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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
d1f53c71-0c0d-4f3d-b7db-fbba6d62ca93 | multi-level-representation-learning-with | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Wu_Multi-Level_Representation_Learning_With_Semantic_Alignment_for_Referring_Video_Object_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Wu_Multi-Level_Representation_Learning_With_Semantic_Alignment_for_Referring_Video_Object_CVPR_2022_paper.pdf | Multi-Level Representation Learning With Semantic Alignment for Referring Video Object Segmentation | Referring video object segmentation (RVOS) is a challenging language-guided video grounding task, which requires comprehensively understanding the semantic information of both video content and language queries for object prediction. However, existing methods adopt multi-modal fusion at a frame-based spatial granularity. The limitation of visual representation is prone to causing vision-language mismatching and producing poor segmentation results. To address this, we propose a novel multi-level representation learning approach, which explores the inherent structure of the video content to provide a set of discriminative visual embedding, enabling more effective vision-language semantic alignment. Specifically, we embed different visual cues in terms of visual granularity, including multi-frame long-temporal information at video level, intra-frame spatial semantics at frame level, and enhanced object-aware feature prior at object level. With the powerful multi-level visual embedding and carefully-designed dynamic alignment, our model can generate a robust representation for accurate video object segmentation. Extensive experiments on Refer-DAVIS_ 17 and Refer-YouTube-VOS demonstrate that our model achieves superior performance both in segmentation accuracy and inference speed. | ['Jianbing Shen', 'Ling Shao', 'Xingping Dong', 'Dongming Wu'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['video-grounding', 'referring-expression-segmentation', 'referring-video-object-segmentation'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 4.14403342e-02 -3.89306039e-01 -6.92465663e-01 -4.54611838e-01
-9.64784443e-01 -4.80872124e-01 3.03534865e-01 7.31543377e-02
-2.75791019e-01 2.31634393e-01 4.67233390e-01 6.13038614e-02
2.18512326e-01 -5.35696208e-01 -8.65763843e-01 -4.04166877e-01
2.59098679e-01 -1.18369423e-01 8.09163332e-01 -5.86322434e-02
2.40610406e-01 2.23968193e-01 -1.41851842e+00 3.35916638e-01
7.47962594e-01 1.10012507e+00 6.54537857e-01 5.74098647e-01
-3.51654023e-01 9.75922883e-01 -1.89553022e-01 -2.50736654e-01
1.83366686e-01 -3.63532156e-01 -8.80251408e-01 4.99249518e-01
7.96445966e-01 -6.03966057e-01 -6.99872434e-01 1.27189302e+00
1.56852096e-01 3.55471969e-01 3.20441067e-01 -1.40379250e+00
-7.95671701e-01 4.01093423e-01 -6.28170907e-01 5.14565587e-01
5.94527721e-01 3.41188550e-01 1.11686766e+00 -7.32807994e-01
6.69226706e-01 1.46721232e+00 3.59310210e-01 3.76862794e-01
-9.09847498e-01 -4.24869955e-01 6.32579625e-01 6.39971554e-01
-1.42676985e+00 -2.81044275e-01 9.41861093e-01 -6.60973370e-01
5.82420111e-01 1.27828002e-01 7.44302452e-01 8.60058427e-01
-8.92762560e-03 1.11334312e+00 7.19889700e-01 -1.11156106e-01
-5.27340099e-02 -1.34509087e-01 1.13883592e-01 1.01526141e+00
-1.23604853e-03 -1.39327630e-01 -6.80813909e-01 6.55471534e-02
1.08068049e+00 3.61243218e-01 -3.56179208e-01 -3.77660871e-01
-1.44074202e+00 7.41191506e-01 6.09273434e-01 3.30629081e-01
-4.57636356e-01 4.69505548e-01 4.82113063e-01 -2.46223465e-01
2.00352773e-01 -1.35533035e-01 -2.51220852e-01 8.13514963e-02
-1.19755864e+00 1.65720433e-01 2.93625355e-01 1.34052086e+00
8.89485836e-01 3.47072393e-01 -5.68797112e-01 6.49885237e-01
7.89305389e-01 5.59549570e-01 3.27531487e-01 -1.40715873e+00
5.47568679e-01 5.72822452e-01 -9.20847058e-02 -1.38682115e+00
9.20055360e-02 9.24513501e-04 -5.24130046e-01 -3.69561553e-01
1.47907406e-01 3.39298069e-01 -1.08915353e+00 1.61304438e+00
4.56577599e-01 7.14347780e-01 -1.61505137e-02 1.22501528e+00
1.11400127e+00 8.58750403e-01 4.01624411e-01 -1.43832669e-01
1.62574530e+00 -1.20741355e+00 -7.68900573e-01 -2.88853526e-01
4.02796984e-01 -7.74854779e-01 1.04081011e+00 -2.80137241e-01
-1.19132638e+00 -8.65258574e-01 -9.40451443e-01 -4.15653586e-01
-1.55365124e-01 4.84365271e-03 4.03704643e-01 1.83857441e-01
-8.88233066e-01 -4.41284552e-02 -6.46081507e-01 -2.16213226e-01
5.62285542e-01 9.29912999e-02 -2.61646152e-01 -4.60875303e-01
-1.24978435e+00 3.54558229e-01 6.89043045e-01 8.41718689e-02
-1.00668287e+00 -5.87382615e-01 -1.19294631e+00 -1.36277333e-01
7.10446179e-01 -8.00727069e-01 9.98866856e-01 -9.06612217e-01
-1.13347411e+00 7.12360740e-01 -5.12704015e-01 -3.93697143e-01
3.43653738e-01 -1.08772464e-01 -1.88490003e-01 7.52899289e-01
4.41324830e-01 1.05571377e+00 9.55286205e-01 -1.39063442e+00
-7.98929870e-01 -3.76116753e-01 3.25036794e-01 3.24638993e-01
-1.17840119e-01 5.57282642e-02 -1.31196213e+00 -9.05630052e-01
2.71783471e-01 -6.02463126e-01 -2.80281693e-01 7.05013350e-02
-1.90494180e-01 -2.85309017e-01 9.82133508e-01 -9.60618377e-01
1.23538601e+00 -2.23785281e+00 2.16320187e-01 -1.85253516e-01
3.41942757e-01 1.43522590e-01 -2.71702230e-01 8.29980448e-02
3.23794961e-01 7.27770478e-02 -2.44674589e-02 -3.25302891e-02
-1.64430335e-01 3.03783745e-01 -3.70520234e-01 3.80388051e-01
1.43316835e-01 1.30579162e+00 -1.15753531e+00 -1.10254169e+00
5.70956051e-01 4.60570395e-01 -8.28209102e-01 4.03909147e-01
-5.17249644e-01 3.67144346e-01 -9.98029768e-01 1.10796046e+00
4.20606732e-01 -3.85456353e-01 -1.54638663e-01 -7.72133589e-01
1.06205150e-01 -2.44598374e-01 -1.08422744e+00 1.98277199e+00
-2.13683367e-01 6.76010311e-01 -8.32065344e-02 -1.01952529e+00
6.27679706e-01 8.11464489e-02 7.36679614e-01 -7.28840530e-01
3.51740122e-02 -7.49685764e-02 -4.32259321e-01 -7.72371113e-01
6.79323435e-01 3.96930605e-01 -3.81003916e-02 2.82317097e-03
1.69327278e-02 -9.15605500e-02 1.36107519e-01 5.23295820e-01
5.85372448e-01 3.42895508e-01 1.52748704e-01 -1.41548198e-02
7.88591862e-01 -8.94963667e-02 8.19271743e-01 5.31897187e-01
-6.09801352e-01 7.00407982e-01 2.83296764e-01 -2.41943821e-01
-9.05784845e-01 -9.40810561e-01 8.98137689e-02 1.17206311e+00
8.60539198e-01 -5.95842421e-01 -7.70302117e-01 -5.26989877e-01
-2.71870960e-02 5.53396404e-01 -3.42668951e-01 -1.26571119e-01
-6.46669388e-01 -1.66419670e-01 2.82597989e-01 7.41245508e-01
7.66968727e-01 -7.87270725e-01 -3.83875102e-01 1.93262026e-01
-6.13438189e-01 -1.80698872e+00 -1.01363599e+00 -5.59879959e-01
-7.86470056e-01 -1.19789648e+00 -6.99662983e-01 -1.02625978e+00
4.54031527e-01 8.02803278e-01 9.96573627e-01 2.46526092e-01
-2.78356016e-01 9.65485811e-01 -4.62329954e-01 1.45021886e-01
-8.69909897e-02 -3.23461562e-01 -2.80008525e-01 1.16920143e-01
3.17202300e-01 -7.24850446e-02 -7.80256212e-01 3.52885365e-01
-1.07594705e+00 1.32268816e-01 3.91268730e-01 6.67013824e-01
9.37220395e-01 -2.34191224e-01 1.95354715e-01 -3.50977600e-01
1.03430450e-01 -5.31816721e-01 -5.30213237e-01 5.69368422e-01
-1.70416042e-01 -2.90778577e-01 2.09240496e-01 -2.57380605e-01
-7.77972281e-01 9.33575928e-02 4.02514897e-02 -9.74257648e-01
-4.14042808e-02 2.66042471e-01 -3.33157122e-01 -9.19846818e-02
2.04207338e-02 7.67394185e-01 -1.35382891e-01 -2.23870322e-01
6.80434763e-01 4.62584764e-01 6.94248974e-01 -6.70646250e-01
6.21720552e-01 4.44427818e-01 -2.45639697e-01 -8.97061825e-01
-9.14149284e-01 -7.36726522e-01 -7.57441521e-01 -4.10816491e-01
1.51173913e+00 -1.38021791e+00 -5.32580256e-01 2.51645863e-01
-1.33703339e+00 -1.12603754e-01 -4.95114587e-02 5.56861937e-01
-7.14684844e-01 8.22626650e-01 -5.86257577e-01 -4.00032073e-01
-1.42977700e-01 -1.63502145e+00 1.48380244e+00 2.94625938e-01
2.28966311e-01 -1.05270255e+00 -4.15267318e-01 8.74967217e-01
-8.39388445e-02 1.35652840e-01 7.45992482e-01 -3.25106591e-01
-1.25366235e+00 1.26946032e-01 -7.00362563e-01 4.53370243e-01
-2.16342136e-02 9.77915972e-02 -5.65929592e-01 -1.74437553e-01
-1.94029972e-01 -2.47792169e-01 8.99079025e-01 5.09417415e-01
1.51611578e+00 -1.70904562e-01 -3.14565688e-01 7.91063190e-01
1.49545252e+00 6.28651865e-03 3.88772398e-01 2.20786542e-01
1.35339522e+00 4.37370479e-01 9.46301103e-01 2.74789810e-01
8.80730867e-01 7.81873941e-01 4.99712348e-01 -2.11766046e-02
-2.70211607e-01 -2.80640095e-01 3.36935252e-01 8.48888040e-01
5.27125830e-03 -1.18127950e-01 -8.38382483e-01 5.67790806e-01
-2.00685143e+00 -1.13284612e+00 3.71174142e-02 1.72448516e+00
6.95282102e-01 -1.37217060e-01 1.70386091e-01 -4.14477199e-01
8.84981453e-01 5.73288262e-01 -5.42182982e-01 2.10185215e-01
-9.40456241e-02 -6.10938132e-01 5.55272937e-01 5.49555659e-01
-1.10417569e+00 1.35494637e+00 6.04037046e+00 9.36397076e-01
-9.18659031e-01 2.76268184e-01 5.42346239e-01 5.46157025e-02
-5.64172149e-01 -1.01756258e-02 -8.70701253e-01 6.97342992e-01
4.01200235e-01 -1.55573964e-01 2.49435127e-01 8.71000648e-01
2.18720719e-01 -5.80436327e-02 -1.06607902e+00 1.39803874e+00
3.69184762e-01 -1.65815878e+00 4.54358429e-01 -2.80757695e-02
6.58184350e-01 6.66826665e-02 -1.37918830e-01 1.72391236e-01
-9.92845073e-02 -7.24720240e-01 9.97332096e-01 5.51070452e-01
5.56563437e-01 -5.17559648e-01 4.58688676e-01 -6.53537065e-02
-1.68036926e+00 7.32928738e-02 -2.18420133e-01 5.17852902e-01
4.17231530e-01 2.03865096e-01 -3.19348127e-01 6.36632919e-01
8.39115858e-01 1.29903173e+00 -5.34534514e-01 7.77781785e-01
1.43173873e-01 4.62452084e-01 -8.57774913e-02 2.64907897e-01
6.80487692e-01 -2.47001946e-01 5.16849220e-01 1.24056888e+00
-2.97633521e-02 3.02284151e-01 8.70538592e-01 7.89702415e-01
6.61279857e-02 5.47521710e-02 -4.12184983e-01 -2.36875430e-01
5.34885705e-01 9.58559930e-01 -7.71196306e-01 -5.38824379e-01
-8.38024139e-01 1.02526331e+00 7.92778358e-02 7.36981392e-01
-1.15090847e+00 9.92013216e-02 7.92689443e-01 6.42708316e-02
4.89447951e-01 -5.79471350e-01 7.71204382e-02 -1.37552178e+00
7.60012493e-02 -7.49027014e-01 3.50769430e-01 -8.91880512e-01
-1.05888164e+00 3.76823008e-01 1.58307537e-01 -1.24886847e+00
-1.58846498e-01 -3.65220338e-01 -3.79007727e-01 5.59780836e-01
-1.71711242e+00 -1.49973774e+00 -5.00512600e-01 1.02334666e+00
1.13328838e+00 -1.94165990e-01 1.37994960e-01 3.80978972e-01
-6.17105067e-01 4.24152523e-01 -1.95294157e-01 3.33734751e-01
4.03327286e-01 -8.76488447e-01 -5.97439483e-02 1.00487149e+00
3.32585633e-01 4.65850562e-01 4.52683836e-01 -7.05844760e-01
-1.68691647e+00 -1.35216022e+00 4.53866005e-01 -4.81834650e-01
7.22989202e-01 -1.79229248e-02 -9.54486132e-01 7.06344724e-01
1.20253168e-01 2.56088883e-01 6.13906384e-01 -4.75386381e-01
-3.26795638e-01 -6.05305061e-02 -7.38534927e-01 7.37787247e-01
1.08525586e+00 -9.21784699e-01 -7.50569701e-01 4.43049580e-01
1.24753940e+00 -5.16358018e-01 -9.31448758e-01 4.55135792e-01
4.05364513e-01 -7.82269895e-01 1.36724365e+00 -5.02207518e-01
5.25683761e-01 -5.94549358e-01 -6.95995092e-01 -4.66060817e-01
-2.56869167e-01 -3.52528602e-01 -2.18435898e-01 1.37918663e+00
-3.27297986e-01 -1.28104284e-01 5.53560317e-01 4.77854788e-01
-1.32124558e-01 -9.24677134e-01 -6.99953794e-01 -5.45363188e-01
-4.14866775e-01 -6.85901642e-01 3.70907903e-01 8.27434123e-01
-5.78704298e-01 2.02963099e-01 -4.46227819e-01 3.87386978e-01
7.43387222e-01 3.58644843e-01 7.57867813e-01 -7.19760239e-01
-4.69977632e-02 -4.34081882e-01 -8.26264679e-01 -1.54352415e+00
4.14768636e-01 -8.10669780e-01 4.09544371e-02 -1.67329645e+00
2.59035647e-01 -1.41138390e-01 -3.91354889e-01 2.33104318e-01
-3.92434418e-01 3.32834244e-01 5.41030467e-01 4.88369554e-01
-1.16000795e+00 7.01454580e-01 1.51054156e+00 -3.64392847e-01
2.31746882e-02 -4.81521130e-01 -5.21218061e-01 8.87597620e-01
3.06775421e-01 -1.98188439e-01 -5.69295883e-01 -7.23970473e-01
-1.80911824e-01 2.59039760e-01 7.29731321e-01 -7.79932141e-01
2.82643467e-01 -5.18833399e-01 3.06698114e-01 -8.40368629e-01
4.85429853e-01 -8.41656029e-01 -2.56172717e-01 2.97329098e-01
-3.37785602e-01 -5.68331741e-02 4.27532867e-02 1.08344865e+00
-6.70527518e-01 1.90402772e-02 6.59799755e-01 -1.47594362e-01
-1.57297826e+00 8.27776432e-01 -8.40566680e-02 2.68029779e-01
1.25822413e+00 -5.49646914e-01 -4.38218638e-02 -1.86271936e-01
-5.69801092e-01 6.25132859e-01 5.90982080e-01 8.08040202e-01
9.38032627e-01 -1.40765166e+00 -5.05976856e-01 1.97820067e-01
2.95096844e-01 3.69460620e-02 6.02492630e-01 8.15581918e-01
-5.82845509e-01 3.59758109e-01 -1.35452941e-01 -1.11720669e+00
-1.16563058e+00 5.67420304e-01 1.60460815e-01 3.51058304e-01
-6.76469624e-01 9.66534436e-01 4.87987638e-01 1.64395779e-01
2.83754379e-01 -3.82664919e-01 -1.94714651e-01 7.86674395e-02
5.07699072e-01 1.89468041e-01 -5.26795447e-01 -1.24977887e+00
-4.32756394e-01 1.05367470e+00 -1.44387454e-01 1.51285768e-01
7.79848099e-01 -6.51777983e-01 -9.78347436e-02 3.93636465e-01
1.41245008e+00 -1.96319506e-01 -1.46742463e+00 -6.13324523e-01
-2.46892631e-01 -8.70861471e-01 3.06129009e-01 -3.26911435e-02
-1.19955456e+00 8.81308675e-01 4.90221947e-01 -1.01971433e-01
1.03621447e+00 2.32195958e-01 1.23313057e+00 1.88556701e-01
4.13390189e-01 -1.07594776e+00 5.45293510e-01 3.85154754e-01
6.37624025e-01 -1.54417479e+00 1.52747497e-01 -6.62174702e-01
-8.22294891e-01 9.29509401e-01 8.40700507e-01 -4.18542176e-02
5.15914977e-01 -3.67758989e-01 8.06073919e-02 -1.20376207e-01
-4.04470533e-01 -4.22645628e-01 6.37892008e-01 4.48552072e-01
1.38852745e-01 -1.80035204e-01 -5.02555780e-02 3.68174404e-01
3.02724361e-01 -1.69462875e-01 1.88637018e-01 7.31158555e-01
-5.78907788e-01 -7.81009138e-01 -2.16070786e-01 2.52242804e-01
-3.49621266e-01 -1.47424296e-01 2.74025183e-02 6.23148024e-01
1.38066277e-01 7.77471542e-01 6.15193285e-02 -2.90825754e-01
1.92772329e-03 -2.49052376e-01 3.93223792e-01 -6.29624426e-01
-1.60699952e-02 3.14926296e-01 -3.90623748e-01 -1.01751900e+00
-8.81475329e-01 -5.95476270e-01 -1.45930040e+00 -3.40971490e-03
-1.11880869e-01 -2.56841164e-02 3.78491521e-01 1.17306507e+00
3.54838371e-01 5.40635109e-01 4.25857008e-01 -9.80006039e-01
-1.70773178e-01 -3.44095886e-01 -5.19133031e-01 5.65026641e-01
4.53928918e-01 -7.23383605e-01 -6.90604895e-02 5.03472030e-01] | [9.733931541442871, 0.5779173374176025] |
d3162d37-60c4-4622-8351-a70571082319 | semi-supervised-learning-in-video-sequences | 2005.10266 | null | https://arxiv.org/abs/2005.10266v4 | https://arxiv.org/pdf/2005.10266v4.pdf | Naive-Student: Leveraging Semi-Supervised Learning in Video Sequences for Urban Scene Segmentation | Supervised learning in large discriminative models is a mainstay for modern computer vision. Such an approach necessitates investing in large-scale human-annotated datasets for achieving state-of-the-art results. In turn, the efficacy of supervised learning may be limited by the size of the human annotated dataset. This limitation is particularly notable for image segmentation tasks, where the expense of human annotation is especially large, yet large amounts of unlabeled data may exist. In this work, we ask if we may leverage semi-supervised learning in unlabeled video sequences and extra images to improve the performance on urban scene segmentation, simultaneously tackling semantic, instance, and panoptic segmentation. The goal of this work is to avoid the construction of sophisticated, learned architectures specific to label propagation (e.g., patch matching and optical flow). Instead, we simply predict pseudo-labels for the unlabeled data and train subsequent models with both human-annotated and pseudo-labeled data. The procedure is iterated for several times. As a result, our Naive-Student model, trained with such simple yet effective iterative semi-supervised learning, attains state-of-the-art results at all three Cityscapes benchmarks, reaching the performance of 67.8% PQ, 42.6% AP, and 85.2% mIOU on the test set. We view this work as a notable step towards building a simple procedure to harness unlabeled video sequences and extra images to surpass state-of-the-art performance on core computer vision tasks. | ['Ekin D. Cubuk', 'Liang-Chieh Chen', 'Jonathon Shlens', 'Bowen Cheng', 'Raphael Gontijo Lopes', 'Maxwell D. Collins', 'Barret Zoph', 'Hartwig Adam'] | 2020-05-20 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/942_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123540664.pdf | eccv-2020-8 | ['patch-matching'] | ['computer-vision'] | [ 4.34522152e-01 1.30478770e-01 -2.32766628e-01 -4.17928189e-01
-1.15548277e+00 -6.87626421e-01 6.44733548e-01 -8.85577649e-02
-6.88492060e-01 6.13646865e-01 -2.96236485e-01 -3.58699203e-01
3.57525557e-01 -6.76674843e-01 -6.97161615e-01 -7.58160174e-01
2.31502242e-02 5.88576496e-01 4.96926606e-01 -4.51490842e-02
1.09253921e-01 3.50565106e-01 -1.58084309e+00 -3.45832780e-02
1.14751029e+00 9.80817199e-01 2.63518453e-01 6.79546297e-01
-9.30134356e-02 8.82461071e-01 -1.40556782e-01 -4.09929961e-01
4.14791703e-01 -3.65654707e-01 -9.72331464e-01 7.41312683e-01
8.23557794e-01 -3.95739377e-01 -1.67315930e-01 1.06352174e+00
1.79752856e-01 3.06792498e-01 6.90662742e-01 -1.07534611e+00
-2.08335474e-01 3.55177283e-01 -7.35970616e-01 1.67893976e-01
-6.24862202e-02 4.29583251e-01 1.05504632e+00 -8.21319997e-01
5.97730339e-01 8.00041616e-01 5.17088056e-01 3.82249057e-01
-1.32955980e+00 -4.70579714e-01 2.44414881e-01 1.19891651e-01
-1.26670182e+00 -5.56845427e-01 7.07668602e-01 -6.60017371e-01
6.14406705e-01 1.27551079e-01 5.25342882e-01 8.26468349e-01
-4.08264697e-01 1.00075781e+00 1.05433774e+00 -4.18076247e-01
3.15518230e-01 2.15301216e-01 1.66775823e-01 9.23908949e-01
1.69572100e-01 1.42076150e-01 -1.70100152e-01 1.57942191e-01
5.79046190e-01 -7.76201189e-02 -2.30275542e-01 -4.50514048e-01
-1.01633084e+00 8.63771796e-01 4.01544929e-01 1.81687310e-01
-2.71777123e-01 -1.99338677e-03 2.69417465e-01 5.76135851e-02
6.66439772e-01 4.90104467e-01 -5.41611791e-01 -1.68309227e-01
-1.37325156e+00 -6.69303834e-02 6.94754243e-01 8.42299044e-01
1.15130424e+00 8.91935602e-02 5.26601560e-02 8.69986117e-01
