544e-01 4.41735536e-01 -4.29202139e-01 2.54523814e-01
5.22204995e-01 -2.52241582e-01 4.26019430e-01 2.83469677e-01
-5.33686578e-01 2.65078217e-01 -1.22729933e+00 1.58066785e+00
-5.30305743e-01 3.38429183e-01 2.81239033e-01 -2.13314798e-02
4.98971194e-01 4.35232818e-01 7.14644670e-01 -6.92974627e-01
3.32539827e-01 1.91468522e-01 -7.05374181e-02 -1.20605700e-01
9.09894705e-01 2.52564222e-01 -2.82761604e-01 5.82480609e-01
-4.46755677e-01 -2.67913669e-01 5.31026542e-01 4.23829615e-01
1.31246078e+00 3.33023578e-01 2.36316681e-01 -3.75489593e-01
3.83503765e-01 5.07339001e-01 7.33422875e-01 7.06760883e-01
-3.38074982e-01 -7.43841529e-02 3.34653586e-01 -4.77152973e-01
-1.10821652e+00 -9.39183474e-01 5.34677267e-01 8.91335249e-01
5.32625735e-01 -9.34458375e-01 -6.34561300e-01 -4.71048176e-01
8.29073861e-02 9.16937649e-01 -3.44705164e-01 1.60994697e-02
-7.69777596e-01 -4.28306490e-01 6.37916267e-01 4.39996064e-01
3.67912054e-01 -6.20821416e-01 -1.25279272e+00 4.95555043e-01
-1.44869074e-01 -1.27283120e+00 -6.17079973e-01 -3.54635090e-01
-3.81227374e-01 -9.42686856e-01 -2.71794915e-01 -3.01763743e-01
-6.75726831e-02 4.68681991e-01 1.14655471e+00 3.56571116e-02
4.01084810e-01 3.90040338e-01 -4.50274527e-01 -3.00836354e-01
-6.99144959e-01 -5.09555228e-02 4.48633254e-01 -5.21147251e-01
-3.24977815e-01 -4.85357612e-01 -6.31708384e-01 3.76254559e-01
-4.86484021e-01 3.28639627e-01 7.88742676e-02 6.64192021e-01
2.53008246e-01 -2.55994052e-01 4.64064538e-01 -1.15686692e-01
8.25357676e-01 -5.30637622e-01 -1.27201796e+00 -8.50059018e-02
-7.11518705e-01 -1.01649994e-02 7.89138675e-01 -5.13027489e-01
-9.47277546e-01 -1.79847717e-01 1.05180405e-01 -3.67219567e-01
3.27767760e-01 6.05816662e-01 4.25481528e-01 -1.24223292e-01
4.83519763e-01 3.67630050e-02 3.76967907e-01 2.62014493e-02
2.93730706e-01 1.59932718e-01 4.18675035e-01 -4.60338116e-01
7.32797682e-01 4.51780260e-01 1.63443968e-01 -4.00605917e-01
3.78851108e-02 4.00411431e-03 3.07174563e-01 -4.96635616e-01
6.47782922e-01 -1.24993026e+00 -1.49517548e+00 4.69587743e-01
-1.10654879e+00 -1.12037158e+00 -1.97439477e-01 7.10104942e-01
-9.98200417e-01 2.39482641e-01 -7.98393786e-01 -7.22008705e-01
-3.65493596e-01 -1.57490277e+00 8.21949005e-01 1.86929151e-01
-2.83029288e-01 -8.33245993e-01 6.67748153e-01 -1.89104639e-02
6.22174859e-01 5.25853395e-01 2.50871390e-01 -3.64823639e-01
-9.44909990e-01 2.54780371e-02 3.74340862e-01 -4.22707260e-01
-5.61697125e-01 7.02413090e-04 -2.75412709e-01 -7.93090284e-01
-3.64790320e-01 -1.95588410e-01 2.26691648e-01 3.19320679e-01
1.63796902e-01 -4.11698788e-01 -3.24605733e-01 5.05877614e-01
1.22471058e+00 3.85086536e-01 4.33649898e-01 7.95667768e-01
1.90330625e-01 6.89143315e-02 9.75403488e-01 1.07843161e+00
1.04775727e+00 9.90192592e-01 7.57692814e-01 3.36956948e-01
1.64070264e-01 -2.09034666e-01 9.07321393e-01 7.78125226e-01
-3.07459801e-01 -4.07620311e-01 -9.20011103e-01 4.32668865e-01
-2.15001440e+00 -9.50386405e-01 -3.04667950e-01 2.12364864e+00
4.31719989e-01 2.38268152e-01 4.88786846e-01 -3.06291014e-01
1.89810142e-01 3.04051191e-01 -3.02558035e-01 -3.33967358e-01
5.49763301e-03 -1.69631213e-01 7.03687668e-01 9.02392149e-01
-7.62513757e-01 1.12920487e+00 7.03007412e+00 8.05385590e-01
-7.74946332e-01 7.28109553e-02 -5.67005277e-02 -1.74723133e-01
-2.20891416e-01 2.48685539e-01 -5.49578488e-01 5.59774756e-01
1.44353783e+00 -7.39431798e-01 5.88219881e-01 8.14478636e-01
8.24991167e-01 -2.26480767e-01 -8.16274583e-01 5.11334360e-01
-3.70603353e-01 -1.78607750e+00 -2.36432582e-01 2.42870912e-01
9.94786322e-01 2.62044072e-01 -3.28570344e-02 6.61362350e-01
1.11735666e+00 -8.74730349e-01 9.42344129e-01 3.78088742e-01
2.34002084e-01 -8.57043505e-01 8.33748221e-01 4.79750186e-01
-1.42113733e+00 -1.79915242e-02 1.37500972e-01 -5.41338503e-01
8.56320739e-01 -1.02698907e-01 -9.05996561e-01 6.30724669e-01
3.27812463e-01 4.46542889e-01 -1.34554729e-01 1.05306733e+00
6.96978765e-03 4.47778285e-01 -5.05970120e-01 -1.90441743e-01
6.07525051e-01 -1.12490930e-01 1.11821651e+00 8.10558140e-01
3.32194120e-01 7.74984881e-02 6.67985857e-01 4.32888448e-01
3.66222441e-01 -3.54761362e-01 -4.17032748e-01 2.23989025e-01
8.16270232e-01 1.05833399e+00 -3.53513479e-01 -7.21154809e-01
-8.62844735e-02 7.38290191e-01 -4.41273563e-02 3.48336101e-01
-1.40092528e+00 1.44871533e-01 1.12008607e+00 -1.04772709e-01
1.85591191e-01 -7.18495548e-01 -9.38278139e-02 -9.81189609e-01
-2.13129759e-01 -1.13769937e+00 1.24052674e-01 -5.83471835e-01
-6.67851150e-01 7.74154007e-01 4.45005037e-02 -1.54475570e+00
-1.01629078e+00 -1.76459819e-01 -6.96969032e-01 3.99562925e-01
-1.21268475e+00 -1.08102024e+00 -3.65261048e-01 3.94676685e-01
5.95883667e-01 -4.27410483e-01 8.06815624e-01 9.39935818e-02
-4.09936726e-01 2.58361816e-01 2.75691509e-01 -5.69011271e-01
2.04894364e-01 -1.00549042e+00 8.42655480e-01 8.33237052e-01
-6.76914513e-01 2.68981338e-01 1.36847925e+00 -7.56442070e-01
-2.00587869e+00 -1.10935998e+00 2.61591882e-01 -4.40658987e-01
9.81490433e-01 5.65314814e-02 -3.84055734e-01 9.42344844e-01
5.17729878e-01 -3.65530491e-01 7.07073361e-02 -3.59607220e-01
4.42580804e-02 1.62378952e-01 -8.37277114e-01 1.06213427e+00
8.61227334e-01 -4.14053425e-02 -1.47482499e-01 2.43362382e-01
1.15960538e+00 -8.54605913e-01 -1.17462599e+00 3.72187674e-01
5.93281627e-01 -9.09506619e-01 9.31614935e-01 -4.42827374e-01
1.23827681e-01 -6.20620251e-01 -1.68319702e-01 -1.56742299e+00
-1.13551185e-01 -1.31955194e+00 -6.68137550e-01 5.69896996e-01
2.55272210e-01 -6.58996940e-01 7.76133120e-01 2.20056042e-01
-3.47999305e-01 -5.00494778e-01 -9.49038625e-01 -1.24959135e+00
-6.85470477e-02 -4.39771801e-01 1.00139296e+00 4.83450294e-01
1.97913095e-01 2.28160203e-01 -6.97676599e-01 4.02082533e-01
8.34983110e-01 -6.32042857e-03 1.30360842e+00 -5.33502042e-01
-5.10345519e-01 -3.54345500e-01 -2.14659616e-01 -1.18206298e+00
1.02290295e-01 -3.98082823e-01 -3.40964235e-02 -1.38216507e+00
-1.77002832e-01 -3.45564008e-01 2.91081578e-01 -9.75779220e-02
4.69743200e-02 -3.04218438e-02 6.91075027e-01 1.04071826e-01
-9.79815066e-01 9.77871418e-01 1.10091937e+00 1.49785861e-01
-4.27392721e-01 -1.35301091e-02 3.63565143e-03 7.95947015e-01
8.91037345e-01 -1.01605669e-01 -5.08354843e-01 -4.52192038e-01
6.40573651e-02 8.13717365e-01 5.73352933e-01 -1.28465176e+00
4.24115151e-01 -6.52447879e-01 -5.56689501e-01 -8.81802440e-01
6.03220701e-01 -6.33742213e-01 7.42138147e-01 9.26486492e-01
1.26323085e-02 7.28506207e-01 2.53358632e-01 5.63516080e-01
1.46707550e-01 3.28377753e-01 4.45929885e-01 1.13177881e-01
-8.43924344e-01 2.67517895e-01 -1.16978621e+00 1.83575302e-02
1.38291228e+00 1.92096159e-01 -7.36948907e-01 -6.53009713e-01
-4.40242022e-01 1.03834605e+00 7.23272085e-01 3.98338586e-01
6.25400662e-01 -1.37154257e+00 -8.03178191e-01 -2.40175575e-01
-1.59053266e-01 -3.27960432e-01 2.36802444e-01 1.00488579e+00
-9.19456840e-01 2.59487480e-01 -1.91529766e-01 -3.63750041e-01
-1.01215649e+00 5.60530305e-01 3.40059876e-01 -7.95569241e-01
-8.52080464e-01 1.40599459e-01 -3.90513152e-01 -4.77021277e-01
9.83931273e-02 -1.54610336e-01 1.27038598e-01 -4.34518397e-01
6.12061083e-01 8.82292569e-01 -1.45556420e-01 -5.52331507e-01
-4.25800055e-01 -5.58699705e-02 1.04801856e-01 -6.80410504e-01
1.12024784e+00 -2.74244815e-01 1.84408560e-01 1.50592104e-01
4.22450930e-01 -5.84368631e-02 -1.62900150e+00 9.66634378e-02
-6.70406073e-02 -2.00099781e-01 3.26665230e-02 -7.62335539e-01
-5.52172661e-01 1.13051347e-01 3.67406487e-01 1.78325877e-01
5.75819314e-01 -4.10402536e-01 1.12692165e+00 2.84766078e-01
1.00729239e+00 -9.60957587e-01 -1.45109631e-02 9.97697532e-01
7.84300625e-01 -1.00489283e+00 -1.90720323e-03 -2.08133817e-01
-1.02113044e+00 6.52427554e-01 7.57606328e-01 -3.86155158e-01
1.46377951e-01 7.10240066e-01 7.69991279e-02 -1.24448918e-01
-1.62684464e+00 -2.29669496e-01 -4.45073187e-01 7.49020278e-01
-1.32603303e-01 4.63663757e-01 -4.08277094e-01 5.79399824e-01
-4.96345997e-01 3.88877653e-02 1.05102420e+00 1.16579580e+00
-3.75178486e-01 -8.47500384e-01 -4.87994999e-01 -1.18756175e-01
1.68709010e-01 2.17826709e-01 1.59005597e-01 9.38221455e-01
-3.94191682e-01 1.17759049e+00 5.20551130e-02 -6.38670921e-01
3.07683825e-01 -5.65705359e-01 2.38071114e-01 1.09334692e-01
-9.07643378e-01 8.05966556e-02 6.29466832e-01 -9.46742892e-01
-1.11427985e-01 -5.72634637e-01 -1.43827522e+00 -1.11529839e+00
1.38625987e-02 2.74510741e-01 3.60342175e-01 7.79375136e-01
5.70438683e-01 8.28239560e-01 6.23460770e-01 -1.26844358e+00
-8.37592363e-01 -6.67543709e-01 -1.70902655e-01 3.39237298e-03
4.10604000e-01 -8.45723391e-01 -1.07292987e-01 -3.97570729e-01] | [5.20245885848999, 1.1606420278549194] |
4f4f1cd0-763a-43a9-8b78-8b9520c930c3 | postech-etris-submission-to-the-wmt2020-ape | null | null | https://aclanthology.org/2020.wmt-1.82 | https://aclanthology.org/2020.wmt-1.82.pdf | POSTECH-ETRI’s Submission to the WMT2020 APE Shared Task: Automatic Post-Editing with Cross-lingual Language Model | This paper describes POSTECH-ETRI’s submission to WMT2020 for the shared task on automatic post-editing (APE) for 2 language pairs: English-German (En-De) and English-Chinese (En-Zh). We propose APE systems based on a cross-lingual language model, which jointly adopts translation language modeling (TLM) and masked language modeling (MLM) training objectives in the pre-training stage; the APE models then utilize jointly learned language representations between the source language and the target language. In addition, we created 19 million new sythetic triplets as additional training data for our final ensemble model. According to experimental results on the WMT2020 APE development data set, our models showed an improvement over the baseline by TER of -3.58 and a BLEU score of +5.3 for the En-De subtask; and TER of -5.29 and a BLEU score of +7.32 for the En-Zh subtask. | ['Jong-Hyeok Lee', 'Young-Kil Kim', 'Baikjin Jung', 'Jaehun Shin', 'WonKee Lee', 'Jihyung Lee'] | null | null | null | null | wmt-emnlp-2020-11 | ['automatic-post-editing', 'automatic-post-editing'] | ['computer-vision', 'natural-language-processing'] | [-1.57109816e-02 7.25951940e-02 -1.86678469e-01 -4.49104816e-01
-1.52277744e+00 -4.64788198e-01 9.08987284e-01 -2.11695731e-01
-7.13661849e-01 9.73881185e-01 4.03641909e-01 -5.96277475e-01
5.70464730e-01 -3.16287577e-01 -1.04595804e+00 -4.30937409e-02
3.07722718e-01 7.66933203e-01 -2.66074896e-01 -3.66189718e-01
-1.98123842e-01 -1.58013135e-01 -6.71065211e-01 6.88006997e-01
1.15403533e+00 5.41112721e-01 1.75440237e-01 6.20394588e-01
-1.42472789e-01 6.05810821e-01 -4.19430196e-01 -8.28198314e-01
2.64001906e-01 -5.43714821e-01 -7.37697005e-01 -4.18267727e-01
4.38714832e-01 2.23597541e-01 -2.35689923e-01 8.76690149e-01
5.11027753e-01 -3.32838744e-02 6.36815369e-01 -1.04393637e+00
-1.03793955e+00 1.21677899e+00 -5.89909852e-01 -1.49244383e-01
1.08636364e-01 -6.44139796e-02 9.46596563e-01 -1.66479921e+00
7.61780858e-01 1.16186273e+00 6.14362478e-01 7.63788044e-01
-1.24234140e+00 -9.68090892e-01 4.16734815e-02 3.62007618e-02
-1.43890035e+00 -7.66964436e-01 3.02637279e-01 -3.25556159e-01
1.53961897e+00 1.57521129e-01 9.11107063e-02 1.36346912e+00
4.03629631e-01 9.70841527e-01 1.11389089e+00 -7.67043233e-01
-2.34973624e-01 3.59982610e-01 1.53653780e-02 6.08260512e-01
9.13463347e-03 5.41911982e-02 -7.50509322e-01 2.17213899e-01
1.60572022e-01 -6.48134351e-01 -1.73335671e-02 3.69662911e-01
-1.63916695e+00 5.75883269e-01 -5.46773300e-02 3.55062127e-01
-3.32220465e-01 5.25068790e-02 4.28205341e-01 6.92562342e-01
7.18195081e-01 5.76578677e-01 -8.23515892e-01 -3.14809918e-01
-1.24556947e+00 1.01611227e-01 6.84101939e-01 1.26110578e+00
5.31990349e-01 3.51316482e-02 -5.28537035e-01 1.13625121e+00
3.08600515e-01 8.02548945e-01 6.68968081e-01 -6.36781812e-01
1.23218286e+00 2.33259305e-01 1.00027941e-01 -2.45994374e-01
1.59955863e-02 -5.85829675e-01 -8.62141132e-01 -3.49423468e-01
1.98940963e-01 -5.14615059e-01 -8.31320465e-01 2.03852582e+00
-3.24822485e-01 -5.89767136e-02 3.81873786e-01 2.27829695e-01
8.00987124e-01 1.09669149e+00 2.75804661e-02 -3.52927655e-01
9.97130036e-01 -1.64155853e+00 -7.85329163e-01 -5.67100406e-01
1.11682355e+00 -1.28779685e+00 1.25806642e+00 2.12334037e-01
-1.37496972e+00 -7.63172984e-01 -9.14827466e-01 -2.06164554e-01
-3.36353362e-01 8.73146713e-01 3.22897844e-02 3.81600499e-01
-1.24732137e+00 3.67722303e-01 -7.33296812e-01 -4.14631754e-01
-1.32615879e-01 1.55649215e-01 -4.82428342e-01 -7.93781430e-02
-1.39443779e+00 1.11183202e+00 3.91662031e-01 2.63894144e-02
-5.76786757e-01 -8.39533269e-01 -7.22675025e-01 -1.20356187e-01
-8.25762376e-03 -5.13629496e-01 1.30849576e+00 -8.59965026e-01
-1.82453895e+00 1.13961840e+00 -3.82352978e-01 -5.20193338e-01
7.96790242e-01 -7.54642248e-01 -9.58018541e-01 -5.57358265e-01
2.86480606e-01 7.47161865e-01 2.64063984e-01 -8.05083156e-01
-5.85569680e-01 4.35584374e-02 -5.50998509e-01 2.82750517e-01
-3.32694799e-01 5.39225698e-01 -8.16176534e-01 -9.63755190e-01
-2.58311987e-01 -1.19942427e+00 -9.37227905e-02 -6.36851788e-01
-7.00885534e-01 -1.88729227e-01 3.75873208e-01 -1.39870703e+00
1.64103293e+00 -1.96883237e+00 1.69759572e-01 -1.48543641e-02
-2.57681698e-01 4.03930664e-01 -6.81877732e-01 6.69271231e-01
-1.98292106e-01 1.59085691e-01 -2.03773290e-01 -1.06108320e+00
1.21563062e-01 -2.27840886e-01 -3.92260492e-01 -5.01638912e-02
1.60595432e-01 1.19942689e+00 -8.29935908e-01 -1.99504137e-01
-1.11974716e-01 2.14516044e-01 -4.17918146e-01 3.06984097e-01
-2.09653363e-01 3.59714717e-01 9.83407721e-02 3.41142327e-01
4.30491894e-01 -7.34760016e-02 2.72798419e-01 8.03781077e-02
-3.92047465e-01 9.18717742e-01 -6.59702957e-01 1.86051774e+00
-7.55936325e-01 7.71968126e-01 -1.78106233e-01 -3.12771887e-01
9.99460101e-01 5.10465801e-01 2.21231312e-01 -7.65319526e-01
-1.24367684e-01 6.52947187e-01 8.48751962e-02 -4.40582447e-02
9.40523982e-01 1.26765057e-01 -4.00099784e-01 7.13449299e-01
2.85986513e-01 4.78690341e-02 3.88378501e-01 1.06018379e-01
7.62473464e-01 3.57696146e-01 1.17913067e-01 -3.92779440e-01
5.02969086e-01 -1.86704457e-01 7.63995171e-01 4.39817756e-01
1.93261415e-01 5.83443165e-01 5.27525321e-02 -9.81184915e-02
-1.12604582e+00 -1.03184307e+00 1.83056951e-01 1.08122599e+00
-4.27699357e-01 -8.41349781e-01 -8.76827657e-01 -6.68514907e-01
-2.94024140e-01 1.33560181e+00 -2.79755920e-01 -3.82254690e-01
-8.94741416e-01 -7.49591470e-01 8.12035620e-01 3.27550918e-01
6.35649502e-01 -9.48715687e-01 4.03721839e-01 5.00671029e-01
-7.03157127e-01 -1.33714139e+00 -1.15784061e+00 -7.73821399e-02
-5.29659152e-01 -4.77680147e-01 -9.28274035e-01 -9.31204796e-01
3.32960665e-01 -2.35189840e-01 1.38823104e+00 -3.60133022e-01
3.97101700e-01 -1.46393925e-01 -2.40323156e-01 -3.54593784e-01
-9.05974388e-01 3.93642366e-01 3.91903847e-01 6.51359260e-02
5.75090349e-01 -1.76823199e-01 -4.23022546e-03 1.94505498e-01
-2.38650754e-01 5.54444730e-01 5.57163954e-01 8.12762558e-01
5.66538393e-01 -7.18847871e-01 4.03541267e-01 -8.26059163e-01
6.05959058e-01 -4.07348424e-01 -6.30324483e-01 8.96169364e-01
-8.40484381e-01 6.94864169e-02 6.00474894e-01 -4.09861594e-01
-1.19477081e+00 -3.05755973e-01 -1.45045042e-01 -1.33790612e-01
5.03671587e-01 7.51226842e-01 -3.51101160e-01 4.10641134e-01
4.09167171e-01 3.55338335e-01 -4.92893755e-01 -8.80668521e-01
4.75975960e-01 1.01490796e+00 6.59984887e-01 -6.97457790e-01
7.22704530e-01 -4.18379128e-01 -6.03099406e-01 -5.31882167e-01
-5.35606205e-01 2.54759770e-02 -6.52762949e-01 1.59825116e-01
8.51324856e-01 -1.45636189e+00 1.37046397e-01 6.90820277e-01
-1.38316572e+00 -6.21394098e-01 1.83334481e-02 7.66381621e-01
-3.92330647e-01 3.40596527e-01 -8.55379760e-01 -3.49537730e-01
-7.41117775e-01 -1.23729265e+00 9.44902360e-01 -1.94945455e-01
-5.39975703e-01 -9.31917429e-01 4.10155922e-01 5.04762769e-01
3.93632650e-01 -1.63323611e-01 1.02568316e+00 -6.38888657e-01
-3.47001702e-01 -6.17491417e-02 -1.02555230e-01 6.50702477e-01
1.39705598e-01 2.11009905e-01 -6.41007185e-01 -4.94577289e-01
-5.96519530e-01 -2.74730355e-01 7.74875402e-01 1.47614703e-01
7.38186002e-01 -2.46812403e-01 -1.76959932e-01 6.97096050e-01
1.16118348e+00 -1.63648631e-02 6.18034482e-01 1.74089536e-01
6.45394504e-01 3.58402818e-01 6.09607756e-01 5.92845120e-03
8.09195757e-01 1.12011468e+00 -4.67775643e-01 -1.37164608e-01
-4.34890807e-01 -7.05629766e-01 1.20893574e+00 1.81224108e+00
1.37770668e-01 -4.85814840e-01 -1.14846170e+00 5.79237938e-01
-1.77200341e+00 -5.69955170e-01 -2.39223078e-01 2.36464119e+00
1.19496560e+00 9.05243605e-02 -1.17416561e-01 -4.49705333e-01
7.32669234e-01 -1.28512979e-01 -5.80821559e-02 -7.74156392e-01
-4.33419347e-01 2.73780137e-01 4.22183186e-01 7.28543818e-01
-8.56491148e-01 1.56374288e+00 5.78118801e+00 9.11810398e-01
-1.13612175e+00 3.21977139e-01 5.54710150e-01 -2.10684799e-02
-3.96518916e-01 4.15783189e-02 -1.12356126e+00 6.88426852e-01
1.52177155e+00 -5.96496940e-01 5.42254686e-01 4.23669070e-01
2.08755344e-01 3.67606759e-01 -1.27852476e+00 9.72639382e-01
4.25758772e-02 -1.26136780e+00 1.90132126e-01 8.30826834e-02
1.12020421e+00 7.15305567e-01 7.68171698e-02 8.95217180e-01
6.43642724e-01 -8.94778550e-01 8.76044273e-01 6.12392068e-01
1.32825077e+00 -6.13203108e-01 6.52115107e-01 4.27801311e-01
-1.00911927e+00 3.90746504e-01 -2.74209321e-01 2.56079286e-01
3.70351613e-01 6.22796834e-01 -8.48376155e-01 7.70406246e-01
4.69971120e-01 8.14362824e-01 -6.05107427e-01 6.43584430e-01
-4.69591469e-01 9.45811391e-01 -2.24560127e-01 2.35992864e-01
1.68874308e-01 -2.87457317e-01 6.34847343e-01 1.58329868e+00
6.82240486e-01 -5.44647515e-01 -1.22084329e-02 8.08765471e-01
-7.40651667e-01 3.75602573e-01 -2.67399460e-01 -4.17357236e-01
6.71275258e-01 1.00492513e+00 6.85075996e-03 -6.23648107e-01
-4.97373611e-01 1.33139062e+00 5.43625176e-01 4.37422812e-01
-8.01807642e-01 -3.35740507e-01 6.66419148e-01 -3.41441900e-01
3.32313478e-02 -2.81291604e-01 -2.32020065e-01 -1.64034772e+00
1.07916325e-01 -1.10509884e+00 5.02141565e-03 -7.42582381e-01
-1.23738444e+00 9.67697084e-01 -2.59377688e-01 -1.32294118e+00
-6.13576710e-01 -3.85990858e-01 -6.01292014e-01 1.40356779e+00
-1.30817544e+00 -1.55776811e+00 5.31427979e-01 3.70685726e-01
6.90885723e-01 -7.62582481e-01 1.01939738e+00 5.88008642e-01
-6.68215990e-01 1.21722901e+00 5.45748353e-01 3.96067351e-01
1.20593202e+00 -1.12815094e+00 1.16279113e+00 1.22757244e+00
4.90258366e-01 7.28611946e-01 3.35611880e-01 -7.64445484e-01
-1.02355552e+00 -1.58828044e+00 2.23187280e+00 -8.71707201e-01
6.11779690e-01 -7.31874287e-01 -6.30509138e-01 1.05863202e+00
5.91418743e-01 -4.06078726e-01 6.21076465e-01 3.19026649e-01
-4.68122303e-01 8.92290473e-02 -6.61770642e-01 8.29206347e-01
8.68902087e-01 -8.47324669e-01 -6.78870738e-01 2.51323014e-01
9.93827701e-01 -4.51841384e-01 -9.88392949e-01 3.30746710e-01
4.91500467e-01 -2.71348417e-01 5.97265363e-01 -7.13226795e-01
5.70075512e-01 -1.73982516e-01 -5.30672312e-01 -1.61217856e+00
-3.53768319e-01 -8.67901206e-01 1.97201986e-02 1.54713666e+00
1.13719273e+00 -3.31758529e-01 2.69945562e-01 2.97166318e-01
-4.44976419e-01 -5.08597136e-01 -9.55568075e-01 -9.75577891e-01
5.67173660e-01 -4.31642503e-01 5.91028869e-01 1.03649080e+00
-1.18608169e-01 7.96332717e-01 -7.31821299e-01 -5.94480559e-02
3.43261927e-01 -1.57455102e-01 8.60461712e-01 -7.08673656e-01
-5.42433739e-01 -4.77345675e-01 4.21009481e-01 -1.04599535e+00
3.24540406e-01 -1.32900739e+00 -4.26266231e-02 -1.38338697e+00
1.17403038e-01 -1.62841991e-01 -4.35251892e-01 5.86044550e-01
-3.91888887e-01 1.20047994e-01 4.11444992e-01 1.70406595e-01
-5.13790429e-01 6.84817791e-01 7.67430604e-01 -1.34075806e-01
-2.40874454e-01 -5.68412356e-02 -6.10224426e-01 2.33808145e-01
7.36872852e-01 -5.04707992e-01 -2.83939838e-02 -1.05685210e+00
2.02201903e-01 -1.48269281e-01 -2.71175921e-01 -7.97789514e-01
1.92152187e-02 1.27644673e-01 9.10109431e-02 -6.69404089e-01
2.84366935e-01 -3.88557851e-01 1.54093489e-01 2.92137116e-01
-5.83731532e-01 5.90254366e-01 3.51641387e-01 4.77619795e-03
-3.33899051e-01 2.56482840e-01 5.64744771e-01 1.47247985e-01
-4.71938610e-01 2.36873999e-01 -2.86741108e-01 2.62837559e-01
5.35492122e-01 1.80927381e-01 -1.32043719e-01 -3.38062137e-01
-5.59439421e-01 4.15583909e-01 3.34574252e-01 9.37605500e-01
3.02380264e-01 -1.68058610e+00 -1.36640084e+00 4.84457344e-01
3.70053768e-01 -5.86316824e-01 -1.09862603e-01 8.59611928e-01
-2.04804227e-01 5.50549269e-01 5.99088427e-03 -1.62232280e-01
-1.21415567e+00 1.94455907e-02 2.08397627e-01 -8.41817677e-01
-1.10344566e-01 1.06592011e+00 -1.45781189e-01 -1.11341381e+00
8.87025744e-02 -2.75679886e-01 2.80258328e-01 -3.58843237e-01
4.88198191e-01 3.57451320e-01 4.05054122e-01 -8.63165200e-01
-3.78230065e-01 3.75547945e-01 -4.18060094e-01 -5.12486935e-01
1.26751792e+00 -2.77778625e-01 -2.17374653e-01 5.11423171e-01
1.37589526e+00 3.72186691e-01 -6.95315361e-01 -6.91005349e-01
4.33101267e-01 1.01508819e-01 4.40847017e-02 -1.38653982e+00
-6.32868588e-01 1.03410554e+00 2.78538585e-01 -7.68302500e-01
8.26970637e-01 -3.10542762e-01 1.02095401e+00 4.24288541e-01
3.00551653e-01 -1.62398231e+00 -3.06372732e-01 1.11660373e+00
1.05641747e+00 -1.14896750e+00 -4.81672674e-01 -6.77941889e-02
-7.56205559e-01 9.38279271e-01 5.31869352e-01 1.85833931e-01
5.33997834e-01 1.40968174e-01 2.93483764e-01 4.03153390e-01
-1.16006029e+00 3.28422487e-01 6.85552359e-01 2.63285846e-01
9.76663530e-01 3.69488776e-01 -4.13165182e-01 9.36299860e-01
-4.69759375e-01 7.25096092e-02 2.34415174e-01 4.72971022e-01
4.13969383e-02 -1.66969872e+00 -4.63339529e-04 1.50251448e-01
-4.13374931e-01 -7.77565122e-01 -5.56388378e-01 6.76336586e-01
1.39484396e-02 9.86145139e-01 1.90228689e-02 -7.24016845e-01
3.48389566e-01 3.34848434e-01 3.69301826e-01 -7.68641949e-01
-7.72238672e-01 2.53767163e-01 5.56037962e-01 -4.38000739e-01
-7.59943649e-02 -5.93116641e-01 -9.06242967e-01 -3.47860992e-01
1.22105805e-02 3.82568151e-01 6.17956102e-01 8.97547603e-01
5.58280885e-01 2.94669986e-01 6.67290032e-01 -3.46360058e-01
-5.24215579e-01 -1.30981433e+00 -2.80292839e-01 1.05030976e-01
2.87590381e-02 5.80286374e-03 -5.86078353e-02 1.31744578e-01] | [11.659098625183105, 10.291336059570312] |
f8b56f49-ad31-4eef-88e3-32c7306cbca0 | learning-robust-representations-for-continual | 2210.04497 | null | https://arxiv.org/abs/2210.04497v1 | https://arxiv.org/pdf/2210.04497v1.pdf | Learning Robust Representations for Continual Relation Extraction via Adversarial Class Augmentation | Continual relation extraction (CRE) aims to continually learn new relations from a class-incremental data stream. CRE model usually suffers from catastrophic forgetting problem, i.e., the performance of old relations seriously degrades when the model learns new relations. Most previous work attributes catastrophic forgetting to the corruption of the learned representations as new relations come, with an implicit assumption that the CRE models have adequately learned the old relations. In this paper, through empirical studies we argue that this assumption may not hold, and an important reason for catastrophic forgetting is that the learned representations do not have good robustness against the appearance of analogous relations in the subsequent learning process. To address this issue, we encourage the model to learn more precise and robust representations through a simple yet effective adversarial class augmentation mechanism (ACA), which is easy to implement and model-agnostic. Experimental results show that ACA can consistently improve the performance of state-of-the-art CRE models on two popular benchmarks. | ['Zhifang Sui', 'Sujian Li', 'Yunbo Cao', 'Binghuai Lin', 'Tianyu Liu', 'YiFan Song', 'Peiyi Wang'] | 2022-10-10 | null | null | null | null | ['continual-relation-extraction'] | ['natural-language-processing'] | [ 1.69938803e-01 6.16034329e-01 -4.31227297e-01 -2.93717593e-01
-3.17639083e-01 -3.44761074e-01 5.60806572e-01 3.89018238e-01
-6.47800341e-02 1.08727372e+00 -4.87049818e-02 -2.61379391e-01
-3.61872092e-02 -1.13277900e+00 -9.68092620e-01 -3.96374673e-01
-2.71863997e-01 5.52968442e-01 4.05113399e-01 -5.65134108e-01
-2.91878492e-01 4.47834849e-01 -1.42820048e+00 2.63203561e-01
8.50555420e-01 7.88142323e-01 -6.48534834e-01 4.35461938e-01
1.64995284e-03 1.48354065e+00 -8.31716120e-01 -9.36249733e-01
1.83587492e-01 -3.46044242e-01 -1.15768290e+00 -2.85114050e-01
1.97263241e-01 -2.60545969e-01 -7.41514087e-01 9.59825456e-01
-6.41746959e-03 8.62871259e-02 4.99690622e-01 -1.62577224e+00
-1.09558773e+00 1.03421187e+00 -5.12629807e-01 4.37010765e-01
1.95904821e-01 -1.77141242e-02 9.07753468e-01 -8.77826571e-01
7.28675604e-01 1.09380770e+00 8.61064494e-01 6.45575047e-01
-1.16540849e+00 -1.08782530e+00 5.99014580e-01 2.71390975e-01
-1.47263312e+00 -5.20028889e-01 7.67417371e-01 -1.13815174e-01
9.97087717e-01 2.83311427e-01 6.06960952e-01 1.32820177e+00
4.26803768e-01 7.10321307e-01 8.21519196e-01 -1.97687075e-01
5.45350388e-02 1.99447215e-01 3.94833356e-01 5.32220602e-01
8.00021708e-01 2.25622296e-01 -7.39981353e-01 -1.71892375e-01
4.38769400e-01 1.30161330e-01 -1.61865503e-01 -4.13694888e-01
-7.76401162e-01 4.85773116e-01 6.67583168e-01 2.51511961e-01
-2.70274192e-01 4.32599969e-02 3.58924329e-01 7.47182071e-01
6.57132149e-01 5.25600255e-01 -7.45254874e-01 -3.83569673e-02
-3.88996363e-01 2.20711529e-01 7.22217441e-01 1.17675567e+00
7.88956940e-01 -1.13000050e-02 1.69503444e-03 5.92095912e-01
-2.03966275e-01 2.08579730e-02 6.30871534e-01 -4.66168016e-01
4.93031591e-01 8.64488840e-01 -3.61320347e-01 -8.70835066e-01
-1.45408347e-01 -7.18781590e-01 -1.04070616e+00 1.55959681e-01
3.11062902e-01 -6.42710626e-02 -9.61852074e-01 1.76034689e+00
2.30944872e-01 4.29638475e-01 3.77135307e-01 3.33444268e-01
8.00656021e-01 2.39767358e-01 2.54148722e-01 -4.21029031e-01
8.93125117e-01 -9.81837273e-01 -9.02889609e-01 -5.99025548e-01
3.38445485e-01 -3.87554348e-01 9.82710242e-01 3.36265266e-01
-9.11159813e-01 -6.06142163e-01 -1.45126116e+00 8.83450285e-02
-5.84164023e-01 -5.50777376e-01 1.20000792e+00 4.05826062e-01
-4.76090640e-01 8.30672920e-01 -8.63043904e-01 -3.37237239e-01
7.37115502e-01 4.17253673e-01 -6.42681479e-01 -8.14504456e-03
-1.44005287e+00 1.02590621e+00 5.32063603e-01 -9.63492766e-02
-9.10522759e-01 -8.04154813e-01 -8.58541608e-01 9.31983516e-02
6.43540502e-01 -6.76473379e-01 1.26997471e+00 -1.05891645e+00
-1.07763088e+00 6.12226307e-01 -6.81761801e-02 -7.93609321e-01
4.62011695e-01 -7.99093783e-01 -7.54992425e-01 -1.98279366e-01
-1.08919464e-01 1.00924954e-01 9.46198285e-01 -1.60539567e+00
-5.71673095e-01 -1.63534239e-01 1.67974979e-01 -1.78250343e-01
-4.96864557e-01 -3.50479752e-01 -1.17141433e-01 -9.65345383e-01
2.20658049e-01 -7.46481895e-01 3.15711312e-02 -1.81841537e-01
-2.96802104e-01 -2.75012434e-01 9.51599061e-01 -2.24560127e-01
1.47077990e+00 -2.20580506e+00 -9.78307277e-02 -2.58944202e-02
4.04929191e-01 4.71302152e-01 -5.80786541e-02 2.83912271e-01
-6.00647748e-01 2.20787421e-01 -1.58120826e-01 -2.27253109e-01
-5.73831975e-01 5.98115265e-01 -8.29065859e-01 1.39231861e-01
7.69580424e-01 1.03129947e+00 -1.19374299e+00 -2.17513829e-01
-4.19687122e-01 3.83409828e-01 -3.37586939e-01 3.74194205e-01
-1.70033142e-01 1.12645961e-01 -1.50008962e-01 6.84788883e-01
7.20187247e-01 -2.49360397e-01 2.38225549e-01 -3.67722511e-02
4.69713807e-01 4.43209261e-01 -1.01083469e+00 1.35655630e+00
-7.99857080e-02 3.76505584e-01 -6.66476488e-01 -8.44552398e-01
1.00190485e+00 2.74462104e-01 2.30671078e-01 -4.70544666e-01
-6.94460273e-02 2.55153980e-02 1.30157590e-01 -3.04979384e-01
4.52643514e-01 -3.01264703e-01 8.62221569e-02 4.29687679e-01
2.32001096e-01 1.77811775e-02 1.89200195e-03 2.79006571e-01
1.42901254e+00 5.29715456e-02 4.60808098e-01 2.08357602e-01
2.95289278e-01 -4.91803735e-02 1.06011915e+00 8.39214265e-01
-4.69339080e-02 4.10599142e-01 4.79173571e-01 -8.34451854e-01
-6.24041140e-01 -1.18026948e+00 7.16904625e-02 8.55563760e-01
1.03868119e-01 -5.96917510e-01 -2.49430612e-01 -1.28756094e+00
2.10802019e-01 6.99690998e-01 -9.81295824e-01 -1.02514184e+00
-7.60641098e-01 -7.69165218e-01 5.97560763e-01 7.80486882e-01
5.65225482e-01 -9.57526386e-01 -1.16849110e-01 4.47907239e-01
4.00739647e-02 -1.04549730e+00 -2.53628381e-02 4.72110480e-01
-1.09024298e+00 -1.39933336e+00 -1.04461648e-01 -6.52861536e-01
7.93702722e-01 2.69687086e-01 1.31753325e+00 3.63716751e-01
7.17038214e-02 -1.29415877e-02 -5.89173853e-01 -6.21629000e-01
-7.13210583e-01 4.83222812e-01 2.38931909e-01 -1.43004969e-01
4.70114321e-01 -8.32786500e-01 -1.38226762e-01 1.15114190e-01
-9.03566539e-01 -9.15832892e-02 5.05209565e-01 9.89979267e-01
3.02377284e-01 5.39887249e-01 9.94306326e-01 -1.54832137e+00
6.85246468e-01 -5.89479029e-01 2.14405935e-02 4.59450990e-01
-1.10539842e+00 1.79527223e-01 7.20868826e-01 -8.82210493e-01
-1.01143420e+00 -1.09450474e-01 2.67406181e-02 -4.93550628e-01
1.40959591e-01 4.60364908e-01 -1.45531878e-01 8.93357769e-02
9.05318201e-01 1.96380392e-02 -2.30547979e-01 -3.93024802e-01
3.95087242e-01 1.99537292e-01 8.04908037e-01 -6.54179096e-01
1.23324251e+00 3.95896137e-01 -1.23456821e-01 -1.35695338e-01
-1.30548525e+00 3.38811940e-03 -6.03622198e-01 1.01734370e-01
1.00049771e-01 -9.45128024e-01 -3.68178576e-01 5.30093789e-01
-1.04828620e+00 -2.76601464e-01 -7.66993642e-01 1.28140092e-01
-2.52843499e-01 5.73691446e-03 -7.91398585e-01 -7.40377128e-01
-3.40394527e-01 -6.34420276e-01 3.99383247e-01 2.91706979e-01
-4.74591583e-01 -7.19824612e-01 1.18401076e-03 2.62397137e-02
3.36695194e-01 3.86941046e-01 1.08962691e+00 -6.24296546e-01
-4.98872012e-01 -3.75434786e-01 8.42407569e-02 5.10535657e-01
5.78748286e-01 2.77207885e-02 -1.28651130e+00 -2.74644881e-01
-3.29926051e-02 -3.70796204e-01 1.04904366e+00 -5.86326241e-01
1.00826859e+00 -5.45368671e-01 -4.26252574e-01 4.79304910e-01
1.04788780e+00 2.35186607e-01 7.50465333e-01 2.47032672e-01
7.02928543e-01 2.95723438e-01 7.45306909e-01 1.75901547e-01
4.74472553e-01 2.60369271e-01 3.68453830e-01 -1.03722755e-02
-4.39203829e-01 -6.60530150e-01 3.34873438e-01 7.22607374e-01
-1.08088322e-01 -6.55598193e-02 -9.53388393e-01 5.53761721e-01
-1.96418560e+00 -8.07774782e-01 2.60990739e-01 2.09510779e+00
1.49335551e+00 7.85636663e-01 -1.78272665e-01 6.32319808e-01
3.72680366e-01 -1.12915672e-01 -6.11842811e-01 -3.81692708e-01
-2.58767515e-01 5.88144064e-01 2.91984081e-01 3.89100671e-01
-1.05223572e+00 1.15514719e+00 6.72487450e+00 4.32211608e-01
-9.56509829e-01 1.25137031e-01 6.92021310e-01 1.77364394e-01
-1.41102865e-01 2.03358471e-01 -7.07547843e-01 1.80363238e-01
7.44775474e-01 -3.53639662e-01 2.18282178e-01 8.42835426e-01
-7.24100292e-01 4.83647175e-02 -1.44843233e+00 6.08070195e-01
-3.35710943e-02 -1.33601344e+00 1.72266185e-01 -3.20532262e-01
7.08413839e-01 -3.44783813e-01 6.42033592e-02 8.03059101e-01
5.83495855e-01 -9.88580108e-01 5.36482632e-01 7.21776962e-01
7.38727510e-01 -9.84170377e-01 9.33946788e-01 2.18673423e-01
-1.07424676e+00 -1.41695216e-01 -1.88705102e-01 -4.61835682e-01
-3.34674269e-01 5.77775002e-01 -8.87874484e-01 6.76076233e-01
7.59900331e-01 6.58358634e-01 -1.15950274e+00 7.25869238e-01
-6.78593576e-01 7.23537087e-01 -4.35131676e-02 4.45532292e-01
-5.11080742e-01 5.42904377e-01 3.42280805e-01 8.02895963e-01
-5.01682870e-02 1.12193018e-01 -1.56040624e-01 5.60007393e-01
-3.75285685e-01 -3.21752340e-01 -7.42513418e-01 -4.81287353e-02
6.19366348e-01 8.32342744e-01 -4.76252943e-01 -3.97774160e-01
-4.68654454e-01 8.55229259e-01 7.55409777e-01 3.90877157e-01
-5.48049569e-01 -1.38718680e-01 7.69081056e-01 2.61111528e-01
1.41884193e-01 -3.63283195e-02 -4.54520941e-01 -1.29078162e+00
1.07822932e-01 -1.04842055e+00 6.42534196e-01 -4.90983367e-01
-1.65933108e+00 8.05802643e-01 -1.23662248e-01 -9.78934109e-01
-1.15519263e-01 -8.32676739e-02 -4.52474743e-01 3.80546033e-01
-1.48044145e+00 -1.11637843e+00 -2.15799659e-01 5.23734450e-01
4.55022961e-01 -2.64445394e-01 1.05748940e+00 4.45640713e-01
-7.98777163e-01 1.13399708e+00 -5.84695518e-01 3.07417750e-01
8.81109595e-01 -1.10634363e+00 7.07858682e-01 9.66334462e-01
2.67552525e-01 8.46929610e-01 7.80180275e-01 -8.91576648e-01
-1.11373603e+00 -1.21099591e+00 9.86874819e-01 -7.14895070e-01
7.37237155e-01 -3.56623948e-01 -1.43834591e+00 1.03710294e+00
1.86191546e-03 4.12503570e-01 6.84843838e-01 3.91562521e-01
-8.71579051e-01 -4.86001611e-01 -1.08272362e+00 5.32101393e-01
1.28377736e+00 -5.01917958e-01 -9.04152632e-01 5.22614680e-02
1.11671007e+00 -4.11360681e-01 -8.61930251e-01 9.84458923e-01
4.93930250e-01 -6.57354891e-01 9.45104957e-01 -9.86103415e-01
3.13773036e-01 -2.81703383e-01 1.70412600e-01 -1.17637682e+00
-3.75264615e-01 -6.39948785e-01 -1.07584107e+00 1.55375481e+00
5.60241520e-01 -7.36399174e-01 6.60658300e-01 5.85164487e-01
2.51280516e-01 -9.23497200e-01 -7.85869420e-01 -9.89294171e-01
2.44227484e-01 -4.00330126e-01 1.15398216e+00 1.33757067e+00
2.77320296e-02 7.10879385e-01 -4.43333775e-01 3.11459601e-01
2.46517360e-01 -6.07504733e-02 8.45288992e-01 -1.34102654e+00
-2.94518322e-01 -1.02420077e-01 -4.62729096e-01 -5.83122432e-01
1.52818918e-01 -6.89391255e-01 -2.00367123e-01 -1.07776153e+00
1.22664869e-01 -7.18116522e-01 -7.48419344e-01 9.14989769e-01
-6.70172393e-01 4.61821444e-02 1.16047487e-01 2.97521204e-01
-4.84902322e-01 6.65568590e-01 1.28612995e+00 -3.35340023e-01
-1.08893670e-01 2.14165926e-01 -1.15630198e+00 5.42060673e-01
8.58466625e-01 -9.22539711e-01 -6.46574318e-01 -1.50761649e-01
6.54978275e-01 -4.71250892e-01 1.41244501e-01 -1.07814169e+00
2.80862302e-01 -1.01864547e-01 5.65049469e-01 -1.98663890e-01
1.25501588e-01 -8.83574188e-01 1.89014912e-01 4.28479999e-01
-2.54239827e-01 7.49814436e-02 2.90569276e-01 6.10634625e-01
-3.04744154e-01 6.33168370e-02 6.26918435e-01 1.11066416e-01
-5.96918344e-01 4.08156812e-01 5.08130975e-02 1.29004836e-01
1.13041496e+00 7.63775781e-02 -6.07751310e-01 -3.08543801e-01
-7.88596570e-01 1.02908202e-01 2.81546235e-01 8.90770197e-01
7.71607578e-01 -1.54205489e+00 -6.34302735e-01 4.65299189e-01
3.72234643e-01 3.44311148e-01 -1.94462597e-01 2.51771450e-01
-1.26536652e-01 -2.38267705e-01 3.46690193e-02 -3.19826864e-02
-1.00297678e+00 1.04226089e+00 3.23763072e-01 -4.45777476e-01
-7.63019383e-01 1.07831645e+00 8.72735307e-03 -1.71070188e-01
4.20229256e-01 -3.91905278e-01 -9.14094895e-02 -8.57136920e-02
6.15557492e-01 8.85699093e-02 2.54753888e-01 -2.89442927e-01
-4.14547652e-01 3.81775722e-02 -6.91701233e-01 4.66637135e-01
1.38560629e+00 1.69651568e-01 1.20600834e-02 6.77039266e-01
6.72989666e-01 -6.29965439e-02 -1.27871954e+00 -6.09248936e-01
9.70057175e-02 -3.80242556e-01 -4.23559785e-01 -8.54654431e-01
-1.12635005e+00 5.21766841e-01 4.01142925e-01 2.39908367e-01
1.20171523e+00 1.42856449e-01 8.66995990e-01 4.25532877e-01
4.67818320e-01 -8.78449738e-01 2.45232999e-01 6.76197469e-01
9.96218622e-01 -1.10386932e+00 3.42935234e-01 -7.96215951e-01
-5.08658350e-01 9.79148388e-01 1.12347388e+00 -1.62843466e-01
8.23449850e-01 5.14377952e-01 4.96869115e-03 -5.63164949e-02
-1.27083039e+00 -9.82733220e-02 1.43972322e-01 7.44025171e-01
3.22347492e-01 -3.69204842e-02 4.13638018e-02 9.45190966e-01
-4.03160006e-01 9.12945122e-02 4.48310852e-01 1.17531085e+00
-3.30618420e-03 -1.42163873e+00 -3.59292328e-02 4.76030529e-01
-3.90304744e-01 -9.23274532e-02 -5.51419199e-01 8.66239309e-01
5.42068005e-01 8.34033310e-01 -1.18956134e-01 -6.74642265e-01
6.19317114e-01 2.78448552e-01 4.45995212e-01 -8.85783970e-01
-6.55945182e-01 -6.11343563e-01 5.32257594e-02 -3.29221755e-01
-1.84738696e-01 -4.29598153e-01 -1.22634614e+00 -3.71327341e-01
-5.25240123e-01 1.11296989e-01 -2.19946161e-01 8.47261548e-01
3.08158457e-01 8.93803596e-01 5.49475789e-01 3.84914838e-02
-4.24410075e-01 -9.57747281e-01 -2.91112721e-01 5.31911552e-01
5.48790514e-01 -8.81061196e-01 -2.80623287e-01 7.91887343e-02] | [9.19062614440918, 8.515437126159668] |
3750afb3-45b4-430a-97fd-5c2bdf4854ff | learn-and-review-enhancing-continual-named | null | null | https://aclanthology.org/2022.findings-acl.179 | https://aclanthology.org/2022.findings-acl.179.pdf | Learn and Review: Enhancing Continual Named Entity Recognition via Reviewing Synthetic Samples | Traditional methods for named entity recognition (NER) classify mentions into a fixed set of pre-defined entity types. However, in many real-world scenarios, new entity types are incrementally involved. To investigate this problem, continual learning is introduced for NER. However, the existing method depends on the relevance between tasks and is prone to inter-type confusion.In this paper, we propose a novel two-stage framework Learn-and-Review (L&R) for continual NER under the type-incremental setting to alleviate the above issues.Specifically, for the learning stage, we distill the old knowledge from teacher to a student on the current dataset. For the reviewing stage, we first generate synthetic samples of old types to augment the dataset. Then, we further distill new knowledge from the above student and old knowledge from the teacher to get an enhanced student on the augmented dataset. This stage has the following advantages: (1) The synthetic samples mitigate the gap between the old and new task and thus enhance the further distillation; (2) Different types of entities are jointly seen during training which alleviates the inter-type confusion. Experimental results show that L&R outperforms the state-of-the-art method on CoNLL-03 and OntoNotes-5.0. | ['Dai Dai', 'Sujian Li', 'Wenhao Wu', 'Yong Zhu', 'Yajuan Lyu', 'Quan Wang', 'Yu Xia'] | null | null | null | null | findings-acl-2022-5 | ['continual-named-entity-recognition'] | ['natural-language-processing'] | [-1.29998565e-01 3.04005388e-02 4.00452651e-02 -5.92113674e-01