1.48358479e-01 5.12369335e-01 1.96403876e-01 -1.25443077e+00
3.87964517e-01 6.10545337e-01 5.34461625e-02 -7.28832304e-01
-2.66212225e-01 -5.17863452e-01 -7.16782033e-01 2.44939253e-01
9.12345707e-01 -2.16842800e-01 -1.30404317e+00 1.52962756e+00
3.67112160e-01 5.07445276e-01 -2.18494628e-02 8.02794397e-01
5.60906529e-01 6.36126757e-01 2.51248062e-01 -1.35330200e-01
1.02734041e+00 -1.43664026e+00 -1.14820816e-01 -4.57141817e-01
8.28716397e-01 -8.12863410e-01 1.03127730e+00 2.91692168e-01
-9.28806841e-01 -8.23214114e-01 -6.69252694e-01 1.07923113e-01
-1.41737372e-01 1.97309971e-01 5.33209801e-01 6.93607271e-01
-8.84849966e-01 5.81923485e-01 -9.77013528e-01 -5.88510096e-01
6.82124078e-01 3.19518268e-01 -3.03163052e-01 -3.82099599e-01
-6.68932319e-01 7.16526151e-01 4.16649759e-01 1.87820628e-01
-9.82605577e-01 -6.41229510e-01 -8.19844007e-01 -4.16902080e-02
5.63583612e-01 -3.49494100e-01 1.11911428e+00 -1.29219055e+00
-1.18734443e+00 1.14996946e+00 -1.94914937e-02 -5.96796632e-01
7.18328178e-01 -1.79857343e-01 -9.85549763e-02 3.42796326e-01
2.27928504e-01 1.05893683e+00 9.11376536e-01 -1.39892375e+00
-9.71630037e-01 -2.95494050e-01 -2.68917740e-03 2.50451770e-02
-1.71136662e-01 -1.76954746e-01 -7.30853021e-01 -4.12890255e-01
1.66948475e-02 -1.26531136e+00 -5.64864635e-01 -2.29403973e-02
-2.70489991e-01 -1.89105883e-01 7.39162028e-01 -5.71299314e-01
8.32837582e-01 -2.10010242e+00 -5.54548316e-02 9.26460177e-02
1.09078646e-01 7.07073092e-01 -2.67157704e-01 -4.07631248e-02
1.99147403e-01 -5.04966788e-02 -6.82145894e-01 -4.12771970e-01
-2.90465981e-01 2.55655169e-01 -1.68029398e-01 5.84218621e-01
4.73052323e-01 9.60931361e-01 -1.03229773e+00 -6.87263668e-01
4.40357506e-01 3.22022825e-01 -6.42769754e-01 2.69909143e-01
-3.60789239e-01 7.78780639e-01 -4.75514799e-01 6.31985426e-01
5.12948096e-01 -4.89543557e-01 4.61109579e-02 2.05127429e-02
-1.30044390e-02 -8.41576327e-03 -1.18470466e+00 1.77041209e+00
-3.42376053e-01 7.76061714e-01 -6.85127676e-02 -1.30691147e+00
6.61059439e-01 1.31464049e-01 7.79682994e-01 -5.52769721e-01
1.09730385e-01 3.90347958e-01 -7.90744722e-02 -6.21028602e-01
3.52030814e-01 6.34167492e-02 2.15734482e-01 3.23059261e-01
2.25318491e-01 -2.17244282e-01 4.63519454e-01 1.20089680e-01
9.74746108e-01 3.47962707e-01 -1.48667730e-02 -2.01233074e-01
7.17307270e-01 4.27919358e-01 4.91314292e-01 7.00318754e-01
-5.03903985e-01 6.54461563e-01 2.74580956e-01 -5.22563398e-01
-1.03210247e+00 -9.08179224e-01 -1.67607710e-01 1.19232154e+00
6.23903386e-02 -1.30096421e-01 -8.40317369e-01 -9.72276568e-01
-2.26538330e-01 5.22538900e-01 -4.22962934e-01 1.41893432e-01
-7.30037332e-01 -7.53897905e-01 4.60517973e-01 5.66091120e-01
6.10573232e-01 -9.72926378e-01 -6.76588058e-01 2.63039529e-01
-2.98159212e-01 -1.48438764e+00 -2.53167689e-01 1.68515146e-01
-8.93863559e-01 -1.24098504e+00 -8.69305193e-01 -8.74507189e-01
8.28826547e-01 3.70045811e-01 1.20907950e+00 1.35526180e-01
-3.37952644e-01 4.01856571e-01 -4.12862718e-01 4.47798148e-03
-2.33063936e-01 1.70372427e-01 -3.84631813e-01 2.38686696e-01
2.25568876e-01 -3.61869097e-01 -7.77695119e-01 4.42540348e-01
-7.99009502e-01 5.23297908e-03 5.74856997e-01 8.82224917e-01
7.42326081e-01 -7.81510249e-02 4.33892399e-01 -1.11102033e+00
-2.10113421e-01 -2.80643970e-01 -7.59509683e-01 1.41558290e-01
-6.18201494e-01 -5.38465828e-02 7.02638984e-01 -4.14158434e-01
-1.04608393e+00 4.64093983e-01 -1.97177097e-01 -3.98180366e-01
-4.94672716e-01 3.82278949e-01 7.48727322e-02 -2.13993490e-01
7.17435658e-01 1.89457670e-01 -1.13003358e-01 -2.38550112e-01
4.86163735e-01 6.76779807e-01 5.71408153e-01 -5.44573009e-01
8.83034825e-01 6.37804747e-01 -5.08432873e-02 -9.49118555e-01
-1.23459518e+00 -8.48650157e-01 -8.02677870e-01 -2.16338664e-01
1.11507285e+00 -1.01302481e+00 -1.46283820e-01 4.70497429e-01
-8.60022008e-01 -7.26683795e-01 -3.43155235e-01 4.82905358e-01
-6.49945080e-01 3.56964499e-01 -5.31476676e-01 -6.76438808e-01
-9.31431428e-02 -1.35626292e+00 1.05014753e+00 2.73760855e-01
-1.67308711e-02 -9.89743888e-01 -6.19686267e-04 8.33019018e-01
1.67195082e-01 1.32875875e-01 4.88878906e-01 -8.01017582e-01
-8.44267428e-01 -2.19802260e-01 -4.63352650e-01 6.34573579e-01
-2.30027754e-02 2.46629165e-03 -1.11432326e+00 -2.55420536e-01
-3.35653186e-01 -6.78547740e-01 9.98243332e-01 3.12551081e-01
1.06089032e+00 2.06958875e-01 -1.43406108e-01 5.96200705e-01
1.49388456e+00 6.61755353e-02 4.93412256e-01 2.24380970e-01
9.87750769e-01 8.01327586e-01 7.58722484e-01 9.78286192e-02
4.20045912e-01 5.29610693e-01 3.17992300e-01 -2.35815376e-01
-2.43392050e-01 -1.58385977e-01 1.55855954e-01 6.29741907e-01
-1.97628796e-01 -1.20355323e-01 -1.13346565e+00 6.92290008e-01
-1.92830908e+00 -8.94057512e-01 -3.41117352e-01 2.20044041e+00
6.96939051e-01 2.87078112e-01 1.92998588e-01 1.25977937e-02
5.55943012e-01 2.00876340e-01 -6.11211181e-01 1.32973507e-01
6.53731674e-02 1.65573344e-01 6.32543266e-01 4.02559072e-01
-1.54292989e+00 1.27245712e+00 5.51917505e+00 8.20999920e-01
-1.21971941e+00 6.74246103e-02 1.03535175e+00 1.80691153e-01
1.35385305e-01 2.48737410e-02 -7.24955499e-01 3.39200616e-01
1.05560553e+00 4.78082031e-01 4.07159001e-01 9.00076330e-01
2.75875181e-01 -2.94070989e-01 -1.01616597e+00 9.87240970e-01
5.07123433e-02 -1.24970818e+00 -1.97786584e-01 1.10071607e-01
1.10533583e+00 4.49145734e-01 -1.68639287e-01 3.78823429e-01
2.77362704e-01 -8.78872454e-01 4.89800185e-01 -6.46410137e-02
6.88615620e-01 -5.34744442e-01 6.49735868e-01 4.63736832e-01
-1.09707952e+00 -1.19241178e-02 -2.76606530e-01 8.33980888e-02
3.06154668e-01 4.85647619e-01 -5.64103782e-01 3.32602054e-01
6.03563488e-01 8.25936556e-01 -7.11167455e-01 1.22309029e+00
-3.02847266e-01 1.02081680e+00 -4.39990252e-01 2.84211993e-01
6.90026641e-01 -3.71132821e-01 1.52228191e-01 1.19206440e+00
-4.94931452e-02 5.95833659e-02 7.13384092e-01 5.04057944e-01
-1.78265214e-01 1.90896139e-01 -5.13791323e-01 3.50111313e-02
1.64015606e-01 1.38723922e+00 -1.05646980e+00 -5.70024967e-01
-6.13119125e-01 8.78880858e-01 4.35777247e-01 5.16733050e-01
-8.05686593e-01 -2.09793244e-02 2.41076931e-01 1.70878112e-01
4.83173281e-01 -2.24488318e-01 -3.88526857e-01 -1.26319551e+00
-2.30543464e-01 -7.89763629e-01 3.78489941e-01 -5.27818143e-01
-1.06420422e+00 5.47475159e-01 -1.14962675e-01 -1.24225867e+00
-3.14048558e-01 -5.45592785e-01 -5.32948494e-01 5.03938556e-01
-1.80950272e+00 -1.20454288e+00 -4.68202412e-01 6.64763689e-01
8.24882269e-01 -2.20671110e-02 6.04362011e-01 5.31124532e-01
-6.45540893e-01 3.38800073e-01 5.18979318e-02 4.49617356e-01
6.85498118e-01 -1.23998606e+00 2.34918445e-01 1.10594249e+00
6.06076837e-01 6.04178198e-02 5.67400157e-01 -2.62274802e-01
-1.06122124e+00 -1.30380821e+00 7.28636324e-01 -4.18050200e-01
8.12066019e-01 -1.35329723e-01 -8.32838953e-01 7.10023880e-01
3.19742784e-02 3.72357965e-01 6.11444890e-01 7.50381127e-03
-3.46866697e-01 -5.55460602e-02 -9.32941020e-01 5.96788764e-01
9.42573249e-01 -4.60167229e-01 -3.69888961e-01 4.18875962e-01
5.83655119e-01 -1.47597939e-01 -6.01962447e-01 4.10470992e-01
2.75247931e-01 -9.57172155e-01 9.38891649e-01 -5.36126196e-01
6.11509800e-01 -3.73788148e-01 -1.70240179e-01 -1.05636680e+00
-2.44232193e-02 -5.49639702e-01 7.36806169e-02 1.13913202e+00
5.86703360e-01 -3.81781131e-01 1.23720288e+00 8.31580997e-01
-1.36902079e-01 -7.74234891e-01 -6.32977068e-01 -6.21632695e-01
7.11858273e-02 -5.41926622e-01 -3.91116627e-02 1.04986715e+00
-4.95528877e-01 4.91946071e-01 -2.36815274e-01 3.48058566e-02
8.09293747e-01 2.12401703e-01 9.33919132e-01 -1.27226746e+00
-3.12805653e-01 -2.88136512e-01 -3.78449976e-01 -1.16601002e+00
4.03881520e-01 -8.18804085e-01 2.01671988e-01 -1.37251604e+00
1.35558650e-01 -6.60908580e-01 -2.03396618e-01 4.51767653e-01
-2.99620181e-01 6.40717626e-01 2.89801538e-01 3.06669742e-01
-8.44389379e-01 3.31009597e-01 1.43647206e+00 -3.01650494e-01
-3.33165497e-01 1.07307285e-01 -4.59006429e-01 8.75749171e-01
9.13177073e-01 -4.16615516e-01 -4.86073583e-01 -4.17501479e-01
-3.78322512e-01 -1.11081988e-01 4.60925549e-01 -1.00031078e+00
2.78829783e-01 -1.55687213e-01 1.34723559e-01 -4.44384098e-01
1.84283465e-01 -9.44947898e-01 -3.29956949e-01 4.03021991e-01
-2.43638873e-01 -2.97118038e-01 -7.21181780e-02 6.02983832e-01
-4.06812012e-01 -2.66664177e-01 9.98363733e-01 -2.78203338e-01
-1.12245393e+00 4.54650789e-01 -1.91411972e-01 4.54950392e-01
1.10295045e+00 -3.72937441e-01 -2.15208128e-01 -1.35092661e-01
-7.29219317e-01 3.36840421e-01 4.24290150e-01 1.45336553e-01
2.97261864e-01 -7.71729589e-01 -7.42441833e-01 1.27327725e-01
2.20437229e-01 4.00434822e-01 1.55034542e-01 9.55145597e-01
-5.89259267e-01 3.06611836e-01 -6.99585304e-02 -9.22873557e-01
-1.02438951e+00 4.47563380e-01 1.69489384e-01 -4.27478731e-01
-6.68402433e-01 7.87677944e-01 1.37174904e-01 -4.65142101e-01
2.93970048e-01 -9.07018557e-02 -9.03449357e-02 1.43452540e-01
3.23162794e-01 3.56571853e-01 -2.91078463e-02 -7.54948616e-01
-9.47457626e-02 7.15500474e-01 -5.45842424e-02 -1.66180339e-02
1.29764128e+00 -1.51297361e-01 1.02502182e-01 2.24730164e-01
1.43243289e+00 -1.39963210e-01 -1.71813726e+00 -2.64182627e-01
1.96248055e-01 -3.84173810e-01 5.96043468e-02 -6.06227338e-01
-1.23724544e+00 9.39498901e-01 5.64382732e-01 -3.41166593e-02
1.01571763e+00 9.68916249e-03 9.12807822e-01 4.82375503e-01
4.39862043e-01 -1.11957097e+00 7.28799552e-02 4.03365135e-01
1.68424740e-01 -1.84796655e+00 -8.87255296e-02 -4.86763000e-01
-7.43404031e-01 8.35521579e-01 5.73573709e-01 -3.72477382e-01
6.23924494e-01 -3.55311632e-02 3.14819634e-01 1.31680677e-03
-4.66495961e-01 -6.45701706e-01 4.63750362e-01 4.65138018e-01
3.82989556e-01 1.62591692e-02 -6.48755431e-02 9.41429473e-03
1.53980002e-01 2.10055429e-03 3.09105158e-01 9.06191945e-01
-6.76346004e-01 -1.11713791e+00 -1.82522535e-01 5.34618199e-01
-4.31130141e-01 -1.17317401e-01 -9.24610272e-02 7.34721005e-01
6.56100586e-02 1.11740625e+00 9.02981609e-02 -2.04766318e-01
1.67163730e-01 7.71785676e-02 3.61520171e-01 -5.88238180e-01
-4.83274579e-01 2.25988418e-01 1.03370123e-01 -5.52404642e-01
-8.27137232e-01 -6.64143622e-01 -1.19112635e+00 -1.07255958e-01
-2.72929639e-01 7.67412130e-04 4.30700809e-01 1.06105828e+00
8.39112997e-02 1.90674350e-01 5.34985840e-01 -1.01436806e+00
-4.77287829e-01 -7.66429186e-01 -3.52717072e-01 6.04124784e-01
3.04188222e-01 -5.22725880e-01 -3.37203920e-01 4.95819241e-01] | [9.281479835510254, 0.5183008313179016] |
db16ef11-2b26-46bc-bc4f-c6efed34f724 | awte-bert-attending-to-wordpiece-tokenization | 2211.14829 | null | https://arxiv.org/abs/2211.14829v3 | https://arxiv.org/pdf/2211.14829v3.pdf | ESIE-BERT: Enriching Sub-words Information Explicitly with BERT for Joint Intent Classification and SlotFilling | Natural language understanding (NLU) has two core tasks: intent classification and slot filling. The success of pre-training language models resulted in a significant breakthrough in the two tasks. One of the promising solutions called BERT can jointly optimize the two tasks. We note that BERT-based models convert each complex token into multiple sub-tokens by wordpiece algorithm, which generates a mismatch between the lengths of the tokens and the labels. This leads to BERT-based models do not do well in label prediction which limits model performance improvement. Many existing models can be compatible with this issue but some hidden semantic information is discarded in the fine-tuning process. We address the problem by introducing a novel joint method on top of BERT which explicitly models the multiple sub-tokens features after wordpiece tokenization, thereby contributing to the two tasks. Our method can well extract the contextual features from complex tokens by the proposed sub-words attention adapter (SAA), which preserves overall utterance information. Additionally, we propose an intent attention adapter (IAA) to obtain the full sentence features to aid users to predict intent. Experimental results confirm that our proposed model is significantly improved on two public benchmark datasets. In particular, the slot filling F1 score is improved from 96.1 to 98.2 (2.1% absolute) on the Airline Travel Information Systems (ATIS) dataset. | ['Qing Li', 'Yu Zhao', 'Shaopeng Wei', 'Huaming Du', 'Leilei Wang', 'Huangen Chen', 'Gang Wu', 'Xingyan Chen', 'Zhilong Xie', 'Yu Guo'] | 2022-11-27 | null | null | null | null | ['intent-classification', 'slot-filling'] | ['natural-language-processing', 'natural-language-processing'] | [ 3.70245397e-01 3.28945875e-01 -5.63769281e-01 -6.41093135e-01
-8.17172170e-01 -1.38533324e-01 3.79082561e-01 2.46279433e-01
-5.51059008e-01 6.10432208e-01 4.62734431e-01 -2.75964260e-01
1.00738198e-01 -6.97282553e-01 -3.95351380e-01 -3.98496509e-01
4.75710273e-01 4.48324233e-01 1.21120550e-01 -4.07718718e-01
2.43677497e-01 -2.56906509e-01 -1.41085291e+00 4.70815092e-01
1.30223441e+00 1.10176480e+00 4.84404296e-01 2.61024654e-01
-8.41368735e-01 6.43135071e-01 -3.87582302e-01 -3.12079042e-01
5.50798476e-02 -1.66504592e-01 -1.01573205e+00 -8.94703120e-02
-1.70242116e-01 -3.18653464e-01 8.36002156e-02 8.33916724e-01
2.57458419e-01 3.34371597e-01 4.67106372e-01 -1.18736327e+00
-4.73318845e-01 6.97373211e-01 -5.44910252e-01 -1.68616503e-01
2.29800314e-01 -1.90744594e-01 1.50511515e+00 -9.36920583e-01
2.39594802e-01 1.34836602e+00 5.78851819e-01 4.89396691e-01
-8.38170946e-01 -5.34616947e-01 3.76417786e-01 2.89579719e-01
-1.34096980e+00 -3.63629222e-01 7.62069762e-01 -2.67363518e-01
1.26267827e+00 4.08302307e-01 9.70321298e-02 7.19865203e-01
-2.15060879e-02 1.24861717e+00 8.89532626e-01 -6.24432683e-01
-4.66717072e-02 4.49065506e-01 7.47975111e-01 6.00392401e-01
-1.69639364e-01 -1.84116974e-01 -3.02219689e-01 4.45299186e-02
2.77455896e-01 3.60201597e-01 -3.39368470e-02 2.25181162e-01
-1.11641192e+00 9.36269104e-01 3.44429612e-01 2.42579371e-01
-2.77728051e-01 -1.54123574e-01 6.09235227e-01 6.10770546e-02
5.49100220e-01 3.44334245e-01 -7.32734144e-01 -2.53149092e-01
-4.92367178e-01 1.31796286e-01 4.60459948e-01 1.07503903e+00
9.70696509e-01 -3.09438944e-01 -6.02554679e-01 1.20744407e+00
4.31842744e-01 1.48219585e-01 8.30645263e-01 -4.40921575e-01
6.91405714e-01 9.65726852e-01 1.05372526e-01 -8.52438390e-01
-6.35564625e-01 -5.98945856e-01 -6.35380805e-01 -4.93314922e-01
1.09305024e-01 -1.59005076e-02 -9.38547313e-01 1.72677374e+00
3.40997010e-01 1.23716079e-01 1.37248933e-01 7.60870636e-01
8.67526293e-01 8.40473950e-01 5.13552547e-01 -1.58577904e-01
1.81780088e+00 -1.41744578e+00 -1.07963216e+00 -5.96483767e-01
1.11363733e+00 -8.56985867e-01 1.35998964e+00 1.37938768e-01
-7.18615115e-01 -7.84264266e-01 -1.02664542e+00 -3.56429040e-01
-4.58677471e-01 3.47389132e-01 7.83068001e-01 6.23588145e-01
-4.06807899e-01 1.93120182e-01 -5.21456540e-01 -1.45016879e-01
2.34532639e-01 3.45664471e-01 -1.16547666e-01 -1.53802425e-01
-1.59276366e+00 5.88653088e-01 5.67807496e-01 1.52000099e-01
-1.48231462e-01 -6.85075521e-01 -1.05315459e+00 3.31761152e-01
6.59798563e-01 -6.58351719e-01 1.34162974e+00 -7.66033530e-01
-1.38385522e+00 5.35325468e-01 -6.57491028e-01 -3.27962965e-01
1.89695448e-01 -3.70007545e-01 -4.44691300e-01 -3.29078287e-01
4.19164181e-01 7.75354445e-01 4.89856780e-01 -1.01629710e+00
-1.00982022e+00 -2.87348419e-01 6.51633739e-02 4.84640419e-01
-5.15396833e-01 -8.87969509e-02 -5.21445692e-01 -6.34842694e-01
2.27520421e-01 -6.99831188e-01 -1.48985967e-01 -6.05559170e-01
-4.31418598e-01 -7.89415956e-01 5.65880716e-01 -6.24669373e-01
1.61434424e+00 -2.20066452e+00 -2.64841259e-01 -1.59474388e-01
1.47165945e-02 3.84350598e-01 -1.99055508e-01 4.32722628e-01
-1.18492236e-02 2.36370638e-01 -1.21501163e-01 -9.03759539e-01
2.28419706e-01 3.49652439e-01 -6.41952515e-01 -1.47765234e-01
2.55850732e-01 1.07857811e+00 -8.14821422e-01 -4.03740078e-01
1.36091828e-01 9.98536497e-02 -7.32387841e-01 3.82646739e-01
-2.69838035e-01 1.98492333e-01 -5.89709640e-01 5.56549072e-01
7.85880029e-01 -5.24742641e-02 1.72436126e-02 -2.20991418e-01
-5.64140938e-02 9.12684917e-01 -1.05132830e+00 1.73358071e+00
-7.89670885e-01 5.81510365e-02 -2.73641765e-01 -1.07321692e+00
1.06556773e+00 4.22432065e-01 4.18829054e-01 -9.24565494e-01
1.13042235e-01 2.57229775e-01 -1.46846667e-01 -5.77618003e-01
7.76045620e-01 -4.33404624e-01 -3.96694899e-01 4.10163611e-01
-2.27054302e-02 3.24028462e-01 3.80489677e-02 1.17463648e-01
6.95631027e-01 -1.90036505e-01 3.21823269e-01 -1.94845334e-01
7.76456654e-01 -1.57231949e-02 9.25361931e-01 6.88169837e-01
-3.06343198e-01 4.60433900e-01 4.07291591e-01 -4.29322153e-01
-6.05525970e-01 -5.69670796e-01 -5.70249930e-02 1.33803499e+00
3.05347711e-01 -6.51302397e-01 -6.77727938e-01 -9.95509744e-01
-9.23555270e-02 9.83802915e-01 -5.56439281e-01 -3.24539006e-01
-4.89941031e-01 -6.78263366e-01 3.22733015e-01 6.76596165e-01
6.02703452e-01 -1.15500641e+00 -1.74834728e-01 3.83282870e-01
-6.97304070e-01 -1.19432914e+00 -7.19326258e-01 2.13541612e-01
-6.04301035e-01 -6.85964167e-01 -3.36326689e-01 -8.65383863e-01
5.40106833e-01 4.97014374e-01 9.19587314e-01 3.68704736e-01
-1.04632415e-02 -1.61076039e-01 -7.52900183e-01 -4.29034680e-01
-2.04170138e-01 4.54275012e-01 4.85882498e-02 2.01776564e-01
7.92654812e-01 -1.74793616e-01 -3.93496901e-01 3.93820316e-01
-9.48477566e-01 5.23756742e-01 5.52737057e-01 1.08097601e+00
5.25470138e-01 1.05383173e-01 9.46935117e-01 -1.00471628e+00
5.70450962e-01 -6.43774092e-01 -5.29569015e-02 2.76656568e-01
-8.21765065e-01 1.49219245e-01 7.29754329e-01 -1.29910901e-01
-1.28344774e+00 -1.12102032e-01 -5.49906552e-01 1.29017860e-01
-2.89680392e-01 6.53017700e-01 -3.49790901e-01 5.56231380e-01
1.01587750e-01 3.01332504e-01 -2.11260468e-01 -7.08094418e-01
1.81178644e-01 1.08998752e+00 1.83843508e-01 -4.57279086e-01
3.68133724e-01 1.82354361e-01 -5.24297893e-01 -6.42989993e-01
-1.35469258e+00 -9.44903255e-01 -5.30565798e-01 3.16135585e-01
7.38360226e-01 -8.31143498e-01 -6.62347555e-01 3.22148979e-01
-1.22550845e+00 -3.28567997e-02 -1.93890631e-01 3.41009259e-01
-3.45966220e-01 5.69802463e-01 -5.06371021e-01 -1.01555073e+00
-4.74936992e-01 -1.18861187e+00 1.16954124e+00 3.35353673e-01
-3.23078841e-01 -9.67514277e-01 -1.59736305e-01 8.30086172e-01
3.66697133e-01 -4.29924011e-01 1.17972016e+00 -9.48837161e-01
-3.24783266e-01 -1.15176946e-01 -4.44223970e-01 2.54525572e-01
2.45892838e-01 -6.11439943e-01 -1.07712770e+00 3.76695907e-03
1.09474987e-01 -3.26478004e-01 1.13892424e+00 2.59376049e-01
1.06118977e+00 -2.89287627e-01 -4.04224068e-01 3.74069393e-01
1.11413574e+00 3.85858625e-01 5.18065035e-01 3.75988275e-01
6.80901468e-01 7.67383039e-01 1.12620878e+00 3.77008319e-01
7.06269801e-01 8.79735827e-01 1.94312379e-01 -1.87615842e-01
1.20104887e-02 -5.45690119e-01 2.73123980e-01 8.92792881e-01
4.19981986e-01 -2.40135968e-01 -7.99219966e-01 6.29607379e-01
-2.10521674e+00 -4.77853417e-01 -1.26268834e-01 1.96631098e+00
9.20771182e-01 3.32028538e-01 -1.20147653e-01 1.99407667e-01
4.86114383e-01 1.61629781e-01 -3.51905167e-01 -4.39505309e-01
2.67864585e-01 4.37965542e-02 3.24154168e-01 6.93908751e-01
-1.19532597e+00 1.25363994e+00 5.14270544e+00 1.21195078e+00
-6.79087996e-01 1.28445044e-01 6.95563257e-01 2.82117486e-01
-5.63479304e-01 7.70978779e-02 -1.20722127e+00 5.51091313e-01
8.11709464e-01 -1.27547935e-01 1.34006530e-01 8.48493516e-01
2.62516618e-01 -6.62749410e-02 -8.59434128e-01 7.03032017e-01
-1.32778734e-01 -1.10970438e+00 1.52742669e-01 -1.70629635e-01
4.41202313e-01 -3.63588929e-01 1.06467372e-02 1.04280198e+00
-1.63039058e-01 -9.62414861e-01 5.64871728e-01 2.78680027e-01
6.29260778e-01 -8.99371743e-01 1.13260043e+00 6.75975084e-01
-1.30743349e+00 -1.63728878e-01 -5.12270331e-01 -3.14219654e-01
3.93799096e-01 6.36814892e-01 -1.16663432e+00 7.37959266e-01
2.66807765e-01 6.03232682e-01 -2.99995542e-01 7.36047983e-01
-9.70805660e-02 7.24332273e-01 -1.33103088e-01 -7.44565427e-02
4.29212064e-01 -1.39765486e-01 4.05009836e-01 1.13827693e+00
3.17619592e-02 1.02594554e-01 4.27184999e-01 9.28237677e-01
2.25740522e-02 4.66307998e-01 -1.51910290e-01 -7.66457990e-02
4.31483775e-01 1.22742760e+00 -5.96221685e-01 -4.17215914e-01
-5.28554976e-01 9.79409039e-01 3.69070590e-01 2.04929098e-01
-8.78417432e-01 -5.79183459e-01 7.73434639e-01 -1.25195861e-01
8.73150602e-02 -7.60967284e-02 -6.37513697e-01 -9.94238675e-01
1.07583269e-01 -6.10079348e-01 4.80165452e-01 -3.79751146e-01
-1.10707402e+00 5.68314612e-01 -1.86271951e-01 -1.03449368e+00
-1.13969900e-01 -3.81615281e-01 -5.02259135e-01 9.12091255e-01
-1.96875083e+00 -1.25470173e+00 -1.35816723e-01 1.77516028e-01
1.13196957e+00 -2.94257458e-02 8.43246996e-01 5.05835354e-01
-8.67135763e-01 8.70870054e-01 -1.27671883e-01 4.95469756e-02
7.03141510e-01 -1.18702996e+00 5.92538059e-01 6.03931367e-01
7.54755735e-02 7.50839353e-01 4.36801404e-01 -6.10217035e-01
-1.00868630e+00 -1.04919505e+00 1.49490881e+00 -2.48411268e-01
3.07926744e-01 -3.90347362e-01 -1.02237642e+00 6.19192839e-01
-2.57305969e-02 -3.16192687e-01 9.19271588e-01 4.50387299e-01
-2.64768541e-01 9.69000608e-02 -8.49554300e-01 4.07433659e-01
8.45358133e-01 -3.46348852e-01 -8.14080477e-01 3.69239867e-01
1.32209837e+00 -3.34763348e-01 -4.34141755e-01 5.03861904e-01
3.20983708e-01 -8.69799316e-01 8.66937637e-01 -6.84041798e-01
3.95795524e-01 -2.04513207e-01 -1.15698263e-01 -1.08593678e+00
-2.70928979e-01 -3.49250525e-01 -3.69759835e-02 1.53697515e+00
4.25629884e-01 -7.32351959e-01 6.61607683e-01 8.34168851e-01
-4.42964733e-01 -1.10601842e+00 -7.51022756e-01 -5.56533098e-01
-1.65583700e-01 -6.38858199e-01 9.07548487e-01 8.50428283e-01
2.35018656e-01 6.13489926e-01 -4.41662818e-01 -1.01820633e-01
6.90749064e-02 2.59550840e-01 5.38972378e-01 -1.04715943e+00
-2.19529524e-01 -3.74964029e-01 6.94175214e-02 -1.62783515e+00
3.74640316e-01 -1.03243625e+00 1.14824757e-01 -1.45589960e+00
3.21599275e-01 -9.74606931e-01 -4.86163139e-01 8.18636000e-01
-7.00770199e-01 -1.37101486e-01 3.89865227e-02 1.67961955e-01
-7.20782399e-01 9.10649478e-01 1.03452766e+00 -1.50247857e-01
-3.42470735e-01 2.43950248e-01 -9.88962591e-01 6.04493737e-01
8.34791243e-01 -5.58937490e-01 -4.21840131e-01 -4.04568672e-01
4.06618603e-02 -7.07093626e-02 -6.97767809e-02 -6.75750375e-01
6.63606972e-02 -2.70991951e-01 -1.18992187e-01 -6.82745337e-01
2.71499366e-01 -8.64188790e-01 -4.52892989e-01 2.79319316e-01
-5.29166639e-01 -3.18837583e-01 1.98704928e-01 5.70999026e-01
-3.57268453e-01 -5.29873252e-01 4.32063848e-01 3.39819714e-02
-9.16643441e-01 1.37847841e-01 -1.81373686e-01 2.52718087e-02
7.59190679e-01 -1.33061931e-01 -1.93937749e-01 -2.04099983e-01
-3.83384794e-01 6.29629374e-01 -4.03691530e-02 7.46110022e-01
3.68311435e-01 -1.37544274e+00 -4.69234854e-01 6.50069296e-01
2.56850451e-01 1.93772823e-01 6.04264915e-01 8.05145919e-01
-3.37758549e-02 9.14785981e-01 1.66165590e-01 -4.13608044e-01
-1.08665180e+00 4.22027588e-01 6.93859234e-02 -8.01562548e-01
-3.16886336e-01 9.72536623e-01 4.51090425e-01 -7.49769211e-01
2.32823193e-01 -4.84699249e-01 -3.40060472e-01 7.31541887e-02
6.70521915e-01 1.16091646e-01 1.75037608e-01 -6.24639869e-01
-3.27258080e-01 3.20984989e-01 -4.90634084e-01 1.21103562e-01
1.16168523e+00 -4.60048348e-01 2.77328845e-02 4.23671037e-01
1.37380815e+00 -2.45709997e-02 -9.26377773e-01 -5.14282405e-01
1.86663747e-01 -4.76911366e-01 1.19570687e-01 -8.18656921e-01
-7.51481295e-01 1.01074946e+00 2.02470496e-01 2.95093417e-01
1.03902972e+00 -1.50562823e-01 1.52410126e+00 2.63477057e-01
2.04116687e-01 -1.17232299e+00 1.74931381e-02 8.29763651e-01
6.37006462e-01 -1.49969828e+00 -3.66217226e-01 -6.50870204e-01
-8.87689352e-01 8.95975769e-01 7.21565068e-01 2.97353089e-01
4.44643915e-01 -4.07322310e-02 1.33241206e-01 -6.50546607e-03
-7.68810987e-01 -2.63542384e-01 5.03331780e-01 1.63030058e-01
5.72751939e-01 1.48218393e-01 -6.36417747e-01 1.39958370e+00
-1.46163434e-01 -2.77152807e-01 1.44421697e-01 7.80162215e-01
-7.56467879e-01 -1.44156218e+00 -1.31359383e-01 4.30881172e-01
-2.89451569e-01 -2.68661261e-01 -4.69999947e-02 5.05928576e-01
1.77757591e-01 1.03823936e+00 -4.64926590e-04 -5.03619492e-01
3.42888713e-01 5.33395112e-01 -2.01794103e-01 -7.92128503e-01
-5.69750607e-01 3.28730121e-02 1.04341611e-01 -4.53499556e-01
-1.17987677e-01 -4.28778172e-01 -1.60094774e+00 3.98795167e-03
-6.49374366e-01 3.39520216e-01 5.01121879e-01 1.22876263e+00
4.96204823e-01 7.31017530e-01 7.60556817e-01 -3.99113029e-01
-6.38040423e-01 -1.04928529e+00 -5.18825173e-01 5.20922959e-01
2.71833181e-01 -8.59865069e-01 -2.40825072e-01 -1.71192676e-01] | [12.515287399291992, 7.332326889038086] |
5da1bb99-f4aa-444f-aa7e-f922118cf61d | text-style-transfer-for-bias-mitigation-using | 2201.08643 | null | https://arxiv.org/abs/2201.08643v1 | https://arxiv.org/pdf/2201.08643v1.pdf | Text Style Transfer for Bias Mitigation using Masked Language Modeling | It is well known that textual data on the internet and other digital platforms contain significant levels of bias and stereotypes. Although many such texts contain stereotypes and biases that inherently exist in natural language for reasons that are not necessarily malicious, there are crucial reasons to mitigate these biases. For one, these texts are being used as training corpus to train language models for salient applications like cv-screening, search engines, and chatbots; such applications are turning out to produce discriminatory results. Also, several research findings have concluded that biased texts have significant effects on the target demographic groups. For instance, masculine-worded job advertisements tend to be less appealing to female applicants. In this paper, we present a text style transfer model that can be used to automatically debias textual data. Our style transfer model improves on the limitations of many existing style transfer techniques such as loss of content information. Our model solves such issues by combining latent content encoding with explicit keyword replacement. We will show that this technique produces better content preservation whilst maintaining good style transfer accuracy. | ['Toon Calders', 'Ewoenam Kwaku Tokpo'] | 2022-01-21 | null | https://aclanthology.org/2022.naacl-srw.21 | https://aclanthology.org/2022.naacl-srw.21.pdf | naacl-acl-2022-7 | ['text-style-transfoer'] | ['natural-language-processing'] | [ 3.93844098e-01 3.31889480e-01 -6.18751466e-01 -5.37247300e-01
-5.40311456e-01 -6.81660235e-01 8.63195300e-01 2.75219142e-01
-7.05636442e-01 7.57345021e-01 6.67649448e-01 -6.54941440e-01
3.01181108e-01 -6.96774781e-01 -3.27889174e-01 -1.57106638e-01
5.86738348e-01 5.05523980e-01 1.60064638e-01 -7.22256362e-01
8.87916267e-01 1.58360228e-01 -1.18652987e+00 5.55098057e-01
9.97454166e-01 4.00183737e-01 -2.85286069e-01 3.57489288e-01
-5.26733220e-01 8.60816360e-01 -8.54063272e-01 -1.26538634e+00
-3.29970606e-02 -3.43870699e-01 -1.04537368e+00 -3.14553976e-01
9.22147214e-01 -4.05578494e-01 -1.29000694e-01 1.39390898e+00
6.88760042e-01 -1.57480136e-01 8.56763542e-01 -9.76616204e-01
-1.08806443e+00 9.06449139e-01 -5.91932654e-01 2.24500731e-01
2.49246016e-01 -9.59517807e-02 8.35340023e-01 -5.60289025e-01
7.41667330e-01 1.96128714e+00 7.10368991e-01 9.13306415e-01
-1.56129718e+00 -1.15009689e+00 -1.75754875e-02 -3.72083396e-01
-8.82270634e-01 -5.19619226e-01 9.05131876e-01 -5.12268364e-01
2.55501360e-01 4.02516752e-01 2.95550525e-01 1.81746471e+00
5.01765192e-01 7.87443221e-01 1.48246646e+00 -4.11755621e-01
-2.44154241e-02 9.88732874e-01 8.25114176e-02 2.48319224e-01
4.29085314e-01 -2.84388185e-01 -6.93571448e-01 -5.87900460e-01
3.04521799e-01 -3.15864444e-01 1.83354050e-01 -1.84065014e-01
-7.41306782e-01 1.32718885e+00 9.22283083e-02 3.62142056e-01
6.85897246e-02 -2.17637792e-01 7.54470885e-01 6.01531386e-01
1.09609199e+00 8.32303464e-01 2.30660867e-02 -2.84660220e-01
-8.02142203e-01 5.97756088e-01 9.77608323e-01 8.68786693e-01
4.87923533e-01 -1.80088550e-01 -3.93866956e-01 1.19599903e+00
1.34725451e-01 5.16038656e-01 8.35889697e-01 -8.45942080e-01
6.11495018e-01 4.66757774e-01 1.34024054e-01 -1.61429870e+00
7.75374919e-02 -2.85650820e-01 -4.62509632e-01 -8.86034518e-02
5.22279203e-01 -2.50167787e-01 -5.33681750e-01 1.89948547e+00
1.15956500e-01 -6.47007167e-01 -4.45057601e-01 5.36406696e-01
3.57920319e-01 3.61437947e-01 6.61745608e-01 1.48567438e-01
1.36197746e+00 -1.93595201e-01 -8.34139705e-01 -6.34852529e-01
8.36424470e-01 -1.15062320e+00 1.54152441e+00 1.77445963e-01
-1.02330375e+00 -2.41442248e-01 -7.46598423e-01 -2.36041903e-01
-4.56741929e-01 -4.13108706e-01 6.74835503e-01 1.50386560e+00
-7.47226059e-01 6.49262488e-01 -1.42623559e-01 -5.38296521e-01
6.51291847e-01 -1.31216720e-02 -4.53652032e-02 1.31961927e-01
-1.43674088e+00 1.10408854e+00 8.06502812e-03 -4.55928504e-01
-3.18517923e-01 -8.92972708e-01 -7.16378927e-01 -2.86473006e-01
1.31428748e-01 -2.79792964e-01 1.46121156e+00 -1.69502568e+00
-1.20944071e+00 1.26891661e+00 3.81621607e-02 -2.13113353e-01
7.69634068e-01 -4.79096949e-01 -3.32050294e-01 -1.13181852e-01
5.79715192e-01 6.58577085e-01 1.22748256e+00 -1.04978347e+00
-3.60828817e-01 -3.61808658e-01 8.08901936e-02 2.17076480e-01
-1.13583660e+00 5.48134327e-01 1.79552466e-01 -1.08623147e+00
-4.64774758e-01 -9.48164701e-01 -4.02596407e-03 -9.46015492e-02
-2.61393130e-01 -1.86288223e-01 8.09544981e-01 -9.33689773e-01
1.31092727e+00 -2.07614684e+00 -1.52722538e-01 1.65904656e-01
1.50665924e-01 3.14676791e-01 1.36522770e-01 5.07159173e-01
1.95421919e-01 8.16894948e-01 2.02815622e-01 -7.59726018e-03
2.89066024e-02 -2.30259925e-01 -4.47152287e-01 3.60403329e-01
2.86466088e-02 7.38187134e-01 -8.40480745e-01 -7.63040006e-01
-3.97386789e-01 2.54857033e-01 -8.04096103e-01 -2.53867716e-01
-1.00628741e-01 2.06250295e-01 -4.24063742e-01 2.88881928e-01
5.55918932e-01 2.44872756e-02 2.24997044e-01 2.60451704e-01
-6.31371792e-03 9.32845473e-01 -5.73560059e-01 1.22522402e+00
-5.03250539e-01 8.70535493e-01 2.20616683e-01 -4.10039723e-01
7.97243237e-01 1.56888328e-02 7.16118962e-02 -6.46145523e-01
4.08768207e-01 1.92139491e-01 9.39781815e-02 -2.62327880e-01
1.10317707e+00 -6.02982402e-01 -4.82334465e-01 7.31865108e-01
-3.69224012e-01 -2.90374935e-01 2.27729511e-02 6.35807753e-01
5.77924907e-01 -1.60740808e-01 -5.49154244e-02 -6.40933335e-01
3.60975802e-01 1.20116957e-01 3.15008163e-01 8.80714357e-01
-2.45146379e-01 2.04817295e-01 6.40855432e-01 -2.64417734e-02
-1.29208279e+00 -5.97489953e-01 -8.85965377e-02 1.48608232e+00
3.10834330e-02 -2.33536333e-01 -8.63087416e-01 -7.74081230e-01
1.83167905e-01 1.26279962e+00 -6.11330986e-01 -5.72076201e-01
-3.01480353e-01 -4.23673302e-01 9.07367527e-01 2.68872052e-01
3.65561247e-01 -9.25208271e-01 -1.93423420e-01 1.39650151e-01
-2.78471738e-01 -7.50148892e-01 -8.39651704e-01 -4.05857235e-01
-9.12221253e-01 -4.30096060e-01 -7.65243053e-01 -6.26044035e-01
5.72620094e-01 3.92674118e-01 1.24676597e+00 2.44224638e-01
-1.70646429e-01 4.58205760e-01 -3.77012908e-01 -9.30982828e-01
-1.00703895e+00 4.92853075e-01 1.82642248e-02 -7.76267797e-02
7.98034191e-01 -3.58566433e-01 -3.20461035e-01 2.57753819e-01
-1.21818924e+00 -5.31589575e-02 5.55613995e-01 8.07484508e-01
-4.69579935e-01 -3.42060417e-01 6.63154125e-01 -1.74832928e+00
1.33067012e+00 -5.42785048e-01 -1.02602817e-01 -1.60042997e-02
-8.54071677e-01 -9.23623666e-02 2.81488746e-01 -7.68979311e-01
-1.44471085e+00 -5.62364459e-01 -8.98997709e-02 1.90847620e-01
-1.40932605e-01 2.19770730e-01 1.11106977e-01 -7.32976198e-02
1.05047297e+00 -4.76744026e-03 3.10056269e-01 -4.67638165e-01
1.98610619e-01 1.26664472e+00 -1.38148442e-01 -8.25317085e-01
9.37869370e-01 4.39783633e-01 -6.46353364e-01 -9.68044162e-01
-9.30665791e-01 -3.52891296e-01 -2.12089181e-01 -2.62168765e-01
5.53721905e-01 -7.26980865e-01 -4.17777687e-01 2.36727104e-01
-8.98315549e-01 -2.51864344e-01 1.14327423e-01 2.50469327e-01
-1.78229332e-01 5.58604896e-01 -9.61665332e-01 -7.81568527e-01
-3.62390935e-01 -6.47097707e-01 8.07266474e-01 2.29449943e-02
-1.03962326e+00 -1.10713172e+00 -2.90395711e-02 6.16358638e-01
7.64991343e-01 -2.19699085e-01 1.29452527e+00 -8.43900442e-01
2.01015741e-01 -2.81925052e-01 -1.02502942e-01 4.06529516e-01
1.97818339e-01 -1.98932394e-01 -9.01441813e-01 -3.61777931e-01
-3.25571885e-03 -7.72419691e-01 4.93516296e-01 -7.91718140e-02
1.09334743e+00 -6.61412477e-01 -3.11744004e-01 -9.60918963e-02
1.09569013e+00 -4.57471833e-02 7.06894696e-01 4.16706532e-01
5.04374444e-01 1.12187445e+00 4.47283596e-01 2.36093923e-01
2.17361107e-01 4.84814137e-01 -1.62414536e-01 8.14393610e-02
1.16828218e-01 -7.06614912e-01 5.60489655e-01 5.72710574e-01
4.41513956e-01 4.21680696e-02 -6.94971502e-01 6.44027352e-01
-1.27064335e+00 -1.30353022e+00 -5.79855479e-02 2.13135290e+00
1.34169519e+00 4.34990764e-01 2.69302130e-01 -7.85946697e-02
8.53799760e-01 2.42737323e-01 -2.95181215e-01 -1.06575096e+00
1.06766753e-01 5.85181601e-02 8.65163028e-01 5.69162011e-01
-9.88313138e-01 1.05926895e+00 6.43859482e+00 7.91926503e-01
-1.13538337e+00 9.44725797e-02 9.88265753e-01 3.55939451e-03
-8.96910787e-01 -1.22872688e-01 -8.96160364e-01 5.87728143e-01
7.98584461e-01 -4.69672740e-01 2.38029197e-01 9.31219101e-01
2.65979990e-02 1.23297192e-01 -9.27330673e-01 7.28624344e-01
3.19290221e-01 -1.00771737e+00 2.77019441e-01 1.62754580e-01
7.28391528e-01 -5.95355511e-01 6.23866975e-01 5.45287251e-01
5.01499832e-01 -8.77905905e-01 8.89704049e-01 -1.43459827e-01
7.34342813e-01 -7.44380534e-01 4.52157527e-01 2.65320182e-01
-1.32838255e-02 5.74091226e-02 -5.86091280e-01 -2.83427417e-01
-2.49080688e-01 5.34442186e-01 -8.80567372e-01 -1.00329868e-01
5.33406317e-01 3.90295804e-01 -9.36758041e-01 4.23069865e-01
1.21220322e-02 8.24970663e-01 1.67976320e-01 -4.83257622e-01
1.79180026e-01 -1.20757289e-01 5.40401459e-01 1.31861067e+00
2.41295006e-02 -3.60868633e-01 -2.80279219e-01 8.66877317e-01
-3.24286312e-01 4.51676846e-01 -1.20004618e+00 -5.19601643e-01
5.43590009e-01 1.19085264e+00 -6.01162612e-01 -3.93402249e-01
-6.38414562e-01 1.04162955e+00 2.37323239e-01 1.60486251e-01
-4.35314327e-01 -4.70669121e-01 7.58235633e-01 4.32568073e-01
-2.22906843e-01 -7.91749582e-02 -7.07354307e-01 -1.14486849e+00
-2.93560088e-01 -1.58874249e+00 2.81779081e-01 -3.94005299e-01
-1.70899796e+00 -1.88807733e-02 2.01586057e-02 -7.80966759e-01
-1.86610743e-01 -6.21614695e-01 -2.49623686e-01 9.19426739e-01
-1.09775794e+00 -9.02510226e-01 1.95432901e-01 3.22454005e-01
6.88326001e-01 -1.87195808e-01 4.94217277e-01 3.81204963e-01
-1.55858144e-01 8.65198612e-01 -7.61285797e-02 4.30295736e-01
1.43874156e+00 -1.15754807e+00 4.59237099e-01 5.40523410e-01
-3.39453191e-01 1.07605767e+00 1.04194582e+00 -1.15651560e+00
-1.07077956e+00 -8.62251759e-01 1.24669898e+00 -7.07720816e-01
8.17147136e-01 -5.55511057e-01 -7.90441871e-01 8.20761979e-01
4.24786538e-01 -1.11844182e+00 1.03394198e+00 4.93522257e-01
-6.66714668e-01 2.20073819e-01 -1.17766571e+00 1.12401271e+00
1.00012505e+00 -7.41719127e-01 -7.96436429e-01 3.46740305e-01
4.00450915e-01 -1.29461408e-01 -4.11580056e-01 -1.74640656e-01
6.69171274e-01 -7.35741675e-01 7.43105233e-01 -9.69838917e-01
8.54924440e-01 5.86202443e-01 2.15819150e-01 -1.40878820e+00
-5.00338078e-01 -7.89686799e-01 5.95193148e-01 1.70479095e+00
6.20497644e-01 -6.95720196e-01 8.66762400e-01 1.25497413e+00
2.68213004e-01 1.11550443e-01 -5.47817111e-01 -6.19093835e-01
7.93302715e-01 -9.51415673e-02 3.58482420e-01 1.35236514e+00
8.15317035e-02 6.98837757e-01 -5.55884719e-01 -4.40683752e-01
6.46732509e-01 -1.98586121e-01 8.13213646e-01 -1.36760473e+00
4.91607971e-02 -6.50179029e-01 -3.68014686e-02 -7.90994287e-01
3.82726908e-01 -9.57619131e-01 -2.95904428e-01 -8.43623996e-01
3.59062314e-01 -5.50212622e-01 2.55607486e-01 5.02936495e-03
-3.06564003e-01 2.52421856e-01 1.12132609e-01 4.04882580e-02
-1.59780905e-01 2.73151100e-01 1.27693331e+00 -2.79625684e-01
-3.94259170e-02 -2.22666785e-02 -1.52673578e+00 6.82727337e-01
1.11767781e+00 -6.61116183e-01 -4.58244294e-01 -3.78316671e-01
7.99869061e-01 -5.23359656e-01 1.47612765e-01 -4.22737539e-01
-3.27833712e-01 -3.08534861e-01 3.54961634e-01 -7.51285953e-03