-6.13658845e-01 -6.98376715e-01 3.86714816e-01 3.17661256e-01
-9.67409968e-01 9.48351264e-01 5.91030531e-02 -2.17371255e-01
1.06986508e-01 -7.88074970e-01 -5.52463472e-01 -5.04032254e-01
4.73024786e-01 4.37526703e-01 3.97104472e-01 -1.31629810e-01
-1.37420714e-01 3.11754525e-01 -1.26995754e+00 1.25261158e-01
1.44064605e+00 7.41406620e-01 5.14977336e-01 1.74955189e-01
-6.00856900e-01 6.29727781e-01 -8.93704057e-01 -7.07258582e-01
7.90432841e-02 -2.91128188e-01 -8.21824551e-01 -1.28358409e-01
1.92955449e-01 -1.21434018e-01 -2.48364523e-01 1.12663090e+00
7.76217997e-01 4.88008082e-01 5.43182492e-01 -1.08434010e+00
-9.41428244e-01 1.02140701e+00 -5.46016812e-01 1.67384550e-01
9.49885696e-02 -3.11967172e-02 7.32321203e-01 -1.32133627e+00
5.04135251e-01 1.09101903e+00 6.15264654e-01 8.07839453e-01
-7.19847143e-01 -9.84557629e-01 4.55808014e-01 2.69155204e-01
-1.36131322e+00 -3.74433845e-01 5.67910373e-01 -1.57586157e-01
5.17272830e-01 -2.75059566e-02 3.75529051e-01 7.64138222e-01
-5.54534018e-01 1.11220455e+00 1.09563398e+00 -4.18516606e-01
1.26406163e-01 4.33568716e-01 4.17304903e-01 3.76383781e-01
3.18391919e-01 -2.86535829e-01 -1.81854293e-01 2.07115576e-01
4.17214066e-01 -8.46334025e-02 -4.06570315e-01 4.90546897e-02
-1.03114331e+00 3.49322975e-01 4.37842757e-01 4.45289671e-01
-4.68940526e-01 -3.82061869e-01 4.55989629e-01 1.59807026e-01
2.54673749e-01 4.83448535e-01 -9.25666094e-01 -8.23376179e-02
-8.46003652e-01 1.16850659e-02 7.89998651e-01 1.04364121e+00
9.17073786e-01 -8.77759326e-03 -1.54327050e-01 1.03654253e+00
2.66380280e-01 2.13793173e-01 6.28201425e-01 -2.91772485e-01
8.84295166e-01 8.82167518e-01 8.42854306e-02 -5.11248648e-01
-1.93361536e-01 -5.95041633e-01 -8.86544943e-01 -2.61454582e-01
2.57091701e-01 -6.59003258e-01 -1.07105112e+00 1.72032201e+00
7.66369879e-01 5.33300459e-01 5.53106904e-01 4.78534073e-01
1.38796926e+00 7.86374509e-01 3.68800789e-01 -2.63330162e-01
1.46759331e+00 -1.09789932e+00 -8.88303220e-01 -3.40583056e-01
7.55786180e-01 -7.26211607e-01 8.19658637e-01 1.20956652e-01
-6.90871418e-01 -7.70088851e-01 -8.33077133e-01 -4.80706058e-02
-6.31206930e-01 4.65305775e-01 2.84785390e-01 6.08541369e-01
-5.12870610e-01 3.12314332e-01 -4.09188718e-01 -7.19822869e-02
2.35359445e-01 1.83382228e-01 -3.24825197e-01 -2.23916590e-01
-1.62636852e+00 7.53530383e-01 9.72342670e-01 3.37986112e-01
-4.76580858e-01 -9.44603086e-01 -9.28397834e-01 1.85327038e-01
6.28120720e-01 -4.05600101e-01 1.37804627e+00 -6.93210125e-01
-1.23006189e+00 5.67230999e-01 -2.43416857e-02 -3.57125327e-02
3.70313853e-01 -3.17414999e-01 -6.33993149e-01 -3.31330776e-01
1.33683383e-01 5.96044779e-01 1.67020991e-01 -1.42007089e+00
-9.53129888e-01 -9.49133709e-02 2.39383951e-01 5.27370930e-01
-4.26267236e-01 -1.56400606e-01 -5.34630656e-01 -7.72377014e-01
-9.42924758e-04 -8.46471369e-01 -1.53992265e-01 -5.59720457e-01
-5.45877814e-01 -6.45584106e-01 7.62981951e-01 -4.78979677e-01
1.44831729e+00 -2.22548771e+00 -2.29586050e-01 -4.78408523e-02
2.92771794e-02 9.07712817e-01 -3.03426325e-01 2.50615180e-01
-4.36465591e-01 3.45691890e-01 -1.64871737e-01 -3.63487214e-01
-2.76975513e-01 1.78525552e-01 -3.13818045e-02 -1.71842366e-01
1.71296552e-01 6.88384712e-01 -1.17440069e+00 -5.47706246e-01
-3.11046634e-02 2.41176754e-01 -1.90159053e-01 5.45165956e-01
6.39928803e-02 4.41015750e-01 -5.12185097e-01 3.83777440e-01
8.08094084e-01 1.66135624e-01 -1.99532900e-02 -3.05080473e-01
-2.72769570e-01 5.28137624e-01 -1.72374868e+00 1.36533439e+00
-5.86605668e-01 1.91357329e-01 -3.18712145e-02 -8.33423674e-01
1.02829635e+00 6.12143695e-01 1.49026841e-01 -2.34717667e-01
4.32318524e-02 3.27197909e-01 9.40853357e-02 -5.99847734e-01
5.70990682e-01 -2.27491677e-01 -8.65419954e-02 2.70965457e-01
2.27021083e-01 1.37947217e-01 4.06037331e-01 1.48071244e-01
8.46468508e-01 5.43359295e-02 3.42897743e-01 1.89431190e-01
6.19760394e-01 -1.68058977e-01 1.20924485e+00 4.48761284e-01
-2.77897328e-01 2.22433791e-01 1.99898794e-01 -2.14126527e-01
-4.43185151e-01 -8.35402608e-01 2.03815103e-02 9.67754066e-01
2.22144619e-01 -3.42561901e-01 -7.70521045e-01 -1.33042216e+00
-3.14473897e-01 9.43075478e-01 -3.76587301e-01 -4.09340858e-02
-6.58650398e-01 -7.04754114e-01 7.56681919e-01 7.26499557e-01
9.71679211e-01 -1.34787047e+00 -1.96042389e-01 3.82411540e-01
-4.01123881e-01 -1.12410474e+00 -7.28204012e-01 1.32700771e-01
-6.27035260e-01 -9.95645523e-01 -8.38851273e-01 -1.12245238e+00
9.58495498e-01 2.50681281e-01 9.03219163e-01 3.30907702e-02
1.76820695e-01 8.04397017e-02 -6.00479782e-01 -4.34248269e-01
-3.36375535e-01 2.10678533e-01 3.16700190e-02 3.63473780e-02
5.79049945e-01 -3.70473832e-01 -3.02946478e-01 2.42377579e-01
-8.70243728e-01 -8.80479254e-03 9.55514431e-01 9.09321249e-01
6.35757267e-01 5.04046082e-01 9.68539715e-01 -1.36604667e+00
5.66404581e-01 -6.42223597e-01 -2.99234957e-01 6.12008274e-01
-5.76388299e-01 1.58260211e-01 8.38552952e-01 -7.55007088e-01
-1.64467776e+00 1.31712586e-01 -1.95692822e-01 -2.88333982e-01
-3.00797164e-01 8.54683936e-01 -7.84701049e-01 3.93167913e-01
3.45108718e-01 2.01857969e-01 -7.24327624e-01 -6.92717791e-01
4.41974193e-01 8.71950448e-01 6.65647924e-01 -7.31330752e-01
8.92665446e-01 -3.45938027e-01 -5.46576023e-01 -6.10614598e-01
-1.14889920e+00 -5.54125488e-01 -7.71862566e-01 -6.14958070e-02
6.73432469e-01 -1.02398241e+00 -2.55314052e-01 7.27303982e-01
-1.37214649e+00 -1.22778907e-01 -4.96263832e-01 6.14576221e-01
2.59759277e-01 3.19815725e-01 -5.44078052e-01 -7.75003195e-01
-3.83523256e-01 -1.04768181e+00 5.27513564e-01 1.02009594e+00
1.86537117e-01 -9.31843400e-01 6.45523369e-02 2.44912639e-01
9.11584571e-02 -8.30453113e-02 8.12897861e-01 -1.26972997e+00
-2.41666481e-01 -1.04479864e-01 -2.33694285e-01 3.82001072e-01
3.07448626e-01 -1.39393464e-01 -8.89435828e-01 9.27405246e-03
-9.98986214e-02 -3.59798491e-01 6.14239275e-01 -3.11023891e-01
9.91888523e-01 -4.07560199e-01 -2.65334100e-01 2.59285569e-01
1.12999344e+00 4.62931782e-01 5.20032167e-01 4.47194040e-01
9.71673906e-01 8.50028217e-01 9.13638055e-01 1.70778275e-01
7.79192090e-01 2.54391432e-01 -3.40811610e-02 -2.36083627e-01
-1.59118161e-01 -4.90930974e-01 1.85091481e-01 1.14173257e+00
1.19489737e-01 -3.19317967e-01 -9.11208808e-01 7.77377844e-01
-1.59928703e+00 -8.07003021e-01 -1.11300543e-01 2.18650270e+00
1.42286074e+00 9.83590111e-02 -1.69443786e-01 4.79655489e-02
1.23330498e+00 -3.38488147e-02 -6.43365383e-01 -2.31081508e-02
1.24947228e-01 2.17975527e-01 2.15546921e-01 1.22072846e-01
-1.05399561e+00 1.02830851e+00 4.25372219e+00 1.02586555e+00
-9.51001048e-01 6.44901320e-02 4.66197163e-01 4.76657093e-01
-1.96112990e-01 6.71685264e-02 -1.34272420e+00 5.02404273e-01
7.22539246e-01 -2.92530119e-01 -8.73339623e-02 8.50126684e-01
-2.05336571e-01 6.35044351e-02 -9.84833121e-01 7.64479399e-01
-1.16018750e-01 -6.69478238e-01 7.04202428e-02 -2.46586144e-01
7.73269355e-01 -2.03707650e-01 -2.94295371e-01 9.43810761e-01
4.63127911e-01 -5.29312789e-01 4.58530575e-01 3.32884371e-01
8.13783884e-01 -9.65132058e-01 9.94588614e-01 6.43230915e-01
-1.53956294e+00 1.80724040e-01 -2.75892138e-01 4.63194370e-01
2.35810816e-01 8.35956633e-01 -8.57288778e-01 8.38581204e-01
4.71531808e-01 5.39771795e-01 -5.89906156e-01 1.23833299e+00
-7.81592250e-01 6.39528334e-01 -2.90443957e-01 -1.08627742e-02
1.99353412e-01 -2.97976490e-02 4.62365955e-01 1.29932106e+00
2.92201251e-01 4.79319304e-01 3.12226027e-01 5.21047831e-01
-5.64603865e-01 1.86464027e-01 -2.48210028e-01 9.70550030e-02
1.15417755e+00 1.44675982e+00 -5.43180525e-01 -5.88283241e-01
-4.40100968e-01 9.04412508e-01 5.76624930e-01 3.18826169e-01
-5.89950502e-01 -9.09087121e-01 2.01083496e-01 -2.76403993e-01
3.56794775e-01 3.59118804e-02 5.26815094e-02 -1.31425309e+00
1.20670879e-02 -7.87405908e-01 6.31852746e-01 -4.89375502e-01
-1.38832617e+00 6.76324904e-01 -3.96121144e-02 -1.17400789e+00
1.26708955e-01 -2.69110918e-01 -9.12952185e-01 8.69183362e-01
-1.75968635e+00 -9.56464291e-01 -2.70443469e-01 3.43095988e-01
7.50756145e-01 7.27828778e-03 7.23071933e-01 6.29785836e-01
-9.65595245e-01 9.07198370e-01 -7.49380440e-02 7.64109433e-01
9.90759134e-01 -1.57441449e+00 4.95477974e-01 9.96455133e-01
8.57378095e-02 8.25320184e-01 3.55972409e-01 -8.51317406e-01
-7.74933755e-01 -1.33051050e+00 1.17151666e+00 -1.09401807e-01
3.61946046e-01 1.23804202e-02 -1.23894262e+00 6.59937680e-01
7.80126676e-02 8.87347236e-02 9.60037351e-01 2.28383973e-01
-4.27340239e-01 -1.85928822e-01 -1.22673047e+00 5.33368409e-01
6.15974844e-01 -2.74354875e-01 -1.12526274e+00 1.39987674e-02
6.34375453e-01 -5.51517129e-01 -9.69831467e-01 5.28031170e-01
2.17754409e-01 -4.21923816e-01 5.48193276e-01 -4.49775547e-01
1.68789789e-01 -5.74585915e-01 2.63165116e-01 -1.77971768e+00
-3.03790718e-02 -3.98409486e-01 -1.01908281e-01 2.16943693e+00
6.59155428e-01 -5.43649912e-01 5.35357654e-01 7.32127726e-01
-3.25252444e-01 -8.81261706e-01 -6.60492420e-01 -8.69299531e-01
1.01514295e-01 -7.91493282e-02 7.96661854e-01 1.44681072e+00
-7.61642754e-02 6.82373941e-01 3.77761992e-03 3.29234749e-01
3.21576804e-01 -1.33391082e-01 6.91893697e-01 -1.17089868e+00
-8.84828418e-02 5.73199764e-02 6.23696446e-02 -1.41806555e+00
5.19297235e-02 -8.06028426e-01 3.54023546e-01 -1.57471097e+00
1.19633585e-01 -1.13002157e+00 -5.11709154e-01 8.32226455e-01
-9.60095584e-01 -3.14052135e-01 2.36696437e-01 1.28717795e-01
-5.68470240e-01 6.89177334e-01 1.19364929e+00 2.20000699e-01
-4.27549154e-01 7.63987824e-02 -7.35781491e-01 7.24033296e-01
6.27209306e-01 -7.53695548e-01 -3.58101815e-01 -4.67212379e-01
1.92626435e-02 -3.36064920e-02 -1.96797878e-01 -8.71320844e-01
5.03348291e-01 -1.23065010e-01 2.93377280e-01 -8.19652200e-01
-9.88300983e-03 -7.38470674e-01 -1.80520311e-01 1.48578301e-01
-2.95571446e-01 8.73715356e-02 1.54835001e-01 4.71679717e-01
-2.62118399e-01 -7.76589751e-01 6.96639419e-01 -1.87189177e-01
-8.80353689e-01 4.22026366e-01 2.28470061e-02 6.35695338e-01
9.69119787e-01 4.91340011e-02 -4.34268892e-01 1.93964113e-02
-5.82801223e-01 7.30518103e-01 1.31676674e-01 3.83322597e-01
3.87617141e-01 -1.28281152e+00 -8.27727914e-01 -1.20180167e-01
1.15896940e-01 6.38717234e-01 3.64716917e-01 5.25577307e-01
-4.52386960e-02 -6.94483221e-02 1.11087263e-01 -1.20931923e-01
-1.23684335e+00 4.38372701e-01 2.37704009e-01 -6.66630447e-01
-2.45177940e-01 1.00058317e+00 3.31443369e-01 -9.54440415e-01
2.59624183e-01 -7.86522180e-02 -7.41927445e-01 4.99944896e-01
5.80659091e-01 3.27864319e-01 2.78772060e-02 -5.46413183e-01
-1.97423920e-01 2.43642151e-01 -6.42205119e-01 9.07403696e-03
1.30704641e+00 -1.62226290e-01 1.43002316e-01 5.13777137e-01
8.23364258e-01 3.80367309e-01 -1.03051734e+00 -5.07918656e-01
1.52244374e-01 -2.18237117e-02 -1.94412947e-01 -9.97893691e-01
-1.17596662e+00 7.85501361e-01 2.52758056e-01 4.28749882e-02
1.06504941e+00 -1.92040145e-01 1.13646865e+00 5.87070048e-01
9.29852203e-02 -1.09885788e+00 -2.13071689e-01 7.87448645e-01
3.40497941e-01 -1.05625987e+00 -9.67484638e-02 -6.47630095e-01
-7.64961123e-01 9.13716555e-01 1.14474738e+00 3.00148189e-01
4.65788186e-01 2.34821588e-01 1.91745594e-01 6.25659004e-02
-5.98922670e-01 -3.50277066e-01 1.98086828e-01 4.91726756e-01
6.26001239e-01 -5.48198000e-02 -3.64239514e-01 1.08724129e+00
-1.02775685e-01 -1.70635402e-01 5.88730037e-01 1.00507581e+00
-3.82051528e-01 -1.31203246e+00 -1.63960785e-01 2.11090431e-01
-5.18468678e-01 -3.45704466e-01 -4.24969137e-01 7.90244579e-01
5.44293582e-01 1.02775049e+00 -3.20685804e-01 -1.92990929e-01
7.47494459e-01 2.88334429e-01 6.15341254e-02 -1.09571898e+00
-8.80545557e-01 -2.49290183e-01 2.93545842e-01 2.29895055e-01
-3.35650414e-01 -5.40998518e-01 -1.54673278e+00 2.07276642e-01
-9.29739237e-01 7.82385051e-01 5.39414525e-01 1.19467676e+00
2.47325629e-01 7.08589613e-01 6.96941018e-01 -3.45809579e-01
-5.38664043e-01 -1.13842440e+00 -3.99827421e-01 3.66371244e-01
5.94841875e-02 -6.32373810e-01 -4.28265810e-01 6.53026700e-02] | [9.659504890441895, 9.494075775146484] |
8e6f96ef-8028-4895-976d-ee176cae3630 | copy-and-paste-gan-face-hallucination-from | 2002.10650 | null | https://arxiv.org/abs/2002.10650v3 | https://arxiv.org/pdf/2002.10650v3.pdf | Copy and Paste GAN: Face Hallucination from Shaded Thumbnails | Existing face hallucination methods based on convolutional neural networks (CNN) have achieved impressive performance on low-resolution (LR) faces in a normal illumination condition. However, their performance degrades dramatically when LR faces are captured in low or non-uniform illumination conditions. This paper proposes a Copy and Paste Generative Adversarial Network (CPGAN) to recover authentic high-resolution (HR) face images while compensating for low and non-uniform illumination. To this end, we develop two key components in our CPGAN: internal and external Copy and Paste nets (CPnets). Specifically, our internal CPnet exploits facial information residing in the input image to enhance facial details; while our external CPnet leverages an external HR face for illumination compensation. A new illumination compensation loss is thus developed to capture illumination from the external guided face image effectively. Furthermore, our method offsets illumination and upsamples facial details alternately in a coarse-to-fine fashion, thus alleviating the correspondence ambiguity between LR inputs and external HR inputs. Extensive experiments demonstrate that our method manifests authentic HR face images in a uniform illumination condition and outperforms state-of-the-art methods qualitatively and quantitatively. | ['Yang Zhang', 'Yawei Luo', 'Changhui Hu', 'Xin Yu', 'Xiaobo Lu', 'Ivor Tsang'] | 2020-02-25 | copy-and-paste-gan-face-hallucination-from-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Zhang_Copy_and_Paste_GAN_Face_Hallucination_From_Shaded_Thumbnails_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Zhang_Copy_and_Paste_GAN_Face_Hallucination_From_Shaded_Thumbnails_CVPR_2020_paper.pdf | cvpr-2020-6 | ['face-hallucination'] | ['computer-vision'] | [ 4.96681750e-01 1.77398011e-01 1.87450260e-01 -3.34929347e-01
-7.39008904e-01 -4.71318245e-01 4.94612217e-01 -1.11994636e+00
5.09903617e-02 7.16933310e-01 3.13890606e-01 2.18821943e-01
4.42663193e-01 -8.85148346e-01 -8.48676443e-01 -8.12412322e-01
5.98881006e-01 -2.81489819e-01 -4.74814951e-01 -2.45720148e-01
-2.77875304e-01 9.02287483e-01 -1.64378905e+00 2.62132406e-01
5.69517314e-01 9.68699574e-01 -3.04663241e-01 3.95020723e-01
3.34150612e-01 6.30877316e-01 -5.90944052e-01 -5.65988541e-01
7.23818064e-01 -5.49139678e-01 -3.03057522e-01 1.10995479e-01
9.70496595e-01 -8.35550487e-01 -9.37030375e-01 1.06231940e+00
6.57007635e-01 2.80561927e-03 4.50513452e-01 -1.25389266e+00
-1.51149011e+00 1.03608608e-01 -8.72824252e-01 -2.90030897e-01
4.08603609e-01 4.94384885e-01 3.65390688e-01 -1.29233444e+00
5.81376731e-01 1.59106982e+00 7.36903012e-01 8.67786109e-01
-1.46750832e+00 -1.06356156e+00 -1.91226915e-01 -2.43829399e-01
-1.58768499e+00 -1.01051617e+00 1.02110219e+00 8.44489634e-02
4.71921444e-01 -1.50461625e-02 2.14457333e-01 1.40746474e+00
6.97374716e-02 2.42378592e-01 1.31079900e+00 -1.17254592e-01
-1.22125700e-01 -6.43337965e-02 -7.43776619e-01 6.42005265e-01
1.80827662e-01 4.48571086e-01 -4.62133348e-01 -6.12272546e-02
1.46266127e+00 1.97227225e-01 -5.80719352e-01 6.22429838e-03
-5.98127007e-01 5.64410388e-01 5.83033204e-01 6.59852028e-02
-3.21568996e-01 8.48534033e-02 -2.72038095e-02 3.82445574e-01
3.01567376e-01 2.55998075e-01 -1.46584719e-01 5.48460186e-01
-8.55044067e-01 7.04566762e-02 5.02847612e-01 8.65380466e-01
8.33333313e-01 5.57096601e-01 -3.68063360e-01 1.10753787e+00
1.05787270e-01 7.60876596e-01 1.42178863e-01 -1.29199803e+00
1.46286890e-01 1.36546493e-01 5.24484329e-02 -1.18718243e+00
1.63322929e-02 -4.45699483e-01 -1.20348787e+00 5.40810227e-01
1.57981589e-01 -1.27146700e-02 -9.92314696e-01 2.10841131e+00
2.58797795e-01 1.47758126e-01 2.90861428e-01 1.01008046e+00
1.02707422e+00 3.48584265e-01 -1.11432984e-01 -3.13174844e-01
1.17990184e+00 -8.89794707e-01 -9.58630919e-01 -1.74347356e-01
-4.96790379e-01 -9.10112321e-01 1.16667640e+00 2.81227350e-01
-1.46342421e+00 -8.38154733e-01 -9.81686234e-01 -4.74873662e-01
-3.70793394e-03 2.72517264e-01 9.58993658e-02 4.95644987e-01
-1.35238779e+00 4.20969695e-01 -3.40782553e-01 3.50538082e-02
7.64978409e-01 2.66191244e-01 -7.00159013e-01 -4.19686913e-01
-1.07314384e+00 4.59206730e-01 -3.60414863e-01 4.29394543e-01
-1.05514097e+00 -8.60956490e-01 -1.06820500e+00 9.55762416e-02
8.32630396e-02 -6.62962079e-01 9.07138169e-01 -1.16243112e+00
-1.78817534e+00 8.74211013e-01 -2.55007327e-01 -2.41112318e-02
5.22632539e-01 -1.63113892e-01 -6.30597353e-01 4.38105851e-01
5.02183884e-02 7.04231739e-01 1.44195426e+00 -1.77246439e+00
5.36546670e-02 -4.74661916e-01 -2.09217817e-01 -1.00485886e-04
-2.68836468e-01 5.52867875e-02 -4.96186167e-01 -9.63545084e-01
9.78374705e-02 -5.93106747e-01 2.50771403e-01 3.68785948e-01
-4.16960329e-01 4.59118783e-01 1.23653150e+00 -8.02175939e-01
6.22320116e-01 -2.19216394e+00 -1.58193335e-01 4.02510799e-02
3.81318688e-01 3.52516353e-01 -5.56064069e-01 7.62484893e-02
-3.32974136e-01 -3.66852842e-02 -1.94797203e-01 -4.28762197e-01
-8.98255482e-02 6.56729266e-02 -5.62689006e-01 5.64626336e-01
5.28751075e-01 1.20161617e+00 -6.16680026e-01 -2.04253286e-01
8.14384893e-02 1.25222325e+00 -5.26112735e-01 4.78430778e-01
1.63802475e-01 7.67719507e-01 7.83380941e-02 1.07642698e+00
1.24294794e+00 -7.56717622e-02 -1.20291079e-03 -5.61860979e-01
1.65852174e-01 -3.09124082e-01 -8.34323943e-01 1.27685404e+00
-4.55610812e-01 6.29630506e-01 5.02647936e-01 -3.24092180e-01
1.23859012e+00 3.28416228e-01 2.25593090e-01 -1.03938997e+00
1.23240963e-01 1.59240857e-01 -3.77151906e-01 -1.52515769e-01
4.08286422e-01 -3.11917067e-01 3.00639629e-01 3.24817479e-01
3.07500623e-02 1.39541447e-01 -3.82461101e-01 -1.49545401e-01
9.55371320e-01 1.70143351e-01 5.66986315e-02 2.47271493e-01
7.18077600e-01 -1.04886937e+00 7.24143505e-01 4.19126421e-01
-1.69401646e-01 1.13983119e+00 3.17879051e-01 -3.24335665e-01
-1.11287916e+00 -1.30219591e+00 -1.87214643e-01 7.77489901e-01
4.60289195e-02 1.54368892e-01 -8.64718080e-01 -4.17196900e-01
-9.33515429e-02 1.80413127e-01 -7.87998378e-01 -2.40968987e-01
-6.91464603e-01 -2.05672428e-01 8.93708706e-01 6.06377184e-01
1.06783462e+00 -1.24525595e+00 -1.71113461e-01 -5.44362739e-02
-1.79247603e-01 -1.37147951e+00 -7.05842853e-01 -4.42865133e-01
-2.64111608e-01 -1.01537311e+00 -8.71517837e-01 -7.30536044e-01
9.96183336e-01 4.37924325e-01 1.01838744e+00 -7.67509863e-02
-4.68836099e-01 3.06963980e-01 6.45784736e-02 4.12946939e-02
-2.57279903e-01 -5.62446713e-01 2.18669653e-01 3.92107129e-01
5.46357520e-02 -8.82298827e-01 -9.53634024e-01 4.75641251e-01
-1.13117456e+00 -3.81870046e-02 7.13472545e-01 1.07606792e+00
6.48496449e-01 6.19970495e-03 6.60973489e-01 -6.76753223e-01
3.88768762e-01 -2.02026889e-01 -4.45741504e-01 3.03606838e-02
-3.22288245e-01 -2.29028821e-01 1.06437302e+00 -5.12686253e-01
-1.40078795e+00 -1.61875948e-01 -2.26408884e-01 -1.03861821e+00
-2.04192832e-01 -4.50960636e-01 -6.72924876e-01 -5.38023293e-01
5.66558063e-01 2.74758637e-01 5.68759330e-02 -3.67498845e-01
4.69232827e-01 4.49892849e-01 1.11782253e+00 -4.98096436e-01
1.37141478e+00 7.93230295e-01 1.01681545e-01 -6.42991602e-01
-6.89874649e-01 2.04372719e-01 -4.48088437e-01 -1.91245511e-01
6.57705188e-01 -1.07506418e+00 -8.57318461e-01 6.54962420e-01
-1.01776445e+00 -3.46070021e-01 -3.29274058e-01 5.38920909e-02
-5.43366790e-01 2.00008765e-01 -8.55853617e-01 -6.83250725e-01
-4.75756645e-01 -1.05145812e+00 1.19652236e+00 5.22158563e-01
2.73265868e-01 -4.97122258e-01 -2.80851513e-01 3.33661169e-01
7.90795326e-01 5.54133177e-01 5.69191694e-01 2.29677930e-01
-6.19626522e-01 2.10871771e-02 -7.69135535e-01 6.44274235e-01
4.80560035e-01 -7.14831203e-02 -1.53140247e+00 -5.69873810e-01
3.17177959e-02 -5.16850710e-01 6.81066155e-01 1.36942685e-01
1.34943724e+00 -6.35274708e-01 1.66087568e-01 1.13437474e+00
1.48209882e+00 -8.23428780e-02 1.13147283e+00 -1.15057938e-01
8.44766796e-01 5.88703275e-01 2.88165927e-01 2.59370536e-01
9.24187973e-02 6.11215532e-01 5.15851259e-01 -7.30583847e-01
-6.80737972e-01 -5.12266099e-01 5.11430085e-01 1.22927979e-01
5.37854545e-02 -2.58236099e-02 -3.01872730e-01 2.48818457e-01
-1.30662441e+00 -1.06661069e+00 4.82728571e-01 2.06312561e+00
1.02678657e+00 -4.56443936e-01 -2.08440080e-01 -7.48044923e-02
6.94570363e-01 2.61836857e-01 -8.68221104e-01 -1.13921151e-01
-5.14010012e-01 5.41100860e-01 2.90456802e-01 3.33844602e-01
-7.39587307e-01 9.42041278e-01 6.11859465e+00 5.61711490e-01
-1.19534397e+00 7.36932680e-02 9.67426300e-01 -7.65065029e-02
-3.98150772e-01 -4.62940037e-01 -6.09378517e-01 2.79658377e-01
5.14833510e-01 1.56977370e-01 6.88800037e-01 8.24135780e-01
1.71355560e-01 4.60607201e-01 -7.84917891e-01 1.09080255e+00
4.10317183e-01 -1.04435372e+00 2.92022616e-01 2.33456209e-01
9.61001873e-01 -4.53081280e-01 7.12697089e-01 6.91790730e-02
1.38796106e-01 -1.50232470e+00 3.69393528e-01 7.18791842e-01
1.61287427e+00 -1.07174766e+00 4.75860536e-01 -3.31964612e-01
-1.22193837e+00 -1.52890712e-01 -4.53633189e-01 3.91944826e-01
-3.93479168e-02 2.42124394e-01 -2.38655344e-01 3.87173206e-01
8.10050249e-01 4.58577216e-01 -4.37246144e-01 2.25825265e-01
-4.18800056e-01 1.02776833e-01 8.60223323e-02 1.04438770e+00
-1.81274936e-01 -1.50487483e-01 5.15344024e-01 8.87369514e-01
2.38090456e-01 3.28741223e-01 -1.74290761e-01 1.42007542e+00
-8.01472127e-01 -3.40217739e-01 -7.80694544e-01 2.58678615e-01
5.49715579e-01 1.51187336e+00 -1.69478312e-01 4.35019210e-02
-3.53616565e-01 1.14378786e+00 1.95618138e-01 7.97524989e-01
-9.11514342e-01 -4.47927088e-01 9.96535540e-01 2.31145501e-01
2.33611614e-01 8.47610980e-02 -7.68501312e-02 -1.08728695e+00
1.63189769e-01 -1.10722530e+00 -1.80182606e-01 -9.65638518e-01
-1.51369810e+00 7.44126916e-01 -4.65623379e-01 -1.06987202e+00
-7.45296665e-03 -4.70807195e-01 -6.37739301e-01 1.10127711e+00
-2.05914664e+00 -1.45401382e+00 -8.26681793e-01 1.11409688e+00
2.80778080e-01 -2.00892642e-01 6.73033655e-01 3.51545185e-01
-5.55381298e-01 1.20338178e+00 -2.21830800e-01 4.63486433e-01
1.11501861e+00 -7.24139094e-01 3.26453656e-01 1.03325427e+00
-4.07585591e-01 8.44950736e-01 3.52972329e-01 -4.55364048e-01
-1.59404302e+00 -1.48070765e+00 3.24428886e-01 -9.15324539e-02
3.30222175e-02 -2.95404583e-01 -1.22723746e+00 5.66040099e-01
1.67323276e-01 6.07993245e-01 5.01435876e-01 -5.90069532e-01
-9.96796668e-01 -3.53575826e-01 -1.73937213e+00 7.29772806e-01
1.09423470e+00 -8.20519149e-01 -1.84903890e-01 5.57654304e-03
8.45071793e-01 -3.13450873e-01 -9.71917272e-01 5.73255718e-01
7.44371474e-01 -1.14224434e+00 1.24600875e+00 -1.94673389e-01
7.90039062e-01 -4.42722499e-01 -3.00520301e-01 -1.00833189e+00
-1.96795464e-01 -1.02653158e+00 -2.62171477e-01 1.32658303e+00
-1.04585916e-01 -9.41521883e-01 5.65084994e-01 4.66912955e-01
6.21221121e-03 -5.92314541e-01 -5.36785901e-01 -6.03939295e-01
-4.84220386e-02 2.86614329e-01 8.76327038e-01 8.87645483e-01
-7.04212070e-01 -9.25568715e-02 -6.44742429e-01 3.35004151e-01
1.11427534e+00 -1.27930380e-02 8.88089061e-01 -7.61307418e-01
-1.85674638e-01 -2.42077723e-01 -7.59547278e-02 -7.21345603e-01
4.29982811e-01 -4.60693181e-01 9.56150293e-02 -9.67847586e-01
9.57147628e-02 -1.42360836e-01 -5.93900792e-02 7.25462854e-01
1.09008253e-02 1.12615836e+00 7.46427774e-02 1.95343018e-01
-7.43717849e-02 8.04303885e-01 1.58776379e+00 8.72070193e-02
-2.59190202e-02 -5.14374793e-01 -1.03001225e+00 5.98643482e-01
7.39440620e-01 -1.03959903e-01 -2.38562912e-01 -3.32926035e-01
-1.17721781e-01 1.06091574e-01 7.59706557e-01 -8.89517009e-01
1.18806057e-01 -1.09851934e-01 1.17651439e+00 -3.26650292e-01
5.66018105e-01 -5.90383768e-01 2.96564579e-01 2.43434757e-02
-3.60508710e-01 -1.25902608e-01 1.64281398e-01 5.52854776e-01
-2.65245527e-01 5.66490471e-01 1.44715440e+00 -1.05211526e-01
-1.65009111e-01 7.29998827e-01 2.83217579e-01 -9.43326801e-02
7.24745631e-01 -3.41185778e-01 -7.43578911e-01 -5.24920702e-01
-4.00294363e-01 -3.01210046e-01 9.42690849e-01 4.82718468e-01
1.06662369e+00 -1.62016201e+00 -7.96798468e-01 9.12972212e-01
-7.54590407e-02 -1.70068696e-01 5.09045124e-01 6.28689766e-01
-3.13574165e-01 2.51981542e-02 -6.64563477e-01 -1.67211771e-01
-1.16321516e+00 5.32058060e-01 4.85070407e-01 1.34629026e-01
-7.57770300e-01 6.30127192e-01 6.49789214e-01 -3.26764733e-01
1.94192439e-01 4.73157734e-01 -2.28874832e-02 -3.40779036e-01
9.00610805e-01 2.55869865e-01 -1.55473381e-01 -9.44286823e-01
2.35602930e-02 8.31225276e-01 -8.68847221e-02 -1.07763574e-01
1.31028008e+00 -2.58871943e-01 -2.71329075e-01 -2.42926076e-01
1.51479983e+00 1.87633783e-01 -1.69636190e+00 -4.13067073e-01
-9.18993771e-01 -1.04265070e+00 -1.76024940e-02 -6.55600786e-01
-1.60427034e+00 6.99492276e-01 5.95509410e-01 -2.72858888e-01
1.62791777e+00 -2.92083919e-01 9.56020415e-01 3.99864875e-02
1.68236256e-01 -7.75295734e-01 5.77284694e-01 2.24538386e-01
1.17369246e+00 -1.13320220e+00 -2.24628434e-01 -3.13742071e-01
-3.82423252e-01 1.15778339e+00 8.17308426e-01 -1.82900280e-01
2.85538226e-01 5.25794029e-01 9.19024646e-02 -8.60150084e-02
-4.46126759e-01 2.81436481e-02 2.47599185e-01 8.72476101e-01
2.79330730e-01 -2.54747421e-01 3.14875424e-01 3.63178164e-01
-1.72069922e-01 -4.72240597e-02 4.56764698e-01 4.70386565e-01
-3.70535553e-02 -7.79928803e-01 -5.71915269e-01 4.53891791e-02
-5.49923539e-01 -1.12146407e-01 -3.97324860e-01 8.29056263e-01
1.16595954e-01 8.79766822e-01 2.40829922e-02 -2.55008638e-01
2.63763547e-01 -3.44129860e-01 5.88515401e-01 -2.70659238e-01
-3.75771195e-01 2.32731923e-01 -5.07048666e-01 -1.00079560e+00
-2.48569340e-01 -1.98237538e-01 -1.02362382e+00 -5.32419682e-01
1.28730118e-01 -4.83018905e-01 4.10174549e-01 3.92649144e-01
5.03536582e-01 5.20585895e-01 1.09861171e+00 -1.13693893e+00
-3.71946901e-01 -6.85692132e-01 -5.77903390e-01 3.50248933e-01
7.31687129e-01 -5.13556123e-01 -5.89480400e-01 7.25941136e-02] | [12.806071281433105, -0.06740497797727585] |
445b43f4-d6de-409a-abfd-13bf981ba1e0 | simipu-simple-2d-image-and-3d-point-cloud | 2112.04680 | null | https://arxiv.org/abs/2112.04680v2 | https://arxiv.org/pdf/2112.04680v2.pdf | SimIPU: Simple 2D Image and 3D Point Cloud Unsupervised Pre-Training for Spatial-Aware Visual Representations | Pre-training has become a standard paradigm in many computer vision tasks. However, most of the methods are generally designed on the RGB image domain. Due to the discrepancy between the two-dimensional image plane and the three-dimensional space, such pre-trained models fail to perceive spatial information and serve as sub-optimal solutions for 3D-related tasks. To bridge this gap, we aim to learn a spatial-aware visual representation that can describe the three-dimensional space and is more suitable and effective for these tasks. To leverage point clouds, which are much more superior in providing spatial information compared to images, we propose a simple yet effective 2D Image and 3D Point cloud Unsupervised pre-training strategy, called SimIPU. Specifically, we develop a multi-modal contrastive learning framework that consists of an intra-modal spatial perception module to learn a spatial-aware representation from point clouds and an inter-modal feature interaction module to transfer the capability of perceiving spatial information from the point cloud encoder to the image encoder, respectively. Positive pairs for contrastive losses are established by the matching algorithm and the projection matrix. The whole framework is trained in an unsupervised end-to-end fashion. To the best of our knowledge, this is the first study to explore contrastive learning pre-training strategies for outdoor multi-modal datasets, containing paired camera images and LIDAR point clouds. Codes and models are available at https://github.com/zhyever/SimIPU. | ['Hang Zhao', 'Bolei Zhou', 'Junjun Jiang', 'Xianming Liu', 'Qinhong Jiang', 'Liangji Fang', 'Ang Li', 'Zehui Chen', 'Zhenyu Li'] | 2021-12-09 | null | null | null | null | ['unsupervised-pre-training'] | ['methodology'] | [ 9.15577561e-02 -2.43490368e-01 -1.56350911e-01 -5.00332475e-01
-6.27094388e-01 -3.74368429e-01 5.96406639e-01 -8.52482840e-02
-4.15464550e-01 2.14027703e-01 -1.85649633e-01 -1.99528024e-01
-2.21659243e-01 -9.30439234e-01 -1.01784348e+00 -6.32369757e-01
1.80093527e-01 3.47377926e-01 1.49134278e-01 -2.05662563e-01
3.12144876e-01 6.51218355e-01 -1.97912157e+00 2.36936063e-01
9.69527960e-01 1.13302422e+00 7.38214850e-01 2.63112605e-01
-2.67489225e-01 2.95810103e-01 3.69489677e-02 -1.67716295e-01
5.01850069e-01 -1.07584864e-01 -4.87577826e-01 2.44680777e-01
6.40495181e-01 -1.46940455e-01 -3.96100283e-01 1.11816823e+00
4.40670311e-01 2.61734128e-01 5.42082429e-01 -1.38705397e+00
-8.81506979e-01 -6.80159479e-02 -6.19975984e-01 2.47907899e-02
2.73976147e-01 1.81557328e-01 8.98882449e-01 -1.11799538e+00
3.04998755e-01 1.29267132e+00 4.10065055e-01 4.61115777e-01
-1.18761706e+00 -6.34449303e-01 1.72837108e-01 3.99744362e-01
-1.48815358e+00 -1.20627172e-01 1.27260995e+00 -4.13551658e-01
7.53730297e-01 2.22340047e-01 7.07299590e-01 9.08749044e-01
-1.16866589e-01 7.95305610e-01 1.35059941e+00 -3.75502646e-01
1.69408962e-01 2.80541986e-01 -2.57326633e-01 5.15600502e-01
-1.55250505e-01 5.52969873e-01 -7.20055819e-01 2.48631626e-01
8.84457052e-01 5.66065609e-01 -3.02277625e-01 -1.01563299e+00
-1.05245054e+00 7.57031202e-01 1.14891040e+00 2.95562446e-01
-3.48734230e-01 -2.59460323e-02 2.05194578e-02 1.54931605e-01
4.00027752e-01 1.43159047e-01 -3.35522234e-01 1.00570105e-01
-8.32214475e-01 2.20270809e-02 9.87174213e-02 1.01045072e+00
1.15383947e+00 -2.12490946e-01 1.87008992e-01 7.89627016e-01
6.31788492e-01 8.01211894e-01 3.25731725e-01 -9.80769813e-01
5.53518057e-01 7.14051783e-01 -7.30534792e-02 -1.05149794e+00
-2.34214753e-01 -5.12374341e-01 -9.52708006e-01 5.05130291e-01
5.05537465e-02 5.34743369e-01 -9.07248497e-01 1.73151684e+00
3.26805204e-01 3.67050976e-01 5.22875078e-02 1.28629172e+00
6.89842522e-01 7.68130660e-01 -6.69614747e-02 1.00884564e-01
1.16149259e+00 -8.44296217e-01 -2.43284598e-01 -4.21301335e-01
3.35530341e-01 -5.84154606e-01 1.47146833e+00 1.47996083e-01
-1.12360668e+00 -8.97831380e-01 -1.12429225e+00 -3.72605115e-01
-6.35131657e-01 1.78490162e-01 5.14253497e-01 3.15023869e-01
-8.70911837e-01 3.55228990e-01 -7.78443575e-01 -3.45770866e-01
5.45924962e-01 7.68444166e-02 -5.29612780e-01 -5.35325766e-01
-1.00013435e+00 9.20358360e-01 3.81272107e-01 1.36890754e-01
-6.83044314e-01 -6.91007793e-01 -1.04779470e+00 -1.01001501e-01
1.35349721e-01 -7.47637928e-01 9.00491238e-01 -9.38732505e-01
-1.24836874e+00 1.11694205e+00 -1.74116373e-01 -1.02800004e-01
2.20154062e-01 -1.63533583e-01 -3.48131172e-02 1.84878170e-01
2.65826225e-01 7.49824822e-01 9.62879241e-01 -1.66892755e+00
-5.52889705e-01 -7.32197881e-01 1.70476556e-01 6.96200252e-01
-2.67145127e-01 -4.69976842e-01 -6.33238018e-01 -3.99404496e-01
5.25564075e-01 -8.30964684e-01 -5.96076846e-02 4.33345973e-01
-1.90331310e-01 -5.85922152e-02 7.87740052e-01 -3.68670374e-01
6.00827456e-01 -2.38789797e+00 4.42232132e-01 1.35200903e-01
1.29206171e-02 1.45275965e-01 -1.94253430e-01 1.35127679e-01
-9.78873521e-02 -1.16946682e-01 -4.39694345e-01 -5.92001021e-01
1.01506919e-01 3.62420052e-01 -3.43358248e-01 5.57963789e-01
2.85030246e-01 7.91743636e-01 -9.91363347e-01 -4.15363520e-01
8.15005839e-01 7.10164011e-01 -4.60052252e-01 3.31798017e-01
-9.88719016e-02 6.80874884e-01 -3.22007596e-01 6.92314744e-01
1.14853644e+00 -1.93259239e-01 -4.09618109e-01 -4.20883477e-01
-2.61941224e-01 7.55382180e-02 -1.05890656e+00 2.40264106e+00
-7.33370543e-01 2.91038603e-01 7.86507223e-03 -1.09827149e+00
9.55582142e-01 -1.14110425e-01 4.35935348e-01 -1.07891679e+00
-3.75254899e-02 2.93418974e-01 -4.75725144e-01 -3.71651918e-01
2.55718231e-01 -3.47371072e-01 8.87333229e-02 -6.20066188e-02
2.28146046e-01 -6.17294550e-01 -1.36501640e-01 -4.96189371e-02
6.52436435e-01 2.70135373e-01 1.92479510e-02 1.76210716e-01
6.71854854e-01 1.03462965e-03 4.07841176e-01 2.80205637e-01
-7.05439970e-02 8.90864313e-01 -1.72494769e-01 -3.01256955e-01
-9.45436001e-01 -1.51397681e+00 -3.41967136e-01 7.25011945e-01
6.31174445e-01 -2.45524362e-01 -1.98037446e-01 -4.21064705e-01
1.09400019e-01 6.96834266e-01 -3.60521317e-01 -3.42892170e-01
-2.56754935e-01 -1.16606787e-01 6.29026592e-02 4.97538745e-01
7.83538043e-01 -7.39079297e-01 -6.17045283e-01 -1.84431329e-01
-1.79629534e-01 -1.06759810e+00 -3.20876062e-01 2.45453835e-01
-1.01377702e+00 -1.02770817e+00 -7.14922130e-01 -7.93677330e-01
7.19486713e-01 8.39032769e-01 1.07083809e+00 -1.05254441e-01
-1.84406325e-01 6.00326896e-01 -3.77148926e-01 -4.03964162e-01
1.46872103e-01 -2.88117468e-01 4.01688255e-02 -9.18153673e-03
5.41881323e-01 -8.84432375e-01 -6.61260605e-01 3.67854595e-01
-9.60623860e-01 2.76069105e-01 8.80278409e-01 7.81776786e-01
1.00506878e+00 6.04199572e-03 -1.34184539e-01 -2.60313183e-01
5.45952171e-02 -4.88209546e-01 -5.44989049e-01 1.24752790e-01
-4.21164274e-01 -4.58773039e-02 6.13721013e-01 -1.14551075e-01
-8.91688406e-01 4.37697917e-01 -1.69771999e-01 -1.08134305e+00
-4.18672979e-01 4.00557429e-01 -3.76210868e-01 -2.07888350e-01
5.14549255e-01 4.76205796e-01 -2.65482906e-02 -5.73279977e-01
5.93013465e-01 4.90130544e-01 5.79698801e-01 -5.64673007e-01
1.17725837e+00 7.09495962e-01 1.21718407e-01 -7.82481909e-01
-7.88233459e-01 -8.04025710e-01 -9.70376790e-01 -2.19754606e-01
9.21007812e-01 -1.15905344e+00 -4.29069042e-01 2.73835272e-01
-1.18457830e+00 -9.98978242e-02 -3.08259517e-01 6.26734495e-01
-8.17766905e-01 4.33398902e-01 -6.53295591e-02 -7.01036155e-01
3.37314443e-03 -9.57466841e-01 1.31336188e+00 2.12996662e-01
4.26731884e-01 -7.59175956e-01 2.04588220e-01 4.71695513e-01
1.60229400e-01 1.08400606e-01 6.78948224e-01 -4.94923741e-02
-8.70044351e-01 8.40578303e-02 -4.61368978e-01 5.86731553e-01
1.54433120e-02 -4.45302218e-01 -1.13967121e+00 -3.38295698e-01
2.23445259e-02 -5.36419392e-01 9.11941469e-01 2.88037807e-01
1.31368482e+00 7.97071457e-02 -2.11933345e-01 1.02796662e+00
1.77195811e+00 -2.57510170e-02 6.30684137e-01 3.45877230e-01
8.65563333e-01 6.26987278e-01 7.63037026e-01 2.60582864e-01
7.10856199e-01 8.70551884e-01 9.25663114e-01 -1.94717214e-01
-2.75492817e-02 -5.50166190e-01 1.86360344e-01 6.28136039e-01
-1.29393071e-01 5.51109426e-02 -9.70799804e-01 4.53302562e-01
-1.78538966e+00 -1.00088513e+00 1.60318077e-01 2.28035665e+00
5.23462355e-01 -4.90698218e-02 -1.86550021e-01 1.84503138e-01
4.41524595e-01 2.19576195e-01 -6.30230784e-01 2.33155172e-02
-2.27051184e-01 1.54765889e-01 3.41826260e-01 3.28530669e-01
-1.18418300e+00 7.72733092e-01 4.39874506e+00 6.95334375e-01
-1.29841125e+00 1.54042616e-01 2.73918986e-01 -5.56106456e-02
-3.44619989e-01 -1.48362936e-02 -3.31026286e-01 5.72085619e-01
5.37382305e-01 2.21362531e-01 4.47626293e-01 8.78998518e-01
4.93150018e-02 8.70598927e-02 -1.19464505e+00 1.65077245e+00
1.53619722e-01 -1.20796132e+00 2.22355276e-02 1.04750998e-01
4.55257863e-01 2.15238914e-01 3.89483422e-01 2.44216368e-01
-2.00954378e-01 -9.53760922e-01 8.84409904e-01 7.15215981e-01
7.19706595e-01 -6.47490621e-01 5.37094772e-01 5.01717448e-01
-1.26462412e+00 -2.17274174e-01 -6.33051991e-01 -2.51185223e-02
1.18351474e-01 5.46754718e-01 -1.74559146e-01 7.77879477e-01
1.10680282e+00 1.06031859e+00 -6.01406157e-01 1.24060810e+00
-1.44884959e-01 -9.22488887e-03 -4.33066756e-01 2.73014456e-01
3.14564407e-01 -4.25029755e-01 4.96027321e-01 8.23462784e-01
6.04059279e-01 2.53050447e-01 2.03824326e-01 1.14074993e+00
1.14523128e-01 -4.86453548e-02 -9.22523916e-01 2.41511613e-01
5.08204043e-01 1.05939984e+00 -2.29020000e-01 -2.60129813e-02
-6.28645182e-01 1.09497452e+00 3.63178611e-01 3.71156514e-01
-7.36286342e-01 -1.98506981e-01 6.98387444e-01 1.18521512e-01
3.62600029e-01 -4.96057034e-01 -4.05359626e-01 -1.26006138e+00
3.10041785e-01 -5.23111880e-01 1.30979195e-01 -1.16373944e+00
-1.43121707e+00 4.92998987e-01 7.92427436e-02 -1.79734278e+00
8.71572923e-03 -8.29553545e-01 -5.14382064e-01 1.08609366e+00
-1.97009599e+00 -1.36899686e+00 -7.38806903e-01 9.93111789e-01
3.60279262e-01 1.13128396e-02 6.37914240e-01 3.89437973e-01
-2.00267106e-01 3.21304470e-01 4.60701436e-02 -1.52505726e-01
5.47634721e-01 -1.26983809e+00 -1.00652270e-01 7.23668993e-01
2.65035510e-01 5.02803266e-01 4.61467683e-01 -1.37558952e-01
-1.60469246e+00 -1.09231400e+00 5.37234128e-01 -4.34365243e-01