8.51252005e-02 -6.60926521e-01 -3.36209506e-01 5.20712137e-01
-1.14024985e+00 2.08773687e-01 2.63728410e-01 4.53496695e-01
7.70177618e-02 -4.84308928e-01 8.58149707e-01 -2.96902269e-01
-1.15833066e-01 -7.44951069e-02 -1.02995312e+00 5.26324272e-01
6.34942353e-01 -1.24330103e-01 -4.27253187e-01 -9.26651716e-01
-9.33007151e-03 -1.43899620e-01 7.39403069e-01 8.10136974e-01
1.94190010e-01 -1.38796997e+00 -7.36304760e-01 -1.65935531e-02
1.13673359e-01 -8.45499814e-01 -9.99277979e-02 5.43284893e-01
-3.56747687e-01 3.43068928e-01 -4.51830238e-01 -8.78081322e-02
-1.44288433e+00 4.25060481e-01 -2.28571687e-02 6.79898560e-02
-2.91477263e-01 8.41276646e-01 2.37280488e-01 -5.81251562e-01
5.48570752e-02 1.17759794e-01 -2.16736406e-01 3.05826724e-01
7.17752516e-01 4.48998690e-01 -1.52723327e-01 -6.12990975e-01
-1.74879842e-02 -1.14899084e-01 -6.66636944e-01 -3.81408632e-01
1.03049827e+00 -2.64375150e-01 -1.95341617e-01 4.70220208e-01
1.14629817e+00 5.57273209e-01 -5.65132678e-01 -2.86261678e-01
2.45917574e-01 -9.38127160e-01 -1.27009541e-01 -6.28764451e-01
-4.79026020e-01 8.82574022e-01 4.07024324e-01 4.38900143e-01
5.28136075e-01 -2.67617851e-01 6.99706614e-01 2.14219034e-01
3.30302685e-01 -1.57192826e+00 2.14621931e-01 4.19971317e-01
7.30388224e-01 -1.19254279e+00 1.59395281e-02 -4.84346569e-01
-9.09540832e-01 7.98522651e-01 7.30810583e-01 2.04851851e-01
3.43304276e-01 -3.17662507e-02 3.31231534e-01 -3.52855697e-02
-6.54204488e-01 3.83748621e-01 4.81012203e-02 5.96786559e-01
1.01164997e+00 -7.45781735e-02 -1.04322898e+00 2.09978670e-01
-4.35338527e-01 -2.98207849e-01 7.07958341e-01 9.11198318e-01
-3.28834236e-01 -1.38537252e+00 -6.14073694e-01 6.88165665e-01
-1.22915864e+00 -2.84216642e-01 -1.11858940e+00 5.39654493e-01
-4.64880079e-01 8.89516354e-01 -9.02777463e-02 -3.56380969e-01
-4.26829606e-02 4.39884305e-01 2.49823958e-01 -8.42965126e-01
-9.35800314e-01 -2.46820733e-01 6.09750032e-01 -2.47008249e-01
-1.32078767e-01 -8.48275721e-01 -3.55245948e-01 -8.23967576e-01
-1.69416979e-01 2.09832892e-01 4.85447764e-01 6.95887804e-01
2.37275168e-01 -3.71944718e-02 4.57669228e-01 -4.88866121e-01
-1.02024972e+00 -1.12255847e+00 -4.93596256e-01 9.72109675e-01
-4.74668108e-02 -5.01317024e-01 -1.07120946e-01 3.97137739e-02] | [9.1639986038208, 10.270920753479004] |
c727827f-3ce3-44ef-a58d-29ef1930640a | radar-based-materials-classification-using | 2202.05169 | null | https://arxiv.org/abs/2202.05169v1 | https://arxiv.org/pdf/2202.05169v1.pdf | Radar-based Materials Classification Using Deep Wavelet Scattering Transform: A Comparison of Centimeter vs. Millimeter Wave Units | Radar-based materials detection received significant attention in recent years for its potential inclusion in consumer and industrial applications like object recognition for grasping and manufacturing quality assurance and control. Several radar publications were developed for material classification under controlled settings with specific materials' properties and shapes. Recent literature has challenged the earlier findings on radars-based materials classification claiming that earlier solutions are not easily scaled to industrial applications due to a variety of real-world issues. Published experiments on the impact of these factors on the robustness of the extracted radar-based traditional features have already demonstrated that the application of deep neural networks can mitigate, to some extent, the impact to produce a viable solution. However, previous studies lacked an investigation of the usefulness of lower frequency radar units, specifically <10GHz, against the higher range units around and above 60GHz. This research considers two radar units with different frequency ranges: Walabot-3D (6.3-8 GHz) cm-wave and IMAGEVK-74 (62-69 GHz) mm-wave imaging units by Vayyar Imaging. A comparison is presented on the applicability of each unit for material classification. This work extends upon previous efforts, by applying deep wavelet scattering transform for the identification of different materials based on the reflected signals. In the wavelet scattering feature extractor, data is propagated through a series of wavelet transforms, nonlinearities, and averaging to produce low-variance representations of the reflected radar signals. This work is unique in comparison of the radar units and algorithms in material classification and includes real-time demonstrations that show strong performance by both units, with increased robustness offered by the cm-wave radar unit. | ['Andrew J. Hill', 'Rami N. Khushaba'] | 2022-02-08 | null | null | null | null | ['material-classification'] | ['computer-vision'] | [ 6.48535371e-01 -3.30571920e-01 4.29696500e-01 -2.50381440e-01
-5.76497138e-01 -2.26258323e-01 4.06538486e-01 -2.48491734e-01
-2.40710378e-01 5.13659179e-01 -5.74976467e-02 -1.65506363e-01
-9.53809679e-01 -1.19183147e+00 -1.23312570e-01 -1.16370428e+00
-5.08428395e-01 4.07360435e-01 -4.93023619e-02 -4.55592215e-01
2.48997301e-01 9.16141033e-01 -1.65842092e+00 3.74086440e-01
1.65700272e-01 1.29252386e+00 1.84975192e-01 4.87482160e-01
1.22434884e-01 1.64771706e-01 -9.99532819e-01 2.05470845e-01
5.65375268e-01 4.87146229e-02 -1.23775043e-01 -1.81964740e-01
2.79629350e-01 -3.89452517e-01 -2.06914961e-01 8.54783833e-01
7.59722233e-01 -7.30678961e-02 7.80733526e-01 -9.39017236e-01
-7.59225786e-01 5.28769553e-01 -4.36561853e-01 2.79038757e-01
4.53499407e-01 6.68776780e-02 5.26039958e-01 -5.34554541e-01
3.92490089e-01 1.08684266e+00 8.74685407e-01 2.12818459e-01
-8.50916922e-01 -8.38613272e-01 -5.04512846e-01 2.26375714e-01
-8.97697151e-01 -2.10890099e-02 1.10362899e+00 -4.96044517e-01
1.04431355e+00 4.39539969e-01 4.79932994e-01 1.34990847e+00
7.38214493e-01 6.06446117e-02 1.51720119e+00 -5.10609210e-01
1.30387545e-01 6.63539469e-02 2.80041873e-01 2.17103317e-01
5.66360056e-01 7.19052315e-01 -2.03755766e-01 -9.05994400e-02
5.59115529e-01 -1.50689825e-01 -3.10620546e-01 -4.03759778e-02
-8.98180664e-01 8.67504835e-01 2.69936889e-01 8.52562785e-01
-7.71472454e-01 -1.44888833e-02 2.55838811e-01 7.03001440e-01
5.69088578e-01 7.44325995e-01 -4.22241390e-01 1.20509766e-01
-7.75465846e-01 3.59432578e-01 7.51434028e-01 4.83666509e-01
2.55184531e-01 7.50040472e-01 7.89743513e-02 5.25435686e-01
4.05264348e-01 9.99320030e-01 1.35594174e-01 -5.08138299e-01
7.68417493e-02 -1.23627283e-01 -5.35791926e-02 -1.18237162e+00
-7.29566395e-01 -7.92716861e-01 -7.15964079e-01 6.90716267e-01
3.54379207e-01 -2.09034503e-01 -1.06597960e+00 1.15839350e+00
1.58992842e-01 -3.24379086e-01 4.29491401e-01 1.05629659e+00
1.00311804e+00 5.67647874e-01 -2.19183519e-01 -6.75221011e-02
1.34423459e+00 -1.33119896e-01 -5.67096353e-01 6.14316612e-02
-2.70311497e-02 -1.09104407e+00 3.94493312e-01 7.31284678e-01
-6.69724047e-01 -7.23403275e-01 -1.29532373e+00 8.80306184e-01
-5.27208865e-01 -1.51704267e-01 6.59152746e-01 1.05945766e+00
-4.29235697e-01 6.72324955e-01 -5.53527117e-01 -2.37210363e-01
2.28787005e-01 2.93058842e-01 -9.19379517e-02 -1.75343826e-01
-1.34740543e+00 1.16293311e+00 -7.21993595e-02 5.73619485e-01
-3.41069281e-01 -9.64163423e-01 -4.74872410e-01 -3.38620454e-01
-1.12214193e-01 -2.78602183e-01 7.28489459e-01 -6.63857877e-01
-1.48333073e+00 3.83143783e-01 6.50007427e-01 -4.97140467e-01
3.37977648e-01 -2.79967815e-01 -9.44725275e-01 5.03484607e-01
-1.18802786e-01 -6.28254637e-02 1.10816431e+00 -1.29824781e+00
-3.32429320e-01 -2.88978457e-01 -2.02673420e-01 -3.87586057e-01
1.08691581e-01 -4.49071489e-02 8.83238912e-01 -7.94630706e-01
4.05932009e-01 -7.05274522e-01 5.43331131e-02 -5.83581448e-01
-1.26587674e-01 1.42943442e-01 1.15481079e+00 -7.36277759e-01
4.12664771e-01 -1.93029177e+00 -4.38432395e-01 5.14075041e-01
-1.39741793e-01 2.14371458e-01 -2.43289128e-01 8.66359293e-01
-2.59234786e-01 -4.33907837e-01 -1.51108161e-01 5.90714395e-01
-1.49303794e-01 -1.94693014e-01 -3.43232691e-01 8.02742422e-01
3.05683583e-01 4.99839216e-01 -3.99364382e-01 2.93219358e-01
4.87246245e-01 7.23634541e-01 6.42124563e-03 -2.51375824e-01
2.98568577e-01 2.71850407e-01 -5.82470655e-01 1.06920922e+00
1.16937721e+00 7.20164895e-01 -2.86615193e-01 -8.11801314e-01
-2.43062139e-01 -1.81107581e-01 -1.00785065e+00 8.21470439e-01
-4.79863018e-01 8.72528434e-01 4.18527663e-01 -1.50482416e+00
1.50001252e+00 3.77024472e-01 1.00887191e+00 -1.10288560e+00
2.59074152e-01 4.11834896e-01 4.06486034e-01 -8.08288693e-01
2.43782669e-01 -5.17724454e-01 9.20761302e-02 1.71364456e-01
-6.11993391e-03 -3.17106068e-01 -1.72202989e-01 -5.89535654e-01
1.20043516e+00 1.72959164e-01 -1.39907151e-01 -4.02231187e-01
2.67495871e-01 1.54009834e-01 8.86860676e-03 7.01194584e-01
1.82588492e-02 4.23876345e-01 -4.87387717e-01 -5.25102615e-01
-7.60607243e-01 -1.20411563e+00 -5.46526074e-01 5.98506987e-01
2.37526819e-02 3.95087481e-01 -2.64595598e-01 -5.99499680e-02
3.71967614e-01 7.72162855e-01 -5.00286400e-01 -4.18980151e-01
-7.77875543e-01 -1.06161249e+00 4.29043949e-01 3.68159860e-01
3.56763303e-01 -1.34929180e+00 -1.37704217e+00 3.73654932e-01
3.32547247e-01 -1.07688248e+00 7.17844963e-01 4.92413759e-01
-8.00000072e-01 -1.04980576e+00 -6.22526884e-01 -5.02651632e-01
2.18284637e-01 3.62038344e-01 8.80956471e-01 -4.30400103e-01
-9.42867875e-01 7.47982442e-01 -6.28582597e-01 -8.98862183e-01
-4.85335469e-01 -3.73079509e-01 5.83235249e-02 -3.21241111e-01
5.76449454e-01 -7.62578428e-01 -3.26725423e-01 1.37729973e-01
-7.37875819e-01 -5.96666336e-01 1.16412234e+00 7.20839977e-01
-1.24243587e-01 5.36085069e-01 9.15179372e-01 -5.30378520e-01
7.98450649e-01 -3.85128051e-01 -4.04554278e-01 -1.16400570e-01
-4.88152176e-01 -1.79651976e-01 3.68311167e-01 -5.37555218e-01
-9.02983248e-01 -4.01552171e-01 -1.58649310e-01 -2.45175824e-01
-5.36257744e-01 6.17416263e-01 1.67630613e-01 -4.51487094e-01
6.88442171e-01 5.42602763e-02 2.05543131e-01 -2.71304578e-01
-2.36613393e-01 6.98737204e-01 3.87523353e-01 -5.73848009e-01
1.27657092e+00 3.81978214e-01 3.12899619e-01 -1.57154393e+00
-3.19824457e-01 -3.52165401e-01 -2.16740638e-01 -5.25287807e-01
6.92451477e-01 -6.13852859e-01 -5.95977545e-01 4.15926069e-01
-8.10839534e-01 1.96862161e-01 -3.40476781e-01 1.02936137e+00
-4.45499927e-01 1.48294345e-01 -4.57733989e-01 -1.04837787e+00
-6.74470961e-01 -8.33229363e-01 8.78422379e-01 -1.44892428e-02
-2.98949927e-01 -7.11142719e-01 -1.11865468e-01 3.59447598e-01
9.59059119e-01 7.57960021e-01 1.01287639e+00 -4.47459042e-01
-2.53223419e-01 -5.72839022e-01 1.02299057e-01 3.92936230e-01
3.38084579e-01 -1.18362658e-01 -1.02438474e+00 -3.20028126e-01
4.86046910e-01 -6.88761398e-02 8.50452721e-01 8.14902663e-01
3.33880961e-01 2.06724972e-01 -1.15865737e-01 3.89220327e-01
1.68467009e+00 6.64522111e-01 5.85298657e-01 6.20080650e-01
1.28409266e-01 8.38002741e-01 7.25899339e-01 2.09909081e-01
-6.15622997e-01 4.61885750e-01 5.91251314e-01 -2.26452857e-01
-6.64403141e-02 7.36420155e-01 5.44316947e-01 5.16418636e-01
-5.90470314e-01 -1.20121241e-01 -5.56869924e-01 1.36133641e-01
-1.08280087e+00 -1.26887250e+00 -2.63835788e-01 1.92652404e+00
-1.67842638e-02 3.85102689e-01 -9.94481519e-02 5.37862360e-01
5.65854907e-01 1.04741722e-01 -2.64482379e-01 -5.28480351e-01
-1.10299148e-01 8.27434897e-01 6.96295261e-01 1.33076608e-01
-1.01117384e+00 7.91405067e-02 6.26778078e+00 5.71246982e-01
-1.26452577e+00 -2.17607915e-01 -9.49178785e-02 1.06375506e-02
-1.57702789e-01 -5.03468990e-01 -2.85776556e-01 2.03487530e-01
8.00674736e-01 2.27154523e-01 1.06666520e-01 6.38441086e-01
6.74479976e-02 -2.11132124e-01 -6.74456596e-01 6.27208054e-01
4.32576537e-02 -1.00719547e+00 -1.68330938e-01 1.92359805e-01
1.83653057e-01 -4.38590385e-02 2.08830103e-01 2.89294332e-01
-1.91985324e-01 -8.32989812e-01 7.79347241e-01 5.76454341e-01
3.24680030e-01 -8.54958177e-01 1.10766065e+00 -7.48175085e-02
-9.96488929e-01 -1.71346471e-01 -4.13089842e-01 -8.05720314e-02
2.26059750e-01 9.17213023e-01 -8.68190527e-01 1.00323594e+00
7.39147663e-01 5.17665558e-02 5.93185313e-02 6.41874313e-01
2.60381699e-01 5.51814139e-01 -4.26660508e-01 -3.05823326e-01
2.79988319e-01 -3.62089455e-01 9.68834460e-01 1.29303002e+00
7.51893222e-01 -1.07976079e-01 2.15843156e-01 8.10204625e-01
8.30001593e-01 -3.62572938e-01 -8.14197183e-01 -1.85075000e-01
2.28478372e-01 1.33349383e+00 -9.26662087e-01 2.78355449e-01
-2.92832911e-01 2.13710397e-01 -7.84184456e-01 3.61582875e-01
-8.11075807e-01 -7.38260984e-01 6.43095255e-01 4.26245421e-01
5.70748568e-01 -4.65694129e-01 -1.99092805e-01 -2.08364025e-01
6.06841333e-02 -7.31710136e-01 3.14519182e-02 -7.09134459e-01
-1.64567471e+00 6.33004785e-01 3.65807682e-01 -1.24809957e+00
-2.09588647e-01 -1.09551597e+00 -4.79418099e-01 8.98916185e-01
-1.35216129e+00 -1.26419246e+00 -3.16171825e-01 1.94182664e-01
4.28199828e-01 -5.90755522e-01 9.28794444e-01 1.73677593e-01
1.76612318e-01 3.37300971e-02 1.21396676e-01 -4.75589186e-03
4.97458130e-01 -8.04385543e-01 -1.17147513e-01 4.01253968e-01
-1.06543064e-01 3.13495338e-01 1.29986334e+00 -6.63545012e-01
-1.67052853e+00 -7.72381723e-01 1.16853304e-01 -1.79372169e-02
5.85419893e-01 -2.58411437e-01 -4.74408895e-01 8.53554234e-02
2.70647913e-01 -3.71398002e-01 6.43052697e-01 -2.91612260e-02
-1.08715862e-01 -2.49955237e-01 -1.53819442e+00 8.26885402e-02
5.69988191e-01 -1.79144725e-01 -8.44588161e-01 2.54572570e-01
2.43162498e-01 -2.55955979e-02 -9.73551512e-01 9.08454835e-01
8.84043694e-01 -1.06176186e+00 1.29775906e+00 -2.19272539e-01
4.22828943e-01 9.92751718e-02 -5.56242049e-01 -1.11209500e+00
-5.25597811e-01 -6.24860786e-02 1.76202625e-01 9.50079978e-01
1.40257224e-01 -9.86535013e-01 7.31143892e-01 -3.18471670e-01
-3.13095421e-01 -7.24179268e-01 -1.07444704e+00 -9.97160852e-01
-2.65210625e-02 -5.46364486e-01 1.98796213e-01 7.45246708e-01
-5.42590082e-01 8.36934298e-02 -4.08781506e-02 4.63105261e-01
8.55168343e-01 5.94012737e-01 3.85333210e-01 -1.63909006e+00
-2.18959630e-01 -3.29119295e-01 -8.38666916e-01 2.12959219e-02
-1.05193462e-02 -5.38800955e-01 1.12231895e-01 -1.57361412e+00
-3.83839637e-01 -3.50099236e-01 -2.46790260e-01 4.45069149e-02
6.00188673e-01 3.39862913e-01 1.20727263e-01 -5.45842163e-02
6.05364203e-01 2.05257386e-01 1.05684471e+00 -4.31185246e-01
5.66829666e-02 3.84440809e-01 -3.16432148e-01 6.06803477e-01
8.52765679e-01 -4.05080020e-01 -2.53954709e-01 -6.38440102e-02
7.15645179e-02 -1.69127226e-01 5.22829831e-01 -1.68786991e+00
-1.87749490e-01 2.28237864e-02 7.45641530e-01 -6.87375009e-01
5.72780907e-01 -1.14504802e+00 4.63923931e-01 8.44538689e-01
1.68759584e-01 -8.52246955e-02 2.29429051e-01 3.39305639e-01
-2.05746591e-01 -2.95265853e-01 7.61686146e-01 -2.08109647e-01
-6.53746426e-01 -2.63286591e-01 -8.33161771e-01 -5.35341382e-01
1.11540997e+00 -8.09496939e-01 -3.01834047e-01 -3.20043087e-01
-7.48855710e-01 -5.91266990e-01 -1.61117688e-01 3.34845811e-01
7.66221225e-01 -1.08860540e+00 -8.27858806e-01 3.37736249e-01
-3.67117345e-01 -5.74610353e-01 4.28807199e-01 7.72309422e-01
-4.84219551e-01 4.52164859e-01 -9.16105330e-01 -5.77959478e-01
-1.31043279e+00 3.44959766e-01 1.72576994e-01 -5.94173893e-02
-7.80022681e-01 4.90017980e-01 -3.12222540e-01 -1.80278704e-01
-1.64114878e-01 -3.27325970e-01 -2.57373929e-01 3.01535189e-01
1.65931836e-01 4.84709531e-01 4.37046051e-01 -4.51087415e-01
-4.64543819e-01 8.94266069e-01 2.04700246e-01 -7.79476613e-02
1.82888365e+00 4.22111124e-01 1.26280442e-01 3.49386752e-01
9.82333183e-01 2.68337846e-01 -6.44326925e-01 1.74678415e-01
-5.15838936e-02 -6.27694726e-02 2.90038079e-01 -9.57499087e-01
-1.09519672e+00 6.64828956e-01 1.03880847e+00 5.44023991e-01
1.30270493e+00 -2.45849237e-01 5.84906876e-01 5.03922403e-01
6.04899466e-01 -1.07303119e+00 -9.21774954e-02 3.46819609e-01
1.04914927e+00 -7.22814441e-01 4.29456919e-01 -4.78386641e-01
2.29316745e-02 1.80507863e+00 1.80730864e-01 -4.23022866e-01
8.11783254e-01 8.10497463e-01 3.52058470e-01 -5.50869584e-01
-1.68671906e-01 -1.32344574e-01 1.62680119e-01 1.00865376e+00
3.55284601e-01 1.70202777e-01 -1.74396411e-01 4.21294987e-01
-3.93393666e-01 -2.52827257e-01 3.30907822e-01 1.37469137e+00
-6.71205521e-01 -8.64031017e-01 -8.90177727e-01 5.93692541e-01
-5.19909084e-01 1.96492359e-01 -1.92659006e-01 1.26307833e+00
2.34338865e-01 9.10677195e-01 9.88051891e-02 -5.90944409e-01
4.76384282e-01 -8.65196139e-02 8.47138822e-01 -2.51639664e-01
-6.56187713e-01 8.21451619e-02 2.83591002e-01 -3.12522590e-01
-8.73374104e-01 -7.20485151e-01 -8.84817362e-01 7.95203727e-04
-5.18389106e-01 1.21478446e-01 1.13401675e+00 8.87866914e-01
-6.83481395e-02 9.59038675e-01 5.56337178e-01 -1.17200077e+00
-8.37507546e-01 -1.18210638e+00 -8.74068320e-01 6.39878213e-02
2.98520118e-01 -8.98427546e-01 -5.42406917e-01 -4.02970463e-01] | [6.884199619293213, 1.0778496265411377] |
72e984de-fcbd-4bae-ae66-676ae2b87599 | a-deep-knowledge-distillation-framework-for | 2112.07252 | null | https://arxiv.org/abs/2112.07252v2 | https://arxiv.org/pdf/2112.07252v2.pdf | A Deep Knowledge Distillation framework for EEG assisted enhancement of single-lead ECG based sleep staging | Automatic Sleep Staging study is presently done with the help of Electroencephalogram (EEG) signals. Recently, Deep Learning (DL) based approaches have enabled significant progress in this area, allowing for near-human accuracy in automated sleep staging. However, EEG based sleep staging requires an extensive as well as an expensive clinical setup. Moreover, the requirement of an expert for setup and the added inconvenience to the subject under study renders it unfavourable in a point of care context. Electrocardiogram (ECG), an unobtrusive alternative to EEG, is more suitable, but its performance, unsurprisingly, remains sub-par compared to EEG-based sleep staging. Naturally, it would be helpful to transfer knowledge from EEG to ECG, ultimately enhancing the model's performance on ECG based inputs. Knowledge Distillation (KD) is a renowned concept in DL that looks to transfer knowledge from a better but potentially more cumbersome teacher model to a compact student model. Building on this concept, we propose a cross-modal KD framework to improve ECG-based sleep staging performance with assistance from features learned through models trained on EEG. Additionally, we also conducted multiple experiments on the individual components of the proposed model to get better insight into the distillation approach. Data of 200 subjects from the Montreal Archive of Sleep Studies (MASS) was utilized for our study. The proposed model showed a 14.3\% and 13.4\% increase in weighted-F1-score in 4-class and 3-class sleep staging, respectively. This demonstrates the viability of KD for performance improvement of single-channel ECG based sleep staging in 4-class(W-L-D-R) and 3-class(W-N-R) classification. | ['Mohanasankar Sivaprakasam', 'Preejith SP', 'Sricharan Vijayarangan', 'Vaibhav Joshi'] | 2021-12-14 | null | null | null | null | ['sleep-stage-detection', 'sleep-staging', 'w-r-n-sleep-staging', 'w-r-l-d-sleep-staging', 'eeg-based-sleep-staging', 'ecg-based-sleep-staging'] | ['medical', 'medical', 'time-series', 'time-series', 'time-series', 'time-series'] | [ 1.26481026e-01 8.82723778e-02 2.76747309e-02 -5.57380080e-01
-6.68838859e-01 -1.41482353e-01 1.00256026e-01 2.73012877e-01
-7.42926896e-01 1.04672039e+00 1.07393321e-02 -3.07263762e-01
-3.57333809e-01 -4.27254945e-01 -2.22666383e-01 -8.54080439e-01
-1.44164652e-01 3.74280334e-01 2.71432623e-02 -1.34749785e-01
1.55332074e-01 2.46372312e-01 -1.28069556e+00 2.02256609e-02
1.19415545e+00 1.06772864e+00 3.48663956e-01 5.02289593e-01
2.26832628e-01 2.79841334e-01 -8.10834706e-01 -2.44315714e-01
3.14995535e-02 -6.35552764e-01 -8.59101355e-01 -1.67748928e-01
9.30560306e-02 1.71380311e-01 4.65790629e-02 6.98442578e-01
7.84354568e-01 2.07565606e-01 5.44769168e-01 -1.03620732e+00
-2.72751421e-01 1.74015790e-01 -2.93055117e-01 7.22344518e-01
4.15627748e-01 1.34791940e-01 6.80329502e-01 -3.66820395e-01
-1.02196261e-01 1.78997070e-01 9.06334341e-01 6.58505976e-01
-1.23757267e+00 -8.53106737e-01 -3.65626842e-01 5.47661841e-01
-1.50919068e+00 -3.16393971e-01 6.15883410e-01 -2.12126493e-01
9.95678306e-01 3.96725804e-01 1.37866795e+00 8.63557935e-01
5.50084710e-01 4.45555151e-01 1.54630959e+00 -3.73903662e-01
4.39815134e-01 6.16997182e-01 5.59190102e-02 4.93877351e-01
1.54421255e-01 -1.15503877e-01 -6.71866059e-01 1.55962124e-01
4.20984536e-01 2.13149637e-01 -4.03684527e-01 3.57441641e-02
-8.20725977e-01 5.24300754e-01 5.45728385e-01 5.65799713e-01
-3.18169385e-01 -1.97001651e-01 2.38710999e-01 4.10599232e-01
4.31424141e-01 7.07304537e-01 -4.94911522e-01 -5.46987653e-01
-1.59612405e+00 -3.05302739e-01 7.49197900e-01 4.91974175e-01
5.09827554e-01 -1.10364430e-01 3.39326076e-02 7.08962381e-01
1.71262711e-01 2.54568279e-01 9.25223410e-01 -6.31463408e-01
1.64118201e-01 7.23641992e-01 -1.67952329e-01 -7.24886656e-01
-9.32548165e-01 -7.58887529e-01 -1.10416102e+00 -2.58860607e-02
1.36926383e-01 -2.62500364e-02 -6.94313169e-01 1.60819745e+00
-4.46113646e-02 3.28745216e-01 2.35765497e-03 7.95672297e-01
7.91566908e-01 2.43005157e-01 2.13063851e-01 -4.87092495e-01
1.48125815e+00 -6.16022050e-01 -4.63068455e-01 -2.05470249e-01
3.92618239e-01 -3.86812270e-01 1.16673207e+00 7.76314139e-01
-1.09010744e+00 -6.76831961e-01 -1.11744690e+00 1.75282136e-01
-2.73262292e-01 9.81161073e-02 4.63340729e-01 9.92522955e-01
-1.30603111e+00 6.63309157e-01 -1.04480731e+00 -7.73817658e-01
5.90549171e-01 1.03193057e+00 -4.58761305e-01 3.16904575e-01
-1.08702683e+00 1.20738423e+00 1.74231678e-01 1.14760734e-01
-6.71031892e-01 -7.33429790e-01 -6.01469696e-01 1.12186126e-01
9.36473981e-02 -8.38857234e-01 9.00465488e-01 -7.46770859e-01
-1.48713350e+00 8.35957110e-01 -1.17239185e-01 -5.10409057e-01
2.53423333e-01 -9.80034322e-02 -5.65311372e-01 3.72005463e-01
2.61507835e-02 5.30372798e-01 7.47129738e-01 -7.44771183e-01
-5.48480630e-01 -4.73503262e-01 1.22278132e-01 2.96625704e-01
-4.14605916e-01 -2.43755549e-01 4.71113771e-02 -4.14177954e-01
-1.38612121e-01 -8.37813079e-01 -2.80380193e-02 -2.79103905e-01
-3.23817171e-02 -3.35474342e-01 2.98696488e-01 -7.23787725e-01
1.28961694e+00 -2.11124349e+00 -3.36552784e-02 1.03472218e-01
5.23332000e-01 2.49468923e-01 3.60514849e-01 2.35360622e-01
-1.99565649e-01 -6.77960739e-02 -2.75024354e-01 -3.28714579e-01
-1.86408341e-01 2.23342761e-01 4.05008584e-01 5.82978010e-01
4.03846316e-02 6.97057128e-01 -7.79984593e-01 -4.31638122e-01
2.10516945e-01 3.48762155e-01 -4.94251549e-01 1.61658406e-01
7.34388530e-01 7.70137548e-01 2.32457034e-02 4.35099274e-01
2.23990351e-01 -3.21163356e-01 -5.75201213e-02 -1.22176871e-01
8.68914276e-03 2.23800451e-01 -6.43842876e-01 1.89630651e+00
-6.76378667e-01 6.17038965e-01 -3.07189673e-01 -1.20490253e+00
7.29670167e-01 5.00199676e-01 3.62609178e-01 -8.09120893e-01
2.59139240e-01 1.57218963e-01 3.66189629e-01 -5.97806871e-01
2.22504646e-01 -7.91003823e-01 -1.75026469e-02 4.12975520e-01
2.60170162e-01 -1.43872172e-01 4.97949608e-02 -7.65100792e-02
1.11906028e+00 -1.80114955e-01 5.86939335e-01 -5.30078948e-01
5.72967589e-01 -3.15601051e-01 5.37009656e-01 6.03032529e-01
-3.31378609e-01 4.44951057e-01 1.68470338e-01 -3.23511958e-01
-4.66920555e-01 -9.93989289e-01 -5.13262212e-01 6.21121526e-01
5.81191443e-02 -5.22866845e-01 -6.86531484e-01 -7.51366913e-01
-3.64851117e-01 7.12859511e-01 -6.03815436e-01 -6.18650794e-01
-2.51843601e-01 -1.10448372e+00 6.23737633e-01 7.15071797e-01
5.33270180e-01 -1.00225115e+00 -1.04954147e+00 1.18318021e-01
-1.03380382e-01 -6.34983838e-01 -1.86825007e-01 7.49239087e-01
-1.14479136e+00 -9.22114909e-01 -9.16853130e-01 -6.81034505e-01
7.31347382e-01 -1.18919142e-01 1.01195538e+00 1.31170690e-01
-4.79526877e-01 4.58156228e-01 -3.67046982e-01 -5.55794537e-01
-3.56122665e-02 1.98182166e-01 4.84614819e-01 -1.43159300e-01
5.35318673e-01 -9.92610216e-01 -9.49412227e-01 3.01332146e-01
-6.21271014e-01 -3.03099640e-02 8.19419146e-01 8.12021732e-01
3.35011780e-01 1.57259420e-01 8.89488637e-01 -5.36636353e-01
6.50291145e-01 -5.01261652e-01 -1.34368196e-01 1.05254807e-01
-1.06049156e+00 -2.11233363e-01 7.22558141e-01 -2.72293866e-01
-7.45634377e-01 -2.20752895e-01 -3.26533169e-01 -8.13121721e-02
-3.53933364e-01 4.07159150e-01 1.48795405e-02 -8.60011354e-02
6.39417589e-01 4.91938055e-01 -8.66408423e-02 -3.75990599e-01
-2.82829642e-01 9.13895011e-01 5.16077757e-01 -1.88973054e-01
5.38302064e-01 2.72246540e-01 2.12619994e-02 -1.06785440e+00
-6.61792099e-01 -5.59072018e-01 -7.45271564e-01 3.21451984e-02
1.02526236e+00 -7.99726665e-01 -7.36521602e-01 3.32051426e-01
-5.06783724e-01 -4.52242255e-01 -4.12652582e-01 5.97816050e-01
-4.58299905e-01 3.27273339e-01 -3.24749388e-02 -6.74252748e-01
-6.82516694e-01 -1.06509292e+00 7.26729274e-01 5.36491513e-01
-7.27210164e-01 -1.14899158e+00 1.01553917e-01 5.20185828e-01
5.37787557e-01 -1.31179154e-01 1.11177015e+00 -8.19346845e-01
-1.80617049e-02 -3.08155179e-01 5.96913993e-02 4.92416769e-01
2.91013032e-01 -8.10259044e-01 -1.17453110e+00 -2.98402190e-01
4.96362627e-01 -2.78818578e-01 3.67710590e-01 3.13425988e-01
1.01668298e+00 1.74567938e-01 -1.27710745e-01 6.57645285e-01
1.23659384e+00 4.63519394e-01 5.84049761e-01 3.83465439e-01
2.82486886e-01 1.24936901e-01 2.04598904e-01 3.72898996e-01
6.46188080e-01 6.34858608e-01 3.81781198e-02 -2.73291528e-01
-2.03173533e-02 2.52160162e-01 3.03901792e-01 9.66239154e-01
-2.77991533e-01 1.36338368e-01 -8.70377064e-01 3.50559115e-01
-1.30410314e+00 -6.89295053e-01 5.08801192e-02 2.12926888e+00
9.35534060e-01 2.64033616e-01 3.15552175e-01 5.86282790e-01
1.33065447e-01 -3.54379803e-01 -4.22022581e-01 -5.85139394e-01
3.67995799e-01 8.18166554e-01 5.96151203e-02 -1.09465450e-01
-6.70609057e-01 3.66655260e-01 5.79566193e+00 6.02361143e-01
-1.24870241e+00 3.96122754e-01 5.12156785e-01 -3.58128875e-01
3.28135252e-01 -3.13735604e-01 -5.58848262e-01 8.51678252e-01
1.50103295e+00 -1.92699805e-01 3.98016214e-01 5.75288534e-01
4.84575152e-01 -7.34170139e-01 -1.28968751e+00 1.39250910e+00
2.41332307e-01 -1.01628506e+00 -4.58535582e-01 -4.69839051e-02
3.12666416e-01 -9.76148099e-02 -8.94573703e-02 5.76329827e-01
-6.79202318e-01 -1.11077762e+00 4.37113464e-01 6.37592733e-01
1.04704428e+00 -7.37536132e-01 1.24170530e+00 4.70732272e-01
-1.04369688e+00 -5.86124845e-02 -1.19533397e-01 -2.69999206e-01
-8.61521810e-02 2.91193426e-01 -9.14185107e-01 5.94179451e-01
9.86294270e-01 7.30832100e-01 -8.66154552e-01 1.36578882e+00
-2.52534986e-01 8.49582851e-01 -2.00170323e-01 -5.23488633e-02
-9.23906267e-02 -4.79234904e-02 1.25705183e-01 1.00111341e+00
3.20314884e-01 3.23644668e-01 -1.70320690e-01 6.06413424e-01
2.42580213e-02 -8.61881152e-02 -2.80870557e-01 2.22436130e-01
-1.02254152e-01 1.45401931e+00 -1.10882020e+00 -1.90484241e-01
-3.56292546e-01 1.10232460e+00 1.03272125e-01 -2.62424294e-02
-6.79630637e-01 -6.24555826e-01 2.65988946e-01 2.32731193e-01
-5.61933331e-02 1.77647799e-01 -4.44519460e-01 -1.06237364e+00
-1.50751710e-01 -7.68430769e-01 3.44070226e-01 -6.43529773e-01
-1.17630231e+00 6.75655544e-01 8.35098773e-02 -1.20765567e+00
2.89737619e-02 -1.78667918e-01 -8.72979641e-01 9.82586443e-01
-1.45530915e+00 -6.70547545e-01 -5.73837399e-01 6.57022715e-01
5.79570532e-01 -9.32001770e-02 1.06529033e+00 5.41699111e-01
-5.99875271e-01 8.08641434e-01 -1.55742809e-01 -2.16135368e-01
6.56527042e-01 -1.51902473e+00 -1.78514704e-01 5.13672948e-01
6.57723323e-02 7.84289598e-01 4.20992374e-01 -2.62643993e-01
-1.08200741e+00 -7.78711021e-01 9.66343105e-01 -6.23605788e-01
4.20995444e-01 -1.83111116e-01 -6.42260551e-01 2.06752002e-01
2.87621260e-01 -3.24850261e-01 1.32669759e+00 1.92175984e-01
5.65465987e-01 -5.53501189e-01 -1.27791762e+00 3.65328908e-01
7.15394258e-01 -5.66811740e-01 -1.08034706e+00 4.08498384e-02
-8.58085528e-02 -1.26086667e-01 -1.00766456e+00 2.40528196e-01
5.99236250e-01 -1.07924235e+00 5.30718684e-01 -1.30296841e-01
1.14465348e-01 -1.48997977e-01 2.89090931e-01 -1.40302110e+00
-2.53338009e-01 -6.40314639e-01 8.88305306e-02 9.07244086e-01
2.96825528e-01 -4.79574144e-01 8.25646698e-01 6.54091120e-01
-4.81456548e-01 -1.13987660e+00 -1.07790506e+00 -6.59749389e-01
-1.60564050e-01 -5.12681723e-01 3.21576893e-01 5.86762488e-01
3.84830564e-01 5.35069942e-01 -1.70255661e-01 8.23029652e-02
1.49013534e-01 -1.44139603e-01 4.56691235e-01 -1.52562904e+00
-3.54255855e-01 -2.40663514e-01 -7.82379746e-01 -5.40862203e-01
-3.93127240e-02 -1.18861747e+00 -9.24487859e-02 -1.70995164e+00
4.18291092e-01 -4.19012040e-01 -8.45989764e-01 5.00698745e-01
-1.17039159e-01 7.44827151e-01 1.06041711e-02 3.45369317e-02
-6.79278433e-01 4.16363060e-01 9.41964149e-01 1.17513202e-01
-4.52942163e-01 4.10086304e-01 -9.21535969e-01 7.10934579e-01
9.06757891e-01 -6.96632206e-01 -7.88592160e-01 -2.00240426e-02
1.99438021e-01 1.12172924e-01 1.46992594e-01 -1.38062525e+00
2.94908434e-01 4.67198998e-01 7.77973473e-01 -2.75616050e-01
5.61649740e-01 -8.91934693e-01 1.32866025e-01 5.77630579e-01
2.53351927e-02 1.89151585e-01 3.30772728e-01 3.91631156e-01
-4.18736488e-02 -2.59181023e-01 8.55541766e-01 -1.74763665e-01
-4.38902706e-01 -6.16697557e-02 -5.43924928e-01 4.44719307e-02
9.88771379e-01 -7.71655083e-01 9.82784182e-02 -2.34721228e-01
-1.12909317e+00 8.23619589e-02 2.19392657e-01 3.28507982e-02
5.21977484e-01 -7.59742677e-01 -1.53828099e-01 5.58300316e-01
-1.95799768e-02 -1.58990085e-01 2.88080335e-01 1.78745925e+00
-3.49744886e-01 3.82385939e-01 -4.94992644e-01 -6.50902510e-01
-1.20850253e+00 6.36978224e-02 3.07906687e-01 -2.37061262e-01
-7.92220771e-01 9.67606485e-01 -7.60541484e-02 1.96416695e-02
2.93359280e-01 -5.72437346e-01 -3.11012983e-01 3.31044942e-01
4.67300832e-01 5.37496269e-01 5.25644481e-01 -2.32706517e-01
-5.47444701e-01 4.80198503e-01 -3.93274650e-02 1.94462806e-01
1.42729640e+00 -1.89815804e-01 -3.59620638e-02 6.44048512e-01
8.87883544e-01 -2.13967562e-01 -7.62489259e-01 3.57798278e-01
9.33420286e-02 1.83229260e-02 1.95424587e-01 -1.20790708e+00
-7.95882285e-01 1.00211108e+00 1.02471232e+00 2.10294887e-01
1.56298423e+00 -7.77622461e-02 7.34826565e-01 3.81087035e-01
5.65985560e-01 -9.64398801e-01 -4.62377965e-02 2.02201144e-03
3.58776897e-01 -1.14409852e+00 1.22307122e-01 3.56113791e-01
-6.71512425e-01 1.08445799e+00 3.68341506e-01 4.33750041e-02
7.92285800e-01 -3.01471427e-02 1.00389674e-01 -3.13950121e-01
-3.77832174e-01 -3.54772843e-02 4.02712226e-01 5.66627502e-01
3.08901429e-01 -1.02305040e-01 -3.39608461e-01 1.03507614e+00
-4.30108279e-01 3.12962174e-01 4.49875265e-01 8.62320602e-01
-2.71195233e-01 -1.02221227e+00 -6.91708401e-02 9.64223444e-01
-5.97335756e-01 -2.52543151e-01 -6.15864694e-02 8.48958075e-01
5.06343901e-01 1.07056904e+00 -1.26213014e-01 -4.60403472e-01
3.16446543e-01 3.74002516e-01 5.15811503e-01 -8.82863820e-01
-9.26571965e-01 -2.90955813e-03 -2.42678240e-01 -3.90497416e-01
-5.98862946e-01 -4.98679399e-01 -1.21323204e+00 9.32682585e-03
-3.44293267e-01 5.89119971e-01 4.77666587e-01 1.21336889e+00
2.44378388e-01 7.47952163e-01 2.97206879e-01 -8.23697925e-01
-2.14443475e-01 -1.05714273e+00 -9.97073054e-01 -6.23047315e-02
3.29895586e-01 -7.11850643e-01 -3.65113735e-01 -3.84499244e-02] | [13.466123580932617, 3.5551228523254395] |
d5683c99-b672-4755-ba31-5ba21b3d454c | deriving-neural-architectures-from-sequence | 1705.09037 | null | http://arxiv.org/abs/1705.09037v3 | http://arxiv.org/pdf/1705.09037v3.pdf | Deriving Neural Architectures from Sequence and Graph Kernels | The design of neural architectures for structured objects is typically guided
by experimental insights rather than a formal process. In this work, we appeal
to kernels over combinatorial structures, such as sequences and graphs, to
derive appropriate neural operations. We introduce a class of deep recurrent
neural operations and formally characterize their associated kernel spaces. Our
recurrent modules compare the input to virtual reference objects (cf. filters
in CNN) via the kernels. Similar to traditional neural operations, these
reference objects are parameterized and directly optimized in end-to-end
training. We empirically evaluate the proposed class of neural architectures on
standard applications such as language modeling and molecular graph regression,
achieving state-of-the-art results across these applications. | ['Wengong Jin', 'Tao Lei', 'Regina Barzilay', 'Tommi Jaakkola'] | 2017-05-25 | deriving-neural-architectures-from-sequence-1 | https://icml.cc/Conferences/2017/Schedule?showEvent=797 | http://proceedings.mlr.press/v70/lei17a/lei17a.pdf | icml-2017-8 | ['graph-regression'] | ['graphs'] | [ 3.23345989e-01 3.21187600e-02 -2.04363897e-01 -3.56222749e-01
-2.79452622e-01 -6.54202402e-01 6.57835364e-01 3.87609363e-01
-6.22575581e-01 4.32076335e-01 7.55609944e-02 -5.51740408e-01
-2.44076490e-01 -7.59523988e-01 -9.75903511e-01 -3.30281168e-01
-5.12659013e-01 8.82037953e-02 2.47893110e-01 -1.66713580e-01
3.75770718e-01 7.74914861e-01 -1.12215507e+00 3.25945675e-01
7.52920628e-01 9.35905039e-01 1.33325964e-01 1.01277471e+00
-3.60596955e-01 9.35461104e-01 -3.24759811e-01 -5.96024632e-01
1.67142525e-01 -1.56299576e-01 -7.33248711e-01 -3.78111333e-01
4.77224231e-01 1.34642467e-01 -5.70345640e-01 1.23289716e+00
3.34204376e-01 5.81621051e-01 9.82512534e-01 -6.89192414e-01
-1.29040205e+00 8.09176445e-01 -2.80953888e-02 2.30720580e-01
-3.20335887e-02 2.28695571e-01 1.25211143e+00 -1.14907837e+00
7.27895379e-01 1.18571687e+00 7.49406517e-01 4.12447780e-01
-1.66173613e+00 -3.26039016e-01 4.70146984e-01 -2.15957072e-02
-1.34749389e+00 -4.67930704e-01 6.85414314e-01 -5.94752371e-01
1.37847447e+00 -1.96380895e-02 3.71566266e-01 1.17280340e+00
4.55206722e-01 5.76439142e-01 5.42417049e-01 -1.22988567e-01
3.35285306e-01 -6.28330186e-02 6.82521164e-01 1.11115527e+00
2.82261342e-01 5.75319352e-03 -5.00299931e-01 -3.31630677e-01
9.63444054e-01 1.46103784e-01 -3.03252369e-01 -4.55217153e-01
-1.11800992e+00 9.81399298e-01 7.54370093e-01 1.73479766e-02
-4.33495641e-01 6.79495692e-01 4.45695877e-01 3.05762887e-01
2.44734704e-01 5.80472291e-01 -3.63285363e-01 2.79559165e-01
-5.59094608e-01 2.07616642e-01 9.18020725e-01 1.05889070e+00
8.06996346e-01 2.52980202e-01 -4.33488965e-01 7.81093895e-01
2.67957270e-01 -9.98488739e-02 2.69063979e-01 -3.14342976e-01
3.28637481e-01 5.41642070e-01 -2.24188566e-01 -8.21715474e-01
-4.49090898e-01 -8.42096865e-01 -9.98301029e-01 -1.95823699e-01
2.63788342e-01 1.56632468e-01 -1.32140720e+00 1.74841833e+00
-1.56591907e-01 1.77401140e-01 1.04372635e-01 7.76128173e-01
1.05498612e+00 6.34899855e-01 4.31976706e-01 7.67558962e-02
1.04430783e+00 -1.02100706e+00 -5.36921799e-01 -1.26571702e-02
8.05608988e-01 -4.94010150e-01 1.22858751e+00 1.89957619e-01
-1.10711646e+00 -5.58829188e-01 -1.01037180e+00 -3.47423166e-01
-8.12180221e-01 3.14833492e-01 9.04825509e-01 3.93571585e-01
-1.17383850e+00 1.02887547e+00 -8.17093551e-01 -3.74074131e-01
4.06727254e-01 6.35165572e-01 -1.39236078e-01 5.29632330e-01
-1.19048619e+00 5.64873815e-01 5.05123675e-01 4.11500007e-01
-1.04874814e+00 -8.85598242e-01 -8.32989693e-01 1.27057344e-01
2.36603707e-01 -9.55642283e-01 1.10353279e+00 -7.69450724e-01
-1.66942358e+00 7.95129120e-01 7.41102919e-02 -8.25280309e-01
1.49735972e-01 -1.96451291e-01 -3.01796407e-01 -1.19326212e-01
-4.58205968e-01 5.20716310e-01 9.63651180e-01 -8.26195836e-01
-8.91689062e-02 4.23032939e-02 2.51629621e-01 -1.21133544e-01
-4.42390293e-01 1.37477547e-01 -4.88738894e-01 -8.52927864e-01
-1.12142563e-01 -8.05275083e-01 -6.12630904e-01 -6.90410733e-02
-7.29866266e-01 -6.83301464e-02 4.62225407e-01 -3.00104648e-01
1.33841968e+00 -1.84387600e+00 4.83414471e-01 4.56606954e-01
6.66935027e-01 3.74132283e-02 -3.72721523e-01 5.41705787e-01
-3.08965087e-01 2.69791305e-01 -2.89895564e-01 -2.95511764e-02
1.65205598e-01 -3.38858785e-03 -4.43761051e-01 5.00227332e-01
5.51722229e-01 1.39607143e+00 -7.94941425e-01 -8.86969920e-03
-3.89556512e-02 6.45781398e-01 -6.17861509e-01 2.46591404e-01
-6.06868684e-01 -1.82589293e-01 -5.63571870e-01 5.95344961e-01
2.28407860e-01 -6.82894707e-01 -1.91416312e-02 -4.98500496e-01