1.53916746e-01 -4.32180583e-01 -9.12586093e-01 3.79344165e-01
3.92242707e-02 2.37307966e-01 6.17972612e-01 -2.49725226e-02
-5.10545194e-01 -3.74681443e-01 -1.08446240e+00 4.18516815e-01
1.20760095e+00 -8.55827510e-01 -5.75788915e-01 2.85992801e-01
6.52733624e-01 -4.61270839e-01 -8.36073875e-01 4.67371434e-01
1.38763249e-01 -1.22531438e+00 1.41744328e+00 -1.44747511e-01
5.34738183e-01 -5.84597886e-01 -6.68993533e-01 -1.36001658e+00
-3.73994231e-01 2.73552705e-02 -6.06147088e-02 1.04248893e+00
9.22977179e-02 -4.00303870e-01 7.93227077e-01 2.57452846e-01
-3.93857569e-01 -7.41195619e-01 -1.12369514e+00 -7.59892464e-01
1.54463008e-01 -6.28095448e-01 4.64286685e-01 8.53426158e-01
-3.24784189e-01 3.58934939e-01 -1.69868976e-01 4.95875448e-01
9.34500933e-01 4.46952790e-01 6.79799795e-01 -1.21513820e+00
-2.11551860e-01 -4.52796668e-01 -7.24819243e-01 -1.33521628e+00
1.69701710e-01 -1.05981863e+00 1.14320412e-01 -1.59962475e+00
8.40463415e-02 -6.33925319e-01 -3.68196040e-01 3.51292074e-01
2.02786341e-01 2.98745036e-01 4.48037118e-01 3.54025155e-01
-4.36264604e-01 9.87512171e-01 1.32745004e+00 -4.11519080e-01
-1.50849134e-01 -3.54784168e-02 -5.75061977e-01 4.90658790e-01
8.14842403e-01 -1.28466949e-01 -7.17147052e-01 -7.66758502e-01
8.23578909e-02 -1.14943258e-01 8.97065938e-01 -1.16385019e+00
3.66823465e-01 -2.71614641e-01 5.32050133e-01 -9.34177935e-01
7.85339773e-01 -1.15143108e+00 -5.19340001e-02 1.35823831e-01
-7.67746717e-02 -8.46774653e-02 1.56742871e-01 6.32654011e-01
-5.64348996e-01 7.93673098e-02 8.57711315e-01 -2.61979848e-01
-1.06332219e+00 6.86944246e-01 2.28651777e-01 -1.99890345e-01
9.69341099e-01 -4.50545818e-01 -1.38273716e-01 -1.48933530e-01
-5.13215542e-01 3.02211314e-01 7.55994558e-01 4.49732423e-01
1.20329714e+00 -1.64541924e+00 -4.39632416e-01 4.03541982e-01
4.94924158e-01 4.73398387e-01 5.34746706e-01 8.29494357e-01
-4.62332547e-01 4.34780866e-01 -6.28171086e-01 -1.04631305e+00
-1.01327610e+00 8.56457591e-01 3.62509787e-01 1.94971368e-01
-6.37471497e-01 8.26155663e-01 4.90601987e-01 -8.21593761e-01
3.43310416e-01 -3.38491440e-01 1.84711965e-03 -2.80736536e-01
3.76708865e-01 7.53740892e-02 8.65366776e-04 -9.55788791e-01
-5.11737168e-01 1.15937185e+00 3.79685253e-01 -8.08323547e-02
1.29924893e+00 -3.87674451e-01 -9.72950906e-02 5.59933543e-01
1.63849318e+00 -4.20942694e-01 -1.48089695e+00 -4.96886700e-01
-2.87133247e-01 -8.88144791e-01 3.22517574e-01 -5.33522189e-01
-1.07865012e+00 1.38439715e+00 1.04255295e+00 -3.98950279e-03
1.47087312e+00 1.85796931e-01 5.68770051e-01 3.29209119e-01
5.07467687e-01 -9.04464066e-01 2.96250850e-01 4.37416911e-01
1.13908947e+00 -1.59580541e+00 -1.28370374e-01 -4.12477732e-01
-5.95767915e-01 8.72699380e-01 7.98481762e-01 -1.72646582e-01
8.82968426e-01 -3.29854995e-01 -3.58091816e-02 -2.47547433e-01
-4.34044302e-01 -5.52897632e-01 4.49769229e-01 9.86273944e-01
4.50794101e-02 2.29479535e-03 2.10079476e-01 3.35134357e-01
-9.72255543e-02 -1.34910479e-01 -1.53940991e-01 8.17648053e-01
-3.74705642e-01 -9.00697231e-01 -3.93557549e-01 5.22820391e-02
3.68635178e-01 2.02031191e-02 -1.99819282e-01 6.37870967e-01
5.04451036e-01 7.92193651e-01 3.06754589e-01 -6.30455017e-01
6.06502056e-01 -2.38898799e-01 6.78535759e-01 -5.03070891e-01
6.02350384e-02 -1.84976399e-01 -5.18208385e-01 -7.71659553e-01
-7.11815000e-01 -5.12259185e-01 -1.21669030e+00 -2.42616057e-01
-2.35381685e-02 -4.64110039e-02 7.90512741e-01 6.32481217e-01
4.04164165e-01 2.14745954e-01 9.01105821e-01 -1.40671539e+00
-3.30074549e-01 -5.81677198e-01 -6.19408965e-01 5.48753619e-01
3.90246511e-01 -1.09726667e+00 -3.62892628e-01 -1.30983561e-01] | [8.352745056152344, -3.093148946762085] |
98e12f51-77bd-43ff-a90e-22f1450e1168 | icdar-2023-competition-on-structured-text | 2306.03287 | null | https://arxiv.org/abs/2306.03287v1 | https://arxiv.org/pdf/2306.03287v1.pdf | ICDAR 2023 Competition on Structured Text Extraction from Visually-Rich Document Images | Structured text extraction is one of the most valuable and challenging application directions in the field of Document AI. However, the scenarios of past benchmarks are limited, and the corresponding evaluation protocols usually focus on the submodules of the structured text extraction scheme. In order to eliminate these problems, we organized the ICDAR 2023 competition on Structured text extraction from Visually-Rich Document images (SVRD). We set up two tracks for SVRD including Track 1: HUST-CELL and Track 2: Baidu-FEST, where HUST-CELL aims to evaluate the end-to-end performance of Complex Entity Linking and Labeling, and Baidu-FEST focuses on evaluating the performance and generalization of Zero-shot / Few-shot Structured Text extraction from an end-to-end perspective. Compared to the current document benchmarks, our two tracks of competition benchmark enriches the scenarios greatly and contains more than 50 types of visually-rich document images (mainly from the actual enterprise applications). The competition opened on 30th December, 2022 and closed on 24th March, 2023. There are 35 participants and 91 valid submissions received for Track 1, and 15 participants and 26 valid submissions received for Track 2. In this report we will presents the motivation, competition datasets, task definition, evaluation protocol, and submission summaries. According to the performance of the submissions, we believe there is still a large gap on the expected information extraction performance for complex and zero-shot scenarios. It is hoped that this competition will attract many researchers in the field of CV and NLP, and bring some new thoughts to the field of Document AI. | ['Xiang Bai', 'Jingdong Wang', 'Xing Sun', 'Dimosthenis Karatzas', 'Min Zhang', 'Shuicheng Yan', 'Jiebo Luo', 'Cheng-Lin Liu', 'Errui Ding', 'Wanxiang Che', 'Yuliang Liu', 'Yuechen Yu', 'Kun Yao', 'Pengyuan Lyu', 'Xiaoguang Hu', 'Shikun Feng', 'Yuning Du', 'Mengjun Cheng', 'Jianfeng Kuang', 'Mingrui Chen', 'MingYu Liu', 'Huang Chen', 'Bohan Li', 'Wei Hua', 'Haoyu Cao', 'Chengquan Zhang', 'Wenwen Yu'] | 2023-06-05 | null | null | null | null | ['document-ai', 'entity-linking'] | ['natural-language-processing', 'natural-language-processing'] | [ 9.12492499e-02 1.40839547e-01 -2.46734381e-01 -9.69019085e-02
-8.93935204e-01 -7.93373048e-01 8.69827569e-01 2.87474573e-01
-5.40632606e-01 5.48767626e-01 3.44299704e-01 -3.05530354e-02
-1.37796223e-01 -4.24203873e-01 -3.72830003e-01 -2.27752432e-01
-7.30070993e-02 6.30547643e-01 2.06551984e-01 4.00031991e-02
2.16986015e-01 2.72957534e-01 -1.32946157e+00 8.44524205e-01
6.79615736e-01 7.87330866e-01 3.02312672e-01 7.72528768e-01
-6.27531707e-01 8.34725201e-01 -9.48190331e-01 -6.63409472e-01
2.25755304e-01 -2.66310900e-01 -1.16020250e+00 2.29831003e-02
6.61692202e-01 -5.00254184e-02 -3.56173486e-01 9.49548900e-01
7.98568070e-01 -1.42694429e-01 6.49131536e-01 -1.33358872e+00
-5.97600877e-01 8.95226300e-01 -7.68618524e-01 2.06317872e-01
3.75697732e-01 -5.17678745e-02 1.22090089e+00 -1.37630975e+00
1.35814726e+00 1.16838479e+00 4.63312060e-01 5.45835614e-01
-6.55323267e-01 -4.47637856e-01 4.55373764e-01 2.20528528e-01
-1.27856481e+00 -3.90827328e-01 3.46258223e-01 -6.33717835e-01
9.52635527e-01 3.31550181e-01 3.96354616e-01 1.26176107e+00
-1.04348846e-01 1.71094048e+00 7.70406544e-01 -6.43819630e-01
5.97860403e-02 3.24456066e-01 5.03591061e-01 4.90512639e-01
5.97498357e-01 -2.31745586e-01 -5.06687760e-01 1.25367597e-01
1.00758243e-02 -3.99922311e-01 -2.35257372e-01 -4.23628569e-01
-1.39193690e+00 6.69669867e-01 6.22825846e-02 5.42011619e-01
-1.05713747e-01 -2.87590295e-01 9.31079268e-01 4.28495987e-04
5.02629399e-01 4.12115842e-01 -1.78230420e-01 -9.50236022e-02
-1.40220582e+00 6.52793348e-01 8.05478990e-01 1.40007257e+00
7.31094703e-02 -2.27643430e-01 -9.21535671e-01 9.04211819e-01
4.48038392e-02 4.33711857e-01 2.26655394e-01 -6.26777589e-01
9.16597843e-01 6.53976977e-01 5.56024946e-02 -8.43360305e-01
-6.10562503e-01 -3.18288624e-01 -9.00892198e-01 8.71465281e-02
3.67967874e-01 -1.85733005e-01 -1.11086369e+00 1.12534249e+00
1.73423678e-01 -6.67798042e-01 -5.32826632e-02 8.30059826e-01
1.49461937e+00 7.72852480e-01 2.94611901e-02 -3.36945355e-01
1.73144639e+00 -1.11417973e+00 -1.07032180e+00 -1.77223250e-01
8.17324340e-01 -1.03252065e+00 1.17242777e+00 4.55546260e-01
-1.25399041e+00 -4.13431257e-01 -9.53261018e-01 -3.39995146e-01
-7.65767038e-01 4.43922520e-01 2.68157363e-01 5.32108545e-01
-6.13402009e-01 8.09459239e-02 -3.78512114e-01 -5.79098403e-01
6.54288292e-01 7.64238928e-03 -2.69188195e-01 -3.59253466e-01
-1.10294139e+00 8.22680771e-01 5.25416911e-01 -5.84656820e-02
-4.77312505e-01 -6.31535292e-01 -5.22140384e-01 -1.28092289e-01
6.99237823e-01 -6.42957091e-01 1.22450149e+00 -2.30365336e-01
-7.34524786e-01 1.23829818e+00 3.28750834e-02 -4.90305632e-01
6.60734773e-01 -2.74643511e-01 -5.04152179e-01 2.00827241e-01
3.53602082e-01 7.41437256e-01 2.96412438e-01 -1.18196750e+00
-8.79180253e-01 -5.54056466e-01 -2.20298633e-01 3.01629722e-01
-5.10834932e-01 4.73136038e-01 -1.15697062e+00 -8.38211775e-01
-5.44922769e-01 -7.46452630e-01 5.78476489e-02 -1.52653247e-01
-8.08297455e-01 -4.80663866e-01 1.08826029e+00 -6.85745418e-01
1.39625382e+00 -2.23534822e+00 -6.46753833e-02 -9.95544121e-02
4.39010054e-01 4.24991667e-01 -2.67176956e-01 5.26544452e-01
1.65317327e-01 4.10365760e-01 -2.00706065e-01 -5.15366316e-01
2.78248906e-01 -3.82073790e-01 -3.71131480e-01 5.59954979e-02
1.77731656e-03 1.10657883e+00 -7.49607086e-01 -8.48997533e-01
-4.61493582e-02 2.99955577e-01 1.05606012e-01 -1.37127757e-01
-4.22332674e-01 -1.07389383e-01 -5.87985218e-01 6.37964785e-01
7.36936271e-01 -2.67274141e-01 2.81409584e-02 -4.87286568e-01
-2.11713254e-01 2.08782166e-01 -1.29944253e+00 1.65257120e+00
6.85516819e-02 1.08277774e+00 1.11538924e-01 -7.12762833e-01
7.95777857e-01 3.89192611e-01 6.51806831e-01 -1.01578581e+00
6.39938787e-02 -1.66491598e-01 -4.05278534e-01 -4.25014108e-01
1.01835179e+00 4.18223351e-01 -3.29994261e-01 2.38206670e-01
2.00770441e-02 -2.10594639e-01 1.07262361e+00 8.23010087e-01
1.30681443e+00 1.90432712e-01 2.23179415e-01 -1.03253148e-01
3.04564506e-01 4.65645254e-01 1.30898893e-01 9.01793897e-01
-1.62591577e-01 7.18299747e-01 5.71170032e-01 -4.37839240e-01
-1.18646812e+00 -9.12105143e-01 -1.22013345e-01 8.24005604e-01
1.48338318e-01 -9.52421904e-01 -8.41100454e-01 -1.07061195e+00
-1.89235568e-01 7.14081943e-01 -5.58863938e-01 2.28220582e-01
-4.75057572e-01 -6.95089996e-01 9.29388821e-01 5.13942897e-01
7.04296112e-01 -1.29089499e+00 -4.20862466e-01 2.61609573e-02
-4.76909518e-01 -1.56973946e+00 -5.81520617e-01 8.48305225e-02
-4.88267601e-01 -1.14053857e+00 -1.09239173e+00 -7.79785514e-01
2.66003966e-01 3.18614751e-01 1.34589291e+00 -1.56599790e-01
-6.96764767e-01 5.04091024e-01 -5.50594032e-01 -8.47805321e-01
-2.05559880e-01 2.96552151e-01 -3.96360725e-01 -3.34061027e-01
5.94022155e-01 5.18169582e-01 -4.85908359e-01 1.87665462e-01
-9.51008260e-01 7.48629197e-02 5.50664067e-01 5.46396971e-01
6.28715813e-01 1.18444987e-01 2.84762472e-01 -1.16565609e+00
9.36344564e-01 -2.12721363e-01 -5.93158066e-01 5.42651713e-01
-8.28077853e-01 -1.94878623e-01 3.07003379e-01 -4.31599049e-03
-1.15321660e+00 -4.75474782e-02 2.33029410e-01 -1.14030331e-01
-1.31707713e-01 4.84970093e-01 -2.20283180e-01 5.72457016e-01
8.48045588e-01 -1.20842345e-02 -6.68201745e-01 -6.52583659e-01
4.93218631e-01 9.00710762e-01 6.20134771e-01 -5.37247181e-01
7.98292518e-01 4.08380777e-01 -2.77883500e-01 -8.70591760e-01
-9.45030808e-01 -7.74114728e-01 -4.75356042e-01 -1.93716034e-01
9.70165730e-01 -8.88359666e-01 -3.33908767e-01 3.76475751e-01
-1.28490329e+00 -1.33881763e-01 -6.42485917e-01 2.40072355e-01
-8.20964351e-02 5.48412681e-01 -5.22335410e-01 -7.61934876e-01
-8.03519130e-01 -9.78709757e-01 1.31287861e+00 8.59249756e-03
-3.68773848e-01 -7.62089252e-01 1.42956838e-01 6.03228331e-01
-7.06759617e-02 2.65853494e-01 7.25296259e-01 -6.50024772e-01
-3.59465092e-01 -4.04869080e-01 -5.36150932e-01 1.49863958e-01
-2.86602139e-01 2.18057111e-01 -8.72975647e-01 -2.46985331e-01
-5.47613084e-01 -3.63476425e-01 1.11483037e+00 4.14145827e-01
1.08587885e+00 -1.08377457e-01 -5.80075741e-01 4.45219763e-02
1.19201100e+00 2.52304226e-01 8.38068187e-01 4.82788771e-01
7.68214166e-01 8.11593354e-01 9.08438861e-01 3.43320787e-01
3.88743877e-01 9.12843823e-01 1.55016437e-01 -1.42538920e-01
-5.92512071e-01 -2.08758712e-01 4.28441912e-01 5.96331835e-01
6.52171718e-03 -8.16425502e-01 -1.08099103e+00 7.11018324e-01
-1.74451518e+00 -1.06133103e+00 -5.64426184e-01 2.07024288e+00
7.69408822e-01 2.80749381e-01 3.87359977e-01 3.27604711e-01
8.52689087e-01 1.10647932e-01 -3.00365895e-01 -3.01569402e-01
-5.33723474e-01 7.01520359e-03 2.03426570e-01 8.87295008e-02
-1.15014470e+00 9.20424759e-01 6.19573069e+00 1.20834744e+00
-6.11482859e-01 -9.81297623e-03 4.21686709e-01 -2.97806054e-01
-1.01846382e-01 -4.24844436e-02 -1.27924573e+00 4.09892648e-01
7.49966502e-01 -2.62821317e-01 -1.56585813e-01 6.97595239e-01
-1.81964990e-02 -1.11417346e-01 -1.07828140e+00 9.12757277e-01
1.48798645e-01 -1.65765417e+00 1.41744539e-01 1.93011969e-01
7.45994151e-01 7.23031461e-02 -1.55033201e-01 4.67556119e-01
2.79335856e-01 -1.03489435e+00 7.20366478e-01 3.85587007e-01
9.29921865e-01 -6.11686647e-01 7.56913543e-01 2.88420558e-01
-1.27170837e+00 1.62516713e-01 -2.01234832e-01 5.55756867e-01
2.11954191e-01 9.08483028e-01 -8.87654424e-01 7.55053461e-01
9.44004595e-01 6.02222025e-01 -7.84136295e-01 1.25574696e+00
-1.20268725e-01 4.19008553e-01 -4.94745141e-03 -3.17465574e-01
2.20078781e-01 1.89612627e-01 7.38656282e-01 1.95463645e+00
-7.05566555e-02 -2.12565288e-01 1.87821910e-01 6.93198621e-01
-5.21761239e-01 4.30487245e-01 -6.37875021e-01 -2.70783454e-01
3.39494884e-01 1.38500869e+00 -9.76566553e-01 -5.71458995e-01
-4.70114142e-01 8.59413862e-01 -5.14116064e-02 2.74441719e-01
-4.67950910e-01 -8.73377919e-01 -1.29763752e-01 -4.31358628e-02
4.32835609e-01 -1.22949764e-01 -6.99985027e-01 -1.11714566e+00
3.36237282e-01 -1.07132792e+00 7.34295547e-01 -9.07544017e-01
-1.18200493e+00 7.38169074e-01 1.85768768e-01 -1.06960750e+00
-6.23811446e-02 -6.39902472e-01 -1.34129718e-01 5.45770347e-01
-1.22287786e+00 -1.09777999e+00 -4.77792114e-01 5.54131508e-01
8.96297753e-01 -3.00890744e-01 7.55265415e-01 2.87446469e-01
-5.91491818e-01 7.70367682e-01 1.91620678e-01 5.38001060e-01
8.59735489e-01 -1.28822207e+00 6.90549850e-01 8.94318163e-01
6.32530153e-01 1.59799054e-01 5.42103231e-01 -7.06941068e-01
-1.18121982e+00 -1.09582126e+00 1.01000762e+00 -7.74898827e-01
3.88159275e-01 -6.53834760e-01 -6.89683139e-01 4.24721777e-01
5.19765139e-01 -1.75691843e-01 4.30488110e-01 -4.38658856e-02
-3.92293304e-01 1.29033834e-01 -1.07543278e+00 6.92991138e-01
1.01138473e+00 -2.59075701e-01 -5.71263790e-01 6.63775086e-01
6.68482482e-01 -4.97648746e-01 -8.09482634e-01 4.25737381e-01
3.54203969e-01 -5.21054208e-01 7.96433210e-01 -5.74571609e-01
5.94298184e-01 -2.36147106e-01 4.49196494e-04 -6.68150842e-01
-1.17312841e-01 -5.12504876e-01 -2.80443162e-01 1.47977769e+00
5.71894765e-01 2.19440162e-01 8.30318034e-01 2.24636838e-01
-1.16096370e-01 -6.83613181e-01 -5.89623272e-01 -9.89400864e-01
2.26961821e-01 -4.76421207e-01 1.96007818e-01 7.41904676e-01
-6.11207895e-02 8.10104072e-01 -3.04169118e-01 -3.51694733e-01
7.56546974e-01 8.45371336e-02 9.33046162e-01 -1.23790264e+00
9.66258794e-02 -5.43636322e-01 -1.61758497e-01 -1.09251952e+00
-2.32689187e-01 -1.06807065e+00 -1.84288919e-02 -2.18212366e+00
5.34097493e-01 2.47489270e-02 7.05985073e-03 5.75017214e-01
-1.74865033e-03 3.17805618e-01 5.60017049e-01 3.06707710e-01
-9.99332607e-01 2.73033679e-01 1.28491926e+00 -5.61747253e-01
-3.89186144e-02 -1.08034931e-01 -6.53927505e-01 4.70383674e-01
5.05090714e-01 -3.88411283e-01 -2.45097727e-01 -4.32822108e-01
3.68102491e-01 -2.75773019e-01 2.63624728e-01 -6.25775993e-01
3.80142808e-01 -2.92164255e-02 5.27283728e-01 -1.27410126e+00
5.95601164e-02 -6.04250550e-01 -4.66395050e-01 6.88043609e-02
-3.83186191e-01 -2.53182620e-01 3.40058684e-01 2.88412154e-01
-3.20643574e-01 -4.51626956e-01 4.81541485e-01 -1.50686517e-01
-7.07307100e-01 3.45076948e-01 -4.42435473e-01 6.29752338e-01
1.01709294e+00 -3.87401432e-01 -6.26010120e-01 -1.02814980e-01
-6.98148072e-01 5.09038687e-01 1.12845123e-01 7.99283087e-01
6.34694576e-01 -1.10914779e+00 -1.11166728e+00 -4.53574628e-01
3.79800886e-01 9.55165923e-02 8.74150321e-02 6.85482860e-01
-2.65441328e-01 7.97154307e-01 6.49962202e-02 -5.39050877e-01
-1.56303585e+00 6.73841178e-01 -1.59829766e-01 -8.51046920e-01
-8.87581646e-01 5.66673815e-01 1.37107939e-01 -2.04242200e-01
6.41296923e-01 -4.79529947e-02 -3.83540839e-01 3.63532096e-01
7.86068439e-01 4.57231373e-01 3.69134188e-01 -2.76015192e-01
-5.03778398e-01 4.23669904e-01 -5.19632220e-01 -1.82908565e-01
1.16658831e+00 -5.71936592e-02 1.52100831e-01 3.14329863e-01
1.02627861e+00 2.78874338e-01 -5.96017718e-01 -2.62783080e-01
5.25111556e-01 -1.08216450e-01 3.73156741e-02 -1.26306260e+00
-9.82616246e-01 8.74442518e-01 5.55888355e-01 2.75157660e-01
1.21191669e+00 1.92453802e-01 8.23212028e-01 4.48712885e-01
9.01202485e-02 -1.48898208e+00 8.54901895e-02 6.48750007e-01
1.16676545e+00 -1.02336943e+00 3.69466990e-01 -4.35679197e-01
-7.97182739e-01 1.03626168e+00 4.04573530e-01 3.05785090e-01
3.25108260e-01 5.31146705e-01 -1.64665908e-01 -5.70815742e-01
-6.93196177e-01 -1.93658009e-01 6.44546330e-01 7.12519765e-01
5.40707052e-01 -2.01433122e-01 -4.58285332e-01 8.14907730e-01
-1.13917738e-01 8.28253180e-02 4.52408612e-01 9.99639690e-01
-3.17286521e-01 -1.17002857e+00 -3.75420094e-01 4.63578939e-01
-4.91282225e-01 -2.48972997e-01 -1.03835380e+00 1.16063070e+00
1.44570455e-01 8.83950353e-01 -1.26618281e-01 -1.42429486e-01
7.12499142e-01 7.39156455e-02 4.57555681e-01 -7.95867205e-01
-7.24171102e-01 4.86294702e-02 4.35108989e-01 -3.65732223e-01
-2.75810540e-01 -6.68397486e-01 -1.43461013e+00 -1.88195959e-01
-3.60244155e-01 2.67073393e-01 8.97022069e-01 7.79806077e-01
4.92245287e-01 5.27367711e-01 2.73383796e-01 -3.55037332e-01
-2.80084103e-01 -9.21667278e-01 -4.75554138e-01 2.65871078e-01
-3.39094289e-02 -3.87079895e-01 -3.65538001e-02 2.62091309e-01] | [11.767138481140137, 2.5922393798828125] |
f3b6d615-08c4-44be-b841-80055c6b1b1c | mutual-exclusivity-training-and-primitive | 2211.15578 | null | https://arxiv.org/abs/2211.15578v1 | https://arxiv.org/pdf/2211.15578v1.pdf | Mutual Exclusivity Training and Primitive Augmentation to Induce Compositionality | Recent datasets expose the lack of the systematic generalization ability in standard sequence-to-sequence models. In this work, we analyze this behavior of seq2seq models and identify two contributing factors: a lack of mutual exclusivity bias (i.e., a source sequence already mapped to a target sequence is less likely to be mapped to other target sequences), and the tendency to memorize whole examples rather than separating structures from contents. We propose two techniques to address these two issues respectively: Mutual Exclusivity Training that prevents the model from producing seen generations when facing novel, unseen examples via an unlikelihood-based loss; and prim2primX data augmentation that automatically diversifies the arguments of every syntactic function to prevent memorizing and provide a compositional inductive bias without exposing test-set data. Combining these two techniques, we show substantial empirical improvements using standard sequence-to-sequence models (LSTMs and Transformers) on two widely-used compositionality datasets: SCAN and COGS. Finally, we provide analysis characterizing the improvements as well as the remaining challenges, and provide detailed ablations of our method. Our code is available at https://github.com/owenzx/met-primaug | ['Mohit Bansal', 'Xiang Zhou', 'Yichen Jiang'] | 2022-11-28 | null | null | null | null | ['systematic-generalization'] | ['reasoning'] | [ 9.51616466e-01 2.73682982e-01 -9.08089429e-02 -4.18641597e-01
-6.88934445e-01 -1.03765023e+00 5.07799387e-01 1.20804369e-01
-4.90594059e-01 1.09778380e+00 2.87977308e-01 -7.84739316e-01
2.34366104e-01 -7.17050314e-01 -1.25559330e+00 -7.77351081e-01
-7.57193342e-02 3.24243993e-01 2.83206373e-01 -3.25703561e-01
2.83048093e-01 1.30264297e-01 -1.64178824e+00 7.45738149e-01
8.72122288e-01 3.42531830e-01 4.02990997e-01 8.64062846e-01
-2.44955346e-01 7.09277034e-01 -6.49352074e-01 -4.44031626e-01
1.10064261e-01 -8.10186386e-01 -8.68919194e-01 -4.75779921e-01
7.53099382e-01 -1.66090310e-01 -3.31613645e-02 9.39172268e-01
6.17867053e-01 1.41117647e-01 5.47319055e-01 -8.73178661e-01
-7.91141927e-01 1.14688003e+00 -4.57355082e-01 4.82654363e-01
1.35356203e-01 3.94373327e-01 1.20774293e+00 -6.50086820e-01
8.03435206e-01 1.18617117e+00 8.27725470e-01 1.02560353e+00
-1.39553297e+00 -5.70609748e-01 2.63116390e-01 -1.47344014e-02
-8.93503785e-01 -6.42086208e-01 2.57769108e-01 -3.22226733e-01
1.37580955e+00 4.97113705e-01 4.59277064e-01 1.58849728e+00
1.21693760e-01 9.37474370e-01 9.77352679e-01 -4.79236871e-01
1.20470919e-01 -1.27775043e-01 2.35866725e-01 6.65940285e-01
1.30960956e-01 4.13238436e-01 -7.87737608e-01 -1.74419492e-01
3.89351457e-01 -3.21819752e-01 -4.60063040e-01 -3.34243965e-03
-1.05120742e+00 5.48820555e-01 2.08457679e-01 3.93998057e-01
-6.00321069e-02 3.33407104e-01 6.53311789e-01 3.36339772e-01
3.03004444e-01 5.98867178e-01 -7.30289340e-01 -2.71080405e-01
-7.30125606e-01 2.92789549e-01 5.26139677e-01 1.10675931e+00
7.14695394e-01 8.74803141e-02 -8.47945288e-02 7.25723863e-01
-1.53783649e-01 2.50415117e-01 9.05279934e-01 -6.65123940e-01
4.90149856e-01 3.16685140e-01 -4.04439479e-01 -3.68063182e-01
-1.61706656e-01 -6.51303113e-01 -4.84429061e-01 -7.38907903e-02
4.59349841e-01 -1.86735138e-01 -7.62690902e-01 2.37425470e+00
4.99529019e-02 1.44320920e-01 8.22606981e-02 5.41870594e-01
5.93049526e-01 7.20907092e-01 1.46175832e-01 -9.60225388e-02
1.17554295e+00 -8.28265250e-01 -3.76739889e-01 -3.57754767e-01
9.87668335e-01 -4.52800095e-01 1.37458229e+00 2.35968024e-01
-1.27887976e+00 -4.28358704e-01 -1.18611574e+00 1.25880521e-02
-4.26497936e-01 -3.82023156e-01 3.56435806e-01 6.24489963e-01
-1.09540987e+00 9.76618290e-01 -6.63497210e-01 -5.36287665e-01
2.39156589e-01 8.55083987e-02 -1.37452185e-01 1.50773197e-01
-1.39003658e+00 8.13217282e-01 7.25336671e-01 -1.89718500e-01
-9.35244262e-01 -1.06608641e+00 -8.28797698e-01 4.44313586e-02
1.28815562e-01 -7.31738985e-01 1.44999230e+00 -1.26727128e+00
-1.21176851e+00 8.38936508e-01 -3.33884835e-01 -6.77661538e-01
4.67856258e-01 -3.49350005e-01 -3.17576170e-01 -2.52953172e-01
-1.08142316e-01 7.62770116e-01 4.24869150e-01 -1.12048984e+00
-5.96178234e-01 -3.54723305e-01 -3.54521543e-01 1.64416656e-01
-2.80178189e-01 -1.09727316e-01 2.08015665e-02 -8.42423081e-01
-2.33514354e-01 -7.92308092e-01 2.93764956e-02 -3.41714829e-01
-5.55129409e-01 -1.58537868e-02 3.54183048e-01 -6.55194283e-01
1.29904175e+00 -2.05326509e+00 2.23393202e-01 -9.59373787e-02
-1.82584114e-02 4.51778978e-01 -5.97075582e-01 6.92976415e-01
-4.04569566e-01 5.06637752e-01 -6.07630312e-01 -9.74390563e-03
-1.48534119e-01 1.70026526e-01 -6.66636407e-01 1.71449780e-01
4.35460091e-01 1.09851134e+00 -1.12193215e+00 -1.45473242e-01
-3.20139140e-01 1.91630095e-01 -5.57853580e-01 -1.41358245e-02
-6.48721755e-01 1.85898900e-01 1.51953861e-01 3.28669846e-01
5.09138227e-01 -1.31903425e-01 6.18229151e-01 2.32528508e-01
-1.56501174e-01 9.20898676e-01 -6.16817236e-01 1.69479311e+00
-3.13688487e-01 3.85873526e-01 -4.38314795e-01 -6.68366790e-01
6.61759138e-01 2.35422552e-01 -3.42809021e-01 -7.68644929e-01
-1.24945000e-01 5.41515887e-01 4.52038169e-01 -3.85568678e-01
5.07845938e-01 -4.12007570e-01 6.57165200e-02 6.76066279e-01
1.90724894e-01 3.02480072e-01 3.48091662e-01 2.24686548e-01
1.28288770e+00 5.67865252e-01 2.58135557e-01 -4.56050545e-01
3.24066609e-01 -1.12745896e-01 7.40386128e-01 9.53346729e-01
9.78495926e-03 3.59484911e-01 4.76496965e-01 -3.78122747e-01
-1.42241299e+00 -1.21061218e+00 -2.69448627e-02 1.51593649e+00
-3.12640965e-01 -4.50762272e-01 -7.42967188e-01 -9.46499825e-01
-7.98661709e-02 1.23451591e+00 -8.61646771e-01 -4.43265021e-01
-1.06728971e+00 -8.29365730e-01 1.08800638e+00 6.81260705e-01
6.24819621e-02 -1.18024147e+00 -4.78836149e-01 1.29161686e-01
-2.19718963e-01 -3.16661924e-01 -6.18669629e-01 6.10085189e-01
-9.18961346e-01 -7.58647382e-01 -5.22135377e-01 -8.73737693e-01
5.41640520e-01 -8.79859030e-02 1.35202301e+00 2.94948727e-01
-1.95813656e-01 -2.56104469e-01 -3.28114331e-01 -4.35811520e-01
-9.25379992e-01 4.25625920e-01 -4.14535180e-02 -5.49562037e-01
3.12140197e-01 -6.71457231e-01 -3.85281056e-01 1.82861164e-01
-1.08509743e+00 1.50282353e-01 6.30608022e-01 1.09594095e+00
3.44366699e-01 -4.09264565e-01 8.66634846e-01 -1.21386337e+00
5.15030086e-01 -5.85563958e-01 -2.44224727e-01 3.75226408e-01
-3.85751575e-01 3.57405037e-01 8.15449595e-01 -4.94827479e-01
-1.04335499e+00 -5.01899421e-02 -3.96009237e-01 3.35178003e-02
-1.56294763e-01 4.57099706e-01 -1.17566273e-01 4.51223642e-01
9.43267643e-01 7.78641641e-01 1.40250981e-01 -4.64563042e-01
4.33838248e-01 2.92238891e-01 3.94497961e-01 -9.06840742e-01
6.13807559e-01 1.20615691e-01 -3.25147659e-01 -7.68799424e-01
-6.86205328e-01 5.72704105e-03 -5.63384414e-01 9.61873159e-02
5.44592202e-01 -5.82696021e-01 -5.16086519e-01 3.79409432e-01
-1.11783695e+00 -8.02444816e-01 -4.27536130e-01 2.00237706e-02
-5.73044181e-01 4.84229624e-01 -8.69588912e-01 -7.22029090e-01
-4.66174364e-01 -8.36741447e-01 7.15958118e-01 1.48825413e-02
-7.62439549e-01 -8.64013195e-01 3.48903656e-01 -1.60444237e-03
4.06839788e-01 -5.33173829e-02 1.49660015e+00 -1.01237464e+00
-4.30397451e-01 3.19458187e-01 2.12122306e-01 3.81435186e-01
5.18904030e-02 9.32447016e-02 -1.00276911e+00 -3.94219130e-01
-2.09064912e-02 -5.71529865e-01 1.20011115e+00 2.00196374e-02
8.79971206e-01 -5.30933142e-01 -3.47667366e-01 5.92088163e-01
1.20294917e+00 3.20809096e-01 7.28021502e-01 4.24801022e-01
3.87237787e-01 7.35071659e-01 3.11396033e-01 4.40619178e-02
-3.13533247e-02 3.00310493e-01 2.36675203e-01 1.06224179e-01
-3.19139391e-01 -7.83558607e-01 5.40414691e-01 8.82769644e-01
2.69907475e-01 -6.79464459e-01 -8.38063359e-01 7.24089742e-01
-1.64722788e+00 -1.14195263e+00 -6.75663054e-02 2.26403928e+00
1.23483813e+00 1.25172675e-01 1.59711823e-01 5.98599650e-02
8.58584702e-01 -1.87831819e-02 -6.63706481e-01 -5.09273946e-01
-4.85802740e-01 3.42722595e-01 4.16023195e-01 6.20745957e-01
-7.33604968e-01 1.03108788e+00 6.72888184e+00 9.35749233e-01
-1.06769264e+00 1.61450356e-02 8.14584315e-01 -3.60142976e-01
-8.47925067e-01 -1.13577038e-01 -8.86821210e-01 6.20015085e-01
1.02932417e+00 2.64434051e-02 4.26464915e-01 4.86809045e-01
-2.20192686e-01 1.72948822e-01 -1.41045284e+00 2.36324027e-01
3.61999720e-02 -1.35559535e+00 4.57062721e-01 -1.17287397e-01
4.64856267e-01 1.02627441e-01 3.30221832e-01 2.75775760e-01
4.37679201e-01 -1.00235140e+00 9.88737524e-01 1.60788581e-01
7.70886898e-01 -6.42783225e-01 3.86622429e-01 3.67160439e-01
-7.42894828e-01 1.41684590e-02 -3.31248343e-01 -1.70384675e-01
1.51359156e-01 3.94671708e-01 -1.07831562e+00 4.16203737e-01
4.30966139e-01 4.41712528e-01 -4.77639645e-01 6.35307252e-01
-4.60006952e-01 9.20759737e-01 -2.79429667e-02 -3.21028739e-01
1.83837950e-01 8.45656767e-02 7.35786021e-01 1.46582222e+00
4.36129868e-01 -1.86819464e-01 -4.28253829e-01 1.10067117e+00
-6.91108927e-02 6.77118525e-02 -7.02708483e-01 -1.00466646e-01
6.42491519e-01 7.83004344e-01 -3.78569990e-01 -4.22260255e-01
-1.80347279e-01 9.98670042e-01 6.90265596e-01 3.62638623e-01
-7.99999475e-01 -4.39252377e-01 8.15235496e-01 1.12555377e-01
3.90350819e-01 1.24972805e-01 -3.60332072e-01 -9.57848907e-01
8.84952694e-02 -1.22838521e+00 6.25050545e-01 -5.59151351e-01
-1.31619918e+00 6.38587415e-01 -3.99708688e-01 -7.35712349e-01
-1.98797166e-01 -5.25110245e-01 -7.20454097e-01 1.01928186e+00
-1.28299296e+00 -9.07954335e-01 4.07858819e-01 -4.79375869e-02
7.43575931e-01 -1.57683462e-01 7.58486867e-01 2.86573887e-01
-5.23004353e-01 8.78568769e-01 6.97643831e-02 3.88836376e-02
7.43561029e-01 -1.33815670e+00 1.15390038e+00 1.03963196e+00
-3.35102305e-02 1.25429106e+00 7.62787700e-01 -8.65488410e-01
-1.23378515e+00 -1.21997881e+00 1.19298923e+00 -6.80007279e-01
5.40796280e-01 -7.46906221e-01 -1.13766694e+00 1.12165689e+00
2.77876168e-01 -6.92564070e-01 1.01921499e+00 1.83015198e-01
-7.60286629e-01 2.35014752e-01 -7.77114987e-01 9.69482362e-01
1.49816549e+00 -3.32909018e-01 -7.69162536e-01 -5.12342378e-02
1.04950190e+00 -1.03858285e-01 -4.02393252e-01 4.28037792e-01
5.57213962e-01 -1.07673168e+00 7.29107678e-01 -1.13805401e+00
7.67303348e-01 -2.51491874e-01 -2.25246862e-01 -1.41845644e+00
-5.79145670e-01 -6.43708467e-01 -1.90378726e-02 1.33610630e+00
1.06029809e+00 -7.99650371e-01 7.20821857e-01 9.42974538e-02
-5.46448708e-01 -7.03550220e-01 -7.56563008e-01 -1.12976491e+00
5.09495795e-01 -1.64353326e-01 7.36277342e-01 1.13039768e+00
2.50918120e-01 5.85649490e-01 -2.79629350e-01 -1.26973718e-01
2.93942392e-01 -3.97905596e-02 4.90357906e-01 -5.72797656e-01
-6.05687141e-01 -6.59190297e-01 -8.75140503e-02 -1.23133099e+00
1.19064376e-01 -1.39207304e+00 3.07412177e-01 -9.90499079e-01
4.62911248e-01 -3.57362062e-01 -4.14609194e-01 5.29144406e-01
-3.93310159e-01 8.81721172e-03 1.41442223e-02 -4.74978723e-02
-3.42659682e-01 4.34287339e-01 9.32429075e-01 6.80346787e-03
-1.15752267e-02 -3.90566081e-01 -9.06406522e-01 4.47256327e-01
1.11517131e+00 -5.38273931e-01 -5.54381192e-01 -7.12624133e-01
5.58458328e-01 -2.31126592e-01 8.71794298e-02 -6.98704183e-01
-1.01495147e-01 -1.39309755e-02 4.46555942e-01 -5.12873292e-01
3.78997289e-02 -2.31056899e-01 1.83490038e-01 9.64124024e-01
-8.57962430e-01 3.11454713e-01 4.75434393e-01 5.17696261e-01
2.38327265e-01 -4.73181665e-01 7.06168950e-01 -3.51057827e-01
-5.77408910e-01 -1.03091605e-01 -6.99414134e-01 2.79040635e-01
7.05095172e-01 -1.06640682e-01 -8.17403555e-01 -8.47028196e-02
-4.88790095e-01 1.53096035e-01 7.13472188e-01 4.72599953e-01
3.55112672e-01 -1.12071145e+00 -9.14703429e-01 2.11330816e-01
1.71266064e-01 -3.23288947e-01 4.37113404e-01 5.46348035e-01
-5.10012209e-01 3.87009084e-01 -2.33259410e-01 -4.19951648e-01
-1.35137904e+00 8.73180926e-01 3.32459837e-01 -9.94958356e-02
-2.49811813e-01 1.34628570e+00 6.57287240e-01 -9.23220694e-01
1.64175630e-01 -3.05436820e-01 2.83064723e-01 -2.10511744e-01
6.16899192e-01 3.70623440e-01 9.48380530e-02 -2.99983948e-01
-3.78407836e-01 8.36348757e-02 -4.35215801e-01 -1.33293226e-01
1.25998747e+00 6.20396659e-02 -2.47507840e-01 7.23370314e-01
1.11569512e+00 -3.74335535e-02 -1.03140223e+00 -1.51283473e-01
2.11567655e-01 -2.46380404e-01 -7.52120733e-01 -1.19789243e+00
-5.20695806e-01 1.05325425e+00 1.82717279e-01 6.01577684e-02
7.68621564e-01 -1.78278759e-02 8.44420314e-01 3.07620555e-01
5.94049394e-02 -8.88309777e-01 4.36685905e-02 8.02710831e-01
7.51618028e-01 -5.72077632e-01 -4.41143066e-01 -4.07804072e-01
-5.55868208e-01 9.16038036e-01 8.11310172e-01 2.76551634e-01
-1.08253390e-01 4.30131078e-01 4.00716364e-02 5.72560132e-02
-1.27928507e+00 3.86727639e-02 -1.40061334e-01 5.55812120e-01
8.59306633e-01 -1.53733730e-01 -3.25295985e-01 5.64037979e-01
-3.67585868e-01 -3.11532229e-01 4.92839664e-01 1.01181042e+00
-4.28134263e-01 -1.28237176e+00 -6.49074763e-02 4.28498536e-01
-5.33313334e-01 -7.22424090e-01 -7.24817157e-01 5.02687871e-01
1.28602535e-01 5.21559715e-01 3.12877566e-01 -4.05023903e-01
1.31907076e-01 7.32024431e-01 6.68285429e-01 -7.64098346e-01
-8.17883968e-01 -8.55384693e-02 3.58133286e-01 -1.75688773e-01
1.27074376e-01 -7.88627505e-01 -1.15447998e+00 -3.22304189e-01
-1.62458748e-01 -3.07313111e-02 3.36606443e-01 6.04723692e-01
4.75339174e-01 7.36464322e-01 1.85081750e-01 -4.27548259e-01
-7.61133432e-01 -9.31006253e-01 -2.73271501e-01 4.02517945e-01
3.71742249e-01 -5.25272563e-02 -2.41384044e-01 1.22693256e-01] | [11.259979248046875, 9.15325927734375] |
c442f94c-4810-4127-82c3-ca37edfca59b | reinforcement-learning-based-speech | 1811.04224 | null | http://arxiv.org/abs/1811.04224v1 | http://arxiv.org/pdf/1811.04224v1.pdf | Reinforcement Learning Based Speech Enhancement for Robust Speech Recognition | Conventional deep neural network (DNN)-based speech enhancement (SE)
approaches aim to minimize the mean square error (MSE) between enhanced speech
and clean reference. The MSE-optimized model may not directly improve the
performance of an automatic speech recognition (ASR) system. If the target is
to minimize the recognition error, the recognition results should be used to
design the objective function for optimizing the SE model. However, the
structure of an ASR system, which consists of multiple units, such as acoustic
and language models, is usually complex and not differentiable. In this study,
we proposed to adopt the reinforcement learning algorithm to optimize the SE
model based on the recognition results. We evaluated the propsoed SE system on
the Mandarin Chinese broadcast news corpus (MATBN). Experimental results
demonstrate that the proposed method can effectively improve the ASR results
with a notable 12.40% and 19.23% error rate reductions for signal to noise
ratio at 0 dB and 5 dB conditions, respectively. | [] | 2018-11-10 | null | null | null | null | ['robust-speech-recognition'] | ['speech'] | [ 1.40483484e-01 -2.27507710e-01 4.32827324e-01 -3.92335832e-01
-8.13045263e-01 8.81553665e-02 1.64310157e-01 -3.52206379e-01
-5.84309042e-01 5.55660963e-01 3.17795038e-01 -4.27358687e-01
-7.08048344e-02 -4.43910152e-01 -3.78733873e-01 -9.38853800e-01
1.73848346e-01 -5.49800396e-01 -2.18400404e-01 -4.58573848e-01
1.40434816e-01 3.66105378e-01 -1.19332671e+00 -1.77205224e-02
1.02516246e+00 1.09707117e+00 8.47346365e-01 7.93176472e-01
1.03054576e-01 6.98434651e-01 -1.05884016e+00 -1.66633222e-02
3.28943655e-02 -5.20369172e-01 -2.30145752e-01 1.81199476e-01
-2.17690647e-01 -3.30831409e-01 -6.40669823e-01 1.45159793e+00
1.09946346e+00 5.12442350e-01 2.67499715e-01 -6.04445219e-01
-7.46674657e-01 7.26713181e-01 -3.68530035e-01 4.39133734e-01
-1.80346921e-01 -1.56447008e-01 6.07017338e-01 -9.73606527e-01
-4.26933095e-02 1.22619414e+00 4.63923156e-01 7.06442595e-01
-8.02476585e-01 -6.71800375e-01 4.93948050e-02 4.25102681e-01
-1.54425263e+00 -1.00600004e+00 7.81679273e-01 1.59454003e-01
9.25880790e-01 2.51497060e-01 2.38452539e-01 6.41201437e-01
-2.28419885e-01 6.67395055e-01 9.58882987e-01 -6.31882370e-01
-2.74567073e-03 1.42687470e-01 1.40428454e-01 3.49481672e-01
-1.63345948e-01 4.13450271e-01 -3.11429441e-01 4.25963819e-01
6.11660242e-01 -4.44291443e-01 -5.64607143e-01 5.89782357e-01
-7.99425781e-01 6.39962316e-01 3.13644856e-01 5.47671199e-01
-6.25164986e-01 -1.46665573e-01 2.34662816e-01 3.38535607e-01
3.14059734e-01 3.87671769e-01 -3.49165499e-01 -3.76800686e-01
-7.06175923e-01 -3.09240699e-01 5.15372097e-01 7.03814864e-01
3.50685716e-01 8.99341941e-01 -1.67060211e-01 1.60246658e+00
6.25862360e-01 7.79731750e-01 7.26125598e-01 -7.03742087e-01
4.98582810e-01 -1.18215524e-01 7.29761645e-02 -1.03574240e+00
-4.49949577e-02 -8.02351534e-01 -9.75730360e-01 9.06880423e-02
-1.07635520e-01 -7.24894404e-01 -1.00595415e+00 1.51779962e+00
1.71210527e-01 -3.73219587e-02 6.22890592e-01 1.15588987e+00
9.08499241e-01 1.35833049e+00 -1.09677434e-01 -5.83140552e-01
8.82719338e-01 -9.85068202e-01 -1.49882448e+00 -2.20746547e-01
3.06547016e-01 -1.06246817e+00 8.90032709e-01 3.52030605e-01
-1.13620675e+00 -7.77549386e-01 -1.31467235e+00 3.93815666e-01
1.45557495e-02 6.34072483e-01 -1.73630208e-01 8.33837450e-01
-9.70344901e-01 4.17815834e-01 -7.44101524e-01 1.30719528e-01
-8.40105116e-02 3.12872231e-01 -2.54854355e-02 2.23851576e-01
-1.53488052e+00 8.56888115e-01 4.27789152e-01 7.35090971e-01
-8.54206026e-01 -3.49333197e-01 -6.69970393e-01 3.73429567e-01
2.98864275e-01 1.10521242e-01 1.53678143e+00 -1.15499687e+00
-2.01785254e+00 9.52197239e-02 -2.27866635e-01 -4.68700588e-01
5.47147505e-02 -3.31303746e-01 -1.05537033e+00 9.76629257e-02
-7.14975357e-01 3.25793415e-01 7.83242047e-01 -1.12755704e+00
-8.06462944e-01 3.26662101e-02 -2.43966326e-01 4.80863273e-01
-5.42514563e-01 3.64622444e-01 -1.98774621e-01 -9.74837482e-01
1.52269453e-01 -5.10250568e-01 -1.31741121e-01 -5.02095044e-01
-1.76625520e-01 -2.34302841e-02 8.51125658e-01 -1.48013043e+00
1.50801206e+00 -2.48055863e+00 -3.12817365e-01 1.53033912e-01
-4.40323919e-01 9.42402005e-01 -1.70549303e-01 -1.32734422e-02
-4.06117290e-02 8.06589553e-04 -7.58271366e-02 -5.77292778e-02
-7.82658681e-02 -1.28264427e-02 1.27890874e-02 3.52305055e-01
1.88089594e-01 2.82625824e-01 -5.26740074e-01 -2.13452607e-01
3.81930500e-01 8.44384849e-01 -3.32240880e-01 5.48147440e-01
2.60302871e-01 -8.79680645e-03 -1.84301600e-01 3.17138374e-01
7.25713134e-01 3.31403345e-01 -6.27198592e-02 -3.21003258e-01
-2.60129005e-01 4.32739019e-01 -1.39347005e+00 1.02886903e+00
-7.35256791e-01 9.67249274e-01 5.55656493e-01 -1.00544155e+00
1.46064806e+00 7.01683402e-01 1.54878050e-01 -1.05977511e+00