-2.00333316e-02 5.13211071e-01 -8.53880405e-01 1.79763138e+00
-2.88150579e-01 5.21914542e-01 -3.32887441e-01 -1.00272286e+00
9.32348192e-01 -7.91765824e-02 -1.14668570e-01 -2.28818789e-01
1.03349060e-01 -1.00377798e-02 3.17402273e-01 -3.97761725e-02
6.12394691e-01 2.07577616e-01 1.02441363e-01 3.62188786e-01
4.72315103e-01 2.48860255e-01 1.73352003e-01 2.29443252e-01
1.14632440e+00 1.35375839e-02 4.39391434e-01 -6.42497361e-01
3.79531384e-01 -2.58720994e-01 2.44362024e-03 1.15173256e+00
1.53926268e-01 2.16862127e-01 6.61643505e-01 -6.49472535e-01
-1.04239690e+00 -1.41663742e+00 -3.64296176e-02 1.49498594e+00
-2.23379791e-01 -6.25509977e-01 -5.67059219e-01 -5.13214231e-01
-3.29817794e-02 4.76694912e-01 -9.16532218e-01 -4.24588054e-01
-3.77581745e-01 -7.40924120e-01 8.96510422e-01 6.56397581e-01
1.28694087e-01 -1.03821242e+00 -4.38136607e-01 5.01604795e-01
8.50925982e-01 -9.90010262e-01 -7.83782601e-01 8.44168723e-01
-1.02452767e+00 -8.91514778e-01 -5.10946453e-01 -8.16924512e-01
7.18267143e-01 -4.71700057e-02 1.28138864e+00 7.14716315e-02
-4.66076702e-01 1.93004221e-01 -4.52145077e-02 -1.50997594e-01
-3.62949431e-01 4.51573730e-01 -6.60136789e-02 8.86742324e-02
-1.88963059e-02 -5.84989846e-01 -6.08276248e-01 -2.34301742e-02
-1.20529830e+00 9.37335715e-02 7.84918189e-01 9.53513205e-01
6.24558389e-01 -5.14210939e-01 3.08509201e-01 -1.12857926e+00
1.24626672e+00 -1.92786291e-01 -9.42824423e-01 5.32132447e-01
-1.70087427e-01 7.89666772e-01 7.79063106e-01 -9.78840530e-01
-5.21724701e-01 1.65194377e-01 8.80350620e-02 -6.35406494e-01
1.24832563e-01 9.13147330e-01 2.28850394e-01 -4.68317419e-01
8.25379729e-01 1.17667168e-01 -2.18411148e-01 -2.79954463e-01
7.88343251e-01 3.17181796e-02 4.36234504e-01 -8.97542238e-01
6.59007251e-01 1.25106871e-01 2.34588817e-01 -9.90185082e-01
-6.45588875e-01 -7.84368217e-02 -5.72438836e-01 2.70930007e-02
7.95835793e-01 -5.22620976e-01 -1.05238152e+00 1.58525333e-01
-1.35947120e+00 -5.16860962e-01 -1.40930146e-01 4.36118484e-01
-5.15819073e-01 -4.77097929e-02 -9.34954822e-01 -6.97659314e-01
-5.32742321e-01 -1.33028674e+00 9.56325233e-01 2.05361605e-01
-7.67180473e-02 -1.35461652e+00 4.25120629e-02 -6.21483803e-01
5.22143662e-01 4.40280288e-02 1.36215794e+00 -9.08937514e-01
-8.67843747e-01 2.31435383e-03 -3.81572276e-01 2.57928312e-01
-1.10123465e-02 3.05170149e-01 -9.18734968e-01 -1.66129515e-01
-4.02302593e-01 -3.77700299e-01 1.18753064e+00 3.56811315e-01
1.44007456e+00 -2.03221321e-01 -2.79881001e-01 1.09861672e+00
1.37039089e+00 4.20480184e-02 5.78552544e-01 -5.82828782e-02
8.18119824e-01 3.85701150e-01 -2.24074170e-01 9.77242142e-02
-9.97493193e-02 3.16508919e-01 2.25578085e-01 -1.86732948e-01
2.18594596e-01 -3.33413035e-01 3.71203125e-01 8.92432928e-01
-7.65149221e-02 -3.01356465e-01 -9.37212050e-01 8.45222473e-02
-1.99002051e+00 -7.58316994e-01 1.70141518e-01 2.04169345e+00
7.62128353e-01 3.11709642e-01 -4.16284427e-03 -6.41970634e-01
6.85839057e-01 2.65363693e-01 -7.65461922e-01 -6.62319720e-01
-2.00147733e-01 6.65002465e-01 9.17600453e-01 5.14700651e-01
-1.11914968e+00 1.30541515e+00 7.24675226e+00 7.78257847e-01
-1.32121682e+00 -1.13843799e-01 7.82628715e-01 -1.19397994e-02
-3.54190111e-01 -3.08743343e-02 -8.32177997e-01 -2.85607483e-02
1.10722220e+00 -6.08796924e-02 6.80206299e-01 8.20390761e-01
2.93966495e-02 5.65528214e-01 -1.49754524e+00 9.06041145e-01
-2.75702149e-01 -2.12764072e+00 5.08240819e-01 3.39510925e-02
5.23737669e-01 1.04354970e-01 3.60293835e-01 3.71567875e-01
6.32677972e-01 -1.52176213e+00 6.86757147e-01 8.85422289e-01
5.88742673e-01 -7.38801003e-01 7.68719092e-02 -2.43060380e-01
-1.34290767e+00 1.05608314e-01 -6.48178577e-01 1.44457385e-01
-2.04522371e-01 3.61005187e-01 -7.86340892e-01 3.57139200e-01
3.29594910e-01 7.52589643e-01 -6.32217586e-01 8.99474859e-01
-1.19771273e-03 5.95095158e-01 -2.60300070e-01 -5.80577731e-01
6.49520338e-01 -3.11433524e-01 3.09554517e-01 1.59799230e+00
-1.65205330e-01 -1.08746037e-01 1.66888684e-01 1.53725040e+00
-5.71291566e-01 1.91455916e-01 -8.71843934e-01 -6.99665368e-01
1.30570576e-01 1.19381118e+00 -7.94956386e-01 -2.09388375e-01
-5.17157316e-01 8.46668005e-01 1.01577866e+00 9.41940069e-01
-8.51549864e-01 -7.58974671e-01 1.01172352e+00 -7.88874030e-02
3.49531293e-01 -6.21163845e-01 -6.89829746e-03 -1.03463745e+00
-2.75615543e-01 -7.20568299e-01 2.55139649e-01 -4.44500238e-01
-1.46825993e+00 6.03455007e-01 9.41329170e-03 -5.65589488e-01
-5.77944703e-02 -1.22618794e+00 -7.38931954e-01 9.98812020e-01
-1.22644854e+00 -9.05590713e-01 1.41757697e-01 4.22593117e-01
1.75029829e-01 -1.78337067e-01 7.88076878e-01 6.31681159e-02
-6.44785404e-01 7.19661236e-01 2.42967699e-02 2.42513061e-01
1.80405155e-01 -1.03201210e+00 9.72197592e-01 6.68856621e-01
4.01678294e-01 1.46868312e+00 4.92848426e-01 -6.19467676e-01
-1.89124846e+00 -1.26245987e+00 1.87532991e-01 -2.37383321e-01
1.19885170e+00 -7.93672860e-01 -1.14458096e+00 6.20715201e-01
4.63096835e-02 2.00644940e-01 6.24342263e-01 1.96976304e-01
-6.62077963e-01 2.03476787e-01 -5.68761110e-01 9.99398828e-01
1.52269363e+00 -1.07016265e+00 -2.36270577e-01 3.72307807e-01
1.05785704e+00 -2.96840191e-01 -8.36962759e-01 4.72977996e-01
5.38537681e-01 -5.14918923e-01 1.23286617e+00 -1.61026692e+00
2.54130393e-01 -2.58418888e-01 -2.07844004e-01 -1.17945230e+00
-5.28622091e-01 -9.46798265e-01 -3.28050792e-01 5.65723956e-01
8.06210220e-01 -6.56566620e-01 8.04177582e-01 3.91340882e-01
-3.12923372e-01 -1.13905823e+00 -3.82655054e-01 -7.52783597e-01
4.11923267e-02 -4.66060698e-01 3.52384686e-01 7.34207392e-01
-4.04054493e-01 4.46662098e-01 -2.71894597e-02 3.12228471e-01
5.20230830e-01 3.54649946e-02 5.34256876e-01 -1.08609605e+00
-5.98077536e-01 -9.51433659e-01 -7.30360925e-01 -1.25777328e+00
4.26736325e-01 -1.26324594e+00 -1.74058840e-01 -1.12502348e+00
5.88761792e-02 -1.17401585e-01 -6.09952033e-01 2.77873516e-01
-1.02724314e-01 -5.09539023e-02 -1.22817338e-01 -1.05365396e-01
-5.71288824e-01 6.32095873e-01 1.05512631e+00 -3.12952846e-01
-3.81534845e-01 -3.88889045e-01 -5.58987081e-01 5.19139409e-01
7.10346222e-01 -2.66185641e-01 -5.79351306e-01 -4.32341188e-01
4.50293392e-01 -2.11406216e-01 4.47442532e-01 -6.57357872e-01
3.53610903e-01 -2.22903937e-01 2.80308753e-01 -4.28361714e-01
2.87628412e-01 -3.54956865e-01 9.19695720e-02 4.34940785e-01
-9.45316851e-01 2.51101851e-01 1.96704507e-01 6.64237082e-01
-8.50937888e-03 -2.46775851e-01 6.93986356e-01 -9.54684392e-02
-5.02594769e-01 6.91387475e-01 -1.87916875e-01 5.84893227e-02
6.28140926e-01 -2.46561900e-01 -3.80046785e-01 -2.81012118e-01
-9.29696858e-01 -1.89222582e-02 2.72464603e-01 2.72782654e-01
9.26905394e-01 -1.12632906e+00 -2.76167005e-01 1.66876331e-01
2.43369922e-01 -2.25823134e-01 -2.28945956e-01 7.18179226e-01
-6.49448335e-01 5.67989409e-01 -7.92382956e-02 -5.20613551e-01
-9.12231803e-01 8.09943259e-01 6.43458188e-01 -2.89224744e-01
-4.06327099e-01 8.78907204e-01 7.68511176e-01 -3.60220909e-01
5.63797116e-01 -9.71093774e-01 1.22002944e-01 -2.15440452e-01
3.19771886e-01 -9.18832645e-02 1.36909023e-01 -9.44270119e-02
-2.14156076e-01 3.29317480e-01 -3.71221870e-01 1.97186366e-01
1.29326987e+00 5.80707133e-01 -1.27423689e-01 5.15917778e-01
1.44377613e+00 -3.90286595e-01 -9.58424807e-01 -3.29065859e-01
3.36703956e-01 1.04083173e-01 -2.78332718e-02 -2.83572823e-01
-9.00095165e-01 8.99597704e-01 4.32638049e-01 5.72563522e-02
6.96738899e-01 8.27505216e-02 3.49850833e-01 1.19999182e+00
5.20513244e-02 -7.83378541e-01 2.40280539e-01 8.89803171e-01
8.79871130e-01 -7.04129815e-01 -2.02019170e-01 -4.64404196e-01
-2.67979741e-01 1.42370546e+00 3.16803455e-01 -5.26707053e-01
9.13791239e-01 3.50658149e-01 -3.41038316e-01 -3.62295359e-01
-1.14671481e+00 -3.98364693e-01 6.96694613e-01 3.84704769e-01
9.19953346e-01 -1.51962349e-02 -6.99551478e-02 5.35826564e-01
1.40357599e-01 -2.67898053e-01 2.25035399e-01 8.77546787e-01
-3.72633845e-01 -9.64742959e-01 5.89655675e-02 7.36588538e-01
-3.26236188e-01 -6.29588783e-01 -4.58315939e-01 6.47920370e-01
-5.52334487e-01 2.40086526e-01 -1.95528846e-02 -3.24989319e-01
4.17915255e-01 5.82678802e-02 5.78331470e-01 -7.52058923e-01
-9.06027138e-01 -1.12818062e-01 8.16063881e-02 -4.98890519e-01
-1.37154117e-01 -5.92134595e-02 -1.28564918e+00 -8.83708820e-02
-2.13729307e-01 -1.06060579e-01 5.89780390e-01 5.63411474e-01
4.66796279e-01 7.52901137e-01 1.12987481e-01 -7.48739958e-01
-7.24989891e-01 -6.46302998e-01 -2.66201079e-01 3.19679320e-01
2.81896234e-01 -6.75670087e-01 1.66916160e-03 -4.26138081e-02] | [6.879095077514648, 6.314600944519043] |
60968768-37c3-4efc-b4df-0a287a6e9b81 | predictive-modeling-of-equine-activity | 2306.05311 | null | https://arxiv.org/abs/2306.05311v1 | https://arxiv.org/pdf/2306.05311v1.pdf | Predictive Modeling of Equine Activity Budgets Using a 3D Skeleton Reconstructed from Surveillance Recordings | In this work, we present a pipeline to reconstruct the 3D pose of a horse from 4 simultaneous surveillance camera recordings. Our environment poses interesting challenges to tackle, such as limited field view of the cameras and a relatively closed and small environment. The pipeline consists of training a 2D markerless pose estimation model to work on every viewpoint, then applying it to the videos and performing triangulation. We present numerical evaluation of the results (error analysis), as well as show the utility of the achieved poses in downstream tasks of selected behavioral predictions. Our analysis of the predictive model for equine behavior showed a bias towards pain-induced horses, which aligns with our understanding of how behavior varies across painful and healthy subjects. | ['Hedvig Kjellström', 'Pia Haubro Andersen', 'Sofia Broomé', 'Ernest Pokropek'] | 2023-06-08 | null | null | null | null | ['pose-estimation'] | ['computer-vision'] | [ 4.61327881e-01 6.85367659e-02 9.04934779e-02 -2.58421272e-01
-4.50395793e-01 -5.66462338e-01 -3.39777023e-02 -1.71639740e-01
-5.83570898e-01 3.99988055e-01 1.12289533e-01 2.91526586e-01
-2.65805572e-01 2.10270453e-02 -8.84801328e-01 -3.04477245e-01
-4.35009271e-01 6.10628545e-01 5.24182558e-01 -3.24666858e-01
3.34328145e-01 7.12451696e-01 -1.86917412e+00 2.06371620e-01
-3.45873423e-02 6.84340954e-01 3.60882819e-01 1.07766891e+00
1.03146505e+00 8.05198729e-01 -5.07045507e-01 -5.54231226e-01
3.26137990e-01 -1.88727483e-01 -6.74574733e-01 3.49418908e-01
7.39107132e-01 -8.89985800e-01 -1.96607590e-01 3.60268533e-01
3.11513484e-01 1.66114718e-01 2.63739437e-01 -1.04674768e+00
2.07114875e-01 -3.73541385e-01 -6.37277544e-01 3.71542573e-01
8.78951132e-01 4.23048347e-01 5.15116036e-01 -3.46584827e-01
1.10422409e+00 9.31745291e-01 1.11815727e+00 4.87286985e-01
-1.16454375e+00 -1.50906771e-01 -8.42981339e-02 2.80917823e-01
-1.11001813e+00 -2.77723491e-01 4.28881496e-01 -6.98394477e-01
1.02165902e+00 4.29140687e-01 1.33706713e+00 1.35258317e+00
6.62861347e-01 5.91604650e-01 1.12278855e+00 -1.07431769e-01
-1.26835704e-03 -1.59761488e-01 -1.68408424e-01 4.93477404e-01
2.17566311e-01 1.92493856e-01 -8.80677462e-01 -3.69982779e-01
8.89835119e-01 1.12229124e-01 -1.94908038e-01 -6.91248894e-01
-1.09744978e+00 5.81315517e-01 1.34621963e-01 -5.87172627e-01
-6.48964167e-01 3.55723023e-01 3.63215327e-01 -5.74026518e-02
4.48734015e-01 5.31769872e-01 -5.92657924e-01 -4.76972431e-01
-6.60392344e-01 7.63132334e-01 4.42954630e-01 7.10293233e-01
3.55673641e-01 -5.52095830e-01 1.20796494e-01 5.58774173e-01
2.50738055e-01 3.60654682e-01 1.10775672e-01 -1.60468781e+00
1.45708919e-01 3.16117883e-01 8.31989050e-02 -1.07498384e+00
-7.26333082e-01 -4.82126027e-02 1.97339922e-01 3.86873454e-01
3.99961561e-01 -2.23278895e-01 -4.68931407e-01 1.29982662e+00
7.76974380e-01 2.27607355e-01 -3.86654377e-01 1.32459331e+00
3.67878646e-01 9.26187560e-02 -2.00096995e-01 -9.20331329e-02
1.41389251e+00 -6.76199019e-01 -4.19399202e-01 -7.05603600e-01
6.01539850e-01 -8.25700641e-01 4.86230731e-01 6.45367086e-01
-1.08405149e+00 -2.89586723e-01 -7.76201487e-01 -3.79586481e-02
2.39975393e-01 6.61219135e-02 5.99127650e-01 2.17740744e-01
-9.75912929e-01 7.78064609e-01 -1.36788917e+00 -6.78171456e-01
1.07232943e-01 4.74589229e-01 -8.01212788e-01 9.49407090e-03
-6.22397304e-01 1.60789347e+00 1.59291491e-01 6.43480182e-01
-7.62347996e-01 -5.20778060e-01 -1.04321516e+00 -6.17275536e-01
6.95097685e-01 -7.10900545e-01 1.38173735e+00 -3.87891918e-01
-1.56355810e+00 1.26470971e+00 -1.03341602e-01 -5.63572526e-01
9.24658477e-01 -9.83781397e-01 2.12478861e-01 5.74679613e-01
4.15363647e-02 7.44453728e-01 5.13245344e-01 -9.91955459e-01
-5.32730401e-01 -6.15915239e-01 1.51041038e-02 5.63231111e-01
5.50079107e-01 2.54969835e-01 -3.54243815e-01 -1.72836602e-01
2.29500979e-01 -1.36446524e+00 -5.10599852e-01 3.00425321e-01
7.96421543e-02 4.42357659e-01 6.58700526e-01 -9.40217555e-01
7.36891031e-01 -1.77908838e+00 3.62888217e-01 1.38986306e-02
2.48359203e-01 -1.18858986e-01 3.47409904e-01 5.80719531e-01
7.38890469e-03 -6.98459744e-01 9.85581875e-02 -1.51747867e-01
-4.14886296e-01 3.25193942e-01 -4.09142300e-02 1.05356550e+00
3.77883315e-01 6.89372957e-01 -8.90115321e-01 -6.42534792e-01
5.35610437e-01 4.32973087e-01 -8.79385471e-01 4.28396553e-01
5.79401776e-02 7.15693295e-01 -1.54652059e-01 5.09896040e-01
4.39984232e-01 1.68340206e-01 3.09054226e-01 -2.15179086e-01
-6.59632087e-02 -5.83124422e-02 -1.02809012e+00 1.73463130e+00
-2.31563020e-02 6.11595333e-01 2.60179639e-01 -3.61047626e-01
4.33454037e-01 1.18848152e-01 6.91705942e-01 1.57428682e-01
2.43935376e-01 2.22278312e-02 -9.61464345e-02 -1.04391253e+00
5.55204749e-01 -2.24480808e-01 -1.50514310e-02 -1.51699251e-02
2.42334709e-01 -2.47625932e-01 -3.58696803e-02 -1.88869774e-01
1.13625669e+00 9.90765810e-01 3.94617110e-01 3.39152925e-02
-9.22073424e-02 5.50515294e-01 2.35847548e-01 1.35369599e-01
-3.02278399e-01 9.65391874e-01 7.30062425e-01 -6.35788262e-01
-7.64030635e-01 -8.02236795e-01 -7.81894922e-02 8.04698944e-01
3.06170255e-01 -4.41840947e-01 -8.67978752e-01 -3.17905962e-01
6.33121058e-02 3.85330856e-01 -9.10349429e-01 -2.53705621e-01
-9.99996722e-01 -5.44983625e-01 3.76667559e-01 5.27008593e-01
-1.95035055e-01 -7.37589955e-01 -1.79120684e+00 1.86483636e-02
-1.31910965e-01 -1.28425348e+00 -8.51571560e-02 3.84362042e-01
-8.18739653e-01 -1.42294955e+00 -3.86530548e-01 -4.97069836e-01
4.61145192e-01 1.72524169e-01 8.37566614e-01 -5.69907874e-02
-5.31848073e-01 5.14631629e-01 -2.11385697e-01 -3.49012971e-01
-1.84123099e-01 -4.94541973e-01 2.14004546e-01 -3.65898818e-01
2.12709934e-01 -2.20501378e-01 -8.28708887e-01 5.46687841e-01
-5.25128365e-01 1.14209782e-02 5.75996161e-01 7.53641546e-01
7.45357692e-01 -4.17780399e-01 -3.84713888e-01 -7.04171002e-01
3.28349590e-01 -4.97420520e-01 -5.38282633e-01 -9.53432098e-02
1.30744457e-01 -4.87220556e-01 8.05491507e-02 -4.55156177e-01
-7.85026729e-01 6.86357319e-01 -8.33164081e-02 -6.77910566e-01
-4.81914520e-01 1.77769110e-01 4.61303860e-01 -7.26315603e-02
5.07638097e-01 -4.13736552e-01 6.22502565e-01 -5.48243463e-01
4.03936133e-02 3.59864771e-01 9.26028371e-01 -3.32469076e-01
3.02099615e-01 6.77243888e-01 3.75672936e-01 -7.65424490e-01
-7.31009901e-01 -5.81804752e-01 -1.22178543e+00 -8.23597968e-01
1.11190927e+00 -7.69973695e-01 -1.19549191e+00 4.27643418e-01
-1.14453125e+00 -3.03682029e-01 -1.41213477e-01 1.03309107e+00
-1.04110920e+00 3.90811831e-01 -6.02039337e-01 -8.34035695e-01
8.27274099e-02 -1.42078161e+00 1.69081712e+00 -1.86452329e-01
-8.65985811e-01 -7.06220388e-01 4.46356654e-01 1.01093674e+00
-9.23460051e-02 9.30363953e-01 2.19999582e-01 -3.59348446e-01
-2.21633792e-01 -6.49068475e-01 4.72905576e-01 1.14537902e-01
-3.85395318e-01 7.33034732e-03 -9.13106382e-01 -2.35020325e-01
2.61379033e-01 -5.89981139e-01 2.89216876e-01 6.54446483e-01
7.36320734e-01 1.42494738e-01 -3.89869153e-01 5.67686856e-01
1.05338955e+00 -2.31533498e-01 3.70754719e-01 4.66061175e-01
5.13114095e-01 1.21037316e+00 1.17836356e+00 3.15597713e-01
2.37749994e-01 1.05339408e+00 6.41151011e-01 8.23592320e-02
1.91267416e-01 -2.91767448e-01 4.50825900e-01 3.66274953e-01
-4.53655392e-01 1.47357360e-01 -8.65933657e-01 2.61915416e-01
-1.74308825e+00 -8.93925369e-01 -3.93304706e-01 2.20494747e+00
3.33318233e-01 1.35848835e-01 5.09217799e-01 -2.56700665e-01
5.74728012e-01 -1.70477122e-01 -5.25740862e-01 -4.12932545e-01
3.84991020e-01 2.29521319e-02 8.98150682e-01 2.99889654e-01
-1.00161207e+00 6.67490840e-01 8.07155991e+00 1.90726697e-01
-9.81331289e-01 -5.72487116e-02 1.26549616e-01 -6.49468541e-01
4.98729348e-01 1.76561013e-01 -4.94413167e-01 1.84437707e-01
9.51706767e-01 3.53120834e-01 1.67975232e-01 7.68412948e-01
3.33692789e-01 -7.34893560e-01 -1.21522748e+00 6.65504575e-01
3.86296749e-01 -8.79383802e-01 -5.98274529e-01 1.34615719e-01
2.18957365e-01 1.41794309e-01 -1.54322222e-01 -1.68337479e-01
1.95000526e-02 -8.22677433e-01 1.01348448e+00 5.86490095e-01
4.67489243e-01 -4.45208490e-01 8.60604048e-01 6.62379503e-01
-5.85147619e-01 1.36986554e-01 -1.30662039e-01 -5.00128984e-01
3.80107343e-01 -2.19004899e-01 -1.07827806e+00 1.74018607e-01
1.01335061e+00 6.37499273e-01 -6.90536618e-01 1.07684040e+00
-7.81418979e-02 3.06153029e-01 -6.28364086e-01 2.62546480e-01
1.10301860e-02 -1.90796882e-01 7.17293024e-01 9.05952156e-01
1.69953287e-01 4.28462476e-02 1.17837206e-01 3.68217945e-01
5.64554513e-01 -1.47713974e-01 -8.43427896e-01 5.00529706e-01
-1.52574182e-01 1.14967787e+00 -7.72024333e-01 2.17102200e-01
-2.04497710e-01 9.93589818e-01 3.44221532e-01 -1.19874746e-01
-9.72289383e-01 2.01162994e-01 7.63843298e-01 5.70999026e-01
2.41846934e-01 -1.44176751e-01 2.73195431e-02 -9.27232504e-01
4.71053869e-01 -7.70459473e-01 3.47267330e-01 -1.15429735e+00
-7.11970568e-01 3.09238732e-01 8.61691177e-01 -1.60996294e+00
-4.81170774e-01 -6.97483242e-01 -3.65748852e-01 5.32568038e-01
-9.21502650e-01 -1.13309097e+00 -1.65018797e-01 6.03401512e-02
4.26669538e-01 3.81612420e-01 7.74173319e-01 -1.79873601e-01
-4.21979308e-01 3.60956937e-02 -2.78984547e-01 -3.65317464e-01
6.98489845e-01 -8.83306742e-01 1.54724419e-01 5.93664050e-01