2.26168722e-01 2.75728434e-01 4.43134010e-01 -4.49377537e-01
3.96918416e-01 -2.56282508e-01 5.98818421e-01 3.92040387e-02
-1.09712675e-01 1.72747865e-01 -2.54821330e-01 -2.95195967e-01
5.83064675e-01 -5.06812990e-01 1.85940579e-01 -1.11684948e-01
9.08102214e-01 -5.54900289e-01 7.27732003e-01 5.95814526e-01
-2.81420231e-01 3.64564568e-01 -1.13038771e-01 1.33478954e-01
-9.77131307e-01 -8.95045161e-01 -1.20497040e-01 7.43325412e-01
7.19244918e-03 1.33401230e-01 -8.64334106e-01 -7.00255930e-02
-7.14658678e-01 9.83092487e-01 1.12785980e-01 -4.42719460e-01
-6.89750016e-01 -7.46248901e-01 5.79713404e-01 3.85456651e-01
8.52883756e-01 -1.07669330e+00 -1.19667714e-02 4.69283968e-01
-2.52834082e-01 -1.04309690e+00 -7.54846692e-01 1.08515993e-01
-4.96133834e-01 -3.79480481e-01 -9.17460680e-01 -1.06728446e+00
6.43139899e-01 2.66377807e-01 2.75947422e-01 6.90373331e-02
3.50421250e-01 1.49263263e-01 -6.18359864e-01 -4.19366449e-01
-8.78241718e-01 -2.76286483e-01 2.89336234e-01 2.64586449e-01
2.26398245e-01 -2.59846270e-01 -3.82473886e-01 5.33688664e-01
-7.24285185e-01 -1.07178167e-01 6.84439301e-01 1.13839948e+00
4.66939181e-01 5.60192049e-01 9.97446358e-01 -7.49395117e-02
1.11523712e+00 3.99634689e-02 -7.52288580e-01 4.27325070e-01
-7.75440335e-01 -1.46469384e-01 6.95469320e-01 -6.91950440e-01
-1.63163292e+00 -2.68350124e-01 -6.66953743e-01 -2.86863089e-01
-1.03439964e-01 7.71745443e-01 -4.94485080e-01 -2.22387221e-02
3.99087608e-01 6.74227655e-01 -9.06705577e-03 -5.87748587e-01
-7.48080090e-02 1.60722971e+00 4.33762997e-01 -2.50115991e-01
5.60170233e-01 -4.18788671e-01 -4.51627433e-01 -1.28926313e+00
-3.16226572e-01 -3.55205208e-01 4.15118132e-03 -3.46379876e-01
6.30502880e-01 -8.91270638e-01 -5.65875947e-01 6.74691916e-01
-1.18522060e+00 -1.91733703e-01 2.34833658e-01 1.10084116e+00
-2.93514848e-01 4.35820520e-01 -7.19668865e-01 -1.23636270e+00
-7.74886250e-01 -1.33003819e+00 4.69890684e-01 5.77410281e-01
2.53440112e-01 -7.44193494e-01 -3.84610593e-01 2.08368629e-01
7.94534564e-01 -5.83235800e-01 5.78019023e-01 -5.43684423e-01
-2.49361932e-01 -1.94103390e-01 -6.02333620e-02 1.21929216e+00
4.05349880e-01 -2.96578202e-02 -1.11003721e+00 -2.02512592e-01
4.80980814e-01 1.22148775e-01 2.96769053e-01 5.49112797e-01
1.12505126e+00 -6.77427113e-01 3.38019848e-01 5.15046477e-01
1.25193453e+00 1.17418253e+00 1.08083594e+00 2.56281823e-01
2.58518398e-01 3.50555807e-01 7.58381009e-01 4.04935539e-01
-3.40419531e-01 6.28763258e-01 -3.39913517e-02 -9.78724062e-02
-2.34852627e-01 -1.41549170e-01 6.98942602e-01 1.55575943e+00
1.46519065e-01 -4.34643328e-01 -6.31762683e-01 3.89410794e-01
-1.27941895e+00 -9.05826688e-01 1.89188972e-01 1.97952449e+00
1.04843462e+00 3.22764926e-02 -2.93291420e-01 3.80072117e-01
1.15844500e+00 1.20424703e-01 -3.47648352e-01 -5.95215023e-01
-1.63147658e-01 8.22908580e-02 3.37326229e-01 7.49264538e-01
-8.98218036e-01 6.66058838e-01 5.75001192e+00 1.30471504e+00
-1.40535355e+00 -7.04875439e-02 6.26186430e-01 1.76986203e-01
3.19509134e-02 -4.66939121e-01 -6.12521470e-01 3.27732056e-01
1.29837942e+00 -3.90076369e-01 8.83204520e-01 7.30393350e-01
8.67872477e-01 3.54242772e-01 -5.18901646e-01 1.18558645e+00
-1.39310673e-01 -8.69397938e-01 -1.56087607e-01 -2.34003991e-01
5.59271276e-01 -1.41973704e-01 7.14781806e-02 4.33160454e-01
-6.23453520e-02 -9.38752890e-01 6.20701432e-01 5.16426444e-01
5.83270907e-01 -1.01008725e+00 7.95490384e-01 4.80019659e-01
-9.53100145e-01 -1.31821513e-01 -4.29614693e-01 2.36294121e-01
2.06435502e-01 3.70166659e-01 -9.14616883e-01 2.97722340e-01
4.49741334e-01 2.61348724e-01 1.67698473e-01 1.19543135e+00
-3.23478997e-01 9.61541116e-01 -2.19370693e-01 -4.49438483e-01
2.50100911e-01 -2.12776631e-01 7.79786825e-01 1.29529428e+00
5.47197521e-01 2.76587844e-01 -3.09523255e-01 6.46420002e-01
-1.53773203e-01 2.29513660e-01 7.69276079e-03 -2.61051267e-01
6.43651247e-01 9.21215713e-01 -4.53465581e-02 -8.48820359e-02
-2.12451458e-01 6.53448403e-01 -1.56023189e-01 7.70718932e-01
-9.31400418e-01 -9.81396079e-01 5.95260382e-01 -3.40542585e-01
4.75760877e-01 -2.90042907e-01 -1.01166680e-01 -6.56694591e-01
-1.01009710e-02 -1.28704143e+00 -1.83217958e-01 -1.05426133e+00
-9.55586851e-01 7.98404753e-01 -4.55919206e-01 -1.15842903e+00
7.51505867e-02 -4.92171794e-01 -5.21210670e-01 1.29608870e+00
-1.48237896e+00 -4.01168227e-01 4.40413319e-02 3.44024241e-01
8.70032787e-01 -4.39543456e-01 6.08209312e-01 6.81051135e-01
-8.42295229e-01 8.88414264e-01 5.70847511e-01 2.64018059e-01
2.44517565e-01 -9.57101822e-01 9.76198241e-02 1.23964703e+00
-7.34449998e-02 5.00547051e-01 7.95796931e-01 -3.62398952e-01
-1.21903098e+00 -1.02173042e+00 7.31880009e-01 6.66985035e-01
4.20926422e-01 -3.86012085e-02 -9.16492701e-01 1.67679846e-01
3.13662797e-01 -3.07058454e-01 3.19589496e-01 -2.32415423e-01
2.53406912e-01 -6.25759006e-01 -1.07132459e+00 8.24573755e-01
5.39113343e-01 -4.77835238e-01 -4.70953882e-01 -7.66342357e-02
1.09225583e+00 -4.91598040e-01 -9.58806276e-01 4.93555367e-01
2.97237098e-01 -3.32983911e-01 8.14294636e-01 -3.10499817e-01
1.07402861e-01 -4.62630749e-01 -5.03846943e-01 -1.78366554e+00
-2.81891882e-01 -6.77444398e-01 -7.36280084e-02 1.52427292e+00
6.98670566e-01 -5.38975358e-01 1.54698074e-01 5.37314773e-01
-4.54640448e-01 -6.04439020e-01 -7.85598814e-01 -8.75655770e-01
-4.63328451e-01 -4.24409181e-01 5.04564822e-01 4.74132419e-01
-1.89944655e-01 3.30685943e-01 -5.94682872e-01 7.07646728e-01
2.87800491e-01 -6.46456420e-01 3.25012445e-01 -2.60578901e-01
-3.20970744e-01 -3.28366578e-01 -7.77480528e-02 -1.45915151e+00
-1.59507077e-02 -2.94458628e-01 5.87584019e-01 -1.56117058e+00
-4.45089370e-01 -1.63727492e-01 -6.77274942e-01 1.07809916e-01
-3.57297838e-01 -6.15276694e-01 1.27910167e-01 -2.32463136e-01
-1.91486478e-01 1.12100935e+00 1.26843596e+00 -4.11229491e-01
-3.29444975e-01 3.88186157e-01 -6.01973057e-01 4.47536945e-01
9.72697020e-01 -4.20898438e-01 -4.05781269e-01 -6.22774839e-01
-5.20751119e-01 3.57211173e-01 -9.34284106e-02 -1.01301801e+00
3.12481493e-01 -1.35105684e-01 1.52373880e-01 -5.31173885e-01
4.21178579e-01 -9.87648487e-01 -3.39741334e-02 4.60796952e-01
-6.33572936e-01 -2.15378255e-01 2.71043658e-01 4.46142524e-01
-6.19444609e-01 -4.49467838e-01 9.88582015e-01 2.90939003e-01
-7.82964289e-01 -1.44427009e-02 -5.48971772e-01 -3.74344766e-01
5.74733853e-01 -1.39106527e-01 -2.36329585e-01 -7.12310970e-01
-5.08529425e-01 1.32353734e-02 -4.76947069e-01 4.05581117e-01
1.09876013e+00 -1.31234384e+00 -8.68227541e-01 1.39205456e-01
-4.31695253e-01 -4.11740452e-01 5.61525047e-01 5.93112886e-01
-3.59439313e-01 4.33848113e-01 1.14745960e-01 -1.84429646e-01
-1.52845776e+00 2.35880241e-01 9.98539686e-01 1.57008678e-01
-2.33913869e-01 7.23501205e-01 -2.38923468e-02 -1.67322636e-01
6.53852105e-01 -1.59284309e-01 -5.26654005e-01 -5.30037999e-01
7.66597509e-01 5.90319991e-01 2.43268386e-01 -5.83577931e-01
-1.08720228e-01 8.36800188e-02 2.29244772e-02 -3.72588128e-01
1.30875516e+00 -5.34877777e-01 1.68325409e-01 1.80701315e-02
1.31472421e+00 3.18311267e-02 -9.89210546e-01 -2.99717605e-01
-1.90850630e-01 -3.99910569e-01 6.92246258e-01 -1.04313695e+00
-1.20537138e+00 9.46423054e-01 1.05892599e+00 1.82772011e-01
1.48951173e+00 -6.01776898e-01 9.06719685e-01 6.00852966e-01
-2.58537661e-02 -1.42735028e+00 -3.01428549e-02 5.87908685e-01
1.09367013e+00 -9.83980536e-01 -4.46864754e-01 8.07049796e-02
-5.59358478e-01 1.24217486e+00 7.49935806e-01 1.99597374e-01
7.30003595e-01 3.56252313e-01 3.57877821e-01 3.87886703e-01
-4.98112112e-01 -2.64758188e-02 2.47532248e-01 5.58791995e-01
4.14536625e-01 6.87523112e-02 -5.14324009e-01 8.38283300e-01
-1.00981101e-01 -2.21922353e-01 3.94763798e-01 5.56542516e-01
-8.98523629e-01 -8.15406740e-01 -6.67609215e-01 2.07290918e-01
-6.67891085e-01 -2.58569837e-01 1.50863364e-01 3.03131223e-01
-2.82789916e-01 1.65394092e+00 -1.36303112e-01 -6.12673521e-01
6.11203492e-01 -3.60701293e-01 -8.13139323e-03 -2.19205871e-01
-2.06835032e-01 5.29285550e-01 2.70843655e-01 1.01753414e-01
-4.23347205e-01 -3.35665941e-01 -1.31144500e+00 -1.88045338e-01
-9.36203182e-01 2.93210119e-01 1.03907335e+00 9.80517387e-01
1.62363410e-01 9.20660198e-01 1.14542603e+00 -7.57125095e-02
-8.94287407e-01 -1.33981287e+00 -6.89959705e-01 3.64726745e-02
4.05488521e-01 -2.59804040e-01 -4.97696966e-01 7.14529604e-02] | [14.793742179870605, 6.017111301422119] |
4aff43b8-99ca-4855-9b21-d5e4f692e2a5 | homotopic-policy-mirror-descent-policy | 2201.09457 | null | https://arxiv.org/abs/2201.09457v9 | https://arxiv.org/pdf/2201.09457v9.pdf | Homotopic Policy Mirror Descent: Policy Convergence, Implicit Regularization, and Improved Sample Complexity | We propose a new policy gradient method, named homotopic policy mirror descent (HPMD), for solving discounted, infinite horizon MDPs with finite state and action spaces. HPMD performs a mirror descent type policy update with an additional diminishing regularization term, and possesses several computational properties that seem to be new in the literature. We first establish the global linear convergence of HPMD instantiated with Kullback-Leibler divergence, for both the optimality gap, and a weighted distance to the set of optimal policies. Then local superlinear convergence is obtained for both quantities without any assumption. With local acceleration and diminishing regularization, we establish the first result among policy gradient methods on certifying and characterizing the limiting policy, by showing, with a non-asymptotic characterization, that the last-iterate policy converges to the unique optimal policy with the maximal entropy. We then extend all the aforementioned results to HPMD instantiated with a broad class of decomposable Bregman divergences, demonstrating the generality of the these computational properties. As a by product, we discover the finite-time exact convergence for some commonly used Bregman divergences, implying the continuing convergence of HPMD to the limiting policy even if the current policy is already optimal. Finally, we develop a stochastic version of HPMD and establish similar convergence properties. By exploiting the local acceleration, we show that for small optimality gap, a better than $\tilde{\mathcal{O}}(\left|\mathcal{S}\right| \left|\mathcal{A}\right| / \epsilon^2)$ sample complexity holds with high probability, when assuming a generative model for policy evaluation. | ['Guanghui Lan', 'Tuo Zhao', 'Yan Li'] | 2022-01-24 | null | null | null | null | ['policy-gradient-methods'] | ['methodology'] | [-7.87532330e-02 2.68947124e-01 -4.03344989e-01 1.81039214e-01
-6.65385842e-01 -8.04044962e-01 4.87761050e-01 2.66438395e-01
-8.36775541e-01 1.16726482e+00 8.64264742e-02 -7.38306999e-01
-5.82184613e-01 -5.41128159e-01 -8.68511140e-01 -9.62290168e-01
-4.94758576e-01 3.25888604e-01 1.16674960e-01 -2.95055062e-01
3.45029175e-01 3.39029819e-01 -1.41853201e+00 -5.74781656e-01
1.10841286e+00 1.35176742e+00 2.58829176e-01 7.45636404e-01
3.05003554e-01 7.23266542e-01 -2.60148168e-01 -2.59543836e-01
6.35749280e-01 -5.11809170e-01 -8.34115088e-01 -1.00378692e-01
7.69726411e-02 -6.42552614e-01 -1.25055671e-01 1.47227204e+00
5.33867955e-01 4.77222741e-01 6.39321923e-01 -1.08448899e+00
-4.45178300e-01 2.90660679e-01 -4.08545643e-01 2.94154912e-01
1.91059679e-01 1.62521601e-01 1.08497632e+00 -3.21743697e-01
5.30805171e-01 9.74773645e-01 5.92293918e-01 5.30698776e-01
-1.13985837e+00 -2.37892091e-01 4.90113378e-01 -1.89707920e-01
-8.07017207e-01 7.68866241e-02 1.28539637e-01 -4.22685087e-01
7.51321018e-01 3.16768974e-01 5.14025450e-01 7.46249437e-01
1.73227414e-01 7.13418245e-01 1.36768556e+00 -4.57548201e-01
6.61115825e-01 1.10029966e-01 -3.37972976e-02 9.12221014e-01
3.77636671e-01 4.95169997e-01 2.00261548e-01 -3.67392004e-01
7.48984337e-01 -4.14049439e-02 -4.84740108e-01 -4.06709790e-01
-9.63734329e-01 9.75047529e-01 3.65770459e-02 3.03568423e-01
-5.78102171e-01 1.40322819e-01 2.55356282e-01 4.49326396e-01
4.01170373e-01 3.19777906e-01 -3.71691465e-01 -5.19607782e-01
-6.57195091e-01 4.83117998e-01 1.22572589e+00 8.61360490e-01
3.97224814e-01 1.72699332e-01 -3.25085580e-01 2.20160529e-01
1.09335510e-02 8.74371171e-01 3.45799893e-01 -1.29792404e+00
3.81774902e-01 9.21109095e-02 7.67461538e-01 -6.68412685e-01
-3.35236311e-01 -4.19766456e-01 -6.52186275e-01 5.02190650e-01
7.42094100e-01 -5.19192815e-01 -3.47759932e-01 2.34854817e+00
2.91887224e-01 -1.16609521e-01 1.51419267e-01 7.93102920e-01
-4.64042515e-01 7.13094294e-01 -1.50066137e-01 -9.28973019e-01
9.61400151e-01 -4.36171293e-01 -4.27449763e-01 2.94004194e-02
6.19261682e-01 -3.57407957e-01 1.32753110e+00 3.66297275e-01
-1.11487365e+00 -3.87152694e-02 -9.79105830e-01 4.10537452e-01
5.41163497e-02 -8.62892345e-02 3.18522274e-01 4.97276783e-01
-1.11460960e+00 1.17801225e+00 -7.50629902e-01 -2.45249391e-01
-1.13677830e-02 2.66222864e-01 6.87616542e-02 1.99844211e-01
-1.05612493e+00 9.71707821e-01 5.25134027e-01 -1.79437831e-01
-1.13465953e+00 -5.90500832e-01 -5.20222723e-01 8.47970974e-03
6.78740144e-01 -4.60366458e-01 1.52772796e+00 -7.74972498e-01
-1.86815345e+00 3.42777520e-01 -3.10463607e-02 -8.09404373e-01
9.12292480e-01 -1.66391447e-01 -7.77781531e-02 1.46833897e-01
7.09484369e-02 -2.14923676e-02 7.52098620e-01 -1.00688028e+00
-9.24503386e-01 -4.07600015e-01 5.76819539e-01 3.53749156e-01
-4.96952474e-01 -1.83579087e-01 3.56859058e-01 -4.11274433e-01
-2.97222733e-01 -9.54111993e-01 -4.11600381e-01 -2.18806565e-01
-9.94427651e-02 -2.56848156e-01 4.26725715e-01 -4.69401836e-01
1.29933512e+00 -1.85358465e+00 3.30194592e-01 3.53369415e-01
-8.34091976e-02 2.59707779e-01 1.76636919e-01 4.55912530e-01
1.31527737e-01 1.22224754e-02 -5.00694990e-01 -1.31723925e-01
3.88301820e-01 3.00173998e-01 -5.28110027e-01 8.35472405e-01
-2.97931701e-01 5.34058571e-01 -1.28184426e+00 -8.13784748e-02
1.08353592e-01 -4.04221043e-02 -6.02021396e-01 -1.56297833e-02
-3.70429277e-01 4.28348005e-01 -6.31349266e-01 -8.52968544e-03
4.74827826e-01 -1.10610932e-01 3.19654882e-01 3.99775237e-01
-4.93936688e-01 -4.02708873e-02 -1.38059759e+00 1.36690903e+00
-3.75826269e-01 6.63448647e-02 5.37875473e-01 -1.34519732e+00
4.46909279e-01 2.93556690e-01 6.76762342e-01 -3.82019043e-01
3.17902893e-01 5.20233095e-01 -4.08501327e-01 -2.52947122e-01
4.44131613e-01 -5.57992280e-01 6.56675128e-03 6.42771721e-01
-8.27409476e-02 9.60922465e-02 3.50714922e-01 -6.11457378e-02
9.54921126e-01 1.52992398e-01 3.48925173e-01 -8.41878891e-01
6.41150475e-01 -9.01949108e-02 5.65667272e-01 8.95839095e-01
-3.63727748e-01 -1.69408455e-01 9.81508315e-01 -4.04255576e-02
-1.08864439e+00 -1.08967376e+00 -8.33270699e-02 1.13635719e+00
1.51538730e-01 -9.59699973e-02 -7.32632697e-01 -5.79958856e-01
2.56432295e-01 9.21783805e-01 -6.72562301e-01 -2.74884552e-01
-3.81381720e-01 -7.99544156e-01 2.89662689e-01 2.48039842e-01
4.81138796e-01 -6.91268325e-01 -9.23098803e-01 1.90320641e-01
2.25271016e-01 -7.16434300e-01 -6.33043528e-01 1.65935323e-01
-9.38018799e-01 -9.78910863e-01 -1.06683671e+00 -3.45577210e-01
5.58158278e-01 -2.03540087e-01 7.08842933e-01 -4.53240663e-01
2.68302798e-01 6.47845984e-01 3.49897034e-02 -3.48527491e-01
-5.07417917e-01 -3.00565183e-01 7.27744520e-01 -8.34748894e-02
-1.63012370e-01 -6.72271371e-01 -7.90883482e-01 1.73810065e-01
-7.69776702e-01 -3.71673465e-01 2.88104862e-01 6.54922485e-01
6.97434604e-01 -9.85036567e-02 3.05798352e-01 -2.36051232e-01
1.04656768e+00 -3.65956992e-01 -1.38136220e+00 1.02217630e-01
-8.74851406e-01 5.98158002e-01 1.02227676e+00 -4.96893942e-01
-8.56505632e-01 -2.95510024e-01 1.69025913e-01 -4.70671356e-01
1.84783936e-01 2.73967355e-01 2.70591378e-01 1.25975862e-01
5.23416996e-01 4.77833480e-01 3.93763393e-01 -5.25747597e-01
4.67302620e-01 3.51940781e-01 6.23552382e-01 -9.81182039e-01
4.38078076e-01 6.85739696e-01 2.21821263e-01 -5.43349504e-01
-9.11321640e-01 -9.34128612e-02 -1.39455497e-01 -4.14536782e-02
7.42653430e-01 -4.50543880e-01 -1.53737187e+00 1.00305893e-01
-8.52717221e-01 -4.88774568e-01 -8.12998354e-01 7.71536052e-01
-1.11068738e+00 6.48781776e-01 -5.67612827e-01 -1.37651896e+00
-2.77012914e-01 -9.32292640e-01 4.58065927e-01 1.81717888e-01
1.77864224e-01 -1.10155869e+00 4.33908492e-01 -3.35519850e-01
3.05663496e-01 2.91990161e-01 8.53229582e-01 -6.32594645e-01
-4.60360311e-02 -5.19285463e-02 7.17845708e-02 7.55690515e-01
-1.36305497e-03 -3.82377744e-01 -3.29738587e-01 -7.30533659e-01
4.14623052e-01 -8.16989318e-02 7.47225106e-01 5.14001489e-01
8.03602517e-01 -8.53899360e-01 -7.57765844e-02 3.71268988e-01
1.49578261e+00 3.25917512e-01 1.42010763e-01 3.56431603e-01
1.73433080e-01 3.41299266e-01 5.84607661e-01 9.41356182e-01
1.06424049e-01 2.63391227e-01 7.04314053e-01 3.30217451e-01
6.77905977e-01 -3.10597062e-01 8.01423967e-01 4.89061654e-01
-3.43966335e-01 4.81560268e-02 -5.06917953e-01 4.75450337e-01
-1.92813051e+00 -1.01867604e+00 3.37015212e-01 3.00702071e+00
1.03670573e+00 7.86247030e-02 4.46694136e-01 -7.50694051e-02
6.82748973e-01 2.96903700e-02 -8.49448562e-01 -6.85975611e-01
-7.59734437e-02 3.33155543e-01 1.03854191e+00 8.06692719e-01
-9.49845791e-01 7.18171597e-01 5.79935837e+00 8.74219120e-01
-1.03735268e+00 1.96128607e-01 3.43149364e-01 -2.18907058e-01
-1.70111015e-01 1.41730219e-01 -6.92457139e-01 7.07420170e-01
9.27285552e-01 -5.98423183e-01 7.09372222e-01 1.06721318e+00
5.09423018e-01 -2.59342372e-01 -8.87728572e-01 5.99097371e-01
-4.77080315e-01 -1.03971720e+00 -4.41204131e-01 4.47080225e-01
9.11840439e-01 -1.54381484e-01 2.03220651e-01 2.41416618e-01
8.03705990e-01 -6.03729844e-01 7.82341719e-01 1.93219125e-01
6.50140584e-01 -1.02424669e+00 4.84079063e-01 6.25579238e-01
-9.88513172e-01 -6.49613559e-01 -4.41462070e-01 -2.76779145e-01
2.81280816e-01 4.56518590e-01 -3.06303591e-01 7.54766822e-01
2.95230418e-01 2.64869362e-01 3.36501807e-01 7.71997273e-01
-1.61364108e-01 2.18093082e-01 -6.61816239e-01 -1.82176933e-01
6.68977439e-01 -6.57362998e-01 9.13979709e-01 7.91585386e-01
4.99613941e-01 1.00391313e-01 3.47412497e-01 7.32220709e-01
1.45617902e-01 1.13239929e-01 -4.59063441e-01 -1.62987038e-02
2.13487878e-01 8.67598653e-01 -4.58845049e-01 -4.12265182e-01
-1.09669052e-01 9.72148180e-01 2.50051528e-01 3.82072210e-01
-8.82733047e-01 -3.42741847e-01 9.81198370e-01 -1.19292729e-01
4.73086774e-01 -4.14110810e-01 1.49547860e-01 -1.07143629e+00
4.02033538e-01 -6.24893546e-01 4.83181179e-01 1.02817062e-02
-9.71326411e-01 3.20947707e-01 2.83905745e-01 -1.22103214e+00
-7.01875567e-01 -6.40651643e-01 -3.81894499e-01 7.21153438e-01
-1.19200993e+00 -1.61945790e-01 4.79674041e-01 7.31853783e-01
4.95899580e-02 -4.98042330e-02 7.20284700e-01 2.82462854e-02
-4.34166670e-01 3.83893222e-01 8.82136583e-01 -3.54697883e-01
1.14588328e-01 -1.43782783e+00 -2.65247077e-01 8.79093707e-01
-4.92255449e-01 3.81017655e-01 1.05532050e+00 -5.29588223e-01
-1.40530658e+00 -8.73940527e-01 5.61732531e-01 1.63688026e-02
1.03043783e+00 -1.97799690e-02 -5.65096498e-01 6.55114770e-01
1.23285688e-01 -1.57210112e-01 -3.64474654e-02 -3.09706032e-01
1.01578258e-01 -5.04915267e-02 -1.20312452e+00 7.09225595e-01
9.30659175e-01 -3.50528896e-01 -5.32370687e-01 5.21249950e-01
6.33483291e-01 -3.46206546e-01 -1.00104976e+00 2.71045953e-01
3.38130713e-01 -9.15001810e-01 6.31062031e-01 -9.46133733e-01
-6.30483851e-02 -1.48981288e-01 -3.54905307e-01 -1.15449429e+00
2.81157321e-03 -1.34490776e+00 -3.83584559e-01 8.03401768e-01
2.64856368e-01 -1.12043262e+00 4.62663740e-01 5.23853242e-01
-9.90765095e-02 -1.01821363e+00 -1.35490787e+00 -1.42746615e+00
6.52878582e-01 -3.89715433e-01 1.73334807e-01 5.76623619e-01
4.04462010e-01 -2.01171100e-01 -4.81980115e-01 1.38713866e-01
7.01893210e-01 5.28600886e-02 1.90881744e-01 -8.48100483e-01
-6.85187817e-01 -7.18319476e-01 2.56880783e-02 -1.29759967e+00
2.15893313e-01 -7.63264000e-01 -5.00033461e-02 -1.20395768e+00
4.37886268e-02 -4.05311018e-01 -5.08633852e-01 1.79481223e-01
5.25043830e-02 -5.71056664e-01 3.24487507e-01 1.50814161e-01
-5.12559295e-01 7.87973046e-01 1.22165835e+00 2.42170930e-01
-4.21956390e-01 3.70211482e-01 -5.55459619e-01 8.04841876e-01
8.31675410e-01 -2.79029787e-01 -5.64442515e-01 -1.40439242e-01
1.65524706e-01 2.19568357e-01 3.16888452e-01 -8.67600083e-01
-4.51513641e-02 -4.71975297e-01 -3.66741687e-01 -1.78024709e-01
1.44299626e-01 -4.37844545e-01 -1.76437393e-01 9.16735470e-01
-4.58912909e-01 2.09407911e-01 -3.30530331e-02 8.32407713e-01
2.87154049e-01 -4.27244604e-01 9.27039266e-01 -1.32418290e-01
-4.76500481e-01 4.59731251e-01 -5.02344966e-01 3.38120759e-01
1.19683611e+00 1.35351539e-01 -6.09259382e-02 -6.51713192e-01
-7.30689645e-01 3.85325730e-01 3.77296239e-01 -5.13103418e-02
2.28203773e-01 -1.21506381e+00 -5.11105597e-01 -2.03178346e-01
-5.34240007e-01 -3.15080494e-01 3.05195488e-02 1.24128616e+00
-1.19949520e-01 4.84970808e-01 -7.08259270e-03 -3.46392721e-01
-7.30978429e-01 7.65711069e-01 5.31377912e-01 -5.96659184e-01
-5.32942712e-01 6.04397535e-01 -9.35948119e-02 -1.18020147e-01
2.69778013e-01 -5.96019924e-01 3.47196817e-01 -4.34790067e-02
4.85826045e-01 7.42491543e-01 -1.37008160e-01 -2.53621638e-01
-1.87072247e-01 3.56186211e-01 2.11106122e-01 -5.62218785e-01
1.08239603e+00 -1.43108845e-01 -3.98683809e-02 2.10846275e-01
1.22794044e+00 -2.89889693e-01 -1.68527377e+00 -1.67218104e-01
9.89500284e-02 -1.46575525e-01 -1.94324344e-01 -3.86205614e-01
-7.46088505e-01 7.29167104e-01 7.91200399e-01 4.45678979e-01
9.04052734e-01 -1.94449872e-01 4.90416557e-01 4.24828619e-01
6.34141624e-01 -1.41996920e+00 -1.68902636e-01 8.51442397e-01
7.02788591e-01 -8.66136074e-01 -4.28577028e-02 4.31388974e-01
-6.61164820e-01 9.15404677e-01 3.94559875e-02 -3.04539561e-01
5.59788644e-01 1.76236778e-02 -5.33428431e-01 3.06243509e-01
-6.13614917e-01 -4.27875012e-01 -1.99282125e-01 2.56255358e-01
1.13150001e-01 1.16359808e-01 -8.68948519e-01 3.73574555e-01
-1.77759618e-01 1.70566589e-02 3.50098968e-01 1.12896550e+00
-6.82235062e-01 -9.48577464e-01 -1.78944439e-01 3.20893675e-01
-6.11659229e-01 -1.48516679e-02 8.22974816e-02 8.54441226e-01
-2.96970963e-01 7.45059490e-01 3.50311995e-02 3.77252996e-02
3.11570525e-01 3.05931363e-02 6.83236480e-01 -8.99707600e-02
-1.58802271e-01 -1.25236109e-01 -1.23404309e-01 -7.21595228e-01
-9.90685299e-02 -6.17359817e-01 -1.30612803e+00 -5.76982498e-01
4.97758575e-02 3.86292607e-01 4.35638726e-01 1.08673573e+00
3.76574188e-01 -3.63846608e-02 8.17761600e-01 -7.25792408e-01
-1.75718617e+00 -7.54783392e-01 -8.10503542e-01 2.49045059e-01
5.04451573e-01 -6.33258522e-01 -6.26884460e-01 -5.20440400e-01] | [4.2863874435424805, 2.6770644187927246] |
ea263c20-6fcc-4e80-b0a9-5c0967b995f9 | large-scale-nonlinear-granger-causality-for | null | null | https://www.nature.com/articles/s41598-021-87316-6 | https://www.nature.com/articles/s41598-021-87316-6.pdf | Large-scale nonlinear Granger causality for inferring directed dependence from short multivariate time-series data | A key challenge to gaining insight into complex systems is inferring nonlinear causal directional relations from observational time-series data. Specifically, estimating causal relationships between interacting components in large systems with only short recordings over few temporal observations remains an important, yet unresolved problem. Here, we introduce large-scale nonlinear Granger causality (lsNGC) which facilitates conditional Granger causality between two multivariate time series conditioned on a large number of confounding time series with a small number of observations. By modeling interactions with nonlinear state-space transformations from limited observational data, lsNGC identifies casual relations with no explicit a priori assumptions on functional interdependence between component time series in a computationally efficient manner. Additionally, our method provides a mathematical formulation revealing statistical significance of inferred causal relations. We extensively study the ability of lsNGC in inferring directed relations from two-node to thirty-four node chaotic time-series systems. Our results suggest that lsNGC captures meaningful interactions from limited observational data, where it performs favorably when compared to traditionally used methods. Finally, we demonstrate the applicability of lsNGC to estimating causality in large, real-world systems by inferring directional nonlinear, causal relationships among a large number of relatively short time series acquired from functional Magnetic Resonance Imaging (fMRI) data of the human brain. | ['M. Ali Vosoughi & Anas Abidin', 'Adora M. DSouza', 'Axel Wismüller'] | 2021-04-09 | null | null | null | null | ['learning-network-representations'] | ['methodology'] | [ 3.21995109e-01 -3.52716029e-01 7.83603787e-02 -1.21759318e-01
-1.32690877e-01 -6.98333979e-01 1.09523070e+00 1.72291785e-01
-8.53007566e-03 1.08457541e+00 3.54123473e-01 -6.21244192e-01
-8.92182708e-01 -4.50964332e-01 -5.25417924e-01 -7.94450402e-01
-1.31142044e+00 4.04833078e-01 -5.41402251e-02 -7.42746592e-02
-3.41547988e-02 6.67507410e-01 -6.90815449e-01 -2.96033144e-01
7.45239079e-01 4.78621632e-01 -3.31676044e-02 8.03387284e-01
7.44636238e-01 7.59922683e-01 -3.96946400e-01 3.92806143e-01
-1.54973567e-02 -6.25120997e-01 -5.02733588e-01 -5.70352793e-01
-4.96912859e-02 -8.33786502e-02 -3.81105393e-01 6.66128576e-01
1.00569487e-01 6.44291341e-02 8.79920959e-01 -1.39604259e+00
-5.63322425e-01 7.00290024e-01 -3.21140826e-01 7.86956072e-01
7.52102882e-02 3.13182086e-01 9.92769718e-01 -6.27785563e-01
4.47144747e-01 1.43217206e+00 7.33004928e-01 -1.00669891e-01
-1.96599221e+00 -8.17103982e-01 1.35828882e-01 1.53083593e-01
-1.24207997e+00 -1.96943805e-01 7.93898463e-01 -8.91163290e-01
8.81239831e-01 5.19521356e-01 8.16048980e-01 1.13183963e+00
8.62257957e-01 -1.81518197e-01 1.43544173e+00 2.87549477e-02
2.21082687e-01 -8.23197007e-01 3.48234773e-01 3.96309346e-01
1.70448855e-01 8.71109545e-01 -7.42826700e-01 -9.70333636e-01
1.00782037e+00 5.98269515e-02 -2.94558942e-01 1.52168170e-01
-1.81913841e+00 8.25563550e-01 3.94626856e-01 2.82311857e-01
-6.55121982e-01 6.02044761e-01 1.81640938e-01 5.90696216e-01
5.70164084e-01 5.57183623e-01 -5.62155604e-01 1.64034754e-01
-6.77007198e-01 3.09637487e-01 7.38141298e-01 4.59266305e-01
2.65391409e-01 5.87803237e-02 2.29982749e-01 1.79930925e-01
2.16855362e-01 1.13670731e+00 -4.15831208e-02 -9.38655138e-01
-3.90406512e-02 1.34819821e-01 -6.08521327e-02 -1.44976699e+00
-8.21209669e-01 -4.28461194e-01 -1.46227300e+00 -9.87839550e-02
2.92295069e-01 -2.83308327e-01 -3.53609979e-01 2.17476463e+00
2.20172822e-01 5.71549654e-01 -3.19996864e-01 7.44897246e-01
2.56422281e-01 6.60301566e-01 -5.19872755e-02 -1.06778312e+00
1.27649820e+00 2.74633430e-02 -8.67730141e-01 1.13291606e-01
2.38573551e-02 -4.80882317e-01 4.68173653e-01 2.36486018e-01
-7.84142494e-01 -1.04583755e-01 -4.57021534e-01 5.38893700e-01
-1.82912588e-01 -2.60839581e-01 9.36364472e-01 -1.40665233e-01
-1.11516035e+00 5.50190032e-01 -1.26448393e+00 -3.56673002e-01
-1.05848178e-01 3.50245833e-01 -3.93716097e-01 2.55301237e-01
-1.47292686e+00 7.32171416e-01 5.92607632e-02 5.11289358e-01
-1.22338355e+00 -1.04627752e+00 -3.61949116e-01 -2.26355702e-01
3.92145991e-01 -6.70642853e-01 5.97747684e-01 -2.01550275e-01
-8.05587530e-01 3.50867421e-03 -4.45274413e-01 -4.45979178e-01
2.85077929e-01 -7.49719143e-02 -8.08643818e-01 1.99226961e-01
2.44626865e-01 1.26693040e-01 7.64337957e-01 -1.01475644e+00
1.51997328e-01 -3.25742722e-01 -4.84200835e-01 -1.77352697e-01
-1.69769526e-02 -4.44115698e-02 4.13819999e-01 -1.02042508e+00
3.76037210e-01 -1.16405606e+00 -3.80761087e-01 -1.38341725e-01
-5.43547928e-01 -2.52738297e-01 7.01144218e-01 -5.75734794e-01
1.30235004e+00 -1.69532299e+00 4.14475143e-01 4.33529973e-01
9.23452079e-01 -3.66698027e-01 -1.90848976e-01 9.41876292e-01
-6.43218219e-01 1.19601049e-01 -6.85141921e-01 3.57598960e-01
-5.76252520e-01 2.09688753e-01 -6.87458754e-01 8.93763840e-01
3.92124832e-01 1.09246147e+00 -1.34935641e+00 -1.22832850e-01
2.95700654e-02 3.24656218e-01 -3.64742428e-01 2.75174737e-01
1.39052138e-01 1.04103994e+00 -3.80881310e-01 2.88541853e-01
2.72222102e-01 -5.64956188e-01 4.29304332e-01 -5.72170913e-02
-3.19913715e-01 1.01337790e-01 -7.72378206e-01 1.05918217e+00
-1.02703705e-01 1.01205075e+00 -3.03206652e-01 -1.35989475e+00
5.38158774e-01 6.10219777e-01 8.39257717e-01 -3.66457224e-01
-1.70102268e-02 4.60454300e-02 5.44644892e-01 -5.39652586e-01
-1.93824857e-01 -4.87187982e-01 -1.84083700e-01 6.98607445e-01
-6.46848604e-02 -3.94069627e-02 -1.82434134e-02 5.08739412e-01
1.66340840e+00 -6.00584984e-01 7.08433807e-01 -8.62725496e-01
6.60915226e-02 -1.99165046e-02 5.03653824e-01 7.06723332e-01
-2.47296602e-01 8.37107003e-02 8.00470889e-01 -3.05926025e-01
-9.05974030e-01 -1.47445035e+00 -4.64900434e-01 3.02274287e-01
-2.06374511e-01 -3.52476478e-01 -1.13935940e-01 -1.37838751e-01
2.60079384e-01 4.30976093e-01 -1.12357330e+00 -3.49926472e-01
-7.93178558e-01 -1.33347952e+00 5.66350460e-01 2.85549641e-01
-1.49457857e-01 -8.99987638e-01 -4.53307599e-01 2.78785139e-01
-2.74903029e-01 -9.32804108e-01 -3.32229823e-01 2.41205305e-01
-1.15548515e+00 -1.24773049e+00 -2.30830431e-01 -1.17995918e-01
7.25584507e-01 1.70766652e-01 1.11635661e+00 -4.62302268e-02
-4.10272658e-01 1.04232237e-01 8.45585316e-02 9.61171240e-02
-3.78146410e-01 -5.11429667e-01 8.13066006e-01 -2.00836718e-01
-1.27497792e-01 -1.16423523e+00 -4.93673235e-01 4.01724339e-01
-7.36313045e-01 9.37804487e-03 3.98827791e-01 1.01795816e+00
2.83983231e-01 3.03242624e-01 9.10400629e-01 -6.35158420e-01
7.40195155e-01 -1.05933428e+00 -7.73735344e-01 8.37810524e-03
-7.91040778e-01 1.02352232e-01 7.93324232e-01 -8.96234095e-01
-7.19422042e-01 -3.73217285e-01 6.80529177e-01 -3.86167228e-01
2.73621958e-02 9.52930629e-01 4.73841488e-01 3.47213954e-01
6.30505264e-01 7.52446800e-02 5.35946526e-02 -1.14411503e-01
5.42146862e-01 -1.22059852e-01 7.18823433e-01 -6.85049176e-01
8.19044292e-01 8.34061742e-01 5.73153257e-01 -8.49231958e-01
-1.87975645e-01 -2.58433044e-01 -8.87671590e-01 -3.74890864e-01
7.19246507e-01 -7.94119716e-01 -1.33019769e+00 3.10657769e-01
-1.15579188e+00 -5.18676579e-01 -5.41092604e-02 8.90068054e-01
-2.63737559e-01 -1.29137188e-01 -7.58290827e-01 -1.10092175e+00
1.02200434e-01 -5.69493771e-01 8.73135567e-01 -4.35151219e-01
-7.93588877e-01 -1.58944631e+00 5.62925279e-01 -2.86426097e-01
2.52978981e-01 7.60258079e-01 1.03896260e+00 -1.78755969e-01
-4.47440743e-01 -7.75551423e-02 -9.26592648e-02 -2.62671947e-01
5.32663107e-01 2.66974211e-01 -3.92188132e-01 -1.99541241e-01
1.59978539e-01 2.23618448e-01 6.83461070e-01 8.84116650e-01
6.87062502e-01 -4.92443800e-01 -6.04428113e-01 2.41330534e-01
1.00141191e+00 3.65656137e-01 8.99058301e-03 -4.72005486e-01
8.59957278e-01 9.46413755e-01 3.09763193e-01 3.53628397e-01
4.05821145e-01 2.27243364e-01 1.70539230e-01 -7.18380511e-02
2.33372182e-01 -1.79561749e-01 3.10214341e-01 1.46121573e+00
-2.76951134e-01 -1.03042677e-01 -1.06460822e+00 6.05466962e-01
-1.87752378e+00 -1.14552724e+00 -9.62280095e-01 2.08315921e+00
9.08756018e-01 -1.49552166e-01 2.01288134e-01 -2.38572612e-01
5.86594820e-01 1.97797492e-02 -9.29634809e-01 3.80516917e-01
-4.63218153e-01 -1.53043956e-01 5.16919792e-01 7.81894684e-01
-8.26269627e-01 4.66514438e-01 7.57106972e+00 2.13022843e-01
-1.09765887e+00 2.24425673e-01 5.53186119e-01 -2.59604305e-01
-4.72749352e-01 4.06116962e-01 -4.03153300e-02 4.16982949e-01
1.36440158e+00 -4.76075590e-01 7.31907487e-01 -2.09281847e-01
1.03933871e+00 -1.81209922e-01 -1.11996770e+00 5.90770960e-01
-4.60811257e-01 -1.17498672e+00 -3.37508321e-01 2.67776728e-01
9.03549910e-01 2.49025941e-01 -4.41798866e-02 -5.31710744e-01
9.30105627e-01 -1.11692786e+00 5.43763161e-01 9.34729993e-01
7.86424518e-01 -1.62308961e-01 1.71025917e-01 4.08208400e-01
-1.27348518e+00 5.61333597e-02 -4.01905999e-02 -6.48496509e-01
5.99182189e-01 1.39876866e+00 -6.67867362e-01 3.77997220e-01
6.94392979e-01 1.19944668e+00 -1.83744997e-01 4.11557823e-01
-2.12986678e-01 1.16778612e+00 -6.10991776e-01 6.32105544e-02
-2.85258293e-01 -4.28118944e-01 8.56304049e-01 7.73876786e-01
1.06145330e-01 7.98967361e-01 -1.28489166e-01 1.22669482e+00
4.01811063e-01 -6.12252235e-01 -8.25210690e-01 -4.15751308e-01
5.75638056e-01 1.01799774e+00 -9.16930079e-01 -3.52801740e-01
-1.14726666e-02 5.04261971e-01 2.16953471e-01 7.32912183e-01
-7.66625524e-01 3.88384879e-01 5.74157596e-01 1.09662926e-02
-4.21522290e-01 -8.82314682e-01 -4.10003066e-01 -1.30544198e+00
-1.48389116e-01 -5.00815332e-01 4.65245664e-01 -7.41192520e-01
-1.80889082e+00 5.03567457e-01 6.02140307e-01 -1.18287516e+00
-3.36597264e-01 -2.30048507e-01 -7.09851623e-01 1.03664136e+00
-9.56335127e-01 -6.85402155e-01 1.59595981e-01 6.27250791e-01
9.94925052e-02 1.88991874e-01 7.08284795e-01 3.17811109e-02
-6.36096716e-01 -3.73553276e-01 1.24920845e-01 -2.05183059e-01
5.87137580e-01 -1.25338864e+00 5.58133900e-01 9.84221637e-01
-2.26936460e-01 1.27156878e+00 1.00858688e+00 -1.33349323e+00
-1.35742259e+00 -9.70777094e-01 9.34692681e-01 -6.47370338e-01
1.52553177e+00 -7.30458856e-01 -9.49679375e-01 6.67528629e-01
3.10032181e-02 3.45369697e-01 6.52970731e-01 3.12699586e-01
-2.90645003e-01 1.14216059e-01 -6.38475418e-01 6.20502412e-01
1.24453604e+00 -7.39685059e-01 -6.40997410e-01 4.87208098e-01
7.42935896e-01 3.15136373e-01 -1.19846797e+00 4.31713432e-01
7.55389214e-01 -4.43945825e-01 1.11669683e+00 -7.15714216e-01
5.87914526e-01 -5.26089907e-01 5.23966700e-02 -1.69205356e+00
-6.00080431e-01 -7.51286030e-01 -1.87449694e-01 7.20881879e-01
2.64866382e-01 -1.16026902e+00 -8.64683390e-02 4.64043200e-01
3.98504376e-01 -1.71472400e-01 -8.24640632e-01 -9.58767951e-01
2.85930753e-01 -4.44226265e-01 2.78416634e-01 1.62791586e+00
1.56057805e-01 3.90846342e-01 -5.09705901e-01 6.03301406e-01
1.07046223e+00 1.56947255e-01 2.69905955e-01 -1.52380097e+00
-3.33286524e-01 -3.33567768e-01 -1.34870127e-01 -3.28573287e-01
2.37305462e-01 -6.86454773e-01 7.17243627e-02 -8.52056623e-01
2.58935928e-01 -5.34833312e-01 -3.26471210e-01 3.53859097e-01
-4.45523381e-01 1.16848767e-01 -2.79975198e-02 6.82019413e-01
8.08956996e-02 6.90299869e-01 1.07023811e+00 8.01617801e-02
-5.08577041e-02 -1.95082754e-01 -2.77253330e-01 5.08898735e-01
4.67582345e-01 -7.78179407e-01 -6.56233191e-01 6.04578257e-02
4.02940452e-01 8.12178075e-01 1.05005121e+00 -3.19604397e-01
3.41943800e-01 -6.31667078e-01 1.43705398e-01 -4.73872423e-01
-3.51579823e-02 -7.54675210e-01 7.47944295e-01 7.64797449e-01
-4.60914612e-01 6.03296340e-01 1.23590633e-01 9.13476229e-01
-1.20241299e-01 7.82795787e-01 1.94911748e-01 1.17082842e-01
-1.85132772e-01 2.41110504e-01 -6.84873164e-01 -5.77522591e-02
7.38435388e-01 6.14919782e-01 -4.87683803e-01 -3.95232350e-01
-9.33297634e-01 3.74067545e-01 -2.29837984e-01 3.06802690e-01
5.57829261e-01 -1.38164854e+00 -9.00706768e-01 1.91317797e-01
-7.10471720e-02 -4.06663835e-01 2.99938142e-01 1.53534079e+00
-1.82953123e-02 7.46651947e-01 5.30478321e-02 -7.94285893e-01
-1.02420712e+00 5.58941960e-01 9.99118909e-02 -2.16121703e-01
-6.24186337e-01 5.33875942e-01 7.19381928e-01 -5.54733992e-01
-7.20852077e-01 -6.87107444e-01 -3.72149087e-02 1.28742278e-01
4.05174434e-01 5.59214711e-01 -2.96478391e-01 -5.45300663e-01
-5.65958738e-01 3.12952816e-01 5.47973394e-01 -4.72675920e-01
1.40785682e+00 -3.09938490e-01 -7.27834702e-01 1.26106191e+00
1.33118749e+00 -2.00148255e-01 -1.35383105e+00 -2.17409372e-01
-5.16869053e-02 -8.44327509e-02 -2.30241101e-02 -8.05606008e-01
-8.75179648e-01 6.20822310e-01 3.43619168e-01 6.50815010e-01
1.12510741e+00 2.76888013e-01 2.58653104e-01 2.30785981e-01
3.85719985e-01 -2.41128951e-01 -1.73749983e-01 2.54694998e-01
1.29806626e+00 -1.09496212e+00 1.27264112e-01 -3.86033833e-01
-2.28712745e-02 9.89666939e-01 -4.88487259e-02 -5.34725010e-01
1.28279412e+00 4.38687205e-01 -3.04802716e-01 -7.67971098e-01
-1.28736997e+00 1.86309874e-01 4.46555465e-01 2.89965242e-01
3.14358830e-01 8.29959750e-01 -4.51946020e-01 1.07459418e-01
-4.06221092e-01 -2.42210180e-01 5.26431620e-01 4.23523933e-01
2.85207123e-01 -6.31190777e-01 -6.30414009e-01 6.45317972e-01
-1.21886000e-01 -6.02741659e-01 -3.48961949e-01 8.02236378e-01
-4.06687170e-01 1.07367849e+00 2.44776949e-01 -1.54037699e-01
1.18234843e-01 -3.13449919e-01 6.54713660e-02 -1.97976038e-01
-1.80224314e-01 3.81097257e-01 -5.45523539e-02 -6.99155331e-01
-6.72852993e-01 -1.35592127e+00 -1.20012307e+00 -6.47809982e-01
-3.78179550e-01 9.39927921e-02 1.73915252e-01 1.21701741e+00
3.25958163e-01 8.55551004e-01 7.29543924e-01 -6.58637881e-01
-9.87294465e-02 -9.31236506e-01 -6.86155796e-01 2.59123117e-01
9.49829936e-01 -7.71298647e-01 -6.42113626e-01 4.14728612e-01] | [7.7075581550598145, 5.18927526473999] |
4a296b33-e0f6-4b6d-8332-6eed86cc4705 | handwritten-arabic-numeral-recognition-using | 1702.04663 | null | http://arxiv.org/abs/1702.04663v1 | http://arxiv.org/pdf/1702.04663v1.pdf | Handwritten Arabic Numeral Recognition using Deep Learning Neural Networks | Handwritten character recognition is an active area of research with