-2.42241219e-01 7.12174535e-01 1.13913774e+00 -7.90495098e-01
-1.74060118e+00 -6.20361686e-01 2.83846229e-01 -1.18160701e+00
7.76258707e-01 -2.82768130e-01 -5.51165342e-01 9.59707499e-01
4.15962338e-02 7.11148009e-02 6.06463671e-01 -1.80390656e-01
3.48260164e-01 2.62861401e-01 -1.06573784e+00 3.79134506e-01
9.45184052e-01 -3.51424813e-02 -7.57424474e-01 4.66100693e-01
2.39359617e-01 -1.27324545e+00 -9.74399984e-01 4.14763600e-01
1.01943946e+00 -1.03426147e+00 8.48306715e-01 -8.17092240e-01
6.72632575e-01 -2.07783729e-01 1.68311194e-01 -1.08036649e+00
4.40170914e-02 -5.47912359e-01 1.20622769e-01 3.61248374e-01
-1.13776311e-01 -2.82296777e-01 9.58302796e-01 4.89900917e-01
-2.63860412e-02 -1.05344236e+00 -1.01027513e+00 -3.91591072e-01
-3.04121464e-01 -4.19185847e-01 -1.66009054e-01 3.37860048e-01
1.06422827e-01 -7.27793388e-03 -7.39571452e-01 3.22886825e-01
5.35135388e-01 9.10629034e-02 1.06167245e+00 -1.02394783e+00
-5.16891241e-01 2.52045870e-01 -1.11442220e+00 -9.03728247e-01
-8.60683806e-03 -2.61368573e-01 2.09832668e-01 -1.16782451e+00
1.82421878e-01 4.12109077e-01 4.51707304e-01 2.56538600e-01
-4.94866818e-02 4.96053010e-01 -3.46671455e-02 3.24316114e-01
-7.19735146e-01 -8.89502168e-02 1.13877594e+00 5.72414219e-01
5.55182211e-02 1.52613908e-01 -2.05906257e-01 1.16429567e+00
4.48175967e-01 -6.53842270e-01 -1.94958776e-01 -4.41587865e-01
3.73137236e-01 3.76454741e-01 8.64220679e-01 -9.19928849e-01
-8.03880543e-02 -2.66213685e-01 5.25020957e-01 -7.03273356e-01
8.48326027e-01 -1.10292161e+00 5.40345490e-01 7.89299250e-01
-2.43922874e-01 2.91861176e-01 2.91012645e-01 6.13597095e-01
9.68709290e-02 -1.69637233e-01 5.61719596e-01 -2.76035905e-01
-7.27937877e-01 -3.46612155e-01 -5.78753531e-01 7.45701313e-04
1.52487898e+00 -5.97156227e-01 -1.18721381e-01 -2.68808842e-01
-1.02793360e+00 3.82219523e-01 8.33489001e-01 3.85341704e-01
5.32625556e-01 -8.62492204e-01 -6.89850867e-01 2.86667287e-01
1.67780265e-01 6.97306991e-02 4.47901726e-01 1.39006603e+00
-1.38684714e+00 1.40913278e-01 -5.77614665e-01 -1.07222927e+00
-1.69133675e+00 2.64624178e-01 3.57150584e-01 -2.26797741e-02
-7.30517089e-01 6.18450284e-01 -6.94220141e-02 -3.65385413e-01
8.88948608e-03 -3.83228332e-01 -2.38749951e-01 -1.47294775e-01
1.88400686e-01 7.90624261e-01 2.05955774e-01 -1.06284022e+00
-5.22834957e-01 7.81544387e-01 4.39930037e-02 8.44716206e-02
1.61429918e+00 -1.76644370e-01 4.58368063e-02 4.36892092e-01
1.10718262e+00 -1.14938408e-01 -1.60741961e+00 3.91562581e-01
-2.29246378e-01 -6.32312894e-01 -2.35847846e-01 -3.05207551e-01
-7.12804854e-01 6.19190991e-01 6.23632669e-01 -3.54928225e-01
8.28765929e-01 1.54870540e-01 6.89384520e-01 3.40221465e-01
5.42929351e-01 -1.07605815e+00 -2.50649065e-01 5.08133397e-02
9.03053701e-01 -1.10044551e+00 5.95458388e-01 -6.51110888e-01
-1.10398018e+00 9.45848465e-01 7.24900007e-01 -4.24342096e-01
3.32778245e-01 3.49274307e-01 3.48457515e-01 -7.42373705e-01
-9.38173294e-01 3.87878492e-02 3.27031255e-01 7.13664472e-01
3.54592592e-01 -1.70348808e-01 -2.16404840e-01 5.89481741e-02
-5.31428158e-01 1.45656973e-01 6.73226058e-01 1.44443107e+00
-1.34327963e-01 -6.81735635e-01 -6.83604181e-01 3.63866657e-01
-6.42260790e-01 5.40199995e-01 -6.30963326e-01 8.82655263e-01
2.10298344e-01 6.99273527e-01 -5.14891781e-02 -5.22597015e-01
9.46637273e-01 -1.70541048e-01 7.17888534e-01 -8.15346599e-01
-7.45856404e-01 3.64293516e-01 2.59387136e-01 -1.07983625e+00
-6.78192794e-01 -1.04294002e+00 -8.58710945e-01 -1.07710347e-01
-4.65110928e-01 -2.90970385e-01 3.44869733e-01 8.18704188e-01
5.53900264e-02 2.20958129e-01 1.40463531e-01 -1.39897072e+00
-5.92535555e-01 -7.20744431e-01 -6.23847604e-01 4.94999886e-01
3.42846841e-01 -1.02057552e+00 -2.93310106e-01 3.98429036e-01] | [7.298370361328125, -0.8791602849960327] |
1bf421bc-8800-4111-8de0-f543cd53a414 | cross-supervised-dual-classifiers-for-semi | 2305.16216 | null | https://arxiv.org/abs/2305.16216v1 | https://arxiv.org/pdf/2305.16216v1.pdf | Cross-supervised Dual Classifiers for Semi-supervised Medical Image Segmentation | Semi-supervised medical image segmentation offers a promising solution for large-scale medical image analysis by significantly reducing the annotation burden while achieving comparable performance. Employing this method exhibits a high degree of potential for optimizing the segmentation process and increasing its feasibility in clinical settings during translational investigations. Recently, cross-supervised training based on different co-training sub-networks has become a standard paradigm for this task. Still, the critical issues of sub-network disagreement and label-noise suppression require further attention and progress in cross-supervised training. This paper proposes a cross-supervised learning framework based on dual classifiers (DC-Net), including an evidential classifier and a vanilla classifier. The two classifiers exhibit complementary characteristics, enabling them to handle disagreement effectively and generate more robust and accurate pseudo-labels for unlabeled data. We also incorporate the uncertainty estimation from the evidential classifier into cross-supervised training to alleviate the negative effect of the error supervision signal. The extensive experiments on LA and Pancreas-CT dataset illustrate that DC-Net outperforms other state-of-the-art methods for semi-supervised segmentation. The code will be released soon. | ['Zhicheng Jiao', 'Xin Li', 'Fan Yang', 'Heng Zhou', 'Chunna Tian', 'Ran Ran', 'Zhenxi Zhang'] | 2023-05-25 | null | null | null | null | ['semi-supervised-medical-image-segmentation'] | ['computer-vision'] | [ 4.70989257e-01 4.86731708e-01 -6.46880686e-01 -7.81444609e-01
-1.29663718e+00 -3.47271383e-01 1.42928571e-01 3.84633839e-01
-4.98491943e-01 7.79778600e-01 -1.02146350e-01 -2.21281543e-01
-8.26989636e-02 -3.58550280e-01 -3.90242457e-01 -1.06487441e+00
1.26266181e-02 6.53725684e-01 2.53652692e-01 4.07731533e-01
-1.76955119e-01 2.04054728e-01 -9.80122566e-01 5.13512969e-01
1.21170771e+00 1.11079037e+00 2.32091114e-01 2.09572852e-01
-1.75952390e-01 6.51238918e-01 -4.42633897e-01 -3.48690093e-01
2.02167436e-01 -5.36169887e-01 -8.62512231e-01 2.83877850e-01
1.09157354e-01 2.00721025e-01 3.38152617e-01 1.18391764e+00
7.50631809e-01 -3.14761609e-01 5.93008280e-01 -1.03526306e+00
-7.69126788e-02 9.40595508e-01 -7.48667240e-01 6.06422946e-02
-1.02844961e-01 -5.76642901e-02 7.17605948e-01 -3.80397528e-01
4.98605520e-01 7.42133737e-01 9.06544805e-01 5.72884917e-01
-1.36177802e+00 -7.35635340e-01 -6.06692284e-02 -9.85509008e-02
-1.24561775e+00 -1.70601428e-01 7.25523293e-01 -4.57197040e-01
3.85267705e-01 3.50241870e-01 5.96385181e-01 8.77646327e-01
2.55485475e-01 1.21895337e+00 1.50005376e+00 -4.50988442e-01
1.93890765e-01 3.69558692e-01 2.10365832e-01 8.52866650e-01
1.09932281e-01 -3.25900107e-03 -1.48657218e-01 -3.35422099e-01
6.01179957e-01 -1.10106923e-01 -5.41970909e-01 -6.95309460e-01
-1.28962076e+00 6.84011042e-01 4.90511477e-01 4.79301274e-01
-3.17592382e-01 -2.89647579e-01 7.40244210e-01 -1.98470592e-01
6.33070409e-01 4.65767235e-01 -5.30962467e-01 2.84679204e-01
-1.26060200e+00 -2.87059009e-01 6.84994042e-01 7.25449264e-01
2.61231303e-01 -2.02003196e-01 -2.94800103e-01 9.82597828e-01
4.69405413e-01 8.48210081e-02 8.19309711e-01 -8.62643480e-01
1.65890589e-01 7.03609824e-01 -3.31862837e-01 -3.91968727e-01
-7.61223853e-01 -8.92307103e-01 -9.82055545e-01 2.51327038e-01
5.73918462e-01 -1.75603986e-01 -1.09607053e+00 1.50113130e+00
6.05963707e-01 2.16679975e-01 8.48940462e-02 9.63365793e-01
9.64347661e-01 1.43153533e-01 2.99399614e-01 -5.28761327e-01
1.37044644e+00 -1.24102557e+00 -1.03650403e+00 5.20492950e-03
8.88220191e-01 -6.45922303e-01 6.85260475e-01 4.55816388e-01
-9.49095964e-01 -3.52801412e-01 -1.16151762e+00 3.67754400e-01
-4.56604697e-02 5.09798169e-01 6.47710204e-01 8.05918694e-01
-6.63286388e-01 6.02668166e-01 -1.10641992e+00 6.05260246e-02
8.53188157e-01 5.80159426e-01 -2.81363517e-01 2.40707099e-02
-1.04699624e+00 9.02305543e-01 5.62958658e-01 3.76880139e-01
-5.79577982e-01 -7.08003283e-01 -9.06086326e-01 -3.40402365e-01
4.19181883e-01 -3.85208607e-01 1.23581934e+00 -1.05742407e+00
-1.57000840e+00 1.24782503e+00 1.47600025e-01 -4.51169103e-01
7.80039430e-01 1.53705388e-01 -3.94479811e-01 3.39161247e-01
3.68873805e-01 7.72138357e-01 6.92299366e-01 -1.40110767e+00
-4.96793926e-01 -6.29373193e-01 -4.77401912e-01 2.29105726e-01
8.59606788e-02 -2.38947943e-01 -5.32186568e-01 -8.17795336e-01
4.63110864e-01 -1.21187198e+00 -5.68200290e-01 1.03579529e-01
-5.17103255e-01 -2.05291957e-01 5.84435403e-01 -5.83186746e-01
1.08408475e+00 -1.99010897e+00 -9.47974101e-02 3.10932755e-01
3.21628034e-01 4.56304580e-01 2.86611587e-01 -4.39248502e-01
-1.47602186e-01 -1.70570631e-02 -6.73454881e-01 -2.65162379e-01
-5.92454851e-01 3.44604492e-01 4.43232983e-01 7.02132106e-01
1.30479172e-01 9.18515027e-01 -9.42277074e-01 -1.08959615e+00
2.46195883e-01 1.67574942e-01 -2.38880932e-01 2.14948401e-01
-2.56240852e-02 9.40149486e-01 -5.10610282e-01 6.88552022e-01
6.73214376e-01 -5.06102204e-01 5.06917775e-01 -4.24314588e-01
1.96858302e-01 -9.04182047e-02 -1.09561932e+00 1.89243448e+00
-1.70408711e-01 1.55914307e-01 3.46252143e-01 -1.29654849e+00
7.42070854e-01 4.45864379e-01 7.99374461e-01 -5.91758549e-01
4.10775065e-01 4.29793715e-01 1.29487723e-01 -7.22072840e-01
-1.92831218e-01 -4.38315004e-01 8.93851966e-02 1.79966882e-01
2.29915649e-01 -9.77836177e-02 5.94612248e-02 -6.59335218e-03
4.68547553e-01 3.03784937e-01 3.92954260e-01 -6.01305604e-01
7.68670022e-01 2.48092830e-01 1.05373824e+00 5.75645626e-01
-7.11034834e-01 6.05300188e-01 3.22250515e-01 -1.68691754e-01
-3.99245948e-01 -7.43699491e-01 -6.78153157e-01 5.76889515e-01
2.63039380e-01 -1.49983997e-02 -8.78506124e-01 -1.23780739e+00
-1.41906127e-01 4.17767376e-01 -6.04713917e-01 7.18183592e-02
-5.91448247e-01 -1.12539351e+00 6.76078856e-01 6.39939308e-01
4.49165344e-01 -7.48354197e-01 -5.26046395e-01 1.87524080e-01
-4.73652214e-01 -1.29570723e+00 -2.99045056e-01 6.43401146e-01
-1.27852464e+00 -1.29031384e+00 -9.86757696e-01 -9.94604349e-01
8.82232785e-01 2.68729962e-02 9.02313113e-01 -1.79586917e-01
-3.30577403e-01 1.33799896e-01 -2.46437222e-01 -4.90648150e-01
-6.83598459e-01 4.65053655e-02 -2.33695954e-01 -6.06169477e-02
1.24187320e-01 -1.28387243e-01 -6.81810498e-01 5.62908053e-01
-7.52300382e-01 1.84321001e-01 6.41237438e-01 1.25524306e+00
1.08211040e+00 -1.30840287e-01 6.64368808e-01 -1.41936767e+00
2.41466314e-01 -2.62373000e-01 -4.70239162e-01 4.28839624e-01
-9.05763447e-01 6.93299482e-03 4.23062623e-01 -2.96008617e-01
-1.40025806e+00 3.68241578e-01 -2.33288974e-01 -1.86875805e-01
-1.75072461e-01 5.45168579e-01 9.27354023e-02 -1.95378974e-01
7.42146730e-01 -2.01855049e-01 4.47706044e-01 -1.37094140e-01
1.94375396e-01 8.00294697e-01 4.91459876e-01 -3.40196192e-01
1.42794266e-01 5.61641574e-01 9.01072174e-02 -3.52451175e-01
-1.19852769e+00 -8.81691456e-01 -7.49782920e-01 -2.57060468e-01
1.04898155e+00 -9.53000486e-01 -2.47127667e-01 4.60068166e-01
-7.36324072e-01 -2.10903674e-01 -4.02696848e-01 8.03194821e-01
-4.31830108e-01 5.68848848e-01 -7.20044076e-01 -4.20210242e-01
-5.17658114e-01 -1.70229125e+00 9.46168363e-01 3.89850259e-01
-1.58907652e-01 -1.17630327e+00 -7.84507170e-02 8.11293483e-01
6.45449460e-02 3.50053638e-01 6.65947318e-01 -1.03693855e+00
-1.75411493e-01 -2.55082756e-01 -1.70191541e-01 5.79242587e-01
2.36909568e-01 -3.21516663e-01 -1.04546845e+00 -2.08283275e-01
2.79729307e-01 -4.17665452e-01 8.25637937e-01 8.27034473e-01
1.19774580e+00 2.12298095e-01 -5.61075032e-01 5.31737566e-01
1.17643964e+00 9.10648406e-02 2.00071618e-01 2.55110979e-01
5.67234814e-01 7.12766409e-01 8.74135256e-01 1.30866334e-01
2.29035854e-01 4.45540905e-01 4.08991158e-01 -6.32750928e-01
-8.18724856e-02 1.22958615e-01 -3.15069556e-01 7.75879860e-01
2.32374713e-01 7.73779303e-02 -1.04625368e+00 4.80803162e-01
-1.95829570e+00 -4.56931829e-01 -4.61587757e-01 1.98039877e+00
1.12509394e+00 1.27757296e-01 -2.42137145e-02 1.57032818e-01
8.56433094e-01 -9.12953839e-02 -4.95112538e-01 9.08571854e-02
-1.67114064e-02 1.06861919e-01 6.16205037e-01 2.39236593e-01
-1.46444237e+00 5.78149319e-01 6.39464617e+00 1.00129437e+00
-1.15091729e+00 3.39004993e-01 1.03250027e+00 3.50977004e-01
1.01859018e-01 -2.43478283e-01 -6.34610057e-01 2.64861643e-01
6.69656754e-01 5.53965211e-01 -3.15146893e-01 7.92697489e-01
1.00506224e-01 -5.57814419e-01 -1.01878095e+00 8.50976169e-01
2.07537383e-01 -1.34135938e+00 -3.50470215e-01 -2.42070302e-01
7.68748343e-01 5.89214042e-02 -1.78869769e-01 1.27230749e-01
1.27933826e-02 -7.06490397e-01 3.35149318e-01 1.85550153e-01
8.31903934e-01 -2.12565124e-01 1.10743618e+00 4.12855297e-01
-8.18954825e-01 1.06149539e-01 9.50445309e-02 5.99597871e-01
3.25852215e-01 9.48087037e-01 -8.29132795e-01 9.31997001e-01
6.05365276e-01 6.61207974e-01 -3.92990887e-01 1.14493406e+00
-2.12202966e-01 6.41341865e-01 -1.17533900e-01 3.16740423e-01
1.30920559e-01 -3.30896199e-01 3.61009806e-01 1.38055480e+00
-1.74532145e-01 8.26425478e-02 4.42603916e-01 5.97830236e-01
1.60924882e-01 1.74885914e-01 2.23436020e-02 1.49958313e-01
-1.19952755e-02 1.54444277e+00 -1.22719443e+00 -5.20458877e-01
-3.32323194e-01 7.12816298e-01 1.28325686e-01 2.17527803e-02
-9.27972794e-01 3.54591310e-02 -6.74710199e-02 -7.12482333e-02
-8.98415744e-02 2.01343358e-01 -6.17296755e-01 -1.11064029e+00
-7.82084018e-02 -7.64468908e-01 7.93592870e-01 -4.44768608e-01
-1.30899322e+00 6.42162383e-01 1.25751384e-02 -1.40590596e+00
-3.18043269e-02 -6.01385295e-01 -2.52040297e-01 6.02733910e-01
-1.79099262e+00 -1.18739390e+00 -2.90873736e-01 3.35068971e-01
4.20905590e-01 5.65373674e-02 8.22007954e-01 4.82000172e-01
-7.08200097e-01 6.01968944e-01 1.29205346e-01 2.02911079e-01
9.82558548e-01 -1.36758399e+00 -5.26939571e-01 4.40646857e-01
-1.58797681e-01 2.88702548e-01 4.04096633e-01 -6.65407300e-01
-7.83678174e-01 -1.07713366e+00 3.74149829e-01 -1.47433519e-01
3.37690026e-01 6.15431704e-02 -8.98386419e-01 5.11459768e-01
5.79254255e-02 4.85331714e-01 1.18743551e+00 -1.04959354e-01
1.19814254e-01 -1.96368039e-01 -1.47809529e+00 1.67042792e-01
6.46362364e-01 -1.80291638e-01 -5.41293979e-01 4.51180547e-01
5.40661514e-01 -9.14052367e-01 -1.23301864e+00 8.91996741e-01
4.95449901e-01 -9.33660984e-01 6.77098572e-01 -2.27268383e-01
1.46180257e-01 -2.69798189e-01 3.63103837e-01 -1.20000434e+00
-6.60494044e-02 -3.67587835e-01 7.67485723e-02 1.19304824e+00
6.60624564e-01 -6.33667946e-01 9.90686059e-01 4.80797410e-01
-4.04121935e-01 -1.10870218e+00 -9.57045555e-01 -5.06803989e-01
3.35201360e-02 -3.18024009e-01 -1.40699334e-02 1.11872923e+00
1.72304273e-01 1.51343703e-01 -8.16753358e-02 1.15209073e-01
8.42402339e-01 1.02225818e-01 1.54203296e-01 -1.24526870e+00
-2.12571844e-01 -3.06583583e-01 -2.21374720e-01 -7.72174239e-01
2.20625132e-01 -1.20979702e+00 3.14609617e-01 -1.41491163e+00
2.70319581e-01 -8.02980959e-01 -3.96356225e-01 4.15273488e-01
-3.78248960e-01 3.18651885e-01 -2.32229218e-01 3.77802312e-01
-7.68671095e-01 4.24631029e-01 1.65535986e+00 -1.47020951e-01
-2.63272136e-01 3.65852267e-01 -5.64369321e-01 7.93793142e-01
8.29576194e-01 -7.25799918e-01 -2.92929530e-01 -7.96805173e-02
-4.19906855e-01 3.55035365e-01 8.98878947e-02 -8.46314847e-01
2.54318893e-01 1.96762890e-01 4.35917258e-01 -5.63916445e-01
-5.58687933e-02 -1.02163792e+00 3.58231030e-02 7.07446873e-01
-5.77553391e-01 -3.49730581e-01 1.70149535e-01 6.48518264e-01
-5.36736012e-01 -4.60759193e-01 1.21194971e+00 -4.17745143e-01
-2.18694299e-01 1.78464413e-01 -2.30923980e-01 1.19865485e-01
1.34597814e+00 -3.05238664e-01 2.30102018e-02 1.17656097e-01
-1.23109972e+00 6.34504676e-01 -5.04855774e-02 8.65702331e-02
3.11247110e-01 -9.87879097e-01 -5.19933224e-01 1.32294551e-01
1.62545905e-01 3.58726293e-01 4.36925977e-01 1.42259908e+00
-4.33567971e-01 3.74105006e-01 -6.50256127e-02 -1.24013042e+00
-1.32607961e+00 3.03527117e-01 6.20637596e-01 -7.40192652e-01
-4.80657637e-01 8.68775070e-01 1.31722540e-01 -7.60406852e-01
3.66521299e-01 -3.02595347e-01 -3.72624964e-01 3.64419930e-02
1.53312787e-01 3.77863675e-01 2.99821526e-01 -6.60811484e-01
-4.63885337e-01 4.05113161e-01 -1.44741699e-01 2.41777942e-01
1.08620501e+00 -9.81929600e-02 -2.33024523e-01 3.65166187e-01
1.05730629e+00 -3.86039704e-01 -1.13059461e+00 -3.41660559e-01
1.66130200e-01 -1.28485542e-02 4.46368337e-01 -1.03539860e+00
-1.23406816e+00 7.77542710e-01 1.02701008e+00 -7.39287436e-02
1.11646569e+00 1.45353712e-02 5.30077219e-01 -1.65979892e-01
2.45453686e-01 -1.26437080e+00 -7.98996240e-02 -1.16811603e-01
4.76479352e-01 -1.79532433e+00 2.04313323e-01 -9.35417771e-01
-1.00477874e+00 1.14745975e+00 6.24377906e-01 1.69082344e-01
7.30340064e-01 4.27551806e-01 5.59949338e-01 -1.53721407e-01
-1.61117658e-01 -1.09375894e-01 4.96210188e-01 5.82261443e-01
6.72181368e-01 2.30323926e-01 -5.68020463e-01 6.24076664e-01
4.59564894e-01 1.30489618e-01 2.11307630e-01 9.86854613e-01
-3.92541215e-02 -1.22959304e+00 -2.61120111e-01 6.16430879e-01
-9.27500308e-01 2.24182934e-01 1.31216526e-01 7.43379176e-01
2.71161824e-01 8.21343482e-01 -4.01197225e-01 7.92850256e-02
1.42288819e-01 6.88816532e-02 4.77704197e-01 -5.56577563e-01
-8.92755985e-01 5.85529804e-01 1.09919712e-01 -4.36235279e-01
-1.03674686e+00 -6.84020579e-01 -1.56015074e+00 5.15565395e-01
-9.80859876e-01 2.45227844e-01 6.46761596e-01 9.90383208e-01
2.22114727e-01 7.98901260e-01 5.33624887e-01 -6.50385737e-01
-6.88696921e-01 -8.60123873e-01 -4.71904457e-01 3.45504791e-01
2.09374830e-01 -6.08493745e-01 -2.19266564e-01 -3.00791077e-02] | [14.713275909423828, -2.1454758644104004] |