applications in numerous fields. Past and recent works in this field have
concentrated on various languages. Arabic is one language where the scope of
research is still widespread, with it being one of the most popular languages
in the world and being syntactically different from other major languages. Das
et al. \cite{DBLP:journals/corr/abs-1003-1891} has pioneered the research for
handwritten digit recognition in Arabic. In this paper, we propose a novel
algorithm based on deep learning neural networks using appropriate activation
function and regularization layer, which shows significantly improved accuracy
compared to the existing Arabic numeral recognition methods. The proposed model
gives 97.4 percent accuracy, which is the recorded highest accuracy of the
dataset used in the experiment. We also propose a modification of the method
described in \cite{DBLP:journals/corr/abs-1003-1891}, where our method scores
identical accuracy as that of \cite{DBLP:journals/corr/abs-1003-1891}, with the
value of 93.8 percent. | ['Akm Ashiquzzaman', 'Abdul Kawsar Tushar'] | 2017-02-15 | null | null | null | null | ['handwritten-digit-recognition'] | ['computer-vision'] | [-2.14698344e-01 -5.41008115e-01 1.48642913e-01 -3.27675074e-01
-2.99848020e-01 -3.29173267e-01 2.58389294e-01 1.46081567e-01
-5.86325884e-01 8.88324976e-01 -1.61998987e-01 -4.71652031e-01
-3.33310962e-01 -9.19138134e-01 -5.84581912e-01 -6.95730925e-01
-1.72933079e-02 2.20110491e-01 2.61140943e-01 -4.11454588e-01
7.66679406e-01 9.15651262e-01 -1.25109565e+00 1.44525796e-01
8.00524175e-01 1.19307899e+00 -1.34312138e-01 6.11201286e-01
-3.37745659e-02 9.08219039e-01 -7.55653679e-01 -6.46783113e-01
3.04781348e-01 -4.42506135e-01 -6.20308816e-01 -7.77852684e-02
1.17946811e-01 -2.46058300e-01 -3.92056763e-01 1.22532845e+00
5.87419808e-01 9.76353139e-02 1.00850332e+00 -6.60503447e-01
-1.09554935e+00 2.85979569e-01 -9.38972414e-01 3.43988508e-01
2.05752984e-01 -4.13710475e-01 5.25042057e-01 -1.08439553e+00
9.89808813e-02 8.63482356e-01 6.80875957e-01 1.08813018e-01
-6.74785018e-01 -7.34502017e-01 -7.26678669e-02 3.17834973e-01
-1.49729395e+00 7.42264464e-02 7.39840508e-01 -3.04979891e-01
8.24751496e-01 2.96192199e-01 1.96393356e-01 6.37957215e-01
5.03034949e-01 9.32830632e-01 1.44474328e+00 -9.58551943e-01
1.13791369e-01 -4.02699858e-02 3.79490882e-01 7.71530092e-01
4.29120928e-01 -3.13126534e-01 -1.65358379e-01 2.10032672e-01
1.10363925e+00 2.86780596e-02 -2.11846991e-03 3.26380312e-01
-9.08420622e-01 1.04303670e+00 2.18015775e-01 6.65155709e-01
-3.71748805e-01 -2.60232449e-01 3.81530434e-01 4.44780916e-01
8.23942944e-02 1.83824614e-01 -1.76421955e-01 -2.08978057e-01
-8.92427146e-01 2.80880392e-01 9.11405861e-01 7.77042150e-01
2.59940565e-01 5.00884056e-01 1.39581501e-01 1.19833755e+00
3.64221483e-01 4.24605399e-01 6.46822214e-01 -5.27350485e-01
3.41456056e-01 7.57560372e-01 1.50237471e-01 -1.34677017e+00
-3.06026101e-01 -2.73939282e-01 -1.12323952e+00 3.86163801e-01
5.52569151e-01 -3.65124613e-01 -7.73918331e-01 9.44071352e-01
-2.33145490e-01 -4.41771537e-01 1.68471411e-01 8.71388137e-01
6.28130794e-01 1.06460321e+00 -2.18437597e-01 -1.23921923e-01
1.13614523e+00 -9.69405651e-01 -5.30190110e-01 -6.05335925e-03
2.14237459e-02 -1.29478645e+00 1.04230773e+00 9.72117484e-01
-6.91709578e-01 -3.89094442e-01 -1.29643893e+00 4.09977168e-01
-4.34955269e-01 6.38918996e-01 4.12079811e-01 7.97555208e-01
-7.91439831e-01 4.06977326e-01 -6.84822440e-01 -4.90054995e-01
1.96967516e-02 5.19279599e-01 -2.42443770e-01 -9.92542282e-02
-8.62187326e-01 1.10478449e+00 5.40793359e-01 2.21176416e-01
-2.35519856e-01 8.95210505e-02 -3.19872588e-01 -3.17083389e-01
2.08263054e-01 5.42515337e-01 1.08249927e+00 -9.57017541e-01
-1.53084838e+00 6.02984846e-01 2.42332026e-01 -1.34966671e-01
4.73375469e-01 -1.53321877e-01 -9.35705125e-01 1.24025457e-02
-3.58404160e-01 1.39102519e-01 5.47491014e-01 -8.26156259e-01
-6.27696097e-01 -5.07715821e-01 -5.92197385e-03 -2.02933088e-01
-4.70622152e-01 4.32380766e-01 -3.69662225e-01 -1.08507979e+00
1.49420127e-01 -7.60375381e-01 2.01411545e-01 -3.18213761e-01
-1.27379343e-01 -1.06793508e-01 9.34125662e-01 -1.04787886e+00
1.29882169e+00 -1.90595257e+00 -4.19277072e-01 4.22289759e-01
-3.88788074e-01 7.40805387e-01 2.17722058e-01 7.18834937e-01
-4.42415513e-02 -1.50337726e-01 -5.05689800e-01 2.42204919e-01
4.33581471e-02 4.00512889e-02 -1.44054234e-01 5.02974749e-01
2.71986574e-01 4.16343898e-01 -3.38097930e-01 -1.92160025e-01
-8.47719163e-02 5.00907123e-01 5.55337854e-02 -6.71432614e-02
5.73854148e-02 -7.54419193e-02 -5.58073401e-01 1.03334057e+00
7.68445611e-01 4.84622531e-02 -1.29181132e-01 -1.75220668e-01
-4.56144452e-01 -4.26032811e-01 -1.54350257e+00 6.18796349e-01
1.60530970e-01 8.72242332e-01 -1.85792983e-01 -1.38279700e+00
1.60597086e+00 1.97051734e-01 1.01076096e-01 -7.32383072e-01
4.74524617e-01 6.96157455e-01 1.95109144e-01 -3.54034305e-01
7.12290645e-01 -3.48160625e-03 1.09888859e-01 4.38472241e-01
-1.79756150e-01 2.67491452e-02 5.29989958e-01 -2.36146748e-01
5.42942643e-01 1.69488281e-01 2.22704098e-01 -2.79717803e-01
8.98828626e-01 4.72709071e-03 2.20706433e-01 7.07730412e-01
-2.97526360e-01 6.14905834e-01 4.83995199e-01 -7.41240382e-01
-1.29323554e+00 -5.50978720e-01 -4.89919364e-01 7.93381751e-01
-2.32021585e-02 2.71667302e-01 -7.70384192e-01 -2.34989822e-01
-5.35296202e-02 4.73581702e-01 -4.27829921e-01 1.06767744e-01
-6.66066110e-01 -1.10683715e+00 1.16050863e+00 5.92853129e-01
1.08218157e+00 -1.42455578e+00 -6.02514446e-01 2.40970671e-01
2.83573538e-01 -7.70903230e-01 -4.22178283e-02 2.86553323e-01
-8.63568187e-01 -8.88355613e-01 -1.27729774e+00 -1.05203879e+00
7.62814164e-01 -3.30580056e-01 7.19409347e-01 2.17362523e-01
-3.26445609e-01 1.50845330e-02 -8.45235169e-01 -5.74506998e-01
-4.23978180e-01 7.54215615e-03 2.02379450e-02 2.47864395e-01
8.25010478e-01 -1.06288418e-01 -3.09865773e-01 2.21385911e-01
-1.00603807e+00 -4.37097788e-01 9.33363199e-01 9.85440075e-01
4.10188347e-01 5.81042618e-02 5.82906425e-01 -6.54084563e-01
1.00556004e+00 -1.05901010e-01 -9.24995005e-01 3.70026767e-01
-6.28269792e-01 -1.28420964e-01 8.97730172e-01 -3.87584358e-01
-7.82721102e-01 3.81623097e-02 -2.97505349e-01 2.36839548e-01
-3.17991078e-01 7.43361831e-01 1.06732130e-01 -3.59768927e-01
6.08365834e-01 7.21071839e-01 -2.16176227e-01 -7.52557337e-01
-4.76560742e-01 1.16000676e+00 5.65826952e-01 -5.67509413e-01
3.02436650e-01 2.92541031e-02 -8.90442729e-02 -1.15545523e+00
-3.51514578e-01 -1.07643031e-01 -5.71848631e-01 -1.49265245e-01
5.88550210e-01 -3.06354821e-01 -7.80689240e-01 1.24283624e+00
-9.70606506e-01 -5.07987700e-02 4.42588568e-01 8.53146493e-01
-6.93223625e-02 6.48384273e-01 -7.27470994e-01 -9.83881116e-01
-3.38520736e-01 -1.14189982e+00 3.83758336e-01 5.64467192e-01
3.85826379e-02 -8.01399052e-01 -2.35089466e-01 5.23333959e-02
4.87120569e-01 3.93458992e-01 9.77331579e-01 -1.01760995e+00
-7.20194504e-02 -7.04941571e-01 -5.73114097e-01 7.38971055e-01
2.35503763e-01 3.82707834e-01 -6.79853141e-01 -1.72171548e-01
-1.17407896e-01 -4.29719359e-01 5.74912012e-01 1.86855122e-01
1.04995298e+00 -1.72715843e-01 3.19186687e-01 -1.71236592e-04
1.64975762e+00 9.83703434e-01 8.51084828e-01 8.32907379e-01
1.55504316e-01 1.40271321e-01 2.59821951e-01 7.56039858e-01
4.92556132e-02 3.27608585e-01 4.65904288e-02 1.26394913e-01
3.73357803e-01 1.99206129e-01 2.69328684e-01 8.85124147e-01
-2.82537609e-01 -2.20141709e-01 -1.28252769e+00 3.68972987e-01
-1.36108875e+00 -6.93923354e-01 -3.74055356e-01 2.08310008e+00
8.28682721e-01 1.12653889e-01 2.62260646e-01 7.22976327e-01
6.51432335e-01 9.40183774e-02 -3.41254920e-01 -1.05310845e+00
-3.55990291e-01 5.48460007e-01 5.04436731e-01 5.24261832e-01
-1.27501333e+00 8.00176322e-01 5.24264336e+00 6.71435714e-01
-1.49101746e+00 -4.20964658e-01 6.58074677e-01 5.52168608e-01
2.88709491e-01 -2.96289712e-01 -7.19765604e-01 5.71899474e-01
7.64330328e-01 -2.31251828e-02 1.95044532e-01 5.94013274e-01
3.52838188e-02 -3.75416458e-01 -6.99502468e-01 8.88207316e-01
5.52838683e-01 -1.00813270e+00 2.24900663e-01 -9.03046280e-02
7.08815932e-01 -3.50801289e-01 1.91081762e-02 1.08196057e-01
-4.30801958e-02 -1.25819099e+00 6.67154789e-01 6.28782153e-01
4.39306378e-01 -1.04432356e+00 1.24070621e+00 2.91269690e-01
-7.49107301e-01 -2.73020744e-01 -4.94292408e-01 -3.55102807e-01
-2.92142987e-01 5.99047065e-01 -5.54931045e-01 5.06770492e-01
8.63994956e-01 4.28921789e-01 -3.94248217e-01 1.29345298e+00
-3.28037776e-02 8.41103792e-01 -6.40748218e-02 -7.13695586e-01
3.37694198e-01 -5.42176545e-01 2.50280201e-01 1.58882022e+00
5.16195655e-01 1.35335311e-01 -9.63590369e-02 5.71594775e-01
9.08572674e-02 5.12744725e-01 -3.00512433e-01 -3.48499894e-01
1.92881793e-01 6.08456552e-01 -9.55165029e-01 -2.79155016e-01
-3.51381123e-01 9.24468100e-01 3.20027731e-02 2.51382232e-01
-5.91235936e-01 -1.25242484e+00 -2.60445893e-01 -1.42621085e-01
4.92723495e-01 -5.28222740e-01 -3.44396234e-01 -9.60521162e-01
3.13782334e-01 -1.19115317e+00 1.56761557e-01 -4.52525079e-01
-1.28491080e+00 8.74598622e-01 -2.37776116e-01 -1.12899017e+00
-2.53836624e-02 -1.27273250e+00 -2.70046949e-01 9.43364680e-01
-1.08027244e+00 -9.21812713e-01 -2.24045366e-01 5.25171340e-01
5.63684702e-01 -9.14358020e-01 9.80189145e-01 5.64216435e-01
-5.52932858e-01 8.60355318e-01 8.09859633e-01 6.54268205e-01
7.17075706e-01 -8.84548724e-01 -3.29360589e-02 9.79973257e-01
-9.76940393e-02 4.44135696e-01 3.68241698e-01 -6.30091906e-01
-1.30491889e+00 -6.71018600e-01 9.96015131e-01 1.49095699e-01
4.86944467e-01 5.56130484e-02 -1.04600692e+00 4.31665897e-01
2.47605130e-01 -3.36098760e-01 4.52641547e-01 -5.95343232e-01
-1.62821889e-01 -3.75369996e-01 -1.38118160e+00 4.83271301e-01
-5.27233407e-02 -2.08288774e-01 -6.17629647e-01 -1.21468212e-02
6.68683946e-02 -4.67507243e-01 -8.25992346e-01 2.18215600e-01
7.14156628e-01 -1.01760018e+00 5.16493857e-01 -3.53372961e-01
7.02297986e-01 -2.96745479e-01 -5.75457692e-01 -6.95267975e-01
-1.79954141e-01 -1.65564511e-02 5.15713133e-02 1.11253226e+00
5.49722791e-01 -8.43702435e-01 6.81988716e-01 5.29712379e-01
-1.91139281e-02 -8.75258684e-01 -6.12242699e-01 -9.03926849e-01
4.25476879e-01 -5.20827211e-02 2.27178946e-01 8.36877942e-01
-3.75534266e-01 -3.55350405e-01 -4.37711537e-01 -1.03968740e-01
3.66855741e-01 -3.67097594e-02 3.02330762e-01 -1.07854331e+00
-8.03549215e-02 -4.91080344e-01 -6.46173179e-01 -7.08862424e-01
-2.19877005e-01 -3.92796010e-01 -1.00329272e-01 -1.65181494e+00
-1.96176499e-01 -3.65791678e-01 -3.78121167e-01 6.44758523e-01
2.90633798e-01 5.14687240e-01 2.33832449e-01 2.31461048e-01
-6.29907474e-02 2.36509368e-01 9.95595276e-01 -3.64717960e-01
-9.72278193e-02 8.66077542e-02 -7.01256752e-01 6.31030738e-01
9.22687054e-01 -2.43378773e-01 7.47824684e-02 -6.51944160e-01
-2.40355991e-02 -2.99205571e-01 -4.84708473e-02 -1.05840623e+00
3.82051289e-01 -3.17463011e-01 7.73019314e-01 -5.56945324e-01
1.46647930e-01 -7.51563549e-01 -4.89228182e-02 5.51762879e-01
-8.98408145e-02 4.98767465e-01 3.67846966e-01 1.46978632e-01
-4.42748457e-01 -6.54237986e-01 8.82924736e-01 -1.54177189e-01
-8.55320811e-01 -1.85013115e-01 -8.89915407e-01 -4.26631980e-02
1.08872712e+00 -5.99395573e-01 -2.19294652e-01 -2.68194199e-01
-2.52796084e-01 -3.00340712e-01 9.03163776e-02 3.76512349e-01
6.81628942e-01 -1.11534417e+00 -9.98800755e-01 7.97926709e-02
-2.44005531e-01 -4.84701365e-01 -1.68377444e-01 7.36000478e-01
-1.41132760e+00 6.68927312e-01 -7.64231324e-01 -8.90895054e-02
-1.05816472e+00 -2.60731634e-02 2.26075590e-01 3.39417756e-02
-4.02431637e-01 6.93969071e-01 -7.16569602e-01 -2.39335209e-01
4.57465768e-01 -2.22891215e-02 -5.21662414e-01 -1.09900296e-01
5.02833426e-01 6.59411907e-01 3.92270327e-01 -7.90735781e-01
-5.63377082e-01 5.77553570e-01 -2.29835257e-01 -2.27669299e-01
1.70870948e+00 4.58506018e-01 -4.93227750e-01 4.41919625e-01
1.04488599e+00 2.17586517e-01 -6.32998288e-01 1.27106339e-01
2.80277908e-01 -5.20707786e-01 -2.71021545e-01 -1.27746212e+00
-9.58325386e-01 8.24244022e-01 9.73944306e-01 -4.67823930e-02
1.06180179e+00 -6.30733967e-01 7.20200479e-01 7.47429311e-01
2.46643484e-01 -1.41678786e+00 3.06678098e-02 1.10083723e+00
1.18247640e+00 -1.17921686e+00 1.66148573e-01 1.72968283e-01
-6.97939694e-01 1.55559957e+00 7.43783593e-01 -6.08819425e-01
6.81777060e-01 2.12593809e-01 2.83657223e-01 1.44454956e-01
1.58425942e-01 4.02150005e-01 1.97375596e-01 4.14000660e-01
9.45956171e-01 -7.32015399e-03 -9.55427825e-01 9.02769387e-01
-6.91320077e-02 2.10173696e-01 7.98972070e-01 1.36817718e+00
-3.20441425e-01 -1.31607926e+00 -7.01724648e-01 6.12533748e-01
-7.62829006e-01 -8.27056095e-02 -7.39945829e-01 9.09013927e-01
-5.93758635e-02 8.75478566e-01 -1.51775077e-01 -3.10357571e-01
3.77592921e-01 2.88815171e-01 4.26989883e-01 6.01270534e-02
-5.04740179e-01 -2.18830824e-01 -1.14965774e-01 2.56928146e-01
-3.27278495e-01 -4.34526145e-01 -1.39055920e+00 -4.79214132e-01
-9.87855867e-02 3.86841208e-01 8.65955055e-01 8.61878216e-01
8.64914805e-03 2.65190363e-01 5.31467795e-01 -6.08699739e-01
-7.03741550e-01 -1.03371298e+00 -8.80510211e-01 -3.02575547e-02
2.15233654e-01 -5.77140927e-01 6.92460090e-02 1.64375067e-01] | [11.868346214294434, 2.7103567123413086] |
690bf77a-5eb5-4575-88f3-aa76c86829d0 | exact-recovery-of-community-detection-in | 2209.14859 | null | https://arxiv.org/abs/2209.14859v1 | https://arxiv.org/pdf/2209.14859v1.pdf | Exact Recovery of Community Detection in dependent Gaussian Mixture Models | We study the community detection problem on a Gaussian mixture model, in which (1) vertices are divided into $k\geq 2$ distinct communities that are not necessarily equally-sized; (2) the Gaussian perturbations for different entries in the observation matrix are not necessarily independent or identically distributed. We prove necessary and sufficient conditions for the exact recovery of the maximum likelihood estimation (MLE), and discuss the cases when these necessary and sufficient conditions give sharp threshold. Applications include the community detection on a graph where the Gaussian perturbations of observations on each edge is the sum of i.i.d.~Gaussian random variables on its end vertices, in which we explicitly obtain the threshold for the exact recovery of the MLE. | ['Sichen Yang', 'Zhongyang Li'] | 2022-09-23 | null | null | null | null | ['community-detection'] | ['graphs'] | [ 1.64320413e-02 1.22254796e-01 1.37085348e-01 1.46083653e-01
-3.64626169e-01 -6.68516457e-01 6.99878782e-02 9.82172266e-02
-1.61910027e-01 5.17769158e-01 -4.60049897e-01 -3.66095871e-01
-2.54812628e-01 -8.72421563e-01 -6.51487947e-01 -1.03307450e+00
-5.29027939e-01 8.12577009e-01 1.38371676e-01 2.05626979e-01
8.42312425e-02 5.73382020e-01 -1.08955204e+00 -4.42248434e-01
9.67337668e-01 5.11297762e-01 1.99906509e-02 1.22421968e+00
-1.07584946e-01 4.69391495e-01 -5.54212868e-01 -2.14071393e-01
3.44259351e-01 -4.33991909e-01 -4.63136852e-01 7.12678015e-01
7.27421641e-02 3.36832792e-01 -4.19017255e-01 1.78417444e+00
2.24953160e-01 -9.55857262e-02 1.27488506e+00 -1.72254205e+00
-6.28738463e-01 7.84274876e-01 -1.39133787e+00 1.74933612e-01
2.91758537e-01 -1.20892920e-01 6.34981096e-01 -5.38606524e-01
5.38215578e-01 1.11489546e+00 5.27114272e-01 2.42539957e-01
-1.56416023e+00 -9.81368899e-01 9.31532681e-02 -3.18092816e-02
-2.23749852e+00 -3.57596427e-01 6.17012143e-01 -6.59142554e-01
5.55857122e-01 2.31838405e-01 2.72426307e-01 5.63875079e-01
9.94833186e-02 4.40873802e-01 8.41059446e-01 -4.47196484e-01
2.05914766e-01 4.07618582e-01 8.51189122e-02 6.73657537e-01
8.54283869e-01 -1.05627887e-01 5.34643568e-02 -6.67208791e-01
7.98013806e-01 -1.07797369e-01 -3.27004820e-01 -4.87518102e-01
-8.92431557e-01 8.92594099e-01 -1.37462363e-01 3.72898817e-01
-3.42161000e-01 1.36548698e-01 -1.64482370e-01 3.70012552e-01
1.96496516e-01 -5.05118594e-02 -8.66810009e-02 3.69333863e-01
-7.71072567e-01 -1.16162412e-01 8.68616045e-01 1.27807331e+00
7.94694185e-01 -1.38152093e-01 3.53929907e-01 2.99302131e-01
6.44539058e-01 1.08664858e+00 -2.98530817e-01 -9.19096231e-01
5.70465267e-01 4.68049765e-01 4.47716743e-01 -1.16033506e+00
-1.10867612e-01 -5.56206346e-01 -1.32549608e+00 8.24315175e-02
4.05740827e-01 -5.05228281e-01 -5.94711602e-01 2.04751039e+00
2.43839264e-01 6.07412517e-01 -1.38117954e-01 3.70613724e-01
2.79456943e-01 5.08567214e-01 -1.99528232e-01 -6.81348324e-01
8.43381941e-01 -1.38490289e-01 -7.60813832e-01 -4.82154191e-02
3.46559644e-01 -7.25290120e-01 1.73234105e-01 2.68470049e-01
-1.05056107e+00 -6.04631007e-02 -9.55276191e-01 8.63436520e-01
1.11102849e-01 3.02907953e-04 2.37949237e-01 9.79525387e-01
-1.27234983e+00 2.06045598e-01 -7.36874521e-01 -1.67353943e-01
-1.94800213e-01 5.15075386e-01 -6.44445777e-01 -2.53554583e-01
-6.75944805e-01 3.09296370e-01 2.22678185e-01 2.29546890e-01
-7.12228596e-01 1.04771830e-01 -7.28548765e-01 5.36965281e-02
2.93502122e-01 -4.08500671e-01 6.33264840e-01 -7.11023331e-01
-7.24163532e-01 8.70800614e-01 -4.06535476e-01 -4.13955972e-02
3.41253579e-01 6.23363554e-01 -5.36987841e-01 2.82000601e-01
3.62924129e-01 -1.71361983e-01 7.86668897e-01 -1.46390128e+00
-4.65703160e-01 -4.13279682e-01 -4.45020050e-01 -5.16303629e-02
6.45437092e-02 1.99097663e-01 -4.87947881e-01 -2.37233043e-01
5.54235935e-01 -1.03845739e+00 -7.09093392e-01 -4.18574572e-01
-8.24607491e-01 3.12487613e-02 4.68393028e-01 -4.27599907e-01
1.39036131e+00 -2.38445377e+00 2.64542311e-01 9.09392715e-01
6.67329133e-01 -1.80705771e-01 -2.13678014e-02 5.97871721e-01
-2.65413254e-01 4.85419035e-01 -2.11332694e-01 -3.88236523e-01
-7.07894266e-02 -1.34334698e-01 8.33576247e-02 1.31597078e+00
-1.17114916e-01 1.24382943e-01 -7.45311737e-01 -3.72535139e-01
-8.95019844e-02 6.82460517e-02 -3.48923922e-01 -1.23577025e-02
2.86476851e-01 2.05162853e-01 -4.75255221e-01 2.61340171e-01
1.08957493e+00 -5.77437580e-01 2.87734270e-01 4.59421545e-01
1.56910434e-01 -4.08301830e-01 -2.06967926e+00 5.55921316e-01
3.78617227e-01 6.05724037e-01 7.63265312e-01 -9.82283413e-01
5.37509918e-01 7.75731504e-01 4.88293409e-01 4.32061225e-01
5.14628828e-01 1.54591873e-01 2.78157920e-01 -1.55742317e-01
8.52853060e-02 -2.46680856e-01 -1.69383660e-01 6.41665280e-01
-1.14953294e-01 1.78956985e-01 2.51962632e-01 7.14297652e-01
1.18453360e+00 -1.06464946e+00 3.32552433e-01 -3.28984618e-01
4.21715856e-01 -4.30703610e-01 7.04662383e-01 1.04961097e+00
-4.22481567e-01 4.19390738e-01 7.41984606e-01 7.03193307e-01
-9.27744567e-01 -1.23904860e+00 -4.14956287e-02 4.03206855e-01
4.28280920e-01 -2.40328088e-01 -8.33673477e-01 -3.01514179e-01
9.63140577e-02 3.94907951e-01 -7.09539294e-01 2.91376375e-02
7.02690855e-02 -1.05162430e+00 2.87016273e-01 1.61781654e-01
2.09014639e-01 -4.85320002e-01 3.26040506e-01 1.04548715e-01
-3.61573875e-01 -9.33025956e-01 -6.84430361e-01 4.59009111e-02
-7.00379670e-01 -1.53889823e+00 -4.26521868e-01 -8.59854937e-01
1.17472088e+00 6.37214124e-01 7.44687319e-01 8.48268718e-02
-4.08546999e-02 5.37839293e-01 -4.11853157e-02 -1.29554287e-01
-5.73116481e-01 -3.60784441e-01 4.24121678e-01 1.13725491e-01
3.19773048e-01 -6.83849573e-01 -2.06389084e-01 4.35294002e-01
-6.13380373e-01 -4.91174102e-01 2.56936222e-01 3.36166143e-01
5.94261169e-01 8.51016998e-01 1.75597861e-01 -5.16235113e-01
7.38600075e-01 -8.43320787e-01 -6.69263542e-01 2.27040559e-01
-3.64495426e-01 -1.20301567e-01 1.97084948e-01 -4.71529067e-01
-6.47992074e-01 -1.70931607e-01 3.37085634e-01 -4.73567516e-01
-3.40028793e-01 5.23385108e-01 -4.47705835e-01 -1.59101292e-01
6.44139171e-01 1.30336955e-01 -1.67879120e-01 -3.76896501e-01
3.97986531e-01 7.05753326e-01 3.72936338e-01 -4.28131878e-01
1.06010592e+00 5.59308708e-01 2.65605867e-01 -1.11408210e+00
1.43369630e-01 -9.34581101e-01 -5.71063519e-01 -1.27825364e-01
5.92624128e-01 -8.61049891e-01 -8.39764595e-01 6.61313117e-01
-9.12109375e-01 -1.24623641e-01 1.71504598e-02 6.16865218e-01
-4.60125268e-01 8.98300350e-01 -9.16186690e-01 -1.40813315e+00
1.41934663e-01 -1.10305023e+00 4.17862922e-01 2.72205442e-01
9.41259190e-02 -8.95515919e-01 2.00444624e-01 -2.33700261e-01
-1.35156721e-01 2.12121904e-01 8.48615348e-01 -7.46571243e-01
-5.26378214e-01 -6.87069416e-01 -2.81977743e-01 1.63495719e-01
-3.99237461e-02 5.33815384e-01 -2.86403954e-01 -5.59681356e-01
8.70094150e-02 4.89298195e-01 3.51625949e-01 7.12950647e-01
4.86327291e-01 -3.31560969e-01 -9.45433319e-01 3.21866572e-01
1.46670163e+00 2.55139261e-01 4.27170157e-01 -3.15968573e-01
4.76740003e-01 3.37395012e-01 1.83985442e-01 5.16541421e-01
6.18297867e-02 2.46327713e-01 3.26311737e-01 1.46176070e-01
4.72955197e-01 2.11016372e-01 3.46479207e-01 9.64784384e-01
-2.34734616e-03 -7.13411212e-01 -9.15641189e-01 8.10685456e-01
-1.78602159e+00 -1.23792565e+00 -9.12562788e-01 2.57206893e+00
6.26636446e-01 -1.88861340e-02 1.82702854e-01 4.68922630e-02
1.39988685e+00 -2.36549124e-01 -2.31328487e-01 4.44482043e-02
-2.84879714e-01 -6.81608245e-02 8.20657730e-01 8.48249853e-01
-9.91034031e-01 4.47573602e-01 7.37000513e+00 7.78344274e-01
-4.87716526e-01 2.40820646e-01 2.79617280e-01 2.80459762e-01
-3.53331745e-01 1.04994951e-02 -7.78030694e-01 3.34245682e-01
8.18705678e-01 -4.68579620e-01 2.39603058e-01 6.48097157e-01
7.80195892e-02 -2.86602050e-01 -7.12729812e-01 8.06852043e-01
-2.21151501e-01 -7.54209995e-01 -4.99933869e-01 7.25322902e-01
1.03677583e+00 8.04165155e-02 1.98925231e-02 2.04511695e-02
9.21774209e-01 -8.38411152e-01 3.08467507e-01 3.17891121e-01
8.47530544e-01 -9.03144658e-01 5.78413010e-01 8.33275855e-01
-1.40012026e+00 1.18606828e-01 -4.71471399e-01 4.00221050e-02
1.84516385e-01 8.11370432e-01 -7.51185596e-01 3.87797594e-01
2.60328948e-01 1.82162702e-01 -1.47731349e-01 1.35280144e+00
-1.32862762e-01 6.57149017e-01 -8.62168193e-01 1.79499879e-01
-1.61669984e-01 -8.82466733e-01 1.01503229e+00 1.06905735e+00
5.06750643e-01 3.60289603e-01 2.68934131e-01 8.00167561e-01
-2.90464666e-02 1.60576522e-01 -5.51335216e-01 -2.02559307e-01
7.04290092e-01 8.51586521e-01 -1.13107622e+00 -3.73383582e-01
-3.80592823e-01 9.70974445e-01 2.44392142e-01 7.17769265e-01
-6.07511461e-01 -4.30390626e-01 1.00208020e+00 1.63215667e-01
5.71549237e-01 -2.17094705e-01 -2.29514346e-01 -1.22690618e+00
-2.48147354e-01 -5.85326135e-01 2.66629159e-01 -2.63688862e-01
-1.55443871e+00 4.24021572e-01 3.41530517e-02 -9.58790779e-01
-4.34357464e-01 -3.63825232e-01 -8.05469334e-01 1.24379230e+00
-4.93314892e-01 -5.11366546e-01 1.84604481e-01 7.75315344e-01
-4.81417716e-01 -4.52872440e-02 7.24865854e-01 2.23322868e-01
-1.01336694e+00 4.25020814e-01 6.18168652e-01 4.11263019e-01
2.65361309e-01 -1.29086161e+00 3.22879761e-01 1.40856349e+00
8.66805017e-02 9.31017637e-01 1.16734195e+00 -1.01448917e+00
-9.67580497e-01 -7.50280142e-01 9.29690123e-01 -2.13220730e-01
7.34772563e-01 -3.77430141e-01 -7.08715200e-01 9.46281075e-01
2.84009669e-02 -1.61839247e-01 9.68357861e-01 1.63207069e-01
-2.81040907e-01 5.39279342e-01 -1.41738045e+00 5.80879807e-01
7.22043276e-01 -5.78055143e-01 -1.36386290e-01 4.38665211e-01
2.93681175e-01 1.66523188e-01 -7.24071324e-01 3.34978580e-01
1.28552720e-01 -8.85198712e-01 9.07735169e-01 -6.67636812e-01
-2.70481914e-01 -7.34480381e-01 -4.10427898e-01 -1.05409849e+00
-7.07632363e-01 -6.19311094e-01 6.58423547e-03 9.70733225e-01
4.18020993e-01 -7.37496853e-01 7.50483513e-01 4.39235985e-01
5.25885224e-01 -3.25836509e-01 -1.06522143e+00 -7.53917575e-01
-1.01287477e-01 -4.53401327e-01 2.30526820e-01 1.06331587e+00
2.76056141e-01 2.03533828e-01 -3.53077680e-01 8.72787297e-01
1.10444677e+00 -1.08766839e-01 5.06477535e-01 -1.42344451e+00
-6.41714334e-01 -5.49444616e-01 -4.68999445e-01 -1.02909601e+00
8.43385980e-02 -4.72655118e-01 3.49653900e-01 -1.36024749e+00
6.46786094e-01 -5.73507190e-01 -8.01851675e-02 -2.37394065e-01
-4.15325046e-01 6.36056140e-02 -8.63644667e-03 1.86114699e-01
-5.63377559e-01 8.05330724e-02 6.38083220e-01 8.72169137e-02
-1.66040793e-01 5.48993051e-01 -6.92373276e-01 8.01160455e-01
5.72206378e-01 -6.52359545e-01 -3.07918489e-01 2.03748643e-01
2.54737169e-01 5.04141271e-01 -1.08355671e-01 -8.82548749e-01
2.18778759e-01 -1.61989719e-01 -2.97966842e-02 -6.35722697e-01
7.79685900e-02 -9.19660926e-01 7.76998818e-01 3.69696140e-01
9.39834341e-02 -1.23164713e-01 -1.70137212e-01 1.17303586e+00
1.71872005e-01 -5.93182921e-01 8.20122004e-01 3.59954014e-02
-1.26422159e-02 5.74728310e-01 -9.57032084e-01 -4.76786569e-02
1.17132473e+00 -2.19400469e-02 -3.82349789e-02 -9.00406599e-01
-9.01924372e-01 1.33086503e-01 6.65998042e-01 -3.36821169e-01
3.01345259e-01 -1.15201724e+00 -8.85329247e-01 7.69030005e-02
-5.32504730e-02 -1.12230122e-01 2.80463010e-01 1.02676880e+00
-2.65546173e-01 4.85673407e-03 3.65370393e-01 -5.83180308e-01
-1.49567688e+00 1.03749406e+00 4.00556654e-01 -1.23925373e-01
6.24769814e-02 1.11757064e+00 3.18559915e-01 -1.67763591e-01
8.09368640e-02 1.44683616e-02 4.57529686e-02 -1.37080103e-01
5.54725289e-01 4.61038977e-01 -4.30892795e-01 -9.53880250e-01
-3.02278757e-01 3.85705441e-01 4.56840359e-02 -4.99591172e-01
7.91071117e-01 -5.39066255e-01 -4.26192939e-01 4.62890059e-01
1.12239075e+00 6.96975529e-01 -9.07817900e-01 -3.67309928e-01
-2.82563150e-01 -4.52867270e-01 -3.33982766e-01 -4.54467908e-02
-1.02358103e+00 5.64849019e-01 3.18057507e-01 6.28716826e-01
8.58728290e-01 4.13484484e-01 1.15275711e-01 -1.07700877e-01
6.35642529e-01 -5.77101409e-01 -4.77447420e-01 2.76840210e-01
3.77484053e-01 -7.20457196e-01 -8.14234391e-02 -9.91721213e-01
-1.99472100e-01 7.24833727e-01 1.44171908e-01 -4.89838272e-01
1.09518290e+00 3.49916071e-01 -2.88041681e-01 -3.05277258e-01
-4.94708121e-01 -3.13641816e-01 5.79202548e-02 8.15573037e-01
2.46383563e-01 4.63546753e-01 -2.39500299e-01 7.23719299e-01
9.13927108e-02 -4.89844024e-01 7.28756487e-01 6.01206958e-01
-7.08560526e-01 -8.85620534e-01 -8.01902413e-01 3.76438677e-01
-2.97938555e-01 -1.32797509e-02 -3.35002005e-01 5.10263622e-01
2.36810312e-01 1.40260339e+00 -4.45325859e-02 -4.88451004e-01
1.71930760e-01 -2.51990799e-02 4.55193192e-01 -6.31677032e-01
7.68156946e-02 4.37944204e-01 -2.06880987e-01 -5.10758571e-02
-1.00953951e-01 -1.00908816e+00 -9.26058412e-01 -8.69126797e-01
-8.47043872e-01 7.02185988e-01 2.39150956e-01 9.16201651e-01
1.74076423e-01 1.48294851e-01 8.08849216e-01 -3.59925717e-01
-5.27766764e-01 -1.14170361e+00 -1.39254427e+00 1.80607781e-01
2.13164955e-01 -4.88760591e-01 -1.14248133e+00 -1.35157295e-02] | [6.946370601654053, 5.1103434562683105] |
a82e25b8-b9c3-4716-80a3-b1cd66d7d665 | pointattn-you-only-need-attention-for-point | 2203.08485 | null | https://arxiv.org/abs/2203.08485v1 | https://arxiv.org/pdf/2203.08485v1.pdf | PointAttN: You Only Need Attention for Point Cloud Completion | Point cloud completion referring to completing 3D shapes from partial 3D point clouds is a fundamental problem for 3D point cloud analysis tasks. Benefiting from the development of deep neural networks, researches on point cloud completion have made great progress in recent years. However, the explicit local region partition like kNNs involved in existing methods makes them sensitive to the density distribution of point clouds. Moreover, it serves limited receptive fields that prevent capturing features from long-range context information. To solve the problems, we leverage the cross-attention and self-attention mechanisms to design novel neural network for processing point cloud in a per-point manner to eliminate kNNs. Two essential blocks Geometric Details Perception (GDP) and Self-Feature Augment (SFA) are proposed to establish the short-range and long-range structural relationships directly among points in a simple yet effective way via attention mechanism. Then based on GDP and SFA, we construct a new framework with popular encoder-decoder architecture for point cloud completion. The proposed framework, namely PointAttN, is simple, neat and effective, which can precisely capture the structural information of 3D shapes and predict complete point clouds with highly detailed geometries. Experimental results demonstrate that our PointAttN outperforms state-of-the-art methods by a large margin on popular benchmarks like Completion3D and PCN. Code is available at: https://github.com/ohhhyeahhh/PointAttN | ['Chunhua Shen', 'Qingshan Liu', 'Junxia Li', 'Dongyan Guo', 'Ying Cui', 'Jun Wang'] | 2022-03-16 | null | null | null | null | ['point-cloud-completion'] | ['computer-vision'] | [-3.94941598e-01 -1.94217995e-01 1.27328187e-01 -4.86487180e-01
-6.72297955e-01 -2.89851278e-01 4.55515176e-01 -8.67980048e-02
4.71824035e-02 1.59544155e-01 1.70002177e-01 -2.60509700e-01
-1.92127481e-01 -8.81930947e-01 -1.25951135e+00 -5.06635129e-01
7.96601847e-02 5.38859904e-01 5.27347326e-02 -2.43714318e-01
2.60201842e-01 1.13054693e+00 -1.54779279e+00 1.02402948e-01
9.96304572e-01 1.01905859e+00 5.79777718e-01 2.66585737e-01
-5.86275697e-01 2.25092649e-01 -2.11075902e-01 -2.60063082e-01
4.59595025e-01 5.26688933e-01 -4.13123369e-01 1.17120193e-02
4.50772017e-01 -6.26946390e-01 -7.01644957e-01 9.66939569e-01
4.58045632e-01 1.19855002e-01 5.97485542e-01 -1.14543319e+00
-1.15556943e+00 1.59387141e-01 -8.32833648e-01 -1.02643937e-01
9.68207791e-02 3.43728960e-01 8.45884025e-01 -1.57377505e+00
2.25134239e-01 1.46224773e+00 8.87766778e-01 3.72375399e-01
-7.41124570e-01 -8.78993332e-01 3.67665380e-01 -1.31878778e-01
-1.64532363e+00 -1.54945508e-01 9.83697355e-01 -3.60522807e-01
1.20316625e+00 2.20579833e-01 8.44855309e-01 6.23326838e-01
3.69061418e-02 7.95935452e-01 5.52817941e-01 5.74103743e-02
-6.45523593e-02 -3.35733056e-01 -3.92538533e-02 5.11694193e-01
2.79534370e-01 1.35543749e-01 -4.45345864e-02 -1.45903632e-01
1.35693038e+00 7.77687073e-01 -2.44602874e-01 -4.30515081e-01
-1.06704843e+00 6.77186847e-01 1.11618590e+00 1.51053905e-01
-5.22918761e-01 3.61861438e-01 4.19390649e-02 -6.30085394e-02
6.25552535e-01 6.43327087e-02 -4.40098792e-01 2.21567467e-01
-5.53878546e-01 4.37169731e-01 2.82412261e-01 1.57504845e+00
1.01124668e+00 8.84092450e-02 4.34558019e-02 7.17342913e-01
5.69672048e-01 9.05333221e-01 -3.39068063e-02 -8.10484827e-01
7.41844118e-01 9.21393514e-01 5.05397730e-02 -1.05435634e+00
-2.69254982e-01 -5.59900582e-01 -1.13960052e+00 3.06708544e-01
-1.75666422e-01 2.84561843e-01 -1.16436660e+00 1.35085130e+00
4.53414977e-01 4.30161655e-01 -2.82549351e-01 1.15975070e+00
1.18268514e+00 1.00102282e+00 -4.74832505e-02 2.93884605e-01
1.14623070e+00 -6.55680478e-01 -2.92606384e-01 -9.24899876e-02
2.15490103e-01 -7.27696121e-01 1.10851109e+00 -2.04747114e-02
-1.21084642e+00 -7.54285574e-01 -8.80818427e-01 -6.76024437e-01
-2.76362747e-01 -3.64413224e-02 8.71142924e-01 -3.82927991e-02
-8.98054302e-01 5.82088232e-01 -9.26304340e-01 3.29694450e-02
1.08039248e+00 6.41770601e-01 -3.61035615e-01 -3.24852288e-01
-7.42753148e-01 4.13195938e-01 -1.81710795e-02 3.71005327e-01
-6.93132102e-01 -1.01469243e+00 -8.43741775e-01 3.63730371e-01
1.45395696e-01 -1.04307508e+00 1.06541610e+00 -4.51436132e-01
-1.08444095e+00 8.04418802e-01 -2.85894156e-01 -7.17878118e-02
2.71618485e-01 -4.09175247e-01 -4.96059246e-02 -1.19964913e-01
2.02874899e-01 9.20621097e-01 7.74289787e-01 -1.52834928e+00
-4.76484239e-01 -6.81640446e-01 -2.06610877e-02 4.92103249e-01
1.44690707e-01 -3.41219544e-01 -8.31876814e-01 -5.46161592e-01
4.84914631e-01 -7.42347896e-01 -3.11113030e-01 2.07162067e-01
-4.40039366e-01 -5.48565567e-01 7.70458519e-01 -3.95422757e-01
6.36418223e-01 -2.28003049e+00 -4.92427982e-02 1.77812755e-01
5.31787097e-01 2.76074499e-01 -2.44199917e-01 3.64649087e-01
-2.45065689e-01 2.09992945e-01 -2.53278136e-01 -6.72180057e-01
1.36133417e-01 2.38592774e-01 -7.30258048e-01 3.40379328e-01
4.99500155e-01 1.24127781e+00 -6.88187897e-01 -2.60884017e-01
6.56027853e-01 8.05550516e-01 -6.74175680e-01 2.72889137e-01
-2.73169875e-01 3.40659499e-01 -8.52143764e-01 9.94866371e-01
1.44584394e+00 -3.46730113e-01 -7.77898908e-01 -2.25517094e-01
-3.20837766e-01 1.65007710e-01 -8.32330287e-01 2.15854549e+00
-4.33498383e-01 2.55846679e-01 1.83576662e-02 -5.96601784e-01
1.19363499e+00 -8.28096271e-03 5.45489073e-01 -3.74277562e-01
1.37515664e-01 2.57327825e-01 -3.57913852e-01 -2.91135460e-01
4.80120987e-01 -9.88776460e-02 1.22176163e-01 -5.07412925e-02
-6.36316240e-02 -4.92465764e-01 -4.88692462e-01 1.50113747e-01
8.59770238e-01 2.24626034e-01 -7.77963996e-02 -2.82888580e-02
3.83778512e-01 -1.84039050e-03 5.55146098e-01 4.16733414e-01
3.98583226e-02 1.02515125e+00 1.01442181e-01 -7.13579476e-01
-1.25796878e+00 -1.22233081e+00 -1.91167012e-01 4.23338085e-01
4.56512928e-01 -2.65007764e-01 -3.55353981e-01 -4.02858913e-01
4.62043852e-01 4.57263112e-01 -4.18626815e-01 -5.27145453e-02
-5.90731859e-01 -2.88145959e-01 1.27717286e-01 6.07464135e-01
5.25422513e-01 -1.10641634e+00 -7.46289864e-02 7.89857358e-02
1.37722537e-01 -1.06490886e+00 -4.76242959e-01 -4.76902500e-02
-1.13438249e+00 -8.02903235e-01 -8.40931654e-01 -7.78160214e-01
8.25128853e-01 8.70178521e-01 1.26835382e+00 3.83709639e-01
-6.20519631e-02 1.52969003e-01 -2.53610909e-01 -7.23155260e-01
2.87402719e-01 1.47810698e-01 -2.49203276e-02 -2.20954239e-01
7.21960187e-01 -9.60244119e-01 -7.49434650e-01 1.39441356e-01
-8.97863805e-01 1.79079026e-01 9.89265084e-01 5.82880318e-01
1.02317047e+00 -1.38448909e-01 3.25347394e-01 -6.56064808e-01
2.94099659e-01 -6.42626047e-01 -6.46399617e-01 -1.37136132e-01
-9.60937887e-02 -1.56755894e-01 6.10026300e-01 5.31056598e-02
-7.75240242e-01 2.07585081e-01 -6.42290473e-01 -1.40023768e+00
-3.38031441e-01 2.43124098e-01 -4.76886630e-01 -2.31647119e-01
1.43665090e-01 3.81921232e-01 -1.22703709e-01 -7.45876670e-01
3.41510504e-01 5.30278981e-01 4.13700968e-01 -5.38674712e-01
1.11812901e+00 7.55051255e-01 5.65730920e-03 -7.59041548e-01
-6.07557595e-01 -4.70247746e-01 -7.15728939e-01 -1.45134032e-02
7.57810950e-01 -1.28832781e+00 -9.43090856e-01 3.76881570e-01
-1.58857656e+00 5.75280897e-02 -3.25377941e-01 3.98796350e-01
-4.47428942e-01 3.69286865e-01 -4.47857857e-01 -6.45010591e-01
-6.14182711e-01 -1.14535427e+00 1.56419289e+00 2.43246138e-01
4.53594744e-01 -6.20970070e-01 -1.41998753e-01 1.43685922e-01
1.95605516e-01 2.22091526e-01 8.81978512e-01 -1.56401753e-01
-1.46610653e+00 -1.91331074e-01 -6.26079798e-01 1.78251162e-01
-3.28858709e-03 -6.73652440e-02 -9.77821767e-01 -1.94778785e-01
8.41842443e-02 -2.65425034e-02 9.04913604e-01 5.67570508e-01
1.61520422e+00 -5.22271097e-02 -3.86056274e-01 1.17779744e+00
1.57895887e+00 -8.33501220e-02 7.38313556e-01 -1.00769803e-01
1.24448526e+00 3.04114133e-01 4.53306645e-01 5.10115802e-01
6.54505134e-01 3.75724107e-01 9.76994336e-01 -2.49884695e-01
-6.87034354e-02 -7.33038247e-01 -2.23350763e-01 1.03658605e+00
-3.50916475e-01 -1.47149442e-02 -9.90500987e-01 4.89536494e-01
-1.79586720e+00 -6.98471904e-01 -3.12839001e-01 1.94754601e+00
1.86415166e-01 4.99819815e-02 -3.90703350e-01 -1.90567821e-01
7.14846611e-01 2.08978534e-01 -8.32004488e-01 -3.50698382e-02
-6.84592724e-02 1.33017465e-01 4.78291839e-01 3.54553312e-01
-9.50924695e-01 1.02107477e+00 4.90152884e+00 1.04770029e+00
-9.67721701e-01 -7.89153948e-02 5.60577989e-01 -3.04044690e-02
-7.10146427e-01 -6.63600639e-02 -7.75322795e-01 5.01776040e-01
1.64515704e-01 2.10286319e-01 3.94069254e-01 9.70899463e-01
1.33441746e-01 5.30840337e-01 -9.56824183e-01 1.53455603e+00
-1.41977981e-01 -1.54775214e+00 4.44630802e-01 1.55378431e-01
6.00492537e-01 6.35896206e-01 1.62172526e-01 3.73724133e-01
9.45895389e-02 -1.08896160e+00 6.78709984e-01 6.60982847e-01
9.42971528e-01 -9.08957541e-01 6.07715487e-01 4.51000124e-01
-1.45029569e+00 5.24581075e-02 -1.03596258e+00 -1.07427046e-01
1.16607890e-01 8.57356668e-01 -4.30488229e-01 8.30819070e-01
8.68526101e-01 1.12303996e+00 -2.48034999e-01 1.21871388e+00
-7.07722828e-02 1.70456201e-01 -5.69625497e-01 5.87639287e-02
4.64302301e-01 -4.13971722e-01 7.74076104e-01 7.51456499e-01
6.44300878e-01 4.69090909e-01 7.92279467e-02 1.37830591e+00
-2.35021651e-01 1.71343740e-02 -8.40646386e-01 2.57772923e-01
6.74747586e-01 1.18146229e+00 -4.97953773e-01 -9.99957621e-02
-6.02541566e-01 5.92505217e-01 6.09014392e-01 3.23558122e-01
-7.06868529e-01 -2.98057854e-01 9.13185596e-01 3.11058998e-01
4.88553733e-01 -5.41287184e-01 -6.47988439e-01 -1.15656602e+00
3.38458538e-01 -3.35808247e-01 -1.56572416e-01 -1.24076581e+00
-1.56417084e+00 6.72609866e-01 -1.65170476e-01 -1.57599604e+00
5.16638696e-01 -5.93000770e-01 -7.55612373e-01 1.17504871e+00
-1.75607252e+00 -1.45511365e+00 -7.13167131e-01 6.66712344e-01
6.24721050e-01 -3.02406587e-02 4.89300936e-01 4.02666360e-01
-3.18321615e-01 1.29895449e-01 -3.48292217e-02 1.17592156e-01
2.12547347e-01 -1.00088573e+00 1.02573848e+00 4.94664580e-01
5.46555184e-02 7.41402388e-01 6.44344687e-02 -7.95659900e-01
-1.66225827e+00 -1.32547891e+00 5.93804657e-01 -5.31722903e-01
1.54237807e-01 -5.70881903e-01 -1.15043139e+00 5.99231064e-01
-1.41266823e-01 2.02144578e-01 2.61084199e-01 5.95829152e-02
-2.31288195e-01 -8.40881169e-02 -8.56853843e-01 4.32291299e-01
1.47820032e+00 -4.58637446e-01 -6.72005832e-01 4.14916068e-01
1.19834518e+00 -7.07398713e-01 -7.72779942e-01 6.33414328e-01
1.07147157e-01 -8.64519656e-01 1.29110861e+00 -3.38493258e-01
6.54841900e-01 -5.27920723e-01 -2.39703178e-01 -1.08246279e+00
-7.23308802e-01 -3.81411433e-01 -4.44318913e-02 1.00596046e+00
1.41239122e-01 -5.20655811e-01 1.04212582e+00 6.44999325e-01
-8.94039035e-01 -1.10120583e+00 -1.00195038e+00 -4.65604663e-01
2.88938075e-01 -5.73242366e-01 1.38863981e+00 8.44832420e-01
-6.79677546e-01 2.98182040e-01 -1.83836773e-01 5.47515154e-01
5.89590669e-01 4.99827772e-01 9.48845863e-01 -1.31517220e+00
2.56516457e-01 -4.69481736e-01 -5.16711056e-01 -1.76913500e+00
1.12362027e-01 -9.92416024e-01 -1.61492094e-01 -1.76362312e+00
-2.14495137e-03 -1.00448930e+00 -2.11619083e-02 3.82480949e-01
-1.52187124e-01 7.24590942e-02 3.15290362e-01 5.39522648e-01
-3.98729861e-01 1.12509143e+00 1.54444838e+00 -1.36999324e-01
-1.77175149e-01 1.06798425e-01 -6.42882884e-01 7.54340053e-01
6.43340766e-01 -3.55134815e-01 -3.08977276e-01 -1.17523527e+00
2.30551288e-01 -7.42741898e-02 6.10219479e-01 -1.05689180e+00
3.12236309e-01 -7.57324815e-05 7.12270379e-01 -1.37134600e+00
7.27331817e-01 -1.14639878e+00 2.13314090e-02 1.27588004e-01
2.64971793e-01 2.63348639e-01 2.54587442e-01 7.51771331e-01
-2.15941384e-01 1.15368225e-01 4.36384261e-01 -3.14557016e-01
-6.88203156e-01 1.17600513e+00 4.33914840e-01 -2.68894196e-01
8.16406190e-01 -2.62335598e-01 -9.00228843e-02 -2.18788490e-01
-4.56565619e-01 4.94670868e-01 6.15371406e-01 5.73560059e-01
1.18072259e+00 -1.58075905e+00 -7.04573095e-01 5.63174129e-01
1.27805188e-01 1.15274310e+00 7.81582475e-01 4.37793404e-01
-8.23257625e-01 4.74766761e-01 -1.70209065e-01 -1.00591755e+00
-6.91995263e-01 6.65451884e-01 1.71031445e-01 1.72252491e-01
-1.02374268e+00 1.01144505e+00 6.74455166e-01 -8.15079868e-01
4.57213372e-02 -7.73489892e-01 6.71233460e-02 -6.08462453e-01
2.91119576e-01 -5.57055362e-02 1.47585735e-01 -6.90499127e-01
-2.78761983e-01 1.06025255e+00 -8.89821500e-02 5.28559387e-01
1.56819987e+00 -1.86087936e-02 -2.20846552e-02 1.26737759e-01
1.14064848e+00 -1.11093543e-01 -1.50408661e+00 -3.02922845e-01
-6.49348974e-01 -9.22114968e-01 1.72585860e-01 -3.04839045e-01
-1.17037094e+00 1.13506162e+00 3.14339966e-01 1.92893781e-02
9.39364195e-01 9.48021635e-02 9.90000844e-01 2.66419649e-01
4.50737327e-01 -4.07643795e-01 -3.01808834e-01 6.38876379e-01
1.24599195e+00 -1.39912891e+00 2.18073428e-02 -7.60459960e-01
-3.84251684e-01 9.26019132e-01 7.56493211e-01 -6.66617930e-01
9.27549303e-01 -1.33953923e-02 -2.94949949e-01 -4.88690674e-01
-6.08335316e-01 -2.01135918e-01 3.20840359e-01 6.71474874e-01
3.05036530e-02 1.05059460e-01 4.01671529e-01 7.68103778e-01
-2.34930038e-01 -8.72106180e-02 3.20058204e-02 5.87222636e-01
-4.15677220e-01 -7.67557502e-01 -4.02598202e-01 6.08565629e-01
2.82170363e-02 -2.51772046e-01 -1.35718837e-01 8.47186327e-01
2.54075676e-01 3.88137579e-01 4.09556091e-01 -4.45503265e-01
5.83517492e-01 -4.45201725e-01 2.15985358e-01 -6.15933895e-01
-3.69039923e-01 2.37169713e-02 -5.24631858e-01 -5.31693935e-01
-1.82120383e-01 -5.06322026e-01 -1.37743938e+00 -5.97600996e-01
-2.99009353e-01 -4.12194692e-02 6.72445774e-01 6.41738057e-01
8.87541354e-01 4.65578139e-01 6.98230028e-01 -1.41840839e+00
-3.60881656e-01 -8.61363769e-01 -5.46899438e-01 4.49812233e-01
4.51167405e-01 -7.28956521e-01 -1.86563909e-01 -4.91625488e-01] | [8.17312240600586, -3.5738234519958496] |
0c4728c8-f353-44c5-8bde-c495434d8e3b | 3d-ldm-neural-implicit-3d-shape-generation | 2212.00842 | null | https://arxiv.org/abs/2212.00842v2 | https://arxiv.org/pdf/2212.00842v2.pdf | 3D-LDM: Neural Implicit 3D Shape Generation with Latent Diffusion Models | Diffusion models have shown great promise for image generation, beating GANs in terms of generation diversity, with comparable image quality. However, their application to 3D shapes has been limited to point or voxel representations that can in practice not accurately represent a 3D surface. We propose a diffusion model for neural implicit representations of 3D shapes that operates in the latent space of an auto-decoder. This allows us to generate diverse and high quality 3D surfaces. We additionally show that we can condition our model on images or text to enable image-to-3D generation and text-to-3D generation using CLIP embeddings. Furthermore, adding noise to the latent codes of existing shapes allows us to explore shape variations. | ['Paul Guerrero', 'Linqi Zhou', 'Alberto Tono', 'Andrew Rodriguez', 'Mariem Khlifi', 'Gimin Nam'] | 2022-12-01 | null | null | null | null | ['3d-shape-generation', 'image-to-3d', 'text-to-3d'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 4.32856828e-01 5.64956784e-01 1.31250873e-01 2.04818752e-02
-8.32498193e-01 -7.19916165e-01 8.70844603e-01 -3.40378165e-01
3.33898783e-01 6.10684156e-01 4.49707150e-01 -1.59101263e-01
3.27061981e-01 -1.20464182e+00 -9.67369735e-01 -6.02944136e-01
2.29918018e-01 5.40901124e-01 -1.79417834e-01 -8.49466547e-02
-2.71112043e-02 6.81257486e-01 -1.26780105e+00 2.77734607e-01
5.81948757e-01 7.98659563e-01 1.34326220e-01 8.47850084e-01
-2.49837577e-01 9.82841402e-02 -7.09926724e-01 -2.70485967e-01
5.49424112e-01 -6.03053033e-01 -3.16030771e-01 3.35004687e-01
6.78675115e-01 -3.66250515e-01 -4.40687239e-01 5.99836528e-01
5.81426382e-01 -1.38321131e-01 1.27351689e+00 -1.02753043e+00
-1.42092943e+00 2.69962270e-02 -5.05763292e-01 -5.53003013e-01
3.88065159e-01 3.20992082e-01 7.81562507e-01 -8.67220104e-01
1.05901754e+00 1.47269058e+00 4.97288465e-01 9.18155193e-01
-1.64832830e+00 -3.88574183e-01 -2.37732619e-01 -5.52486777e-01
-1.20313227e+00 -4.24976528e-01 9.58995104e-01 -4.77146864e-01
1.00347900e+00 5.10470495e-02 9.08161044e-01 1.42093301e+00
2.35204235e-01 8.32071781e-01 9.82840598e-01 -3.50838721e-01
2.71367759e-01 -5.42070717e-02 -7.60683715e-01 5.65071404e-01
7.28772655e-02 1.58186823e-01 -4.55444336e-01 -9.23194587e-02
1.55438423e+00 -2.19557896e-01 -2.01800734e-01 -4.29030061e-01
-1.18477023e+00 9.57617700e-01 4.67493623e-01 1.09689608e-01
-4.73829597e-01 7.06960440e-01 -9.58020166e-02 1.96244135e-01
6.67698264e-01 5.83592594e-01 -1.34558395e-01 -7.39393709e-03
-1.04326618e+00 5.93216836e-01 4.94716257e-01 1.18988013e+00
5.28897405e-01 6.58342957e-01 -5.13084888e-01 6.90439761e-01
2.66603231e-01 9.34569061e-01 2.10092738e-01 -1.29543579e+00
2.02673703e-01 3.79397303e-01 1.33878902e-01 -7.59735644e-01
1.00648880e-01 -2.47046426e-01 -9.41469073e-01 6.03174627e-01
8.29985887e-02 -1.14528999e-01 -1.42766750e+00 1.63632059e+00
9.04306769e-02 1.23713523e-01 5.35670035e-02 7.85703778e-01
7.03996956e-01 9.79375601e-01 -2.96729743e-01 2.30693743e-01
8.54670227e-01 -5.26612461e-01 -5.25307894e-01 7.05715362e-03
2.62067348e-01 -6.54237509e-01 1.02158022e+00 1.18511699e-01
-1.62395465e+00 -4.53070819e-01 -8.94806147e-01 -2.93369472e-01
-1.58191681e-01 -1.72857314e-01 3.96528393e-01 7.05457270e-01
-1.48351061e+00 3.88303161e-01 -7.12862611e-01 -1.62606165e-02
8.33676994e-01 1.86047211e-01 -2.46032551e-01 -7.13211000e-02
-8.48086715e-01 7.72807300e-01 -1.24418564e-01 -3.50272685e-01
-1.07067430e+00 -6.82121098e-01 -1.03970635e+00 -1.07840560e-01
-1.92352280e-01 -1.33647275e+00 1.00869977e+00 -6.38196886e-01
-1.77493596e+00 9.23839271e-01 -2.27042049e-01 -4.68809038e-01
4.87371355e-01 1.24201410e-01 1.13228209e-01 1.29684359e-01
-9.53020602e-02 1.43697000e+00 1.12138402e+00 -1.46242023e+00
8.04822296e-02 -1.84516519e-01 1.74488556e-02 2.28168309e-01
-4.91050594e-02 -7.34836578e-01 -3.74393761e-01 -1.07360506e+00
1.34431347e-01 -9.30870533e-01 -3.57048780e-01 5.08356571e-01
-3.58916700e-01 1.91088185e-01 9.54371929e-01 -5.32968700e-01
4.79030013e-01 -2.09246230e+00 4.43260759e-01 1.28389984e-01
1.79097623e-01 -5.80548793e-02 -5.04336953e-01 4.20500666e-01
1.90305695e-01 5.91948450e-01 -3.83824259e-01 -6.63384259e-01
1.56232893e-01 2.95989662e-01 -4.12027627e-01 2.36509815e-01
6.68735862e-01 1.43148494e+00 -6.71668828e-01 -2.26528272e-01
2.28982225e-01 9.72696424e-01 -9.53201413e-01 6.37689903e-02
-5.96496403e-01 7.00008094e-01 -3.83381695e-01 4.31999832e-01
5.70940137e-01 -1.38546705e-01 -2.81529129e-01 8.74100402e-02
2.80042648e-01 7.65142739e-02 -6.53539658e-01 1.98154950e+00
-7.90600300e-01 6.78383648e-01 -6.96577430e-02 -4.88471419e-01
1.11117435e+00 4.43902582e-01 4.46307331e-01 -5.79820335e-01
-1.04425080e-01 2.35783383e-01 -2.73302436e-01 1.94887202e-02
6.06902599e-01 -3.66983652e-01 -9.62353405e-03 4.16282237e-01
2.12149192e-02 -1.28400910e+00 -1.74291342e-01 1.93376124e-01
8.28722477e-01 3.84556949e-01 -3.67327988e-01 9.73066017e-02
-3.69730331e-02 -2.34181166e-01 4.43315655e-02 5.27984560e-01
5.86153626e-01 1.28159213e+00 4.65125024e-01 1.60580687e-02
-1.78436863e+00 -1.57141495e+00 -1.09072953e-01 -1.88738797e-02
-1.24898106e-01 -1.35386363e-01 -7.61084199e-01 -3.85899782e-01
9.65463594e-02 1.00199306e+00 -6.57021701e-01 -1.67694435e-01
-4.37213838e-01 -2.83057660e-01 5.81974328e-01 4.17934835e-01
9.98172387e-02 -8.88483286e-01 -4.41443354e-01 2.12435961e-01
1.77464455e-01 -1.02139330e+00 -6.36108696e-01 -1.17429957e-01
-1.13298106e+00 -6.43452764e-01 -1.51635253e+00 -7.00892687e-01
8.96421671e-01 -4.81423475e-02 1.13257837e+00 -3.02706480e-01
-2.88718045e-01 6.04429185e-01 -1.90530866e-01 -3.00951183e-01
-8.34907770e-01 -1.25997707e-01 -1.60364032e-01 -2.30447382e-01
-3.84414047e-01 -8.53389800e-01 -6.95223808e-01 9.76304039e-02
-1.20086896e+00 3.54440778e-01 4.70071197e-01 8.43977213e-01
7.44573772e-01 -9.75773931e-02 5.44173360e-01 -5.79886615e-01
6.87981248e-01 -3.06111604e-01 -3.53608817e-01 -1.34869143e-01
-3.18061262e-01 2.81624198e-01 3.23677182e-01 -4.88764107e-01
-9.04600799e-01 -9.59421396e-02 -3.56785893e-01 -8.50187540e-01
-1.63481116e-01 8.28633904e-02 -2.17929538e-02 2.16744021e-02
7.49582767e-01 4.32263792e-01 3.00564378e-01 -3.82683009e-01
9.52020466e-01 4.13861096e-01 3.48903775e-01 -5.66576004e-01
9.52949464e-01 5.26534081e-01 1.95563540e-01 -9.48009193e-01
-1.96449026e-01 3.98452044e-01 -4.83368129e-01 5.49958460e-02
1.09022117e+00 -9.34546530e-01 -2.19654590e-01 4.56485838e-01
-1.29517567e+00 -7.48458207e-01 -7.94279039e-01 -8.72742534e-02
-1.03238463e+00 1.82093948e-01 -7.66203225e-01 -7.95191050e-01
-3.38509202e-01 -1.10556817e+00 1.48041141e+00 1.78301090e-03
-3.33341390e-01 -9.44152296e-01 -1.48222670e-01 -8.52838252e-03
6.71227634e-01 6.18354559e-01 1.21498287e+00 3.36153597e-01
-9.14273083e-01 -1.60014704e-01 -5.99815883e-02 3.71747702e-01
2.44613305e-01 -3.47193033e-02 -8.51234496e-01 -1.44839734e-01
-2.93258250e-01 -2.57296145e-01 7.97479987e-01 7.60462344e-01
1.05896246e+00 -4.50610042e-01 -3.26423973e-01 7.22824037e-01
1.26885772e+00 4.66633551e-02 7.72119343e-01 -2.69326776e-01
7.26486921e-01 3.33449483e-01 -2.74464786e-01 2.49881610e-01
1.97981864e-01 7.72842646e-01 2.60446995e-01 -2.07117334e-01
-8.23961496e-01 -7.26877332e-01 2.69533038e-01 6.73038661e-01
1.05067894e-01 -6.13575935e-01 -6.21829152e-01 6.69429600e-01
-1.27723742e+00 -7.23222673e-01 -1.08350463e-01 1.98489738e+00
7.66998827e-01 1.57686740e-01 -1.17810294e-01 -7.96015412e-02
4.64960098e-01 1.53536856e-01 -6.11702204e-01 -5.21697283e-01
-4.37697887e-01 5.30952513e-01 4.49077040e-01 4.61115450e-01
-5.01384199e-01 9.45821643e-01 7.54358006e+00 7.88977265e-01
-1.01726079e+00 5.29654883e-02 8.63156319e-01 -2.65922725e-01
-1.22854447e+00 -2.09408134e-01 -4.63079214e-01 2.48040080e-01
5.77910900e-01 -7.93524384e-02 4.26870167e-01 4.13664669e-01
1.64592192e-01 1.94485232e-01 -1.05353010e+00 1.20127869e+00
7.01673552e-02 -1.67977214e+00 6.76747859e-01 3.14009994e-01
1.39722896e+00 -2.47165367e-01 6.05688751e-01 1.09982081e-02
4.24489558e-01 -1.43177128e+00 9.27708566e-01 5.99004567e-01
1.46009672e+00 -5.05764246e-01 6.64594248e-02 3.55633527e-01
-7.18228638e-01 3.32677692e-01 -2.97210485e-01 3.26796710e-01
5.48162580e-01 6.16254270e-01 -8.95877600e-01 3.19343776e-01
1.77503496e-01 5.68587244e-01 -2.17177197e-01 7.13072121e-01
-2.26177037e-01 2.83814907e-01 -3.79548937e-01 1.17330037e-01
3.18799973e-01 -2.59546846e-01 6.89552486e-01 8.30912530e-01
9.10041034e-01 -2.33325958e-02 -2.46329054e-01 1.63255274e+00
-2.25963771e-01 -2.60152370e-01 -1.23019314e+00 -3.39957237e-01
3.38284999e-01 5.62473357e-01 -5.25580943e-01 -3.31307471e-01
-1.02516592e-01 1.44501019e+00 4.10424266e-03 5.10537624e-01
-5.66824019e-01 -1.39110714e-01 6.84101462e-01 4.17329788e-01
4.77716118e-01 -7.79359460e-01 -6.29302979e-01 -9.40032721e-01
-1.38536617e-01 -6.61378920e-01 -3.75539690e-01 -1.23369348e+00
-1.32001066e+00 5.34439623e-01 -1.26972824e-01 -9.07401681e-01
-3.59338075e-01 -4.55997497e-01 -4.39108402e-01 1.18347991e+00
-1.09987938e+00 -1.34435272e+00 5.22741564e-02 3.25461805e-01
5.10999560e-01 -5.30992523e-02 9.57686841e-01 -6.05626404e-02
1.86839730e-01 4.86004859e-01 7.62453228e-02 -8.47656280e-02
3.39670420e-01 -1.19956386e+00 1.16029024e+00 4.16011155e-01
3.69572252e-01 2.52949893e-01 4.39434171e-01 -8.75609219e-01
-1.56266963e+00 -1.00313544e+00 3.80275846e-01 -6.84188128e-01
3.45726172e-03 -4.97587740e-01 -6.69611990e-01 5.13846040e-01
2.99792856e-01 -3.07365030e-01 2.78503001e-01 -3.28179061e-01
-2.01388970e-01 4.33340281e-01 -1.25089121e+00 9.41162348e-01
1.37277770e+00 -6.30070210e-01 -2.20999047e-01 1.77140683e-01
9.60340858e-01 -7.23269939e-01 -8.52173090e-01 2.36900121e-01
3.06501299e-01 -6.06500626e-01 1.13951874e+00 -2.50093490e-01
8.32036614e-01 -2.00042635e-01 -2.05290392e-01 -1.50773537e+00
-3.77812415e-01 -7.24309921e-01 -2.99150497e-01 9.94800210e-01
5.00713646e-01 -3.90054822e-01 1.12577355e+00 4.92200226e-01
-2.19433323e-01 -7.57928908e-01 -8.71480465e-01 -8.07246864e-01
5.10162175e-01 -4.41048712e-01 8.21761668e-01 5.92539787e-01
-5.97160578e-01 1.69768155e-01 -3.33115399e-01 -1.20534636e-01
5.49396396e-01 9.28559154e-02 7.52426803e-01 -7.68735349e-01
-3.10629785e-01 -8.37725639e-01 -4.11348164e-01 -1.62088895e+00
-1.18431039e-02 -1.17633784e+00 -1.23420082e-01 -1.93756735e+00
-2.04990461e-01 -6.84342980e-01 4.00093406e-01 1.43174753e-01
2.32710585e-01 6.70966268e-01 3.16948146e-01 6.68331012e-02
2.52222478e-01 1.07084620e+00 1.93649399e+00 -3.96993428e-01
-2.25471571e-01 -3.47332329e-01 -6.25993848e-01 2.58183986e-01
6.23389065e-01 -2.68005013e-01 -4.47851241e-01 -8.54596317e-01
2.94672996e-01 2.54582435e-01 4.88849729e-01 -9.30703759e-01
-2.50563353e-01 7.49198496e-02 6.79657519e-01 -5.31639099e-01
8.19829702e-01 -6.21987581e-01 5.14885128e-01 2.18193516e-01
-3.18545580e-01 -1.99128017e-01 2.24249005e-01 6.35325491e-01
-9.11044404e-02 -8.51407722e-02 6.95752919e-01 -3.08639199e-01
-2.50340253e-03 6.05311096e-01 -3.31803620e-01 7.96399638e-02
8.63338947e-01 -6.16099119e-01 1.92103371e-01 -7.91123629e-01
-7.10401118e-01 -1.20930836e-01 1.05216777e+00 5.20057619e-01
9.40088928e-01 -1.81004155e+00 -9.40702260e-01 3.54535818e-01
-2.11235613e-01 3.83418590e-01 2.34583706e-01 1.46209434e-01
-6.65449023e-01 3.46762717e-01 -2.05750540e-01 -7.25855768e-01
-6.94239438e-01 4.01687056e-01 3.45554262e-01 1.71292722e-01
-8.16047609e-01 8.28363657e-01 4.30633336e-01 -3.51752639e-01
-6.97970688e-02 -3.76512975e-01 2.85311103e-01 -2.79141426e-01
7.12990910e-02 -1.29074708e-01 -3.03261161e-01 -4.89931405e-01
8.28177780e-02 9.24599707e-01 4.39159811e-01 -5.90927005e-01
1.34746325e+00 7.61796609e-02 2.76842415e-01 3.39129984e-01
1.25632071e+00 1.12047002e-01 -1.63966572e+00 3.38798128e-02
-6.13059103e-01 -5.21290243e-01 4.06971686e-02 -5.99024892e-01
-1.14108801e+00 1.10460031e+00 3.89124036e-01 4.38194722e-02
8.55041564e-01 8.27330351e-02 1.14231420e+00 -4.87302020e-02
3.86289716e-01 -6.79075599e-01 3.77878010e-01 5.08802891e-01
1.20937049e+00 -7.64496505e-01 -2.31936336e-01 -2.96244353e-01
-6.13337457e-01 9.90184724e-01 1.45146176e-01 -3.43812466e-01
6.05167031e-01 3.55423659e-01 -1.97858483e-01 -1.45775869e-01
-7.99941301e-01 1.13737300e-01 4.92043912e-01 1.04108095e+00
4.20721382e-01 1.06960312e-01 1.07462861e-01 1.14793517e-02
-3.78269494e-01 -6.24127351e-02 5.61315775e-01 4.87381607e-01
9.31876304e-04 -1.27368641e+00 -3.79729897e-01 4.07060176e-01
-2.09044099e-01 -2.04408526e-01 -3.15253049e-01 3.95449072e-01
7.94897974e-02 5.32424629e-01 3.47690701e-01 -1.11017570e-01
1.73577413e-01 1.31597832e-01 1.02152169e+00 -7.35331237e-01
-1.16696678e-01 1.75600663e-01 -8.75963643e-02 -2.97763586e-01
-7.48146921e-02 -5.31232297e-01 -1.04938245e+00 -2.54492044e-01
-1.08957782e-01 -1.66612431e-01 7.90801883e-01 3.06589484e-01
7.34578013e-01 4.76499051e-01 5.33963144e-01 -1.18055749e+00
-1.68001547e-01 -6.92865312e-01 -5.81976414e-01 3.98462236e-01
3.10166866e-01 -4.72384185e-01 -1.31304979e-01 3.05029988e-01] | [9.082757949829102, -3.541922092437744] |
f0618763-6bae-4ddc-aba8-a72fcfa5233a | moving-average-options-machine-learning-and | 2108.11141 | null | https://arxiv.org/abs/2108.11141v1 | https://arxiv.org/pdf/2108.11141v1.pdf | Moving average options: Machine Learning and Gauss-Hermite quadrature for a double non-Markovian problem | Evaluating moving average options is a tough computational challenge for the energy and commodity market as the payoff of the option depends on the prices of a certain underlying observed on a moving window so, when a long window is considered, the pricing problem becomes high dimensional. We present an efficient method for pricing Bermudan style moving average options, based on Gaussian Process Regression and Gauss-Hermite quadrature, thus named GPR-GHQ. Specifically, the proposed algorithm proceeds backward in time and, at each time-step, the continuation value is computed only in a few points by using Gauss-Hermite quadrature, and then it is learned through Gaussian Process Regression. We test the proposed approach in the Black-Scholes model, where the GPR-GHQ method is made even more efficient by exploiting the positive homogeneity of the continuation value, which allows one to reduce the problem size. Positive homogeneity is also exploited to develop a binomial Markov chain, which is able to deal efficiently with medium-long windows. Secondly, we test GPR-GHQ in the Clewlow-Strickland model, the reference framework for modeling prices of energy commodities. Finally, we consider a challenging problem which involves double non-Markovian feature, that is the rough-Bergomi model. In this case, the pricing problem is even harder since the whole history of the volatility process impacts the future distribution of the process. The manuscript includes a numerical investigation, which displays that GPR-GHQ is very accurate and it is able to handle options with a very long window, thus overcoming the problem of high dimensionality. | ['Antonino Zanette', 'Andrea Molent', 'Ludovic Goudenège'] | 2021-08-25 | null | null | null | null | ['gpr', 'gpr'] | ['computer-vision', 'miscellaneous'] | [-5.38974464e-01 -2.91698277e-01 1.47823423e-01 3.12717676e-01
-5.05569875e-01 -4.61056888e-01 5.37327230e-01 1.18077442e-01
-5.15352190e-01 9.19393182e-01 -4.11909759e-01 -4.01466668e-01
-2.71662533e-01 -1.20287621e+00 -5.44430137e-01 -1.00593913e+00
-2.43258119e-01 6.31905079e-01 -2.61509046e-02 -7.06119835e-02
1.31876707e-01 5.67796409e-01 -1.16505730e+00 -6.61036134e-01
8.36845517e-01 1.25893891e+00 5.37930429e-02 5.23956180e-01
-3.64017606e-01 4.39110875e-01 -2.62244064e-02 -5.40621102e-01
4.18758631e-01 -1.44626006e-01 -2.25592583e-01 -3.33945155e-01
-6.39227808e-01 -1.07291408e-01 1.88145548e-01 1.22844350e+00
2.42384493e-01 4.04445797e-01 1.05577934e+00 -1.28141761e+00
-2.57958025e-01 4.71607298e-01 -6.08181775e-01 1.72608588e-02
-2.02197507e-01 3.83190848e-02 9.60066080e-01 -8.81378055e-01
1.29762188e-01 1.04201376e+00 6.92193925e-01 2.66733944e-01
-1.35986972e+00 -4.67978865e-01 1.01611037e-02 -8.63596201e-02
-1.33874035e+00 3.01042289e-01 6.65080905e-01 -7.32628644e-01
3.48345309e-01 2.14932650e-01 9.32618916e-01 6.41267478e-01
4.16646779e-01 5.32902658e-01 1.06581414e+00 -1.23698011e-01
6.46403313e-01 -1.24213427e-01 1.84506267e-01 1.66167155e-01
1.95655331e-01 2.19301313e-01 1.65851966e-01 -4.98124987e-01
7.29202390e-01 3.57701600e-01 -3.14356647e-02 -9.58581045e-02
-6.85683131e-01 1.08178282e+00 -1.33945450e-01 4.68482465e-01
-8.88830066e-01 6.56024218e-02 1.40157804e-01 2.38667995e-01
7.46825993e-01 -1.13961615e-01 -4.75562066e-01 -2.93408751e-01
-1.31535077e+00 5.40024638e-01 1.19189072e+00 5.44197679e-01
5.48748255e-01 9.74605083e-02 -4.37561512e-01 2.17399627e-01
4.04009014e-01 8.89871180e-01 2.07693473e-01 -7.23917603e-01
2.98717499e-01 3.52365486e-02 5.46657503e-01 -6.88588798e-01
-4.03367281e-01 -3.85878146e-01 -1.24647439e+00 6.84554338e-01
6.48921907e-01 -4.57579792e-01 -3.22448552e-01 1.75968528e+00
3.45806360e-01 2.49738574e-01 9.01567936e-02 4.66359049e-01
-2.32094169e-01 1.05213177e+00 1.56076983e-01 -7.34881639e-01
1.36634552e+00 -5.02119601e-01 -7.04513907e-01 6.46426976e-01
1.32855415e-01 -6.44582570e-01 6.84498191e-01 6.51206791e-01
-1.27423751e+00 -3.24799657e-01 -6.34831071e-01 5.90267837e-01
-2.28902191e-01 -5.20510338e-02 2.21286550e-01 6.70259774e-01
-8.83767188e-01 1.01300871e+00 -7.71622598e-01 2.77728736e-01
2.31006652e-01 -6.96720183e-02 2.45515883e-01 4.23042566e-01
-1.23690319e+00 8.10094297e-01 3.43322307e-01 2.80527830e-01
-6.24955833e-01 -9.37143624e-01 -3.36075842e-01 4.16638345e-01
3.50745171e-01 -6.53733850e-01 8.96613181e-01 -6.06402397e-01
-1.80025077e+00 1.52473792e-01 -1.41343586e-02 -7.61285245e-01
1.38079965e+00 5.46706542e-02 -2.67120689e-01 5.21824397e-02
-1.32322460e-01 -2.54662484e-01 1.24817145e+00 -6.15899384e-01
-5.78027368e-01 -3.22141856e-01 -2.52040863e-01 -7.41642714e-02
8.59344900e-02 -2.05148742e-01 1.33818567e-01 -9.15702343e-01
-1.29818127e-01 -9.30501938e-01 -4.72457409e-01 -1.95778519e-01
-7.43269771e-02 -3.84591311e-01 2.80902803e-01 -8.37636471e-01
1.19496417e+00 -2.27599144e+00 -4.24245074e-02 4.67310190e-01
-2.54842967e-01 -1.34773687e-01 4.83147472e-01 5.46977460e-01
-9.88136604e-02 3.75351399e-01 -5.80689192e-01 -3.18638295e-01
2.21227750e-01 -3.94166484e-02 -5.45536220e-01 4.15652305e-01
6.94065392e-02 5.50695598e-01 -6.19308114e-01 -1.18476443e-01
5.58818355e-02 3.47859532e-01 -3.29145700e-01 -1.03640452e-01
-4.66378540e-01 4.54140067e-01 -5.01097500e-01 9.11358595e-02
1.05633676e+00 3.66306417e-02 -2.97757059e-01 1.85825959e-01
-7.39912748e-01 -5.39659441e-01 -1.71241546e+00 1.11765707e+00
-6.29142582e-01 5.99471554e-02 9.56198350e-02 -9.46662188e-01
9.90469277e-01 4.32683319e-01 6.76768720e-01 -2.96380192e-01
-1.00618079e-01 4.71849561e-01 -4.68053818e-01 -1.31144911e-01
3.25972199e-01 -7.71953762e-01 1.42119601e-01 3.59355599e-01
-4.71115649e-01 -2.67400265e-01 4.85756367e-01 -3.35417449e-01
7.29393423e-01 -1.03212986e-03 1.93303332e-01 -5.32081604e-01
7.26647019e-01 -2.81329006e-01 5.03928185e-01 4.87614095e-01
2.76332527e-01 1.78589493e-01 9.47887719e-01 -2.04794675e-01
-1.11236525e+00 -1.19989002e+00 -2.76368111e-01 3.81776005e-01
-2.23543331e-01 2.17237458e-01 -5.27925789e-01 -1.89578384e-01
2.18957707e-01 1.27005947e+00 -4.95644182e-01 2.43569866e-01
-1.85599312e-01 -1.16899681e+00 1.66453246e-03 6.15780465e-02
5.60652733e-01 -1.01809108e+00 -6.48239553e-01 5.98285139e-01
3.32959861e-01 -5.96642315e-01 -2.20745966e-01 9.63434298e-03
-1.03374577e+00 -9.17510331e-01 -1.45779085e+00 -6.83761537e-02
1.17225036e-01 -3.75102967e-01 7.79067874e-01 -4.67041016e-01
-7.40720481e-02 3.73160124e-01 -3.50371860e-02 -4.78533089e-01
-3.95603210e-01 -4.13474381e-01 -2.42696747e-01 4.74257112e-01
-2.34470759e-02 -6.21386230e-01 -6.61992073e-01 1.10424280e-01
-9.46418524e-01 -3.73275459e-01 4.21050966e-01 7.71465003e-01
7.17490733e-01 6.27991438e-01 4.89202350e-01 -5.23289919e-01
8.02836061e-01 -6.50774658e-01 -1.29280913e+00 2.33230665e-01
-7.25194275e-01 2.67779708e-01 6.20895863e-01 -4.54016656e-01
-1.21333063e+00 -2.82495528e-01 -1.32953197e-01 -3.59937012e-01
3.52138519e-01 5.53200126e-01 4.48348597e-02 1.92328691e-01
-2.21672341e-01 2.62975693e-01 2.14930288e-02 -7.33600736e-01
2.37292603e-01 8.77678841e-02 1.41269699e-01 -6.69780254e-01
8.58700633e-01 6.47921324e-01 6.52104616e-01 -8.84606957e-01
-1.19766548e-01 -2.25624591e-01 -3.14870268e-01 -2.97459587e-02
1.03767574e+00 -6.47073150e-01 -1.11763263e+00 8.75062943e-01
-1.20233786e+00 -3.80894512e-01 -7.87522137e-01 8.18819284e-01
-9.29799914e-01 2.51347721e-01 -8.43243837e-01 -1.56537843e+00
-3.39970917e-01 -9.16395485e-01 3.31974328e-01 2.90162057e-01
1.59586087e-01 -1.37557185e+00 4.19841886e-01 -3.25882703e-01
3.94947648e-01 4.88719761e-01 1.26449108e+00 -7.04549491e-01
-4.65761304e-01 -9.62151140e-02 -4.69266772e-02 5.13692796e-01
-3.19465727e-01 3.82436961e-01 -3.52517009e-01 -4.80748080e-02
6.67487025e-01 5.52731335e-01 5.62990844e-01 7.18259633e-01
8.19560945e-01 -2.38708064e-01 5.51595241e-02 3.53732288e-01
1.63752818e+00 3.20807666e-01 4.90835786e-01 2.28286594e-01
9.53335091e-02 5.34408152e-01 6.98098242e-01 8.70646954e-01
1.94335759e-01 3.48804623e-01 3.91438782e-01 5.33050857e-02
8.44570577e-01 -1.20636806e-01 3.69027019e-01 5.58889270e-01
-4.04709190e-01 2.18010694e-01 -7.11601675e-01 2.26645052e-01
-1.79796088e+00 -1.12661707e+00 -5.84402621e-01 2.65947580e+00
6.14304423e-01 1.71408311e-01 3.68161231e-01 1.21415921e-01
6.97508991e-01 -4.69422303e-02 -4.22429085e-01 -3.96928847e-01
-7.36002848e-02 4.60759252e-01 6.52848780e-01 6.21089518e-01
-8.95562470e-01 2.47262999e-01 5.31167126e+00 1.16261303e+00
-9.47351813e-01 2.26493374e-01 6.40134871e-01 2.89525717e-01
-3.45433474e-01 1.00209460e-01 -8.24919224e-01 9.90552068e-01
8.38961542e-01 -7.36480832e-01 2.74296850e-01 9.39005077e-01
5.24920344e-01 -2.79090673e-01 -7.68314004e-01 9.03523445e-01
-6.47055626e-01 -8.95119667e-01 -4.04410094e-01 4.20013130e-01
6.19382143e-01 -2.74103552e-01 3.75520855e-01 2.89101005e-01
2.72355407e-01 -7.61153638e-01 6.95964992e-01 1.16149318e+00
2.69857496e-01 -1.09935808e+00 6.73281193e-01 6.51520789e-01
-1.25808811e+00 -4.10753675e-02 -4.00930285e-01 8.28200653e-02
5.69884896e-01 1.16739666e+00 -1.40776247e-01 7.76800752e-01
3.84745300e-01 3.62664074e-01 4.81914729e-02 1.25320578e+00
-2.16202403e-04 4.25828815e-01 -6.46673858e-01 -1.48777649e-01
2.40741402e-01 -1.11325002e+00 7.64049649e-01 8.35459828e-01
1.13753593e+00 3.44338901e-02 -8.48393515e-02 1.18572879e+00
4.26406413e-01 3.78087938e-01 -3.48718286e-01 7.81955719e-02
9.34449397e-03 1.07221079e+00 -7.42233634e-01 -2.69367009e-01
-4.47483867e-01 6.93801284e-01 -3.87524992e-01 6.41751587e-01
-8.45896602e-01 1.64723229e-02 5.12558043e-01 -4.60611023e-02
5.96095920e-01 -4.17661071e-01 -1.28544316e-01 -1.04763484e+00
4.91168462e-02 -3.47391278e-01 4.94978070e-01 -3.89070898e-01
-1.48247266e+00 3.09780240e-01 1.64036691e-01 -1.06085670e+00
-4.88307148e-01 -5.36478102e-01 -9.62734103e-01 1.44115257e+00
-1.61215878e+00 -6.36932969e-01 4.17193681e-01 7.05667317e-01
3.14261585e-01 -8.93057976e-03 6.62960470e-01 1.54142097e-01
-5.77308416e-01 -1.29347324e-01 8.31772983e-01 -3.04371625e-01
7.58569464e-02 -1.33900356e+00 2.53082633e-01 5.35036862e-01
-4.57175858e-02 2.83631235e-01 8.77057791e-01 -8.73920977e-01
-1.04456782e+00 -7.41425514e-01 5.21993518e-01 -6.70613488e-03
1.12047970e+00 1.86277868e-03 -1.14717984e+00 2.39984483e-01
-3.54021154e-02 -2.10347608e-01 4.34005260e-01 -2.94540107e-01
3.26985300e-01 -1.45109206e-01 -1.25488281e+00 4.27751422e-01
4.65561524e-02 -2.58992046e-01 -3.60832244e-01 1.44961908e-01
3.91357839e-01 1.75582826e-01 -1.03924787e+00 2.56909430e-01
3.80583704e-01 -1.01650405e+00 6.99218750e-01 -2.26470828e-01
-4.60004471e-02 -2.12162435e-01 5.11470400e-02 -1.26256537e+00
-3.99016812e-02 -1.08831656e+00 -2.77778059e-01 1.48533964e+00
3.02564979e-01 -1.09145164e+00 4.11009878e-01 8.58204722e-01
6.64338529e-01 -6.38005614e-01 -1.35765529e+00 -9.77092206e-01
5.44053018e-01 -5.16867578e-01 7.72686660e-01 4.53059644e-01
-4.79899108e-01 -3.73195887e-01 -2.66034186e-01 1.12740792e-01
1.06265688e+00 1.63100481e-01 3.79382998e-01 -1.46956873e+00
-5.36184430e-01 -6.80577219e-01 6.94614127e-02 -7.60044813e-01
1.40266925e-01 -5.43227375e-01 -1.35672629e-01 -9.01144505e-01
-8.91021714e-02 -5.06458461e-01 -2.36491635e-01 -4.32241350e-01
-1.31481634e-02 -3.50476444e-01 5.27101517e-01 1.85089827e-01
1.65013552e-01 7.81790316e-01 9.92933810e-01 1.57211851e-02
-2.44543180e-01 9.02802169e-01 3.80349271e-02 8.24043751e-01
6.28688216e-01 -4.19135362e-01 -2.64999092e-01 4.39933330e-01
5.14400423e-01 5.09536982e-01 4.19083953e-01 -9.20419216e-01
2.74327785e-01 -1.88361973e-01 3.33087258e-02 -9.54659343e-01
2.89934307e-01 -9.24146771e-01 7.30605841e-01 7.30629027e-01
7.63300657e-02 2.83113748e-01 -7.58513138e-02 8.58325481e-01
-1.90798938e-01 -7.85152853e-01 7.02807903e-01 -2.07809374e-01
-2.82738149e-01 6.82661951e-01 -4.31902379e-01 1.44713059e-01
1.24857378e+00 2.93961376e-01 3.42640787e-01 -4.51633096e-01
-1.16041756e+00 3.29239696e-01 3.47550064e-01 -3.47992033e-01
2.07918048e-01 -1.32736254e+00 -6.44385159e-01 1.24770850e-01
-7.56551564e-01 -1.81241613e-02 6.02124691e-01 1.08498704e+00
-4.68512744e-01 7.42976293e-02 1.72474071e-01 -3.17064703e-01
-7.01928556e-01 8.22602570e-01 2.07407087e-01 -7.33897030e-01
-5.33579707e-01 2.69054174e-01 1.04739264e-01 2.48460859e-01
-1.12565361e-01 -5.22393107e-01 -2.51368970e-01 7.45252907e-01
5.40438354e-01 8.96296620e-01 -2.72993088e-01 -3.10814530e-01
2.00166941e-01 8.46417904e-01 5.30313313e-01 -6.42734408e-01
1.37294042e+00 -1.63934514e-01 -3.57114822e-01 1.05573440e+00
9.89348292e-01 8.32657889e-02 -1.50582862e+00 3.69450971e-02
2.65237868e-01 3.03446297e-02 -2.04516724e-01 -1.07177421e-01
-8.60309005e-01 1.07248402e+00 3.91540587e-01 8.07563663e-01
9.03162062e-01 -6.15925193e-01 6.89447582e-01 -3.57806645e-02
4.50191081e-01 -1.14563596e+00 -4.50084358e-01 5.54594100e-01
8.55000675e-01 -7.72923946e-01 -1.93054110e-01 -1.55567855e-01
-6.23002529e-01 1.41668606e+00 -3.33614528e-01 -2.85418600e-01
1.35139787e+00 1.14035502e-01 -4.20192957e-01 2.55522668e-01
-4.17701691e-01 -1.36037692e-01 -1.51747232e-02 -4.38366793e-02
-5.73456772e-02 3.56406987e-01 -7.44815648e-01 7.93765128e-01
-1.50630489e-01 2.28480890e-01 4.22845066e-01 5.52011609e-01
-1.52005166e-01 -1.01945150e+00 -5.01820266e-01 1.82861105e-01
-7.16180444e-01 -1.79572836e-01 5.41290164e-01 7.50258565e-01
-1.52719066e-01 4.34986115e-01 8.35964158e-02 5.05463660e-01
3.26182276e-01 5.30747056e-01 5.21243587e-02 -1.77040562e-01
-3.49628448e-01 3.17903459e-01 -4.70530212e-01 -1.79655060e-01
-2.08116472e-01 -9.58292603e-01 -1.06491697e+00 -4.62224633e-01
-9.66955423e-02 5.66792607e-01 9.23676312e-01 8.97176385e-01
-2.13506728e-01 2.63815761e-01 9.69529212e-01 -7.39824176e-01
-1.32524562e+00 -6.22788608e-01 -1.23562300e+00 1.72361627e-01
2.63785183e-01 -5.38833737e-01 -7.74833918e-01 -3.65162522e-01] | [5.700814723968506, 3.854480266571045] |
67aaf371-45b6-4c00-a844-b89e11d8ea74 | predator-registration-of-3d-point-clouds-with | 2011.13005 | null | https://arxiv.org/abs/2011.13005v3 | https://arxiv.org/pdf/2011.13005v3.pdf | PREDATOR: Registration of 3D Point Clouds with Low Overlap | We introduce PREDATOR, a model for pairwise point-cloud registration with deep attention to the overlap region. Different from previous work, our model is specifically designed to handle (also) point-cloud pairs with low overlap. Its key novelty is an overlap-attention block for early information exchange between the latent encodings of the two point clouds. In this way the subsequent decoding of the latent representations into per-point features is conditioned on the respective other point cloud, and thus can predict which points are not only salient, but also lie in the overlap region between the two point clouds. The ability to focus on points that are relevant for matching greatly improves performance: PREDATOR raises the rate of successful registrations by more than 20% in the low-overlap scenario, and also sets a new state of the art for the 3DMatch benchmark with 89% registration recall. | ['Konrad Schindler', 'Andreas Wieser', 'Mikhail Usvyatsov', 'Zan Gojcic', 'Shengyu Huang'] | 2020-11-25 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Huang_Predator_Registration_of_3D_Point_Clouds_With_Low_Overlap_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Huang_Predator_Registration_of_3D_Point_Clouds_With_Low_Overlap_CVPR_2021_paper.pdf | cvpr-2021-1 | ['deep-attention', 'deep-attention'] | ['computer-vision', 'natural-language-processing'] | [ 1.20838694e-01 2.61661917e-01 -2.12460399e-01 -4.82231565e-02
-1.22858799e+00 -5.04445672e-01 7.34636605e-01 5.51220357e-01
-2.43289843e-01 5.37356995e-02 1.82223722e-01 1.74110606e-01
-3.17478888e-02 -7.06054091e-01 -1.12996030e+00 -5.60390532e-01
-4.01831001e-01 8.80137801e-01 4.83718127e-01 -1.82315916e-01
2.96174765e-01 8.75953853e-01 -1.63391459e+00 3.19068104e-01
5.25765479e-01 1.01426911e+00 3.12226176e-01 3.58055204e-01
-6.12230413e-02 5.51430061e-02 -4.21145439e-01 -4.48307782e-01
5.22907555e-01 4.29813594e-01 -8.27789903e-01 -8.81637782e-02
1.07407391e+00 -2.57051438e-01 -2.90697515e-01 9.01112616e-01
1.95248351e-01 -4.48320881e-02 5.72456717e-01 -1.26578057e+00
-6.01026595e-01 1.79873377e-01 -9.95826125e-01 2.13869855e-01
3.44077855e-01 8.89489874e-02 1.17198467e+00 -1.12332821e+00
6.80021286e-01 1.24523330e+00 7.21531332e-01 1.30880535e-01
-1.13742781e+00 -9.08724308e-01 2.86859751e-01 3.39460783e-02
-1.65233004e+00 -5.49494147e-01 5.56101978e-01 -4.81277913e-01
1.18754482e+00 2.41137713e-01 6.68864965e-01 5.62750518e-01
7.35647827e-02 5.31098843e-01 4.36358750e-01 -1.46664083e-01
-5.40054850e-02 -2.25097433e-01 9.33152661e-02 4.73545104e-01
9.39873084e-02 2.55872101e-01 -5.23324609e-01 -3.39071035e-01
7.78745294e-01 1.91976786e-01 -7.05611482e-02 -6.14435911e-01
-1.46530116e+00 6.33174837e-01 1.00645375e+00 2.68821597e-01
-4.70741570e-01 4.25046027e-01 -1.46109104e-01 2.45896861e-01
5.41181207e-01 4.20911014e-01 -1.86128423e-01 -6.93674386e-02
-1.04204309e+00 3.71369928e-01 5.07120073e-01 1.25871205e+00
1.05730128e+00 -4.88990396e-01 -1.75180420e-01 2.58647025e-01
5.00323355e-01 6.60000741e-01 -3.00912797e-01 -7.20365524e-01
7.92746007e-01 6.89052343e-01 1.87267035e-01 -1.20705760e+00
-5.75029217e-02 -4.62843567e-01 -4.26158160e-01 2.50884056e-01
-1.79348048e-03 3.19663912e-01 -9.87992585e-01 1.57896650e+00
4.48767871e-01 6.67060673e-01 -3.30387130e-02 9.78327155e-01
8.26517045e-01 7.19584763e-01 -9.62906852e-02 1.69288814e-01
1.40829492e+00 -7.58791327e-01 -1.39660522e-01 -4.34601754e-01
4.08252090e-01 -1.03024399e+00 5.91868103e-01 -2.72128165e-01
-1.22003984e+00 -5.24229050e-01 -9.46173668e-01 -3.21519136e-01
-2.18992084e-01 -1.61391005e-01 4.98365015e-01 2.11192537e-02
-1.38205934e+00 6.66856766e-01 -7.86302328e-01 -3.13193917e-01
5.71958244e-01 8.71118665e-01 -4.83220458e-01 -6.41778708e-02
-7.71396577e-01 8.31679225e-01 4.50089350e-02 -9.51863453e-02
-5.45318007e-01 -9.39683080e-01 -8.93355548e-01 2.98139095e-01
7.43034929e-02 -7.88165748e-01 8.22032630e-01 -7.83025980e-01
-6.94774330e-01 1.20541215e+00 -4.56286371e-01 -4.85515952e-01
4.32050675e-01 -3.87562633e-01 -2.87362076e-02 6.96754009e-02
4.86733437e-01 1.39243782e+00 8.38634193e-01 -1.57509625e+00
-6.67433918e-01 -4.12775248e-01 -1.82634637e-01 3.98076475e-01
1.51789159e-01 7.35654756e-02 -1.08411038e+00 -4.58395958e-01
5.06093323e-01 -1.08172882e+00 -1.33227229e-01 3.48439783e-01
-2.75630087e-01 -2.19057277e-01 1.01724339e+00 -3.79761696e-01
5.30095577e-01 -2.48869014e+00 -5.28505519e-02 4.34341520e-01
4.03022677e-01 2.06450429e-02 -3.53567332e-01 4.12178487e-01
-3.44409168e-01 9.61439312e-03 -2.23773345e-01 -7.63911426e-01
-1.68457016e-01 1.06620185e-01 -7.95739651e-01 6.80150986e-01
5.11519432e-01 1.08305347e+00 -9.83995259e-01 -2.81494021e-01
3.68395925e-01 6.11588955e-01 -4.56532001e-01 1.06879719e-01
-2.83972006e-02 4.09175575e-01 -2.79288948e-01 5.10704935e-01
9.39080834e-01 -4.56518322e-01 -4.07156199e-01 -2.18326390e-01
-1.90412104e-01 2.75841832e-01 -9.94144380e-01 1.81670582e+00
-1.57233804e-01 7.70002842e-01 -1.87351361e-01 -2.31338233e-01
1.00404549e+00 2.16436535e-01 8.66093576e-01 -5.39373755e-01
-2.48562515e-01 2.03049436e-01 -1.94262147e-01 5.04663065e-02
7.43776679e-01 -4.64453781e-03 -3.51969823e-02 1.15643620e-01
-1.04471952e-01 -1.59278944e-01 -3.53723228e-01 1.75002709e-01
9.45026457e-01 -1.18461169e-01 -8.23579961e-04 -5.89978509e-02
1.66863352e-01 7.30224773e-02 3.33208203e-01 6.00402415e-01
9.09130350e-02 1.09028077e+00 3.18211347e-01 -4.22473460e-01
-1.11901605e+00 -1.22058916e+00 -1.93220198e-01 5.47432721e-01
8.37122083e-01 -4.10010099e-01 -2.79752046e-01 -4.96408582e-01
3.73988897e-01 4.31110740e-01 -5.58738708e-01 -1.91551432e-01
-5.73063374e-01 -7.46723711e-02 9.11444649e-02 5.13925552e-01
1.37168109e-01 -8.29218984e-01 -4.48439151e-01 1.01372488e-01
-1.50242582e-01 -1.27466071e+00 -4.35319573e-01 8.25965628e-02
-9.54616129e-01 -1.09818637e+00 -8.16132843e-01 -9.54309285e-01
8.61627758e-01 5.39541006e-01 1.33500457e+00 3.28560978e-01
6.06614761e-02 3.33531410e-01 -2.13330582e-01 -3.25720459e-01
-2.85847876e-02 1.26996309e-01 -1.01733230e-01 5.38229831e-02
4.63829488e-01 -5.14458239e-01 -5.49730301e-01 4.39591080e-01
-4.91199613e-01 6.74490258e-02 5.46827197e-01 4.71804470e-01
8.85649741e-01 -3.32994252e-01 -1.77082736e-02 -3.68417710e-01
-1.21202707e-01 -5.68072796e-01 -6.65890336e-01 -2.77279299e-02
-1.73723131e-01 6.07849434e-02 -9.19884667e-02 -1.53909951e-01
-3.59586418e-01 2.39463419e-01 -7.28487149e-02 -9.10660863e-01
-6.84396029e-02 9.94035229e-02 7.51518384e-02 -3.56779605e-01
2.65459925e-01 -1.10157125e-01 -1.57112226e-01 -6.24784887e-01
2.78465658e-01 4.01652008e-01 6.40801013e-01 -4.37055200e-01
1.12968040e+00 8.46511483e-01 5.67625687e-02 -6.11290932e-01
-4.03634429e-01 -7.41374850e-01 -8.16777527e-01 1.11793242e-01
7.82002091e-01 -1.19859266e+00 -8.38862300e-01 1.49257585e-01
-1.57682538e+00 3.20385918e-02 -4.52962071e-01 3.68103743e-01
-4.94253457e-01 4.10410643e-01 -3.64936709e-01 -5.76003611e-01
-3.65668207e-01 -1.36565089e+00 1.88032115e+00 9.26661585e-03
-2.53058016e-01 -5.91830492e-01 1.55340463e-01 -1.91929787e-02
1.67033654e-02 2.28651464e-01 7.76715100e-01 -5.95930755e-01
-1.22155833e+00 -3.59810829e-01 -5.35020649e-01 -3.25954795e-01
-1.24935851e-01 8.18350017e-02 -1.13042021e+00 -5.86204946e-01
-1.64845616e-01 2.37135455e-01 8.60206842e-01 3.07470769e-01
7.74278402e-01 -6.10690713e-02 -9.01969969e-01 6.74284160e-01
1.32529569e+00 -7.05458447e-02 7.28424430e-01 1.95992813e-01
7.91670859e-01 4.86286074e-01 6.69531167e-01 1.67821527e-01
5.46517551e-01 1.23398817e+00 9.56420183e-01 -4.05874580e-01
-5.28009683e-02 -2.86660016e-01 2.27263764e-01 6.73749626e-01
1.68571830e-01 -1.09454826e-01 -1.08641756e+00 8.50723207e-01
-1.96531725e+00 -7.51398027e-01 -3.17823231e-01 2.48935890e+00
3.93248886e-01 -1.70761049e-02 -2.20077872e-01 -4.18781757e-01
8.54131401e-01 3.45836818e-01 -3.02698433e-01 8.75892341e-02
-2.76305191e-02 2.80262023e-01 6.37147427e-01 6.09825492e-01
-1.36045313e+00 1.09227300e+00 6.35875320e+00 6.71824753e-01
-1.08755004e+00 3.18415835e-02 3.86101305e-01 -4.09010738e-01
-3.21634710e-01 2.93101072e-01 -8.86188567e-01 4.41740066e-01
6.94333315e-01 1.92547128e-01 4.56538573e-02 7.07824767e-01
-3.31941217e-01 2.99840104e-02 -1.35364580e+00 1.11983705e+00
-1.14545405e-01 -1.60678697e+00 1.43402323e-01 4.18513149e-01
7.69491434e-01 6.60263479e-01 3.55558425e-01 -2.46907454e-02
2.00777560e-01 -9.61883307e-01 8.09846878e-01 5.05195796e-01
8.53447556e-01 -8.42935503e-01 7.36309707e-01 1.78421304e-01
-1.41831577e+00 2.15232655e-01 -6.30157411e-01 1.91901594e-01
2.00070351e-01 4.27824736e-01 -8.47002327e-01 5.24446309e-01
7.83012450e-01 9.12870228e-01 -5.27880371e-01 1.46637273e+00
-5.42279482e-02 2.63145380e-02 -7.38336682e-01 4.67570961e-01
4.74692434e-01 1.01853438e-01 9.34224427e-01 9.42416072e-01
4.43229377e-01 -1.35350272e-01 2.22291961e-01 9.94995832e-01
-6.61635771e-02 -9.14535448e-02 -8.53073299e-01 4.04187769e-01
7.00338483e-01 8.80257964e-01 -6.55344009e-01 -2.35739499e-01
-2.73483545e-01 9.54667211e-01 4.40726876e-01 1.82851836e-01
-7.67660737e-01 -6.80932328e-02 1.17338002e+00 2.41751656e-01
6.78816974e-01 -3.52339774e-01 -3.26454192e-01 -8.45454276e-01
2.74803638e-01 -4.46872592e-01 8.94439295e-02 -1.00863469e+00
-1.06284833e+00 6.76856279e-01 4.50464562e-02 -1.59009707e+00
-9.83365066e-03 -2.22352475e-01 -7.46306121e-01 1.18168485e+00
-1.58231413e+00 -1.33773196e+00 -3.46284717e-01 4.76418763e-01
1.40462741e-01 1.44850299e-01 7.53516614e-01 3.62038046e-01
3.14305052e-02 4.69241619e-01 7.16458932e-02 8.72273967e-02
5.56560874e-01 -9.98337150e-01 1.34722054e+00 7.64049828e-01
6.43298328e-01 7.06368744e-01 4.77968901e-01 -8.66654038e-01
-1.12806892e+00 -1.11677253e+00 1.09364581e+00 -8.30104172e-01
2.97847271e-01 -5.48009396e-01 -1.03351283e+00 7.10503936e-01
-2.12963484e-02 4.89794239e-02 2.39003018e-01 2.15269417e-01
-5.61368823e-01 1.21257799e-02 -9.68449712e-01 7.15491533e-01
9.80256200e-01 -6.69513822e-01 -6.23457015e-01 3.87439013e-01
9.75797534e-01 -8.43910038e-01 -8.66911590e-01 5.18709004e-01
2.16741547e-01 -8.15950930e-01 1.40760422e+00 -3.41587454e-01
2.88739353e-01 -4.45774108e-01 -1.66731477e-01 -8.60244393e-01
-4.52201456e-01 -6.68782592e-01 -1.02893740e-01 1.03725100e+00
2.03050762e-01 -3.56028557e-01 9.74074960e-01 5.05829275e-01
-1.75812453e-01 -7.30776489e-01 -1.26294768e+00 -5.55928171e-01
-2.34961584e-01 -4.33185548e-01 9.43594158e-01 8.40823472e-01
-2.20730662e-01 2.07297966e-01 -1.28332525e-01 7.88567841e-01
5.64131021e-01 4.52675313e-01 9.04305935e-01 -1.45514917e+00
-3.81480269e-02 -4.74267364e-01 -9.31963861e-01 -1.45744908e+00
1.33803800e-01 -1.07628536e+00 9.18083712e-02 -1.42823386e+00
5.86992316e-03 -7.61520922e-01 -8.01410079e-02 6.22056663e-01
-1.77236572e-01 4.65217471e-01 5.60291767e-01 7.87469685e-01
-5.36439300e-01 5.16187787e-01 8.25157702e-01 -1.78748190e-01
-1.36182100e-01 4.49049659e-02 -6.06831074e-01 4.49649602e-01
2.62061507e-01 -7.70378351e-01 1.29259169e-01 -6.57632768e-01
9.37501416e-02 -9.50060785e-02 7.46422231e-01 -1.22455740e+00
4.47043985e-01 1.74148366e-01 3.83796126e-01 -1.03065383e+00
8.14126015e-01 -1.25911033e+00 5.00641942e-01 3.08322698e-01
-1.80894598e-01 3.66116196e-01 4.15385902e-01 7.63763428e-01
-2.66061097e-01 5.26498556e-02 6.88192785e-01 1.23674162e-01
-8.47168326e-01 7.49679983e-01 2.81285912e-01 -2.49554604e-01
1.11739850e+00 -5.36140442e-01 -2.84534425e-01 -4.20274317e-01
-6.07221782e-01 3.69913071e-01 9.11963105e-01 6.79497480e-01
6.14786088e-01 -1.46514761e+00 -6.60598695e-01 4.53826904e-01
5.35420835e-01 5.34873605e-01 3.51953022e-02 7.91329384e-01
-3.64613980e-01 3.75241369e-01 -6.02251142e-02 -1.08848047e+00
-1.40511274e+00 5.82137406e-01 1.82682350e-01 -1.83815151e-01
-9.07400966e-01 1.04760301e+00 5.67006767e-01 -2.34012559e-01
2.73282975e-01 -4.67375845e-01 9.57396254e-02 -1.19027838e-01
3.61832887e-01 -2.82826424e-02 2.80151814e-01 -1.25710464e+00
-6.79341733e-01 1.11432040e+00 -3.09934407e-01 -7.64184296e-02
1.27202535e+00 1.10139452e-01 -1.01892799e-01 2.55989015e-01
1.51593125e+00 8.09537619e-02 -1.50191379e+00 -4.92183983e-01
3.79493088e-02 -8.72806847e-01 -1.91787202e-02 -4.23937231e-01
-1.23751652e+00 8.02655280e-01 6.90577626e-01 -1.49740815e-01
5.95686972e-01 3.73351187e-01 8.02776456e-01 1.82452232e-01
5.79768538e-01 -4.00377482e-01 -2.10441083e-01 6.56327009e-01
9.46371496e-01 -1.18372691e+00 4.90020886e-02 -5.31402230e-01
-5.49595833e-01 8.56864333e-01 4.07677829e-01 -4.10066575e-01
5.46898484e-01 2.51339883e-01 -6.13529831e-02 -7.42772400e-01
-6.83823645e-01 -2.97680467e-01 7.22746611e-01 7.07343996e-01
1.47569418e-01 -1.62304163e-01 4.54207242e-01 -1.11244626e-01
-1.80518448e-01 -3.82884055e-01 -1.71026990e-01 7.72569299e-01
-4.17095482e-01 -8.84170532e-01 -5.01127601e-01 3.10482472e-01
4.15574294e-03 -2.67396897e-01 -5.38815975e-01 7.34152496e-01
3.50061501e-03 5.58181822e-01 8.72211397e-01 -3.89034241e-01
4.65712190e-01 -3.56249303e-01 2.25227594e-01 -7.21348941e-01
-9.60633576e-01 7.69501403e-02 -3.50647449e-01 -7.79275298e-01
-2.70991147e-01 -6.80903912e-01 -1.35278630e+00 -4.30897713e-01
-3.11704129e-01 1.86247349e-01 5.31151354e-01 7.85720408e-01
9.92443204e-01 6.46216720e-02 4.83802885e-01 -1.23812091e+00
-3.45971026e-02 -4.39018518e-01 -2.44580194e-01 5.31210005e-01
6.79593623e-01 -8.58810604e-01 -2.33586747e-02 -3.92588913e-01] | [7.677492141723633, -3.0915026664733887] |
8ad0dc59-3626-4eb0-abcf-041990fa2de8 | dmrf-unet-a-two-stage-deep-learning-scheme | 2205.07567 | null | https://arxiv.org/abs/2205.07567v1 | https://arxiv.org/pdf/2205.07567v1.pdf | DMRF-UNet: A Two-Stage Deep Learning Scheme for GPR Data Inversion under Heterogeneous Soil Conditions | Traditional ground-penetrating radar (GPR) data inversion leverages iterative algorithms which suffer from high computation costs and low accuracy when applied to complex subsurface scenarios. Existing deep learning-based methods focus on the ideal homogeneous subsurface environments and ignore the interference due to clutters and noise in real-world heterogeneous environments. To address these issues, a two-stage deep neural network (DNN), called DMRF-UNet, is proposed to reconstruct the permittivity distributions of subsurface objects from GPR B-scans under heterogeneous soil conditions. In the first stage, a U-shape DNN with multi-receptive-field convolutions (MRF-UNet1) is built to remove the clutters due to inhomogeneity of the heterogeneous soil. Then the denoised B-scan from the MRF-UNet1 is combined with the noisy B-scan to be inputted to the DNN in the second stage (MRF-UNet2). The MRF-UNet2 learns the inverse mapping relationship and reconstructs the permittivity distribution of subsurface objects. To avoid information loss, an end-to-end training method combining the loss functions of two stages is introduced. A wide range of subsurface heterogeneous scenarios and B-scans are generated to evaluate the inversion performance. The test results in the numerical experiment and the real measurement show that the proposed network reconstructs the permittivities, shapes, sizes, and locations of subsurface objects with high accuracy. The comparison with existing methods demonstrates the superiority of the proposed methodology for the inversion under heterogeneous soil conditions. | ['Abdulkadir C. Yucel', 'Mohamed Lokman Mohd Yusof', 'Genevieve Ow', 'Hai-Han Sun', 'Yee Hui Lee', 'Qiqi Dai'] | 2022-05-16 | null | null | null | null | ['gpr', 'gpr'] | ['computer-vision', 'miscellaneous'] | [ 2.21266776e-01 -1.08259737e-01 8.75116765e-01 -4.90803510e-01
-6.85425997e-01 2.19891995e-01 -1.26676053e-01 -2.42482185e-01
-1.39486343e-01 8.19379389e-01 -2.65488611e-03 -3.21499079e-01
-5.36575198e-01 -1.40327299e+00 -8.09290290e-01 -1.22072875e+00
-4.16162640e-01 5.22039533e-01 6.04341738e-03 -4.20620650e-01
1.35119054e-02 3.75621885e-01 -1.41853726e+00 3.23870540e-01
8.91194046e-01 1.44234133e+00 6.30370617e-01 1.92478016e-01
1.40181378e-01 3.53267491e-01 -1.18533023e-01 6.51162341e-02
4.05573398e-01 1.87612966e-01 -2.49974489e-01 -5.10820270e-01
-1.60932839e-02 -8.25131714e-01 -4.23223436e-01 1.22414792e+00
9.88371134e-01 3.00518364e-01 6.63052201e-01 -5.64832628e-01
-3.60230774e-01 3.71793121e-01 -1.04793882e+00 4.01303358e-02
-1.07948251e-01 -3.33248973e-01 2.18312427e-01 -1.14801884e+00
1.16830468e-02 1.38273561e+00 1.28469765e+00 9.42088291e-03
-9.43681180e-01 -9.09843624e-01 3.06471307e-02 1.69239998e-01
-1.33587921e+00 -4.04596537e-01 7.35402107e-01 -1.96881667e-01
5.47578573e-01 7.18477219e-02 7.07767606e-01 9.78802621e-01
2.85852849e-01 6.13930106e-01 1.03928959e+00 -2.90021569e-01
2.58887917e-01 -3.34733695e-01 1.77337006e-01 3.38344514e-01
8.03986371e-01 3.88196349e-01 -6.31653355e-04 -9.88611951e-02
9.55085635e-01 6.66541904e-02 -5.37033737e-01 -1.73221484e-01
-8.50943923e-01 6.81435883e-01 5.70514560e-01 8.64729881e-02
-1.07747293e+00 -1.63675070e-01 4.32239592e-01 2.60094970e-01
6.18202329e-01 5.38774133e-02 -4.03688043e-01 5.35419285e-01
-8.56669605e-01 4.92477328e-01 8.16340327e-01 6.00980520e-01
6.99671209e-01 4.47117239e-01 5.05773723e-02 1.07369959e+00
7.53892481e-01 1.08252406e+00 1.42516583e-01 -5.64423800e-01
4.91249084e-01 -1.62933409e-01 4.38317418e-01 -1.30646884e+00
-4.08930719e-01 -7.81944633e-01 -1.32012427e+00 1.26215443e-01
-1.84688598e-01 -4.31382835e-01 -1.11167669e+00 1.51379728e+00
6.04144275e-01 3.90833229e-01 5.77922046e-01 1.12919438e+00
1.13321877e+00 9.59650576e-01 -1.52860601e-02 -5.39630316e-02
1.01982200e+00 -3.15236598e-01 -6.41062438e-01 -3.97464991e-01
1.36635050e-01 -6.70683444e-01 3.79552841e-01 4.37998950e-01
-8.91472042e-01 -4.44956154e-01 -1.18446219e+00 4.81286168e-01
1.79609552e-01 -8.20141360e-02 6.72121763e-01 4.26336735e-01
-8.37374151e-01 3.16428691e-01 -9.27742779e-01 1.95315704e-01
3.91112149e-01 7.11198002e-02 -1.35741815e-01 -6.61575317e-01
-1.64177036e+00 6.29634738e-01 1.83865502e-01 1.18961334e+00
-1.20297396e+00 -9.17810380e-01 -7.82802045e-01 1.33376149e-02
-7.63938501e-02 -6.46006942e-01 1.03358734e+00 -7.74341762e-01
-1.29925442e+00 2.88799435e-01 3.06939155e-01 -1.03418678e-01
4.53087747e-01 -4.20076460e-01 -5.95120549e-01 1.40593499e-01
2.83784240e-01 -7.32976124e-02 7.80891120e-01 -1.66903281e+00
-4.62646872e-01 -4.00641918e-01 -2.96923250e-01 2.61828780e-01
3.54292542e-01 -5.04790425e-01 2.42548570e-01 -5.19691288e-01
1.02466154e+00 -3.13851714e-01 -2.24985570e-01 -2.29510814e-01
-4.32334602e-01 6.45067573e-01 8.32617640e-01 -1.11171150e+00
4.50650632e-01 -2.09117484e+00 -3.34418088e-01 5.63214123e-01
5.01556471e-02 2.36788332e-01 -3.30392063e-01 2.39851445e-01
-3.31720948e-01 -6.38197958e-01 -4.32168961e-01 8.65955129e-02
-2.67829269e-01 -1.56029999e-01 -2.65384823e-01 7.34807730e-01
-1.33760557e-01 4.67169136e-01 -7.85848975e-01 3.31656933e-02
1.39951512e-01 8.60194743e-01 -4.70144272e-01 2.29763374e-01
1.19058855e-01 5.39860547e-01 -8.86612356e-01 8.69593322e-01
1.88914132e+00 1.83943063e-01 1.37811497e-01 -6.25517011e-01
-1.80999428e-01 -2.00175509e-01 -1.39499164e+00 9.31211829e-01
-8.38965714e-01 2.71743953e-01 9.50168192e-01 -1.69162393e+00
1.36163616e+00 3.29534322e-01 4.03284371e-01 -1.24917865e+00
1.69913560e-01 5.98139524e-01 3.59264053e-02 -9.84034419e-01
9.93251279e-02 -7.15975225e-01 3.16785961e-01 1.86251953e-01
-2.52610534e-01 -2.36840285e-02 -5.19351482e-01 -3.08092564e-01
8.96003306e-01 2.10256292e-03 -1.41035289e-01 -5.13930619e-01
3.32352877e-01 -4.30894077e-01 6.85787201e-01 9.49401379e-01
2.82438934e-01 5.25631130e-01 -2.90953577e-01 -7.16104388e-01
-8.57505500e-01 -1.28168643e+00 -4.88624662e-01 6.52019858e-01
3.94528955e-01 9.86177087e-01 -3.78277242e-01 2.09644854e-01
1.72157973e-01 6.09092534e-01 -6.34908020e-01 -1.83727145e-01
-4.73147184e-01 -1.31436932e+00 4.85710233e-01 2.76563019e-01
1.22383857e+00 -1.01701820e+00 -5.03838420e-01 5.85955799e-01
-5.70769191e-01 -9.43048418e-01 6.03066266e-01 3.66734922e-01
-9.15803194e-01 -8.42594326e-01 -8.29306304e-01 -8.48374784e-01
5.22888303e-01 4.25068498e-01 8.85677814e-01 -1.37648851e-01
-8.81038234e-02 7.45128375e-03 -3.88944745e-01 -5.26414633e-01
-1.15688302e-01 -5.06258428e-01 -1.41480386e-01 2.59063631e-01
2.97936708e-01 -1.00771606e+00 -7.40627170e-01 2.49254078e-01
-8.45190763e-01 -8.51098448e-02 8.74606788e-01 9.95954156e-01
5.09568393e-01 4.50598300e-01 7.49768794e-01 -6.08754158e-01
4.44144070e-01 -9.32914734e-01 -6.15427077e-01 -7.49271642e-03
1.24309145e-01 -2.34180868e-01 2.90287465e-01 -3.87407690e-01
-1.66064882e+00 -4.84343410e-01 -5.63870788e-01 -3.46778296e-02
-1.32810548e-01 1.04076135e+00 -5.18273592e-01 -2.54813641e-01
5.21074593e-01 4.56565470e-01 -1.89450651e-01 -5.07355034e-01
-4.47619230e-01 6.96219802e-01 7.26646423e-01 -7.17136502e-01
5.99323213e-01 5.73410153e-01 5.41157462e-02 -1.25791991e+00
-9.03074026e-01 -1.95799470e-01 7.71160722e-02 -2.37640679e-01
3.43643486e-01 -1.27671421e+00 -4.16123211e-01 1.00842679e+00
-1.19391930e+00 -2.75107324e-01 1.32972747e-01 9.21360016e-01
-2.74375677e-01 3.16369712e-01 -5.14823139e-01 -1.08487570e+00
-7.80576944e-01 -8.92168164e-01 9.19039786e-01 2.38153309e-01
4.27901447e-01 -7.69859135e-01 -9.47115272e-02 1.47915676e-01
7.52419174e-01 4.34842139e-01 9.81468379e-01 -2.43271664e-01
-4.42457050e-01 -3.29076529e-01 -5.81087351e-01 4.45236802e-01
-3.54041010e-02 -7.29316294e-01 -1.15443075e+00 -4.34950173e-01
6.49773598e-01 -2.91107625e-01 8.83004487e-01 1.08494782e+00
9.76087630e-01 -9.81963053e-02 -2.95112163e-01 1.13411748e+00
1.93413568e+00 3.77349705e-01 1.15687752e+00 5.41123390e-01
4.19514686e-01 4.71169114e-01 6.78542674e-01 5.37188530e-01
1.10323071e-01 -6.20067976e-02 7.95395970e-01 -2.88489550e-01
2.27678180e-01 6.19823635e-02 -7.97430426e-02 6.58131897e-01
-2.98584729e-01 -3.41651976e-01 -8.14497471e-01 6.54043853e-01
-1.50851893e+00 -8.82776320e-01 -1.39889747e-01 1.92689204e+00
3.02052438e-01 -6.98306598e-04 -8.41326177e-01 2.11267367e-01
8.85669291e-01 -2.46077776e-02 -6.03717387e-01 9.54573676e-02
-2.22190440e-01 4.27440792e-01 5.21722436e-01 4.40243006e-01
-8.66908133e-01 2.16538012e-01 5.30828762e+00 4.69238222e-01
-1.43613875e+00 7.59430528e-02 4.42677468e-01 4.26978469e-01
-5.84770620e-01 -4.26215500e-01 -4.04546380e-01 2.31380910e-01
4.82252747e-01 5.15682817e-01 1.82146445e-01 5.54030120e-01
4.15534586e-01 -2.93110311e-01 -3.89416099e-01 9.27856147e-01
-3.96794170e-01 -9.55651104e-01 1.14948422e-01 -3.33564758e-01
5.06277204e-01 3.09792191e-01 -5.98159879e-02 4.61740941e-01
3.51887137e-01 -7.75672257e-01 8.51319730e-01 9.03635561e-01
7.98342645e-01 -7.42216825e-01 1.27134657e+00 3.01508874e-01
-1.09438813e+00 -2.56229162e-01 -5.90683401e-01 -1.35981530e-01
3.20183873e-01 1.39739406e+00 -5.48700571e-01 9.44655955e-01
1.08068812e+00 3.79505634e-01 4.17360842e-01 1.18569314e+00
-5.94439134e-02 6.30467117e-01 -5.96179008e-01 2.67999560e-01
4.07548934e-01 -4.39864278e-01 6.79940760e-01 1.12717462e+00
9.20919359e-01 4.54603583e-01 -5.76686598e-02 8.68137479e-01
2.07342938e-01 -2.18496516e-01 -5.98835170e-01 4.34418887e-01
5.60160160e-01 1.09109366e+00 -3.94463539e-01 -6.57169744e-02
-1.60881415e-01 2.62140810e-01 -3.32432896e-01 8.33813488e-01
-6.57445014e-01 -6.69372082e-01 3.63985717e-01 5.38794398e-01
5.51007390e-01 6.55134022e-02 -3.04005742e-01 -6.73483729e-01
1.05571158e-01 -8.02704036e-01 -2.88903583e-02 -1.00475287e+00
-1.42925572e+00 6.26305699e-01 1.24177054e-01 -1.07768142e+00
2.96085984e-01 -5.18046677e-01 -7.28383303e-01 1.28083956e+00
-1.97828281e+00 -1.08908617e+00 -9.27566707e-01 4.02410448e-01
-5.55908196e-02 -1.46119684e-01 7.53492057e-01 7.61151254e-01
-1.72863454e-01 1.23098031e-01 5.81999660e-01 5.35200490e-03
1.76842198e-01 -5.19518554e-01 2.41299748e-01 7.28484273e-01
-1.16391492e+00 5.18349946e-01 1.01428652e+00 -8.76359224e-01
-1.17011178e+00 -1.24526298e+00 1.77940711e-01 7.65532970e-01
2.02372804e-01 -1.19633220e-01 -1.13789129e+00 4.25685734e-01
-3.25583637e-01 3.34961563e-01 2.73742288e-01 -2.16669649e-01
-1.04217380e-01 -3.88554722e-01 -1.63004827e+00 1.31569907e-01
6.82694972e-01 -5.92733286e-02 -4.47910547e-01 5.73360734e-02
4.35502172e-01 -4.52019632e-01 -6.13021493e-01 1.03029168e+00
8.22862744e-01 -1.01886439e+00 1.25823176e+00 -9.76439044e-02
5.39403260e-01 -1.25283286e-01 -7.92644799e-01 -1.47198021e+00
-4.46743011e-01 2.00952198e-02 3.60130638e-01 5.89853942e-01
1.15096994e-01 -1.11992788e+00 5.91153204e-01 -8.04599077e-02
-5.25721848e-01 -6.12229764e-01 -9.43521738e-01 -3.71761322e-01
-2.88546272e-03 -3.00387412e-01 7.62501895e-01 9.34242725e-01
-8.69389951e-01 -5.25095277e-02 -1.73083067e-01 8.45506549e-01
1.31865203e+00 2.88450241e-01 2.46337175e-01 -1.50899315e+00
-1.44616112e-01 3.22880477e-01 -6.77234605e-02 -1.12206089e+00
-4.94188033e-02 -6.18658125e-01 7.10037351e-01 -1.68146574e+00
-1.03939725e-02 -1.03000546e+00 -4.01443541e-01 1.42612740e-01
1.12426795e-01 1.16723955e-01 -6.91140354e-01 -5.05711734e-02
1.02848165e-01 7.92487919e-01 1.54549670e+00 -4.18414444e-01
-2.46783178e-02 1.52973130e-01 -5.17896831e-01 7.60353804e-01
5.00105560e-01 -4.63976890e-01 -3.12917441e-01 -9.73710060e-01
1.87698781e-01 5.07991493e-01 6.19383395e-01 -1.18823922e+00
-5.03227115e-02 3.75690274e-02 7.70746112e-01 -8.56711388e-01
3.43085349e-01 -9.13311481e-01 3.53389412e-01 4.24292386e-01
3.54615510e-01 -5.37022591e-01 3.39755744e-01 5.58934629e-01
-4.49035943e-01 -1.59401402e-01 1.14376652e+00 -3.72557640e-01
-4.41557169e-01 3.96145076e-01 -3.31875354e-01 -2.18113109e-01
5.36348224e-01 -4.39922929e-01 3.43467332e-02 -3.49450409e-01
-5.87275207e-01 -2.48817075e-02 -2.83786982e-01 -2.57285148e-01
9.62565303e-01 -1.22585630e+00 -1.09015501e+00 6.51773870e-01
-3.64255697e-01 6.40522003e-01 1.01525593e+00 9.02828217e-01
-8.10157239e-01 -5.31434752e-02 -4.38899547e-01 -5.61271250e-01
-5.46171129e-01 -1.61500558e-01 8.56452763e-01 -3.20825338e-01
-8.11045647e-01 9.81964886e-01 4.26798493e-01 -8.00823331e-01
-2.89454125e-02 -2.78114080e-01 -1.66698858e-01 -7.58210346e-02
5.33974111e-01 4.90165800e-01 6.20726585e-01 -4.77569491e-01
-1.50725633e-01 4.61776793e-01 8.51240382e-03 1.03423275e-01
2.21862507e+00 -9.39480066e-02 -2.11557537e-01 -1.08223110e-01
1.15273225e+00 -4.02959347e-01 -1.11432183e+00 -3.01049083e-01
-5.63290298e-01 -2.80466765e-01 6.59399867e-01 -7.11134315e-01
-1.47501051e+00 8.38487685e-01 8.52864981e-01 -1.13675609e-01
1.26256645e+00 -6.93992794e-01 9.66942906e-01 7.55437493e-01
5.76943994e-01 -7.02686906e-01 -1.35742426e-01 7.88408697e-01
9.91073191e-01 -8.89088929e-01 3.54082100e-02 -2.44438171e-01
-1.23741217e-02 1.08983850e+00 3.43822062e-01 -3.07880729e-01
1.10164523e+00 7.34171629e-01 1.35695800e-01 -3.93663824e-01
1.27033936e-02 5.20002246e-01 -4.82949078e-01 9.58701432e-01
-7.70128742e-02 -2.20664628e-02 1.69668660e-01 9.22696650e-01
3.50479893e-02 2.62706369e-01 2.67285019e-01 1.12541103e+00
-6.56084001e-01 -2.43491337e-01 -8.10931742e-01 5.99546850e-01
-3.45247358e-01 -1.74729899e-01 7.62653172e-01 6.20619655e-01
2.27580026e-01 6.64453864e-01 -1.70064741e-03 -4.92675193e-02
4.57189322e-01 -4.45173115e-01 4.78715450e-01 -3.21917325e-01
5.88473305e-02 1.30466580e-01 4.92978618e-02 -2.83430159e-01
-3.64800602e-01 -4.69316006e-01 -1.22400975e+00 -1.02316208e-01
-4.72256303e-01 3.36371481e-01 6.44021571e-01 8.49764824e-01
-1.06599808e-01 6.38973653e-01 6.24961555e-01 -1.32879853e+00
-7.35828280e-01 -1.23879731e+00 -1.36379433e+00 -9.57620665e-02
7.42745042e-01 -1.05028319e+00 -5.76192796e-01 -4.50215966e-01] | [6.894832134246826, 1.5023306608200073] |
f9722646-9130-4b72-937d-23586fef0b20 | deepsinger-singing-voice-synthesis-with-data | 2007.04590 | null | https://arxiv.org/abs/2007.04590v2 | https://arxiv.org/pdf/2007.04590v2.pdf | DeepSinger: Singing Voice Synthesis with Data Mined From the Web | In this paper, we develop DeepSinger, a multi-lingual multi-singer singing voice synthesis (SVS) system, which is built from scratch using singing training data mined from music websites. The pipeline of DeepSinger consists of several steps, including data crawling, singing and accompaniment separation, lyrics-to-singing alignment, data filtration, and singing modeling. Specifically, we design a lyrics-to-singing alignment model to automatically extract the duration of each phoneme in lyrics starting from coarse-grained sentence level to fine-grained phoneme level, and further design a multi-lingual multi-singer singing model based on a feed-forward Transformer to directly generate linear-spectrograms from lyrics, and synthesize voices using Griffin-Lim. DeepSinger has several advantages over previous SVS systems: 1) to the best of our knowledge, it is the first SVS system that directly mines training data from music websites, 2) the lyrics-to-singing alignment model further avoids any human efforts for alignment labeling and greatly reduces labeling cost, 3) the singing model based on a feed-forward Transformer is simple and efficient, by removing the complicated acoustic feature modeling in parametric synthesis and leveraging a reference encoder to capture the timbre of a singer from noisy singing data, and 4) it can synthesize singing voices in multiple languages and multiple singers. We evaluate DeepSinger on our mined singing dataset that consists of about 92 hours data from 89 singers on three languages (Chinese, Cantonese and English). The results demonstrate that with the singing data purely mined from the Web, DeepSinger can synthesize high-quality singing voices in terms of both pitch accuracy and voice naturalness (footnote: Our audio samples are shown in https://speechresearch.github.io/deepsinger/.) | ['Tie-Yan Liu', 'Yi Ren', 'Zhou Zhao', 'Xu Tan', 'Tao Qin', 'Jian Luan'] | 2020-07-09 | null | null | null | null | ['singing-voice-synthesis'] | ['speech'] | [-0.04668039 -0.39242712 0.0787253 -0.10447875 -1.5172086 -0.99946386
0.08901045 -0.77775896 0.07694104 0.26530343 0.5786697 -0.17955017
0.06036597 -0.2675305 -0.49941227 -0.51033044 0.33105296 0.465656
-0.15624598 -0.307801 -0.2009144 0.13566008 -1.6965176 0.3426024
0.56762284 0.70885205 0.38454092 1.281238 0.04631856 0.49294135
-0.8720796 -0.05073847 0.42250988 -1.0102679 -0.60560834 -0.07741708
0.64211524 -0.3494543 -0.06306673 0.5717342 1.0324353 0.07292581
0.21343906 -0.85925555 -0.34242362 1.1893814 0.05945577 -0.15913
0.33911714 0.5096363 1.5220526 -1.0743015 0.23369181 1.1235229
0.8542427 0.7834003 -1.1483234 -0.98595077 -0.7361372 -0.05534118
-1.3383266 -1.1341617 0.923357 -0.42184493 0.83205855 0.64362
0.7307622 1.18101 -0.59282523 0.8482677 1.0007033 -0.5156946
-0.17747858 -0.3874952 -0.2792337 0.435614 -0.8710704 0.21587223
-1.0216564 -0.11667336 0.73647505 -0.64333874 -0.23650365 0.6158789
-1.3033309 0.64036196 -0.28983924 0.5040235 -0.1727577 0.17395444
0.54652095 0.4312962 0.17775092 0.71681243 -0.43180993 -0.57412565
-1.5715235 0.42733157 1.0054001 0.8164957 0.29188728 0.70651156
-0.2520785 1.3800312 0.10540292 0.6715811 0.79971284 -1.2541658
0.42192873 -0.27011004 -0.2184942 -0.19274104 -0.12260499 -0.5117565
-0.42255434 -0.2687113 0.3342943 -0.32039925 -0.51918554 1.8952355
0.09951689 0.3551668 -0.14949208 1.1111581 1.1427776 0.9107765
-0.51152605 -0.49528563 1.3114239 -1.339473 -0.91470164 -0.02983097
0.15479554 -1.2412184 1.7171445 0.56374353 -1.4766771 -1.0886848
-0.9775425 -0.41152367 0.27267972 0.42186478 0.314143 0.5715015
-0.9530711 0.7579215 -0.56090206 0.12258543 -0.10795894 0.10838177
-0.11227018 0.50270426 -1.3096778 0.23707424 0.08031049 -0.13054447
-1.2985234 -0.90182436 -0.70827997 0.03290418 0.2921665 -0.5670805
1.8606522 -0.7450319 -2.1066003 0.60915583 -0.32911405 -0.4235352
0.09386193 -0.3473056 -0.73882765 -0.09303605 0.14321654 0.29701158
1.1258948 -0.7494375 -0.517714 0.02027192 -0.6818254 0.366081
-0.39387587 0.39836165 -0.4389017 -0.9904146 -0.07298394 -0.86328816
0.3131911 -0.7566883 -0.59730965 -0.2689444 0.7773582 -1.1388707
1.7507329 -2.3345475 0.2400934 -0.21791004 -0.12115644 0.48803398
-0.2593247 0.55064446 0.05540968 -0.12467923 -0.29543895 -0.5989371
0.18866113 0.09501285 -0.68137467 0.18002312 -0.03551253 0.90226936
-0.8853845 -0.3212061 0.19308968 0.5016259 -0.64264715 0.7235372
-0.14109343 0.6582673 0.340028 0.7679275 0.07992855 0.52802527
0.07656269 -0.16343151 -0.42301205 1.2727834 -1.2002432 1.9272822
-0.90645903 0.4641657 0.5562056 -0.39870933 1.1371027 0.8459122
0.42135504 -0.1518158 -0.11652976 0.7561197 0.20294505 -0.42034903
0.53191227 -0.5134316 -0.17960463 0.54334724 0.5926444 -0.94598544
0.23392129 -0.33739582 1.0431288 0.19602707 0.04972959 0.04637615
0.35672116 -0.26843664 0.7190747 0.34234017 0.1696722 0.96138036
-0.01264156 -0.02427167 -1.06213 -1.23503 -0.02042777 1.4209087
-0.8707002 -0.9805519 -1.0602893 -0.04860935 -0.18282022 0.723916
0.09697402 0.28667682 -0.9071493 -0.14602807 1.363018 0.2722909
0.12541796 -1.571483 0.05023754 0.53514594 -0.5677841 -0.83049935
-1.4098084 0.13524483 -0.5852356 -0.7855918 -0.5593453 -0.8910149
-0.3633423 -0.03346311 1.1254326 -0.22560816 -0.28049737 0.05236305
-0.10961031 -0.4601792 -0.8886446 0.32251593 0.5699599 -0.06665452
0.18004563 -0.8602314 -0.3120238 0.27794746 -0.48345184 0.08654594
0.28145885 0.8292662 0.841716 0.08366379 0.9713046 -0.54966
0.8126287 -0.21003488 -0.5423507 -0.31145272 -0.28651476 -0.41952455
1.2196084 -0.40477726 -0.67444915 0.18765575 -0.74610865 -0.65178734
-0.1142288 0.1528006 -0.39505607 0.68442315 0.672559 0.42144397
0.04531477 -1.0343488 0.59228754 1.4279686 1.2262785 -0.49746773
0.9165336 -0.02972809 -0.43254504 -1.2292477 -0.9636165 -0.6546279
-0.54417086 -0.137098 0.6270953 -1.0263445 -0.8338375 0.60069585
-0.9889238 -0.3997906 -0.7529835 0.5883548 -1.0100482 0.23409155
-1.0822806 -0.95613194 -0.9348202 -0.97859514 1.0528992 -0.04441976
-0.6250902 -0.50876415 0.29523212 0.7088591 0.37080812 -0.13986643
0.44700503 -0.4680494 -0.15009812 0.16357271 0.5839787 0.95332915
0.43072203 0.07237024 -1.44815 -0.190969 0.30171368 -0.48773465
0.5965052 0.43925077 1.0414449 -0.60898286 0.5967185 0.9272783
0.5731705 -0.06327704 0.32296088 -0.26822793 0.83881223 0.585656
0.6476705 0.2694701 0.21211767 0.82222307 0.06708173 -0.06908836
-0.853457 -0.743617 0.9107043 2.0675735 -0.00696716 0.17075002
-0.35188955 0.9487861 -1.352855 -1.075313 -0.17832924 2.268779
1.4657149 -0.2518265 0.67832065 0.5670218 0.6795667 0.37570763
-0.5147769 -0.61031973 -0.15872173 0.76245695 0.05776954 0.6896729
-0.90691495 1.2485498 5.860519 1.1775146 -1.081271 0.27768517
-0.00733838 -0.7159061 -0.3909024 -0.05250555 -0.95831466 0.50396824
1.3532513 0.06216619 1.3398536 0.72287846 0.5735379 0.73092216
-1.0408729 1.1759119 0.07189588 -1.3206288 -0.20389211 -0.2271131
0.48176566 0.19730356 0.0072101 0.3875669 0.07711559 -1.2479751
1.1218216 0.17896703 1.2737492 -0.78917044 0.17856877 0.39419863
-1.4893556 0.1462313 0.13599165 0.0731278 0.47488004 0.38831934
-1.3068055 0.43970227 0.5077078 0.67823464 0.03748668 0.5158193
-0.4828958 1.5708735 -0.37361076 0.39398262 -0.05081039 -0.17376198
0.8249016 1.4603578 0.58070207 -0.15998602 -0.06305571 1.0198892
-0.19961567 0.2804256 -0.28695747 -0.55583274 0.8812184 1.5051594
0.03488271 -0.04597095 -0.10964167 0.79894704 -0.1524682 0.06593215
-0.6070159 -0.46247774 0.7772437 0.33543164 0.24956326 -0.43571794
-0.14257693 -0.730708 -0.11133539 -1.3375596 0.19507614 -0.59787154
-1.2503669 0.8097736 -0.6285439 -1.3304977 -0.90099126 -0.42261508
-0.80011016 1.3014592 -1.2314711 -1.1863348 0.13401628 0.6435457
1.0572287 -0.4339713 1.152156 0.46904096 -0.32293114 0.4813104
-0.28622562 0.2368429 0.9899119 -1.4366566 0.8783845 0.6584873
0.8263371 0.47824746 0.5534389 -0.29236302 -1.4898521 -1.0308768
1.0899074 -0.5524947 0.799722 -0.687976 -0.72428346 0.32966143
0.1831249 -0.184651 1.1062539 0.15115476 -0.2617848 -0.25250015
-0.5379935 0.47994605 1.0622439 -1.1175054 -0.7287895 0.29613832
0.83697045 -0.51333135 -1.0352864 0.1949169 0.7522637 -0.9486798
0.75563097 -0.33220595 0.27883345 -0.65962 -0.23701188 -1.59313
-0.14636351 -1.4510041 -0.27933535 1.7184776 0.44597712 -0.0590203
0.21104185 -0.3652162 -0.7248768 -0.29792258 -0.8867016 -0.8920533
-0.13585997 -0.84942454 0.86790746 0.8654043 -0.16984348 0.765232
-0.7931466 -0.0139866 0.3903069 0.37378883 1.0302992 -0.91423887
-1.1199472 -0.23762578 0.492937 -1.0097572 0.13505512 -1.0850359
0.3490993 -1.1739821 -0.424204 -0.34510592 -0.16955224 0.638751
0.05918891 0.41775262 0.38991174 0.36412662 -0.07600809 0.4851908
1.306454 0.10644856 -0.5346815 0.4213314 -0.5062955 0.81889427
0.754746 -0.33379105 -0.1823031 -0.28418013 -0.18736812 0.49200773
0.02388235 -0.99686724 0.14760977 0.17255893 -0.21186772 -0.66735315
0.82019824 -0.16063164 0.3411801 0.158738 -0.24357207 -0.3490793
0.14405708 -0.24136195 -0.4984178 -0.2511497 0.6957106 -0.1281262
-0.13530283 0.13631034 -0.2727554 0.30526686 0.14003298 0.14965706
-0.05378369 -0.5174676 -0.69208753 -0.29864144 0.06577252 0.64636
0.23356356 -1.5842239 -1.092792 0.53372246 -0.03192634 -0.04977624
0.21422714 0.599632 -0.20101358 0.36221734 0.2523095 -0.44764003
-1.3452725 0.10505536 0.20711279 0.12013377 -0.48966068 0.9497232
-0.0496969 -0.8778574 0.18297215 -0.29712555 0.0743908 0.04431647
0.5561793 0.5369634 0.11432336 -0.8682318 -0.17761885 0.4612618
0.44654703 -0.5517383 1.2469269 -0.02476994 -0.1079205 0.99417806
0.94193876 0.8901586 -1.1871935 0.01289273 -0.27624425 -0.20481704
0.16334867 -0.8199442 -0.6689195 0.94757855 -0.04878701 0.35077512
1.1689922 0.03916403 1.4765087 -0.06427211 -0.02926036 -1.3532197
-0.12011698 0.7568779 1.2013396 -0.6983484 -0.63177633 -0.08841204
-0.8001718 1.1492469 0.2347016 -0.07678234 0.41205165 0.64366305
0.43893528 0.06866016 -0.6961837 -0.2792377 0.46292415 0.2776666
0.83970344 0.43884045 -0.09271121 1.233037 -1.3600568 -0.18551423
0.13298248 -0.07739242 -0.3293365 -1.4260126 -0.6076042 0.16418067
-0.62588894 -0.65671074 -0.58421314 0.19893426 0.27101636 1.2812285
-0.06293708 -0.6500608 0.39785346 0.37254775 0.35552284 -0.9714909
-1.040466 0.8110771 0.36254922 -0.38748458 0.01197302 -0.6474865
-1.2356181 -0.02935161 -0.30861765 0.43162537 0.7644966 0.72210807
0.2361505 0.81982 1.0995994 -0.9838881 -0.6941159 -1.3869083
-0.90983206 0.13915022 0.43931338 0.0243753 -0.6311956 0.13783814] | [15.504436492919922, 6.106106758117676] |
a6be4b14-7145-458f-b864-a0ebcd8c0297 | diffusion-and-volume-maximization-based | 2203.09992 | null | https://arxiv.org/abs/2203.09992v3 | https://arxiv.org/pdf/2203.09992v3.pdf | Unsupervised Diffusion and Volume Maximization-Based Clustering of Hyperspectral Images | Hyperspectral images taken from aircraft or satellites contain information from hundreds of spectral bands, within which lie latent lower-dimensional structures that can be exploited for classifying vegetation and other materials. A disadvantage of working with hyperspectral images is that, due to an inherent trade-off between spectral and spatial resolution, they have a relatively coarse spatial scale, meaning that single pixels may correspond to spatial regions containing multiple materials. This article introduces the Diffusion and Volume maximization-based Image Clustering (D-VIC) algorithm for unsupervised material clustering to address this problem. By directly incorporating pixel purity into its labeling procedure, D-VIC gives greater weight to pixels that correspond to a spatial region containing just a single material. D-VIC is shown to outperform comparable state-of-the-art methods in extensive experiments on a range of hyperspectral images, including land-use maps and highly mixed forest health surveys (in the context of ash dieback disease), implying that it is well-equipped for unsupervised material clustering of spectrally-mixed hyperspectral datasets. | ['James M. Murphy', 'David A. Coomes', 'Aland H. Y. Chan', 'Robert J. Plemmons', 'Kangning Cui', 'Sam L. Polk'] | 2022-03-18 | null | null | null | null | ['image-clustering'] | ['computer-vision'] | [ 7.72829056e-01 -2.72026956e-01 -1.53168216e-01 -5.04131280e-02
-4.15613919e-01 -5.93473434e-01 4.04409260e-01 7.59915635e-02
-8.44068080e-02 6.51479840e-01 -8.06303024e-02 -1.93784252e-01
-5.18197417e-01 -1.10808992e+00 -8.00179094e-02 -1.09135699e+00
-3.09938401e-01 6.15484834e-01 -6.51359111e-02 2.31363606e-02
-1.59833506e-02 7.39346743e-01 -1.79591441e+00 2.14610711e-01
1.02441072e+00 8.50063562e-01 6.58248305e-01 3.12031239e-01
-2.24139586e-01 1.79926157e-01 -1.79131761e-01 3.02367270e-01
5.01952052e-01 -3.65198076e-01 -7.86133766e-01 9.70417619e-01
5.35890698e-01 -6.63970113e-02 2.93288171e-01 1.51483977e+00
-6.48632571e-02 -6.39226800e-03 1.11004806e+00 -9.61939573e-01
-3.00161660e-01 4.01318103e-01 -1.07251501e+00 -1.81127816e-01
-2.29593545e-01 -2.00336669e-02 1.24682677e+00 -6.71711624e-01
5.86956918e-01 1.00684345e+00 5.45553923e-01 -8.00919831e-02
-1.58536673e+00 -1.79605022e-01 1.10083684e-01 -9.28128883e-02
-1.56700504e+00 -1.45188153e-01 6.29372478e-01 -8.80485535e-01
6.64135218e-01 6.62787676e-01 8.09972644e-01 5.48922196e-02
-1.78715929e-01 5.41543365e-01 1.31412172e+00 -2.53562093e-01
4.28091019e-01 1.18491836e-02 1.75682560e-01 2.71533221e-01
4.53162938e-01 1.19546056e-01 1.42947122e-01 -2.46013850e-01
4.53384697e-01 1.60750225e-01 -4.70793843e-01 -6.83706284e-01
-9.75196123e-01 7.50656545e-01 7.71711946e-01 4.83882606e-01
-8.03253233e-01 -5.91372848e-01 6.79831719e-03 -7.72452727e-02
6.92307711e-01 6.44415438e-01 -1.83670044e-01 5.96393168e-01
-1.46268415e+00 -3.40288542e-02 3.86544764e-01 4.80984867e-01
1.23677838e+00 1.51650906e-02 2.56166667e-01 1.21744382e+00
3.65936786e-01 7.62672365e-01 2.45993435e-02 -9.94425535e-01
6.93685338e-02 8.52388799e-01 4.11871001e-02 -9.29777205e-01
-4.99343723e-01 -6.48630142e-01 -1.01924765e+00 5.18631637e-01
1.89978331e-02 1.81774393e-01 -1.16470385e+00 1.34783292e+00
2.43115395e-01 -3.03635001e-01 2.77746141e-01 1.00755763e+00
4.30237889e-01 8.19703043e-01 -9.88663081e-03 -5.93998134e-01
1.25087762e+00 -6.68192625e-01 -2.76221246e-01 -5.26665628e-01
2.66994447e-01 -3.88997763e-01 7.81916678e-01 3.88598293e-01
-5.39867163e-01 -2.13577211e-01 -9.71223533e-01 6.62100077e-01
-5.88627279e-01 -1.37009155e-02 5.80684721e-01 7.74153292e-01
-8.21313858e-01 4.34518546e-01 -5.18864632e-01 -5.93153715e-01
5.14440477e-01 5.30616567e-02 -2.78162301e-01 -8.58063325e-02
-6.27791047e-01 5.26178479e-01 6.04356647e-01 1.78326532e-01
-5.94867468e-01 -5.97571492e-01 -6.01664484e-01 1.08578406e-01
3.13294411e-01 -1.32851928e-01 5.19139171e-01 -1.09744751e+00
-9.46423471e-01 1.20100379e+00 -1.18766382e-01 -2.73130506e-01
6.78763390e-02 3.34320724e-01 -4.72232550e-01 4.93996024e-01
3.43957007e-01 6.19395137e-01 8.36381078e-01 -1.85356700e+00
-7.33421683e-01 -8.54461372e-01 -3.35911393e-01 5.18554807e-01
-4.45089310e-01 -2.15432823e-01 6.49463460e-02 -3.76695484e-01
7.99212098e-01 -1.05004168e+00 -4.22749490e-01 -2.07763523e-01
-4.92593765e-01 3.96610707e-01 1.01277542e+00 -4.82415169e-01
8.83144379e-01 -2.29656911e+00 2.78970510e-01 5.65938532e-01
1.83599710e-01 2.40037650e-01 -7.94863850e-02 3.49138707e-01
-2.35457152e-01 2.30129465e-01 -1.06337976e+00 2.72637725e-01
-9.40812007e-02 2.95805603e-01 -6.30066916e-02 5.87686896e-01
1.33243531e-01 2.42821068e-01 -9.35913622e-01 -3.26278895e-01
4.38074052e-01 1.63861126e-01 5.37226275e-02 -3.13720852e-01
-3.36570203e-01 1.88269988e-01 -2.04287052e-01 9.19573128e-01
1.25472236e+00 -1.91826597e-01 3.21824372e-01 -2.58002698e-01
-5.50822198e-01 -2.84712911e-01 -1.15602124e+00 1.24593997e+00
1.25106394e-01 6.09418809e-01 6.72057092e-01 -1.04874396e+00
6.09327495e-01 7.66351447e-02 9.02319670e-01 -3.14197838e-01
-2.38744318e-01 1.47946522e-01 -5.93270138e-02 -2.30562180e-01
6.04931474e-01 -3.24882329e-01 4.90474194e-01 4.17232513e-01
-2.66797215e-01 -5.34313083e-01 3.07955563e-01 -1.52737051e-01
7.13931799e-01 -3.20098251e-01 1.30767554e-01 -8.18585634e-01
3.86042953e-01 5.43559611e-01 3.79205883e-01 3.83154929e-01
-1.82957262e-01 3.47969979e-01 7.86837339e-02 3.39964293e-02
-9.84347403e-01 -1.10478544e+00 -7.03695953e-01 9.35240388e-01
1.41592070e-01 7.24340305e-02 -5.96008658e-01 -8.86387080e-02
3.66602391e-02 5.94537795e-01 -3.81766617e-01 2.28319317e-01
2.43095726e-01 -1.41184151e+00 1.71726078e-01 3.92470770e-02
5.68234384e-01 -8.73960078e-01 -5.05538404e-01 6.49919063e-02
-1.86918184e-01 -8.79204094e-01 2.42123589e-01 3.53905529e-01
-1.07426953e+00 -1.10745728e+00 -8.14437568e-01 -7.00803876e-01
8.20766449e-01 8.21495712e-01 9.15618598e-01 -3.31271857e-01
-4.94977206e-01 3.32865864e-01 -4.62197006e-01 -1.85047209e-01
-3.48573983e-01 1.56883094e-02 -2.30624139e-01 2.63222009e-01
4.24250454e-01 -5.36292136e-01 -3.55144560e-01 4.16207671e-01
-1.19070470e+00 5.88461058e-03 5.84449768e-01 6.75611854e-01
8.85704458e-01 9.42182422e-01 -7.09252506e-02 -7.98408866e-01
2.41954401e-01 -4.66337532e-01 -6.51256025e-01 3.74324352e-01
-6.93200529e-01 -3.17719191e-01 1.35991186e-01 8.09004456e-02
-8.69704902e-01 3.16493779e-01 5.68755507e-01 -1.65922672e-01
-5.22957265e-01 1.06927931e+00 -1.36999875e-01 -3.98286954e-02
8.33626330e-01 3.13766122e-01 2.52643198e-01 -2.28086129e-01
3.52894843e-01 9.79566276e-01 4.90567386e-01 -1.46061644e-01
8.38347971e-01 7.61874557e-01 1.67344823e-01 -1.74784768e+00
-6.30246043e-01 -8.22766840e-01 -7.62684762e-01 -3.75987530e-01
8.35967958e-01 -8.33962739e-01 -1.32782996e-01 5.79242647e-01
-3.62186491e-01 -2.20977232e-01 -2.49349505e-01 5.51522315e-01
-3.65463734e-01 5.49648881e-01 -2.50674427e-01 -1.11006808e+00
-8.72449130e-02 -1.05071342e+00 8.96694064e-01 -4.53576781e-02
1.95454493e-01 -9.56423521e-01 -7.09908828e-02 4.68160033e-01
2.05094159e-01 3.62879395e-01 1.02045953e+00 4.51409742e-02
-5.11858404e-01 4.52894426e-04 -4.91426647e-01 5.43999732e-01
5.71433663e-01 3.18879783e-01 -1.06766069e+00 -3.56957972e-01
1.65809304e-01 -6.32829517e-02 1.23776233e+00 8.68949592e-01
9.44574416e-01 -7.84407780e-02 -3.04867715e-01 6.22800946e-01
1.70396435e+00 2.27219075e-01 7.43548274e-01 3.51559818e-01
6.13161683e-01 1.16707301e+00 6.41906381e-01 3.88990343e-01
-2.27185503e-01 9.75117385e-02 9.05757844e-01 -6.61570907e-01
2.99339026e-01 5.37320554e-01 6.04294688e-02 2.57834166e-01
-2.28360042e-01 -1.59991518e-01 -1.02600217e+00 6.31034255e-01
-1.57310796e+00 -1.42094934e+00 -6.02023721e-01 2.27684569e+00
5.33570349e-01 -3.42197001e-01 2.18060493e-01 4.35869068e-01
1.13445103e+00 4.53019142e-01 -6.36749685e-01 2.85740823e-01
-6.36578023e-01 -2.66252086e-02 9.50562358e-01 4.48342592e-01
-1.43362749e+00 8.75647426e-01 6.88233614e+00 6.03065610e-01
-1.08667803e+00 -3.50552231e-01 6.05286717e-01 1.87688753e-01
-2.04446480e-01 -1.06484443e-02 -2.81057000e-01 1.51674777e-01
5.02817690e-01 1.47644192e-01 6.14706874e-01 6.23387575e-01
3.20719570e-01 -4.83505100e-01 -4.04659897e-01 8.41370344e-01
-2.76650488e-01 -1.00277352e+00 4.52308282e-02 5.49080908e-01
8.80245626e-01 3.10352147e-01 1.49362311e-01 -5.73707640e-01
3.27136636e-01 -9.29027677e-01 3.96641791e-01 1.77978843e-01
8.06437731e-01 -5.74449837e-01 3.93138438e-01 2.95222968e-01
-1.32521200e+00 -1.02904029e-01 -6.28829300e-01 1.85168818e-01
-1.86902195e-01 1.01942480e+00 -6.33067548e-01 7.27118492e-01
7.61764944e-01 9.18978274e-01 -3.72438133e-01 1.08270061e+00
1.64240375e-01 5.12281060e-01 -5.08610487e-01 4.26699311e-01
5.39636612e-01 -8.93599749e-01 5.85487008e-01 1.08289599e+00
4.69468266e-01 9.52661708e-02 4.08223808e-01 1.01292884e+00
4.59631264e-01 1.87676460e-01 -6.36218011e-01 -6.06433332e-01
4.09319431e-01 1.42242360e+00 -1.12239099e+00 -2.38778964e-01
-2.70996600e-01 7.39341736e-01 -3.33425671e-01 4.61879581e-01
-3.18942517e-01 2.83138221e-03 6.15875125e-01 2.68496811e-01
3.11369956e-01 -2.70823181e-01 -4.38459992e-01 -7.76620269e-01
-2.54555970e-01 -7.03016579e-01 3.41201127e-01 -1.02354753e+00
-1.39700234e+00 5.57045400e-01 3.52075845e-02 -1.61177170e+00
2.55363196e-01 -8.82848740e-01 -1.08819805e-01 9.64053512e-01
-1.55069172e+00 -1.06664526e+00 -6.50002062e-01 5.37571549e-01
1.74147591e-01 -2.67943025e-01 1.00364220e+00 -7.26518333e-02
-4.40514982e-01 -4.85112548e-01 6.93537116e-01 -1.65335581e-01
2.89685816e-01 -1.31232250e+00 -3.33382308e-01 8.64814699e-01
1.33088389e-02 5.86267076e-02 5.83446920e-01 -6.48049116e-01
-9.70541060e-01 -1.29140794e+00 1.65638819e-01 3.50398660e-01
6.54364705e-01 3.09968889e-01 -6.88614786e-01 2.73167223e-01
6.24940023e-02 -4.13307995e-01 1.09498954e+00 5.84009513e-02
-2.02313602e-01 -1.89746439e-01 -1.29866397e+00 2.15150461e-01
7.12917387e-01 -6.31030142e-01 -1.28886953e-01 8.18331480e-01
-2.20034309e-02 4.06939358e-01 -9.87345934e-01 5.25333643e-01
2.16054410e-01 -1.12566900e+00 9.45224643e-01 1.97200361e-03
3.43670368e-01 -5.52714765e-01 -5.44939458e-01 -1.45943260e+00
-7.91626632e-01 -4.11260054e-02 6.55297935e-01 9.15040553e-01
4.67481166e-01 -5.90564370e-01 8.89753878e-01 1.44189343e-01
-1.00554988e-01 1.37452543e-01 -5.40264726e-01 -9.94785190e-01
-7.15158433e-02 -1.36040449e-01 5.46672881e-01 1.18257964e+00
-2.12913290e-01 2.22587287e-01 2.87910495e-02 5.45718849e-01
1.18132377e+00 5.52803636e-01 2.69711316e-01 -2.00844598e+00
7.49409432e-03 -8.12161088e-01 -3.66963863e-01 -5.16025722e-01
1.14499092e-01 -8.86943221e-01 2.52220660e-01 -1.61330712e+00
3.52433413e-01 -5.39826632e-01 -1.54394865e-01 4.66499954e-01
-6.88729659e-02 3.40406060e-01 -1.01858616e-01 6.05203569e-01
9.49047431e-02 3.27265054e-01 1.03420639e+00 -6.25502348e-01
-5.03044009e-01 1.02656027e-02 -5.48301637e-01 5.33113897e-01
8.00166249e-01 -2.06334680e-01 -4.21042234e-01 -2.65112817e-01
-1.44442201e-01 -4.08174545e-02 3.97527248e-01 -1.06837535e+00
-2.20034763e-01 -5.79444230e-01 2.28690252e-01 -1.04088330e+00
4.42775369e-01 -1.20275819e+00 6.33178830e-01 4.68684077e-01
-1.13154221e-02 -7.35012114e-01 6.47743940e-02 3.16489369e-01
-2.96671271e-01 -3.87456924e-01 1.16335702e+00 -4.07355607e-01
-9.71438110e-01 3.67685378e-01 -8.43095064e-01 -4.39842254e-01
1.06929708e+00 -6.40626013e-01 -3.89503032e-01 -7.76335150e-02
-7.20608354e-01 -1.73275098e-02 9.28184807e-01 -1.40735894e-01
3.73333275e-01 -1.01065445e+00 -8.12831581e-01 2.79734999e-01
3.89597595e-01 2.11457182e-02 2.48910427e-01 6.85009599e-01
-9.50751543e-01 3.87848884e-01 -4.52002823e-01 -1.12604010e+00
-1.39958334e+00 4.15757954e-01 4.39665645e-01 -4.21875715e-02
-3.96770120e-01 6.06467962e-01 3.60447228e-01 -4.22096938e-01
-1.76632658e-01 -7.91717693e-02 -2.99261838e-01 4.59506035e-01
2.55964458e-01 4.11283851e-01 4.95524444e-02 -1.07328022e+00
-3.24285656e-01 6.20308816e-01 4.62644786e-01 -2.18731984e-01
1.54936540e+00 -8.07769150e-02 -8.27059865e-01 4.97974724e-01
8.80760789e-01 -2.62027293e-01 -1.16224611e+00 -2.23717794e-01
-3.91592979e-02 -5.76593280e-01 7.24732637e-01 -7.21766412e-01
-1.24607790e+00 6.39531255e-01 7.79153883e-01 8.95555198e-01
1.47894585e+00 -1.37601823e-01 -8.11661184e-02 3.59125644e-01
3.12239438e-01 -1.21060479e+00 -4.73375112e-01 2.35817671e-01
6.19624376e-01 -1.26651847e+00 4.38566267e-01 -7.62130022e-01
-5.84422708e-01 1.04718268e+00 8.84206742e-02 1.90149397e-01
7.43643761e-01 9.49310511e-02 9.89553928e-02 -4.90281314e-01
-2.30195895e-02 -7.98636198e-01 3.10693592e-01 8.67956102e-01
1.60442188e-01 6.86293006e-01 2.31043883e-02 -3.40210110e-01
-7.90871456e-02 -4.37779158e-01 2.44604483e-01 8.95442605e-01
-1.07077098e+00 -8.00000906e-01 -8.68867457e-01 8.34670067e-01
4.36792895e-02 -1.55579582e-01 -7.64737487e-01 5.04391134e-01
1.03316242e-02 1.08331168e+00 1.52562320e-01 -2.96486646e-01
-1.36681095e-01 -5.17835431e-02 3.39115500e-01 -7.09909856e-01
-1.13076083e-01 6.20634735e-01 -1.48769662e-01 -1.89804628e-01
-1.18942440e+00 -8.39044690e-01 -9.69488561e-01 -2.27883518e-01
-5.65504074e-01 2.19986349e-01 7.28088796e-01 6.73985541e-01
-2.09964011e-02 2.83171833e-01 7.50466764e-01 -1.07644403e+00
-9.01733562e-02 -8.85796487e-01 -1.58396924e+00 2.10390657e-01
2.19568521e-01 -5.46485841e-01 -4.58634973e-01 9.06576663e-02] | [9.891876220703125, -1.868166446685791] |
82cdba52-746c-4201-9316-acd87054e8c0 | interactive-open-ended-learning-for-3d-object | 1912.09539 | null | https://arxiv.org/abs/1912.09539v1 | https://arxiv.org/pdf/1912.09539v1.pdf | Interactive Open-Ended Learning for 3D Object Recognition | The thesis contributes in several important ways to the research area of 3D object category learning and recognition. To cope with the mentioned limitations, we look at human cognition, in particular at the fact that human beings learn to recognize object categories ceaselessly over time. This ability to refine knowledge from the set of accumulated experiences facilitates the adaptation to new environments. Inspired by this capability, we seek to create a cognitive object perception and perceptual learning architecture that can learn 3D object categories in an open-ended fashion. In this context, ``open-ended'' implies that the set of categories to be learned is not known in advance, and the training instances are extracted from actual experiences of a robot, and thus become gradually available, rather than being available since the beginning of the learning process. In particular, this architecture provides perception capabilities that will allow robots to incrementally learn object categories from the set of accumulated experiences and reason about how to perform complex tasks. This framework integrates detection, tracking, teaching, learning, and recognition of objects. An extensive set of systematic experiments, in multiple experimental settings, was carried out to thoroughly evaluate the described learning approaches. Experimental results show that the proposed system is able to interact with human users, learn new object categories over time, as well as perform complex tasks. The contributions presented in this thesis have been fully implemented and evaluated on different standard object and scene datasets and empirically evaluated on different robotic platforms. | ['S. Hamidreza Kasaei'] | 2019-12-19 | null | null | null | null | ['3d-object-recognition'] | ['computer-vision'] | [ 9.81029272e-02 1.24050051e-01 2.30609089e-01 -4.83839720e-01
4.26591814e-01 -7.10662603e-01 7.14691818e-01 3.60978603e-01
-4.89974350e-01 5.31160057e-01 -4.70597416e-01 6.92429021e-03
-3.71933550e-01 -7.41714895e-01 -5.27815044e-01 -4.92228478e-01
-4.08854574e-01 6.67315245e-01 4.75284576e-01 -2.54023790e-01
5.06957352e-01 9.73914981e-01 -2.31565905e+00 1.12945847e-01
5.58774769e-01 8.78298640e-01 8.22557330e-01 3.67352009e-01
-2.56410480e-01 5.58275044e-01 -4.22160387e-01 2.46023521e-01
3.35002422e-01 8.38747546e-02 -8.75283360e-01 5.90650618e-01
7.65989646e-02 -1.76708207e-01 1.53073087e-01 9.71718132e-01
1.95261583e-01 4.49678034e-01 7.60376096e-01 -1.08948457e+00
-7.22708225e-01 4.20942485e-01 4.50449735e-02 2.71017477e-02
5.08646846e-01 1.87400982e-01 3.40965152e-01 -1.11338496e+00
5.78812659e-01 1.42223239e+00 2.98633873e-01 6.98767245e-01
-8.54089200e-01 -3.46287012e-01 4.37258869e-01 4.13716078e-01
-1.28114319e+00 -3.06613315e-02 8.04942429e-01 -8.13254952e-01
7.24293649e-01 -1.01338968e-01 8.70325446e-01 7.26233780e-01
3.20271514e-02 8.21792364e-01 1.30471408e+00 -7.32077837e-01
6.05777740e-01 6.83499753e-01 4.79632169e-01 3.95997882e-01
3.40334088e-01 3.46238673e-01 -2.98369050e-01 4.40291315e-01
6.61154687e-01 1.75532565e-01 6.94220960e-02 -9.58633602e-01
-9.67854977e-01 3.48927468e-01 6.43767178e-01 7.96357870e-01
-5.17789125e-01 -2.65794903e-01 2.44879693e-01 4.67996180e-01
-1.62360430e-01 4.19367403e-01 -3.67852867e-01 7.58894458e-02
-2.47159123e-01 -1.45535488e-02 7.14981198e-01 1.08704078e+00
1.06277692e+00 -3.16365063e-02 2.40752697e-01 6.46126509e-01
3.74827087e-01 5.38276434e-01 6.38791442e-01 -6.87660038e-01
-3.19973677e-02 9.17147815e-01 1.71783477e-01 -9.51453924e-01
-4.57714140e-01 -3.83085847e-01 -4.61775422e-01 8.44843388e-01
2.55358964e-01 1.36581495e-01 -1.21682978e+00 1.42255354e+00
5.49507201e-01 -1.12316273e-01 3.61637652e-01 8.66888106e-01
8.26264918e-01 4.74169195e-01 1.66349173e-01 -1.62491918e-01
1.16071308e+00 -5.93865752e-01 -4.34795648e-01 -3.52515459e-01
1.84415713e-01 -4.73269969e-01 9.69002068e-01 6.19561136e-01
-7.17216313e-01 -1.46632898e+00 -1.03933084e+00 2.73786247e-01
-8.86628270e-01 6.02018349e-02 6.40798509e-01 5.48118114e-01
-8.68939877e-01 5.70701659e-01 -7.60433257e-01 -7.83407092e-01
3.46852362e-01 4.04903799e-01 -2.34883562e-01 -8.60906094e-02
-8.08489025e-01 1.09603071e+00 1.09148169e+00 3.18378538e-01
-1.35305560e+00 -2.57594645e-01 -5.08571029e-01 -1.89809889e-01
4.86807734e-01 -4.67534035e-01 1.22259235e+00 -1.18856335e+00
-1.37951434e+00 9.22243297e-01 2.91804433e-01 -3.37895960e-01
2.58320957e-01 -3.34533781e-01 -2.71890014e-01 8.42642710e-02
-7.39579424e-02 7.04596937e-01 6.54811621e-01 -1.81600630e+00
-8.86750877e-01 -5.62364280e-01 3.70906591e-01 4.40295637e-01
-3.58217120e-01 -4.42796052e-01 -1.50947213e-01 -2.19469056e-01
4.56446975e-01 -1.00564706e+00 -1.47739440e-01 -2.61096414e-02
2.65571684e-01 -4.48240221e-01 1.02635813e+00 2.21308530e-03
3.44897002e-01 -2.26057792e+00 2.47528002e-01 3.76254112e-01
-2.55327448e-02 4.06215429e-01 3.41232046e-02 3.26824456e-01
-2.79591815e-03 -3.42930436e-01 9.41093266e-02 2.03988962e-02
1.78587157e-02 4.41926152e-01 -1.83519125e-01 -8.88683051e-02
-3.12915370e-02 5.52708268e-01 -1.07073438e+00 -3.46699268e-01
5.48160195e-01 2.43210942e-01 -3.65195870e-01 3.56650352e-01
-3.25331688e-01 5.98693430e-01 -6.27017558e-01 5.81588626e-01
5.68506479e-01 1.98839232e-01 1.04353391e-01 5.27062677e-02
-2.64369547e-01 -2.40362987e-01 -1.51130569e+00 1.75136542e+00
-4.56302583e-01 3.86699915e-01 -1.67747438e-01 -1.30366325e+00
1.30370927e+00 2.33326256e-01 2.22091362e-01 -5.83046496e-01
3.60591650e-01 2.20319480e-01 1.06756194e-02 -8.10363352e-01
3.71401697e-01 -1.93957388e-02 4.78960685e-02 2.21467972e-01
3.18908215e-01 -3.67086977e-01 3.11379939e-01 -2.44977012e-01
5.76439142e-01 2.21804291e-01 3.00395936e-01 -1.64349958e-01
8.17728460e-01 2.56861806e-01 2.18943000e-01 9.03244197e-01
-1.93244278e-01 -1.70841992e-01 -1.93779811e-01 -7.65124023e-01
-7.03676105e-01 -1.36524892e+00 -5.29101938e-02 1.07670259e+00
4.60383862e-01 2.64331698e-01 -4.41309392e-01 -5.74336708e-01
-9.09596384e-02 6.86661839e-01 -6.02074623e-01 -2.01581463e-01
-3.26813906e-01 -9.94731933e-02 -2.13185206e-01 4.21205580e-01
7.27934837e-01 -1.68530786e+00 -1.36468649e+00 3.78688097e-01
3.81171972e-01 -7.25530565e-01 5.14786243e-01 3.91180187e-01
-1.09330189e+00 -1.05652142e+00 -3.48786861e-01 -1.21429777e+00
9.21323597e-01 5.15255451e-01 8.75944793e-01 1.59315109e-01
-4.31102455e-01 1.06891620e+00 -7.74748623e-01 -8.90006721e-01
-4.85725492e-01 -2.03362942e-01 2.60012984e-01 -9.30315554e-02
6.11589491e-01 -7.18454897e-01 -1.99832082e-01 3.03753555e-01
-9.79458392e-01 -1.86530009e-01 9.77553964e-01 6.73978746e-01
6.00878417e-01 4.31216598e-01 6.18127286e-01 -5.49541533e-01
5.50004065e-01 -2.87874341e-01 -6.35875583e-01 3.51573437e-01
-5.52003920e-01 2.54508443e-02 3.62924963e-01 -7.55613685e-01
-1.38905072e+00 2.88172305e-01 1.07607029e-01 -2.72924930e-01
-9.18620706e-01 4.89579618e-01 -2.58358419e-01 -9.00805146e-02
7.44248092e-01 3.83058429e-01 1.15575098e-01 -5.04552841e-01
5.11075079e-01 8.69535685e-01 5.42446136e-01 -7.22294986e-01
8.89050126e-01 2.61021376e-01 -5.34357093e-02 -9.57944214e-01
-6.63370430e-01 -5.35535336e-01 -1.32679808e+00 -7.33634055e-01
7.00175464e-01 -7.02032268e-01 -7.19792902e-01 5.89856744e-01
-8.50924373e-01 -2.90319473e-01 -5.52052736e-01 6.22596502e-01
-6.15089595e-01 1.47152856e-01 -1.46200091e-01 -1.00967717e+00
1.02548674e-01 -8.83836567e-01 5.46749830e-01 6.54674709e-01
9.37353224e-02 -8.79299402e-01 -1.71125516e-01 -1.53815411e-02
2.74914265e-01 1.44114971e-01 8.61344278e-01 -6.62035048e-01
-7.30025709e-01 -2.23749280e-01 4.80525345e-02 4.02191609e-01
3.82398993e-01 -3.80593210e-01 -9.05343354e-01 -3.81204307e-01
1.74464926e-01 -4.74785030e-01 3.50612640e-01 -1.73182696e-01
9.14885819e-01 6.00986928e-02 -5.34959078e-01 -1.28397807e-01
1.41112030e+00 8.56471896e-01 4.49122667e-01 4.47167724e-01
2.21881241e-01 7.87084639e-01 1.05000257e+00 4.11589801e-01
1.99203953e-01 5.29827833e-01 5.58162808e-01 3.33007157e-01
-1.71934485e-01 -6.61198469e-03 3.47456522e-02 6.33224308e-01
-3.55827421e-01 2.63237864e-01 -8.50563347e-01 5.24719357e-01
-1.59394193e+00 -1.01487184e+00 1.94348142e-01 2.10002518e+00
4.58313316e-01 4.25027311e-01 4.15231958e-02 4.06547666e-01
5.06166458e-01 -5.82315266e-01 -6.34921074e-01 -4.38455760e-01
1.66299105e-01 6.01094142e-02 -7.62534365e-02 2.91959614e-01
-1.01970279e+00 8.76457453e-01 5.76421976e+00 2.51766801e-01
-1.15545189e+00 -1.20466106e-01 -8.40604305e-02 4.57349896e-01
2.62043774e-01 -4.80820015e-02 -6.25569284e-01 1.32881388e-01
4.10622656e-01 -1.28682435e-01 4.09581602e-01 1.32112551e+00
-1.20402277e-01 -3.37427229e-01 -1.33163583e+00 8.80479872e-01
1.50178179e-01 -8.21378946e-01 1.11720666e-01 -1.85014963e-01
5.23589551e-01 -4.03245121e-01 1.26568154e-01 6.01782620e-01
1.78413391e-01 -5.67980707e-01 1.01450920e+00 9.07890856e-01
3.98265496e-02 -5.16866803e-01 8.14272106e-01 7.26593494e-01
-1.20059466e+00 -6.27738118e-01 -5.93481302e-01 -3.47173870e-01
-2.44376421e-01 1.87711731e-01 -1.31481576e+00 5.88812411e-01
1.03264678e+00 6.13004148e-01 -7.89383411e-01 1.30275404e+00
-3.30555767e-01 8.97701234e-02 -2.03358471e-01 -6.06221616e-01
1.07285246e-01 1.17107995e-01 5.04094720e-01 8.45915020e-01
3.54235679e-01 4.88173753e-01 3.29468429e-01 8.47884893e-01
5.34940422e-01 -7.82332495e-02 -7.37529159e-01 3.28551978e-01
5.17068744e-01 9.79189098e-01 -9.56415951e-01 -3.95718694e-01
-2.22335055e-01 7.02644408e-01 1.85663849e-01 2.19957337e-01
-3.72159451e-01 -4.44266707e-01 2.07927272e-01 -6.35046661e-02
4.10449564e-01 -5.99781632e-01 -1.58475526e-02 -5.65457940e-01
-7.10160732e-02 -6.10066414e-01 2.99684376e-01 -1.00973308e+00
-1.11155462e+00 6.69774890e-01 4.83778596e-01 -1.37995601e+00
-2.24797398e-01 -9.16128635e-01 -2.23623261e-01 4.43772793e-01
-1.37437069e+00 -8.58267784e-01 -8.02690864e-01 5.86571932e-01
7.15839922e-01 -4.73748296e-01 8.13017011e-01 2.26419475e-02
3.80477495e-02 9.01381373e-02 -1.05624959e-01 -9.85613391e-02
2.30570361e-01 -1.14163244e+00 -3.36630553e-01 5.89292049e-01
1.64545029e-01 6.53694034e-01 7.34867096e-01 -5.35878301e-01
-1.24898136e+00 -8.02551270e-01 3.40303898e-01 -4.37080741e-01
2.66055822e-01 -2.39728570e-01 -1.01118600e+00 4.57186013e-01
-5.23253642e-02 -2.72128314e-01 4.42593157e-01 3.20969485e-02
-1.52935728e-01 -3.76484364e-01 -1.23189640e+00 4.19786274e-01
1.11972225e+00 -2.31613681e-01 -1.49166656e+00 1.12834111e-01
4.08654541e-01 -1.85831308e-01 -7.83249378e-01 5.56984961e-01
5.15736520e-01 -9.50511277e-01 9.79412973e-01 -3.71274889e-01
-2.04803735e-01 -6.32560551e-01 -1.17918737e-01 -1.27149415e+00
-3.95027041e-01 -3.54135819e-02 5.38683906e-02 1.17184341e+00
6.35306239e-02 -5.40546119e-01 6.39631748e-01 3.56112510e-01
-4.25535530e-01 -3.13486427e-01 -6.41831219e-01 -1.02109909e+00
-6.69069886e-02 -2.84390479e-01 6.81650519e-01 5.93744695e-01
2.17464147e-03 7.37636685e-02 1.73479423e-01 3.35760564e-01
7.59059370e-01 3.44373405e-01 9.39570844e-01 -1.72346067e+00
-5.89097701e-02 -3.12034726e-01 -8.32307816e-01 -1.08566523e+00
-5.66560365e-02 -7.26969063e-01 3.58478189e-01 -1.57769012e+00
4.14428227e-02 -7.97691286e-01 -3.81750286e-01 6.02630854e-01
1.40280381e-01 8.92717466e-02 5.36640823e-01 3.17737103e-01
-6.93201959e-01 4.28557366e-01 1.31241822e+00 -2.01371074e-01
-5.43221533e-01 3.02843302e-01 -3.65755528e-01 8.19075644e-01
7.69504130e-01 -1.59645557e-01 -6.65678799e-01 -3.28062534e-01
-9.68296081e-02 -3.67714912e-01 4.81742203e-01 -1.60072827e+00
3.44688982e-01 -1.86695114e-01 5.86769283e-01 -8.01943004e-01
3.78838032e-01 -1.36105621e+00 1.30367950e-01 7.64303565e-01
-7.02787861e-02 -3.63017559e-01 3.38019609e-01 4.86777127e-01
-1.35231778e-01 -5.44211924e-01 6.67839348e-01 -4.86037105e-01
-1.75826871e+00 -2.04497978e-01 -6.24072433e-01 -5.03753543e-01
1.52727342e+00 -9.11248446e-01 1.14731997e-01 9.11916047e-02
-1.38042831e+00 2.29808822e-01 3.05374473e-01 6.95616841e-01
8.15160930e-01 -1.25042951e+00 -1.64734706e-01 3.08584869e-01
3.62229824e-01 1.92717388e-02 2.83162802e-01 2.09642425e-01
-3.29863548e-01 4.20808673e-01 -9.93610024e-01 -9.22244668e-01
-1.08333027e+00 1.10127318e+00 2.02010557e-01 3.25397253e-01
-3.92265499e-01 5.91801107e-01 -7.08636269e-03 -5.80169380e-01
7.15889394e-01 -3.19725752e-01 -7.82948852e-01 3.98897892e-03
3.46697897e-01 2.60829628e-01 -7.40065426e-02 -4.38046932e-01
-2.31591851e-01 9.66128647e-01 1.65135667e-01 2.17542693e-01
1.34382081e+00 -4.54137057e-01 4.21044156e-02 8.22646379e-01
6.05787873e-01 -4.80944991e-01 -1.31055367e+00 -3.56787175e-01
2.85245538e-01 -4.32402760e-01 -4.94578838e-01 -9.80641067e-01
-5.85034966e-01 8.01074207e-01 1.38892186e+00 4.08371955e-01
1.27258730e+00 1.73265666e-01 -1.11666098e-02 1.02053189e+00
9.80356395e-01 -1.24675560e+00 7.44055331e-01 5.45592308e-01
9.92166400e-01 -1.12171006e+00 -3.99759337e-02 -3.85957509e-01
-4.94200498e-01 1.38892102e+00 8.44532490e-01 -4.44669873e-02
8.46602917e-01 -7.65316188e-02 1.12527892e-01 -2.81950861e-01
-2.20363840e-01 -4.58766162e-01 2.63958484e-01 1.09932327e+00
-1.92073762e-01 -2.07832865e-02 -7.34704658e-02 4.21123087e-01
-1.40949875e-01 1.58078492e-01 5.17915308e-01 1.24040818e+00
-1.10585594e+00 -1.10685587e+00 -6.36857629e-01 -1.46044856e-02
2.73593992e-01 7.47165620e-01 -2.67944396e-01 1.03995907e+00
8.08936596e-01 1.02280700e+00 4.26548049e-02 -2.71186471e-01
6.87659740e-01 6.54026121e-02 6.26927435e-01 -8.66953373e-01
-1.45904824e-01 -4.06492859e-01 -3.46308202e-01 -1.92965925e-01
-8.65047395e-01 -7.22646296e-01 -1.46675861e+00 3.33388239e-01
-2.30104551e-01 3.23081195e-01 8.19096327e-01 9.82680082e-01
1.42278627e-01 5.58092594e-01 6.49149299e-01 -1.19437027e+00
-4.29961532e-01 -9.77195144e-01 -3.52412462e-01 5.41837990e-01
1.30733743e-01 -1.25784934e+00 -2.25936949e-01 3.02233607e-01] | [7.605556011199951, -1.1871367692947388] |
7de52007-fcf6-44fb-98d4-26a18cb0efce | uncertainty-aware-gait-recognition-via | 2211.08007 | null | https://arxiv.org/abs/2211.08007v1 | https://arxiv.org/pdf/2211.08007v1.pdf | Uncertainty-aware Gait Recognition via Learning from Dirichlet Distribution-based Evidence | Existing gait recognition frameworks retrieve an identity in the gallery based on the distance between a probe sample and the identities in the gallery. However, existing methods often neglect that the gallery may not contain identities corresponding to the probes, leading to recognition errors rather than raising an alarm. In this paper, we introduce a novel uncertainty-aware gait recognition method that models the uncertainty of identification based on learned evidence. Specifically, we treat our recognition model as an evidence collector to gather evidence from input samples and parameterize a Dirichlet distribution over the evidence. The Dirichlet distribution essentially represents the density of the probability assigned to the input samples. We utilize the distribution to evaluate the resultant uncertainty of each probe sample and then determine whether a probe has a counterpart in the gallery or not. To the best of our knowledge, our method is the first attempt to tackle gait recognition with uncertainty modelling. Moreover, our uncertain modeling significantly improves the robustness against out-of-distribution (OOD) queries. Extensive experiments demonstrate that our method achieves state-of-the-art performance on datasets with OOD queries, and can also generalize well to other identity-retrieval tasks. Importantly, our method outperforms the state-of-the-art by a large margin of 44.19% when the OOD query rate is around 50% on OUMVLP. | ['Xin Yu', 'Robby T. Tan', 'Lincheng Li', 'Chen Liu', 'Beibei Lin'] | 2022-11-15 | null | null | null | null | ['gait-recognition'] | ['computer-vision'] | [-3.61466147e-02 -2.41084367e-01 -3.57265592e-01 -2.02874362e-01
-1.31262827e+00 -4.44859982e-01 3.67851436e-01 -1.20009094e-01
-2.35199910e-02 7.70481050e-01 3.17938216e-02 3.02051485e-01
-1.74935181e-02 -9.59785104e-01 -7.88170159e-01 -7.85314023e-01
-7.61433765e-02 8.34677756e-01 2.43815869e-01 3.30723584e-01
2.14466721e-01 2.03993499e-01 -1.90212798e+00 -7.09974952e-03
7.84347534e-01 1.46710908e+00 -3.23022276e-01 2.38014996e-01
7.77924284e-02 2.97283232e-01 -1.09857500e+00 -6.51723742e-01
-1.52114043e-02 -7.08397478e-02 -5.15585959e-01 1.27731636e-01
3.33377033e-01 -4.77235138e-01 -3.30320626e-01 1.06829083e+00
6.45174980e-01 -4.85263281e-02 9.54737186e-01 -1.45070696e+00
-7.79698670e-01 2.17019498e-01 -6.23222709e-01 9.92524773e-02
5.77413917e-01 -2.22460721e-02 8.79330695e-01 -1.14381766e+00
4.48687315e-01 1.17747033e+00 6.06328011e-01 2.60226697e-01
-1.07357156e+00 -7.12407231e-01 2.00561047e-01 6.91192389e-01
-1.90507126e+00 -4.73906875e-01 5.99020183e-01 -1.61952183e-01
4.42317694e-01 2.08902240e-01 3.09687883e-01 1.29808140e+00
-3.04120481e-02 1.06806719e+00 6.89058304e-01 -2.55994946e-01
6.13686442e-01 -2.26360694e-01 2.64086992e-01 6.36889219e-01
6.69581056e-01 3.97370197e-02 -1.02099860e+00 -5.71499169e-01
4.95052963e-01 -3.60158712e-01 -2.12444276e-01 -1.64373219e-01
-7.65309095e-01 4.99262571e-01 -4.08000611e-02 -9.03814211e-02
-3.94919038e-01 1.23952411e-01 1.39248759e-01 -8.22895989e-02
2.44030625e-01 -1.71772316e-01 7.98570141e-02 -3.47977459e-01
-8.16244543e-01 1.08946249e-01 1.07433724e+00 7.48113632e-01
7.41224825e-01 -1.27909482e-01 -2.78856754e-01 9.61766601e-01
4.37416464e-01 8.19754362e-01 2.30240315e-01 -8.72003675e-01
3.05827737e-01 3.72397959e-01 3.57190877e-01 -1.17901003e+00
4.37871665e-01 -2.27374747e-01 -6.34775102e-01 -1.46506473e-01
5.60793757e-01 1.45579472e-01 -8.57628047e-01 1.67896235e+00
3.12217623e-01 6.03501379e-01 -1.28785878e-01 1.03916240e+00
5.01299858e-01 3.58663231e-01 -5.96059263e-02 3.84804532e-02
1.52569032e+00 -3.12604874e-01 -6.29841387e-01 -3.44952494e-01
1.55728832e-01 -6.45044148e-01 6.44176066e-01 5.57877243e-01
-5.17421007e-01 -3.00976247e-01 -1.22129202e+00 5.61559260e-01
-4.90321703e-02 2.91481823e-01 3.11268091e-01 8.75932813e-01
-6.63135767e-01 4.17674899e-01 -7.17285633e-01 -2.95658380e-01
2.92888403e-01 1.45755380e-01 -2.83264786e-01 -5.35046160e-01
-1.40970910e+00 4.89511520e-01 1.28302947e-01 2.37018794e-01
-8.65549803e-01 -3.94439816e-01 -9.37978089e-01 -1.82673365e-01
8.36599231e-01 -4.90322620e-01 6.43066704e-01 -2.33646795e-01
-1.12645078e+00 8.73150885e-01 -5.15510201e-01 -2.69463152e-01
5.50862014e-01 -2.90471852e-01 -6.08292639e-01 1.52067721e-01
4.46136981e-01 1.95893362e-01 1.05271983e+00 -1.36526322e+00
-4.67213184e-01 -4.96486336e-01 -3.45779300e-01 -2.48022273e-01
-3.20593715e-02 -4.23347652e-01 -1.18664348e+00 -7.21016347e-01
2.64023006e-01 -8.42685819e-01 3.03667605e-01 1.54185221e-01
-5.92452765e-01 -3.33361655e-01 6.19866550e-01 -5.23021579e-01
1.32171512e+00 -2.26958251e+00 -3.35473120e-01 5.44939041e-01
-1.83965728e-01 9.33497399e-02 3.41904163e-03 3.86751324e-01
5.27442276e-01 5.78949973e-02 -2.85833925e-01 -4.55841839e-01
4.98377115e-01 5.62599242e-01 -5.77078283e-01 5.44567645e-01
2.58738697e-01 6.44318581e-01 -9.08017576e-01 -6.55517042e-01
7.17597678e-02 4.83239472e-01 -2.08955705e-01 1.04759462e-01
-7.66101703e-02 1.66176222e-02 -6.06605291e-01 1.38562667e+00
7.97772586e-01 -2.48101249e-01 1.69850379e-01 -3.46910626e-01
5.50283015e-01 -1.84032530e-01 -1.78282738e+00 1.35802674e+00
1.64645419e-01 2.13272154e-01 -1.19664863e-01 -8.46785367e-01
1.09680939e+00 4.41557020e-02 5.47104537e-01 -3.65844488e-01
-1.00739449e-01 3.43474090e-01 -6.31083429e-01 -3.84285480e-01
5.93971312e-01 2.03753486e-01 -4.88957584e-01 5.50045669e-01
-1.93058878e-01 3.49806428e-01 2.36540064e-01 -4.88849990e-02
9.89508092e-01 -1.12340925e-02 -4.78391908e-02 1.86328322e-01
4.67838407e-01 -3.65054280e-01 8.94539893e-01 1.18952537e+00
-5.31109989e-01 6.34415686e-01 4.21108305e-01 2.16520228e-03
-5.21682620e-01 -1.57384658e+00 -3.68371189e-01 6.57350838e-01
6.11448288e-01 -4.05146092e-01 -6.55188620e-01 -5.73602080e-01
5.04465044e-01 4.71689880e-01 -6.25599325e-01 -4.07966942e-01
-1.17206335e-01 -7.76055574e-01 9.49907541e-01 5.28719902e-01
8.58144403e-01 -5.88202119e-01 -8.98553580e-02 1.32904485e-01
-7.56791353e-01 -1.23774743e+00 -7.05751777e-01 -4.60725605e-01
-4.88312393e-01 -1.39465570e+00 -5.12064874e-01 -4.50672716e-01
6.20886266e-01 -5.59816211e-02 1.03623283e+00 -2.82761380e-02
-3.99823219e-01 8.16577613e-01 -2.49518260e-01 -1.15376659e-01
5.86401438e-03 -2.67581046e-01 2.93471903e-01 4.64847624e-01
9.13676560e-01 -4.32272494e-01 -6.63203001e-01 8.89326751e-01
-7.71649897e-01 -9.10085678e-01 5.09404063e-01 8.60271633e-01
9.68616545e-01 2.01269627e-01 6.11053765e-01 -7.78414011e-02
6.28329933e-01 -7.38369524e-01 -2.64331341e-01 4.35986936e-01
-5.40485740e-01 3.83065701e-01 -1.12652928e-01 -6.60516500e-01
-8.06802332e-01 -1.30567864e-01 2.29371771e-01 -5.70836008e-01
-1.59795716e-01 5.02120852e-01 -4.86581594e-01 4.28015739e-02
3.45766187e-01 4.24453229e-01 -4.23620716e-02 -4.24522996e-01
2.00974226e-01 8.51583362e-01 8.57329786e-01 -1.18964446e+00
6.34145737e-01 7.27479458e-01 -1.03633562e-02 -7.85664678e-01
-5.16751647e-01 -4.63172913e-01 -1.92085624e-01 -4.82279986e-01
3.29554766e-01 -8.98723841e-01 -9.15735543e-01 7.52457142e-01
-6.79352462e-01 2.66815484e-01 -1.21259615e-01 4.45022136e-01
-3.69460940e-01 8.49231243e-01 -4.92007285e-01 -1.17936051e+00
-1.38632223e-01 -1.08077502e+00 1.46587098e+00 1.03243113e-01
-2.92236984e-01 -3.94324839e-01 -1.94795579e-01 4.72131133e-01
-5.73368219e-04 2.45667636e-01 5.01751244e-01 -5.97565174e-01
-8.25609863e-01 -6.57679200e-01 -1.83321506e-01 7.40232319e-02
3.10152858e-01 -1.39091089e-01 -1.02555025e+00 -2.44573262e-02
-1.57728940e-01 -4.01961476e-01 9.62846577e-01 1.65535718e-01
1.03106058e+00 -1.09774940e-01 -5.31936407e-01 1.01324640e-01
1.13190842e+00 -1.61990121e-01 8.89637649e-01 2.36819401e-01
2.47280717e-01 4.33224171e-01 1.02706277e+00 8.59352708e-01
5.13413727e-01 7.57610917e-01 4.10753876e-01 6.51018798e-01
9.91159603e-02 -3.70983660e-01 5.09699464e-01 2.59219855e-01
2.53737956e-01 -3.56522769e-01 -7.23397970e-01 7.42404819e-01
-1.91991556e+00 -8.95079911e-01 2.41376966e-01 2.58035278e+00
8.62467289e-01 9.59344134e-02 5.00697494e-02 2.95300543e-01
1.09068215e+00 1.77547053e-01 -8.41952562e-01 2.91684687e-01
-1.97207034e-01 1.58279929e-02 3.78305882e-01 4.31160182e-01
-1.12078011e+00 6.04652464e-01 6.21094513e+00 1.09145749e+00
-5.60490787e-01 -2.74826109e-01 5.15139461e-01 -7.22235953e-03
-7.96722807e-03 -2.51545221e-01 -1.08816409e+00 8.57269108e-01
5.72081864e-01 -2.09301621e-01 1.65253475e-01 7.75539219e-01
-1.97460920e-01 -4.80437547e-01 -1.16291237e+00 1.15164316e+00
4.59377706e-01 -7.68722057e-01 9.53050032e-02 3.67422611e-01
3.44408333e-01 -4.75792229e-01 1.58270702e-01 1.59637079e-01
3.55979770e-01 -7.61994243e-01 6.80292070e-01 8.96413445e-01
7.89310694e-01 -7.01060891e-01 8.55048954e-01 1.93377078e-01
-1.44576132e+00 1.98611990e-01 -2.82415301e-01 2.29415312e-01
2.71508455e-01 9.87844229e-01 -6.72588527e-01 6.69988990e-01
6.87476158e-01 4.87730116e-01 -3.60422850e-01 1.12096095e+00
-3.15581083e-01 6.97048724e-01 -7.96286583e-01 -4.59754989e-02
-3.02170873e-01 9.67561752e-02 6.87679708e-01 7.67284453e-01
5.64891517e-01 -4.86789159e-02 2.95632541e-01 9.62678850e-01
-1.20298624e-01 -2.62600631e-01 -3.03347558e-01 -2.66613699e-02
1.11794043e+00 5.00305414e-01 -3.96636933e-01 -3.30065012e-01
1.35026276e-01 9.98821557e-01 9.37997624e-02 5.09694755e-01
-8.32547426e-01 -2.57950753e-01 9.62135553e-01 -1.92194715e-01
6.62845612e-01 -5.43086268e-02 -8.55473801e-02 -1.25303888e+00
6.32030189e-01 -6.97380185e-01 6.80972219e-01 -6.82698429e-01
-1.91607094e+00 1.31654814e-01 3.94107886e-02 -1.31495881e+00
-4.78002071e-01 -4.61962074e-01 -2.64142394e-01 8.08514118e-01
-1.23010194e+00 -8.47257257e-01 -2.52845615e-01 4.54571009e-01
2.88015348e-03 7.10486397e-02 9.83473599e-01 4.38947380e-01
-5.36554337e-01 9.33862031e-01 1.27523646e-01 4.57868844e-01
1.05176628e+00 -9.26413178e-01 2.10619405e-01 8.71609986e-01
1.43750189e-02 6.00596726e-01 6.71387076e-01 -1.01093900e+00
-1.62084985e+00 -8.32672179e-01 5.85039258e-01 -4.83542025e-01
7.17481613e-01 -1.05539843e-01 -9.76921082e-01 2.90315479e-01
-6.12536967e-01 1.48753643e-01 8.81835520e-01 1.20679297e-01
-7.35377967e-01 -1.47195026e-01 -1.32786381e+00 3.31706434e-01
1.07335746e+00 -7.12059259e-01 -6.95817947e-01 -7.36597180e-02
6.07883595e-02 -3.03223610e-01 -1.04984295e+00 4.63562697e-01
8.77315164e-01 -6.84757113e-01 1.17929065e+00 -1.11965261e-01
-1.11493960e-01 -6.53592765e-01 -6.15291536e-01 -9.67688501e-01
6.88492730e-02 -1.67620823e-01 -6.63536668e-01 1.55655646e+00
-7.04714358e-02 -7.81217992e-01 1.00288892e+00 7.89145648e-01
3.76342714e-01 -5.53817153e-01 -1.45285511e+00 -1.07969928e+00
-4.58198518e-01 -7.44977295e-01 8.80463898e-01 5.48984170e-01
-1.03353806e-01 -3.20898265e-01 -5.05677283e-01 6.57870710e-01
1.29262602e+00 1.95233077e-01 7.32256055e-01 -1.26124954e+00
-3.90072554e-01 -1.36593804e-01 -7.49273300e-01 -1.08984661e+00
2.18440950e-01 -3.69608402e-01 3.31139475e-01 -1.23976064e+00
7.61348233e-02 -4.43387628e-01 -3.39259505e-01 4.02433544e-01
-3.99732679e-01 3.96660626e-01 -8.94229040e-02 5.26424408e-01
-7.32945085e-01 5.62244296e-01 6.62406206e-01 -4.79110211e-01
1.08122237e-01 -4.77649011e-02 -6.32798553e-01 3.66448462e-01
6.39539719e-01 -4.53179032e-01 -1.01132005e-01 -1.55268580e-01
-1.55704618e-01 -2.38265783e-01 5.75520754e-01 -1.02441263e+00
3.65593582e-01 1.68794215e-01 3.66063744e-01 -7.04682052e-01
6.45043194e-01 -6.28239632e-01 3.73115987e-01 1.89995036e-01
2.05797017e-01 -5.61620474e-01 2.03953698e-01 1.22360516e+00
-2.04661384e-01 -1.23009712e-01 1.06505319e-01 2.04194471e-01
-8.23000550e-01 2.35812709e-01 -4.73479182e-01 7.79460594e-02
7.84521937e-01 -5.30103087e-01 -4.54291731e-01 -4.54342455e-01
-7.12786436e-01 4.68478113e-01 5.27923226e-01 3.94492507e-01
6.86081111e-01 -1.61898530e+00 -6.44260943e-01 2.50339925e-01
5.43919325e-01 -2.16587484e-01 1.84345156e-01 5.28042495e-01
1.26373321e-01 1.37609795e-01 2.24020630e-01 -1.01440775e+00
-1.22104263e+00 2.26868108e-01 3.33549500e-01 -2.93589473e-01
-2.16290340e-01 6.29799366e-01 -1.81088090e-01 -8.29392150e-02
5.71666777e-01 5.34239486e-02 1.90361172e-01 1.99250996e-01
6.10543311e-01 6.44174576e-01 -2.40416490e-02 -7.10175395e-01
-7.50739992e-01 8.20285857e-01 5.04214875e-02 -2.63877273e-01
7.55575478e-01 -9.73629877e-02 8.73894468e-02 5.31226218e-01
8.65720868e-01 -2.78513189e-02 -1.21911716e+00 -4.85971034e-01
-1.07732147e-03 -8.17904413e-01 -3.15422386e-01 -8.12016964e-01
-8.57599735e-01 3.91023308e-01 6.61011279e-01 -1.69072106e-01
9.77113724e-01 1.75937533e-01 8.07506144e-01 4.78955626e-01
9.80776787e-01 -1.16703451e+00 2.03980461e-01 1.94218531e-01
7.00247109e-01 -1.26473176e+00 6.45178463e-03 -5.88554263e-01
-4.39902425e-01 7.90508986e-01 3.91193509e-01 -1.81259923e-02
6.56607807e-01 2.38839567e-01 -2.54680067e-01 -2.68989969e-02
-4.14964318e-01 -2.37722680e-01 2.94842452e-01 7.96417892e-01
-2.06252724e-01 1.34717733e-01 -8.83695707e-02 9.27333117e-01
8.93039256e-02 3.57047409e-01 6.82017356e-02 9.95270014e-01
-5.24441600e-01 -1.12955856e+00 -9.92724836e-01 4.23497468e-01
-2.93702245e-01 2.07373694e-01 -4.73244488e-01 4.21292931e-01
2.71671824e-03 1.23966289e+00 5.94715364e-02 -5.11586070e-01
3.78381401e-01 2.62906790e-01 4.31968600e-01 -1.59077302e-01
2.16409028e-01 -1.48887545e-01 2.39683062e-01 -5.42371333e-01
-1.89281285e-01 -8.11878502e-01 -1.18939781e+00 -4.36934471e-01
-3.80294830e-01 2.83566982e-01 2.94559300e-01 9.30025697e-01
7.29530215e-01 -7.79279647e-03 4.08581555e-01 -5.56408882e-01
-7.20799685e-01 -8.19935501e-01 -8.32085073e-01 3.70033026e-01
-3.95941027e-02 -1.21462631e+00 -5.82683921e-01 -1.07384855e-02] | [13.989349365234375, 1.2563611268997192] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.