5a69aa0f-0ba5-4232-a102-ac016102b9f2 | v1net-a-computational-model-of-cortical | null | null | https://openreview.net/forum?id=Hyg4kkHKwH | https://openreview.net/pdf?id=Hyg4kkHKwH | V1Net: A computational model of cortical horizontal connections | The primate visual system builds robust, multi-purpose representations of the external world in order to support several diverse downstream cortical processes. Such representations are required to be invariant to the sensory inconsistencies caused by dynamically varying lighting, local texture distortion, etc. A key architectural feature combating such environmental irregularities is ‘long-range horizontal connections’ that aid the perception of the global form of objects. In this work, we explore the introduction of such horizontal connections into standard deep convolutional networks; we present V1Net -- a novel convolutional-recurrent unit that models linear and nonlinear horizontal inhibitory and excitatory connections inspired by primate visual cortical connectivity. We introduce the Texturized Challenge -- a new benchmark to evaluate object recognition performance under perceptual noise -- which we use to evaluate V1Net against an array of carefully selected control models with/without recurrent processing. Additionally, we present results from an ablation study of V1Net demonstrating the utility of diverse neurally inspired horizontal connections for state-of-the-art AI systems on the task of object boundary detection from natural images. We also present the emergence of several biologically plausible horizontal connectivity patterns, namely center-on surround-off, association fields and border-ownership connectivity patterns in a V1Net model trained to perform boundary detection on natural images from the Berkeley Segmentation Dataset 500 (BSDS500). Our findings suggest an increased representational similarity between V1Net and biological visual systems, and highlight the importance of neurally inspired recurrent contextual processing principles for learning visual representations that are robust to perceptual noise and furthering the state-of-the-art in computer vision. | ['Virginia R. de Sa', 'Vijay Veerabadran'] | 2019-09-25 | null | null | null | null | ['boundary-detection'] | ['computer-vision'] | [ 4.30519193e-01 -1.58649904e-03 3.05594116e-01 -1.10641479e-01
2.48952210e-01 -4.65550274e-01 6.21539831e-01 1.72686521e-02
-5.26938856e-01 2.85196483e-01 1.50069982e-01 -1.42069116e-01
-3.65080461e-02 -6.28274441e-01 -1.00751150e+00 -5.68042696e-01
-3.43789279e-01 1.23889726e-02 7.32128739e-01 -5.32458186e-01
3.71955752e-01 6.17346108e-01 -1.84014964e+00 6.83962345e-01
4.36474353e-01 1.17731500e+00 4.75694358e-01 8.62298846e-01
2.19578400e-01 5.16534507e-01 -5.23560882e-01 2.71044642e-01
4.37761366e-01 -4.87136692e-01 -6.23504639e-01 -1.58821598e-01
7.82705843e-01 2.06686705e-01 -9.97775048e-02 7.47646391e-01
4.64423388e-01 6.57943115e-02 7.94751763e-01 -8.95527005e-01
-1.14770436e+00 2.59545386e-01 -3.38333637e-01 8.31032455e-01
7.13508651e-02 4.92946714e-01 1.12274623e+00 -6.62752509e-01
1.04864311e+00 1.16514421e+00 7.52900720e-01 7.60079741e-01
-1.91754842e+00 -4.51482952e-01 4.02854532e-01 9.74037498e-02
-1.24191082e+00 -3.33676666e-01 4.74782318e-01 -7.21849322e-01
1.72834039e+00 1.24148779e-01 1.10198343e+00 1.34267962e+00
6.51829183e-01 5.66572189e-01 1.05129659e+00 -2.46855706e-01
1.80803940e-01 -5.72202682e-01 9.66783911e-02 3.85328233e-01
3.36422354e-01 4.23049927e-01 -5.31208277e-01 2.45242208e-01
1.21466446e+00 -2.33882204e-01 -4.27711099e-01 -2.56299049e-01
-1.35448992e+00 4.77116019e-01 1.20942819e+00 5.61522722e-01
-3.56339782e-01 5.21422803e-01 2.38966882e-01 2.52112359e-01
-1.02824785e-01 6.72875464e-01 -3.92092854e-01 4.74417031e-01
-7.57949471e-01 1.21041059e-01 5.03124356e-01 6.10397220e-01
6.56243086e-01 2.68081903e-01 -1.16103917e-01 7.52903759e-01
4.49108750e-01 1.87953681e-01 6.46354079e-01 -7.99475610e-01
1.16334751e-01 6.45252943e-01 -3.92414331e-01 -1.03130710e+00
-8.85257483e-01 -4.87809181e-01 -1.02128315e+00 5.78094304e-01
4.21967864e-01 2.92500108e-01 -1.33303893e+00 1.85063493e+00
-1.51354149e-01 -8.09219480e-02 2.02536359e-02 1.00324953e+00
8.71412516e-01 3.33198935e-01 2.88937509e-01 1.36350721e-01
1.16686678e+00 -7.22200096e-01 -1.64085671e-01 -6.91155255e-01
2.47419238e-01 -5.04995465e-01 9.26926017e-01 1.69877902e-01
-9.31285977e-01 -8.45429599e-01 -1.29070139e+00 -2.23639295e-01
-6.75302982e-01 -2.60814399e-01 6.10677481e-01 2.46763676e-01
-1.55986834e+00 4.89656061e-01 -5.92525899e-01 -7.40115285e-01
6.85457528e-01 6.05852365e-01 -5.09123683e-01 4.20631588e-01
-8.43356967e-01 9.11884189e-01 4.65104938e-01 2.73910224e-01
-1.20606506e+00 -6.00343168e-01 -7.69842088e-01 4.90713455e-02
-1.46958992e-01 -1.04224372e+00 8.66288364e-01 -1.18291068e+00
-9.93883967e-01 1.26771545e+00 -2.80284416e-02 -7.35522449e-01
1.84017882e-01 -5.62743656e-02 -5.12872934e-02 3.41447562e-01
1.51879787e-02 1.50149632e+00 7.04332709e-01 -1.42601621e+00
-1.90861613e-01 -3.03158611e-01 -4.76716816e-01 3.19750085e-02
3.87831777e-01 -2.49500811e-01 -7.74599761e-02 -9.64839697e-01
4.97143686e-01 -9.23919976e-01 -2.79553026e-01 1.59208789e-01
-3.37038606e-01 -1.92041978e-01 6.48820043e-01 -3.39430898e-01
5.00831068e-01 -2.28456187e+00 2.67148286e-01 1.99030489e-01
2.01137871e-01 2.32294440e-01 -3.82329553e-01 1.38119668e-01
-3.20938587e-01 5.51625490e-01 -5.25655508e-01 1.69905536e-02
-3.83110642e-01 2.41869003e-01 -5.68246961e-01 4.48479444e-01
5.82229495e-01 1.33249342e+00 -4.59595382e-01 -7.89033547e-02
-1.98589321e-02 5.52944303e-01 -6.82118118e-01 -2.18540430e-03
-5.32074928e-01 5.03731072e-01 7.30824322e-02 5.00126898e-01
3.68455887e-01 -6.59289956e-02 -1.81885123e-01 4.25862633e-02
-2.62486696e-01 9.72499400e-02 -7.16578662e-01 1.80034614e+00
-9.75460559e-02 1.06588209e+00 1.64620548e-01 -8.91727507e-01
8.68738651e-01 2.06285447e-01 -9.16112959e-02 -1.16975200e+00
4.75180924e-01 9.31373984e-02 6.35716081e-01 -1.64899737e-01
9.03212205e-02 1.62435845e-01 2.20935509e-01 2.35759735e-01
3.64070565e-01 -2.87972778e-01 1.26052529e-01 7.18526393e-02
8.61830533e-01 1.50427684e-01 1.39267921e-01 -6.05821609e-01
1.73466086e-01 -2.59476513e-01 5.01938581e-01 8.97787035e-01
-4.20605779e-01 1.13240397e+00 5.41487157e-01 -5.49650490e-01
-1.09606349e+00 -1.17797160e+00 -3.62315506e-01 1.18943667e+00
1.59263045e-01 6.81419820e-02 -7.28795528e-01 2.49891188e-02
-1.36512090e-02 4.40144956e-01 -1.26980424e+00 -3.56457829e-01
-4.92862642e-01 -7.03588486e-01 8.12090755e-01 7.09973156e-01
7.83865511e-01 -1.79726923e+00 -1.34205914e+00 3.38228941e-01
2.94865936e-01 -9.81910586e-01 -5.34297200e-03 6.43696189e-01
-6.80832446e-01 -1.12811637e+00 -6.36130810e-01 -1.14063025e+00
4.60111588e-01 1.90325230e-01 1.00588822e+00 3.33511502e-01
-7.92831123e-01 4.30588514e-01 -4.08004634e-02 -3.11385483e-01
-3.40085812e-02 -5.59959910e-04 -1.66921802e-02 -1.71291962e-01
9.82255712e-02 -9.33158219e-01 -7.57105887e-01 6.12081945e-01
-1.03768909e+00 1.68346867e-01 4.81495738e-01 9.32250500e-01
7.02274382e-01 -6.13506079e-01 4.96403873e-01 -4.45287675e-01
5.31111419e-01 -2.08068758e-01 -5.27693987e-01 1.46301582e-01
-7.18301162e-02 9.77968648e-02 4.83649880e-01 -4.65607077e-01
-8.04519475e-01 -2.04280615e-01 -7.04175010e-02 -4.03939784e-02
-4.67363805e-01 7.86181763e-02 6.33300245e-02 -3.49500626e-01
1.06012344e+00 2.97622412e-01 -3.41123819e-01 -6.65328130e-02
4.97774661e-01 -9.50272754e-03 8.79375875e-01 -5.42337894e-01
3.60918581e-01 7.29456902e-01 -5.77908196e-02 -1.28921127e+00
-3.58987212e-01 -3.89397889e-01 -9.15253222e-01 -1.70105398e-02
1.36021113e+00 -5.51783860e-01 -8.63904715e-01 8.11212718e-01
-1.36912179e+00 -9.01449680e-01 -3.44494194e-01 1.29121467e-01
-8.69518876e-01 -1.77687138e-01 -6.45200074e-01 -3.48564655e-01
-2.43979275e-01 -1.06666028e+00 7.85719693e-01 5.55017471e-01
-2.72111923e-01 -7.94124186e-01 2.38143504e-01 -1.99169237e-02
6.15158737e-01 3.47928524e-01 1.18819761e+00 -5.31275928e-01
-6.40074551e-01 4.95689660e-01 -4.86023694e-01 9.71145183e-02
-3.42444032e-01 2.30356783e-01 -1.33153069e+00 -3.39656584e-02
-1.82003409e-01 -5.22972763e-01 1.59625399e+00 6.48446262e-01
8.45231771e-01 6.80894703e-02 -3.02152306e-01 1.02519000e+00
1.37760913e+00 1.84599787e-01 7.94511378e-01 5.20331979e-01
5.47387302e-01 6.77501142e-01 -3.71651828e-01 -1.16665155e-01
-5.55093437e-02 1.98097363e-01 7.61454761e-01 -4.40141797e-01
-4.96088386e-01 -3.30170579e-02 -9.53705981e-02 3.55109692e-01
-1.96477622e-01 -1.92098580e-02 -1.12440693e+00 8.08606386e-01
-1.54597819e+00 -1.02798712e+00 9.52536538e-02 1.77087724e+00
6.28182769e-01 4.71603960e-01 -1.55610055e-01 -2.48085186e-02
5.82678139e-01 1.46528631e-01 -7.72021055e-01 -7.87944555e-01
-6.63823068e-01 3.61151099e-01 3.21747065e-01 2.42529735e-01
-1.19270015e+00 1.36468887e+00 7.03294086e+00 2.18486741e-01
-1.39410651e+00 -1.94112465e-01 7.58039236e-01 7.00049102e-02
-1.41729593e-01 -1.81268841e-01 -6.49815857e-01 -1.69924974e-01
8.16044092e-01 4.17893261e-01 5.83073378e-01 6.39405131e-01
-6.85926452e-02 -3.62575650e-01 -1.23048985e+00 7.72985220e-01
1.22120798e-01 -1.65065002e+00 1.60525456e-01 1.35235488e-01
9.02090013e-01 7.52369702e-01 4.30572748e-01 1.24620711e-02
3.71000379e-01 -1.53814876e+00 9.05447006e-01 4.53304350e-01
5.83044708e-01 -2.05159128e-01 2.58601993e-01 1.21966809e-01
-1.20509696e+00 -2.03444257e-01 -6.24940634e-01 -1.71730682e-01
-1.45749182e-01 -3.84794585e-02 -5.71585357e-01 -1.55189067e-01
1.13338566e+00 6.15663707e-01 -8.86595368e-01 1.18943167e+00
-2.39288285e-01 3.22286278e-01 -4.04109299e-01 1.51930556e-01
3.69982272e-01 2.08000839e-01 3.97259712e-01 1.43567753e+00
-2.20567390e-01 2.52507567e-01 -9.13087800e-02 1.27690637e+00
-1.01455890e-01 -2.17410084e-03 -9.23673093e-01 2.03501523e-01
1.19975701e-01 9.81680393e-01 -1.30530989e+00 -5.51099144e-02
-2.19191849e-01 9.14690733e-01 5.44472098e-01 7.43840456e-01
-4.21759516e-01 -1.79779619e-01 7.54175901e-01 -4.56043743e-02
6.53994441e-01 -6.38318360e-01 -6.54420972e-01 -8.43257189e-01
-2.77747989e-01 -6.40228450e-01 1.60265386e-01 -1.02940476e+00
-1.28918993e+00 9.40969586e-01 -2.12758332e-01 -5.25774777e-01
9.15583596e-02 -9.24332678e-01 -7.57592499e-01 7.61822939e-01
-1.25357068e+00 -1.15762579e+00 -2.46847123e-01 7.49506891e-01
5.19496381e-01 1.80257391e-02 8.11227679e-01 -4.14655507e-01
-3.30213577e-01 2.99117535e-01 -2.51187533e-01 3.01007718e-01
3.27699125e-01 -1.09952080e+00 8.59378278e-01 9.73525047e-01
4.24749583e-01 8.08253825e-01 4.74394530e-01 -6.04748547e-01
-7.39660919e-01 -9.76580024e-01 3.34701866e-01 -1.93574369e-01
5.43876469e-01 -6.52336240e-01 -1.15961528e+00 7.92971849e-01
5.46310544e-01 4.43584502e-01 6.25207901e-01 -1.22671194e-01
-7.89284229e-01 7.77501091e-02 -9.51051116e-01 8.16233873e-01
1.42254150e+00 -5.19812763e-01 -1.05151248e+00 -1.66904286e-01
7.23332226e-01 3.81177589e-02 -3.95968080e-01 4.06626254e-01
5.82971811e-01 -1.31919527e+00 9.82121229e-01 -5.38435340e-01
1.68071344e-01 -3.56476963e-01 -2.12187022e-01 -1.28219080e+00
-6.08108222e-01 -4.94118303e-01 4.74297881e-01 8.30754578e-01
4.41004187e-01 -8.05897653e-01 2.74750769e-01 2.47631878e-01
-2.41272673e-01 -5.13878226e-01 -1.22702193e+00 -5.32840014e-01
2.86529928e-01 -4.47356343e-01 -1.27476335e-01 5.88581502e-01
-2.93713450e-01 3.88280660e-01 4.48409766e-01 -7.57303387e-02
2.74056494e-01 3.13542783e-02 3.54431540e-01 -1.32176650e+00
-1.83015671e-02 -8.99037302e-01 -8.91389251e-01 -8.90509069e-01
1.23273566e-01 -1.00710905e+00 1.45718113e-01 -1.72274363e+00
-1.31388471e-01 -2.02864826e-01 -3.41962308e-01 6.77350163e-01
3.47636104e-01 6.85399473e-01 3.90176386e-01 3.16636711e-01
-3.18703443e-01 4.70937371e-01 1.29030359e+00 -1.51164815e-01
-2.31768578e-01 -5.14335096e-01 -6.03322506e-01 8.21511924e-01
7.70275831e-01 -3.11581224e-01 -2.68824399e-01 -6.08861268e-01
3.48267347e-01 -7.16181159e-01 6.79654777e-01 -1.30655837e+00
3.15834522e-01 3.71203311e-02 8.04398239e-01 -2.05330029e-01
1.44346327e-01 -7.03295052e-01 -3.02428961e-01 6.68848515e-01
-4.56223518e-01 1.83782429e-01 7.43158102e-01 6.25061035e-01
-4.94929291e-02 2.98654050e-01 9.70826149e-01 -4.30149972e-01
-1.07309544e+00 -9.25896037e-03 -9.42972064e-01 3.11975628e-01
1.05571389e+00 -5.45971692e-01 -8.41591656e-01 1.94283456e-01
-9.88453150e-01 1.85194127e-02 5.52888513e-01 5.18906116e-01
8.17629397e-01 -9.71861839e-01 -6.33883238e-01 6.22534513e-01
1.29915625e-02 -3.22281122e-02 7.17108846e-02 6.15653753e-01
-9.48340118e-01 5.47486305e-01 -9.25703049e-01 -9.43169951e-01
-7.59348094e-01 6.59243166e-01 6.93163097e-01 2.60508120e-01
-6.86394095e-01 1.25106549e+00 8.93280864e-01 -1.92530513e-01
1.03668027e-01 -8.98677051e-01 -3.77878636e-01 6.04749843e-03
2.28039801e-01 -2.47079477e-01 -1.89358816e-02 -6.98706448e-01
-4.35217023e-01 8.00655782e-01 1.33472472e-01 -1.36519492e-01
1.24092686e+00 1.07294694e-01 -2.30837643e-01 6.51342154e-01
8.78739834e-01 -4.69682306e-01 -1.45765305e+00 4.49947901e-02
1.42484933e-01 6.67747781e-02 -2.45830670e-01 -8.69196296e-01
-9.23903465e-01 1.29418111e+00 7.30712116e-01 1.91018879e-01
1.09745944e+00 3.15999836e-02 3.30379605e-01 4.52767730e-01
2.03345597e-01 -9.27441239e-01 3.03719968e-01 8.20769310e-01
1.43423128e+00 -1.07412827e+00 -3.70323092e-01 -1.01683982e-01
-5.30271113e-01 1.02934361e+00 8.89828324e-01 -5.88447273e-01
6.49882495e-01 1.70743257e-01 2.58458287e-01 -2.57626861e-01
-9.89322603e-01 -5.21470129e-01 3.88013661e-01 8.99779022e-01
3.75998884e-01 -1.02896959e-01 9.11489502e-02 2.17144340e-01
-2.47490689e-01 -4.72950548e-01 3.43702465e-01 7.50900149e-01
-6.27941906e-01 -4.61496055e-01 -2.72416562e-01 1.07786745e-01
-1.86450511e-01 -3.35729390e-01 -7.82198668e-01 9.10011590e-01
4.89301622e-01 6.21196568e-01 4.08153772e-01 -3.34396780e-01
2.13497445e-01 1.15247846e-01 5.53200424e-01 -7.24937379e-01
-1.03113842e+00 1.95819482e-01 -3.30111951e-01 -5.55572391e-01
-3.94679129e-01 -4.92447466e-01 -1.42383659e+00 2.32024983e-01
-5.46698757e-02 -3.59324455e-01 5.31738818e-01 7.95993268e-01
4.12921995e-01 7.87356973e-01 -2.33126819e-01 -1.37513673e+00
1.14331350e-01 -9.53626633e-01 -4.19434339e-01 4.79064107e-01
4.66836542e-01 -7.16453254e-01 -2.94368744e-01 3.08223128e-01] | [9.683012962341309, 2.434170722961426] |
467e50a1-25d7-4235-adca-9a734c9e1ad7 | off-apexnet-on-micro-expression-recognition | 1805.08699 | null | http://arxiv.org/abs/1805.08699v1 | http://arxiv.org/pdf/1805.08699v1.pdf | OFF-ApexNet on Micro-expression Recognition System | When a person attempts to conceal an emotion, the genuine emotion is manifest
as a micro-expression. Exploration of automatic facial micro-expression
recognition systems is relatively new in the computer vision domain. This is
due to the difficulty in implementing optimal feature extraction methods to
cope with the subtlety and brief motion characteristics of the expression. Most
of the existing approaches extract the subtle facial movements based on
hand-crafted features. In this paper, we address the micro-expression
recognition task with a convolutional neural network (CNN) architecture, which
well integrates the features extracted from each video. A new feature
descriptor, Optical Flow Features from Apex frame Network (OFF-ApexNet) is
introduced. This feature descriptor combines the optical ow guided context with
the CNN. Firstly, we obtain the location of the apex frame from each video
sequence as it portrays the highest intensity of facial motion among all
frames. Then, the optical ow information are attained from the apex frame and a
reference frame (i.e., onset frame). Finally, the optical flow features are fed
into a pre-designed CNN model for further feature enhancement as well as to
carry out the expression classification. To evaluate the effectiveness of
OFF-ApexNet, comprehensive evaluations are conducted on three public
spontaneous micro-expression datasets (i.e., SMIC, CASME II and SAMM). The
promising recognition result suggests that the proposed method can optimally
describe the significant micro-expression details. In particular, we report
that, in a multi-database with leave-one-subject-out cross-validation
experimental protocol, the recognition performance reaches 74.60% of
recognition accuracy and F-measure of 71.04%. We also note that this is the
first work that performs cross-dataset validation on three databases in this
domain. | ['Yen-Chang Huang', 'Wei-Chuen Yau', 'Sze-Teng Liong', 'Y. S. Gan', 'Tan Lit Ken'] | 2018-05-10 | null | null | null | null | ['micro-expression-recognition'] | ['computer-vision'] | [ 1.63517177e-01 -3.40194225e-01 -3.19127798e-01 -4.24291968e-01
-3.26581866e-01 -1.49531022e-01 4.89064485e-01 -2.89607733e-01
-5.58901370e-01 6.52530849e-01 -7.30499923e-02 4.49497283e-01
2.25717098e-01 -3.82244498e-01 -3.17755699e-01 -1.04916179e+00
-2.32362106e-01 -5.38600504e-01 -3.98826927e-01 -2.08086684e-01
1.14376970e-01 8.46782625e-01 -1.69168198e+00 3.10323328e-01
4.58696693e-01 1.55689394e+00 -2.31935248e-01 4.71070349e-01
-4.65322323e-02 1.04462385e+00 -7.45375693e-01 -5.24915993e-01
-3.21323499e-02 -5.45264602e-01 -4.83444840e-01 2.09404558e-01
4.17588770e-01 -4.82590765e-01 -2.05293626e-01 9.32170808e-01
4.96257544e-01 -3.38173807e-02 4.34063941e-01 -1.47841942e+00
-2.46057868e-01 -3.01774502e-01 -7.21456230e-01 3.90473515e-01
3.46593261e-01 9.59680006e-02 7.67859101e-01 -1.12686145e+00
9.09503639e-01 9.01524842e-01 5.76984942e-01 4.85162526e-01
-8.34625840e-01 -9.05767143e-01 -1.40488744e-01 3.92329633e-01
-1.55160701e+00 -7.57872701e-01 1.11534035e+00 -5.06079733e-01
7.07856178e-01 8.18596557e-02 9.07510459e-01 1.20016789e+00
2.19066203e-01 8.07295024e-01 1.14529002e+00 -2.75440127e-01
7.17632547e-02 3.67632508e-02 -1.41265720e-01 8.45401764e-01
-2.59534150e-01 2.13454232e-01 -7.53193974e-01 1.94567833e-02
4.85317677e-01 -1.19327322e-01 -3.92369151e-01 7.78416321e-02
-7.36424148e-01 5.37057161e-01 2.48104334e-01 7.11437523e-01
-7.23797977e-01 3.79991271e-02 6.65727198e-01 2.88983971e-01
5.10099769e-01 -1.76891731e-03 -2.85168707e-01 -5.76085508e-01
-1.05873513e+00 4.76851203e-02 6.60315573e-01 3.93202096e-01
8.11089933e-01 4.44608271e-01 -7.89803341e-02 8.82198930e-01
6.80867806e-02 4.39074248e-01 5.51392019e-01 -9.33107257e-01
6.14524893e-02 5.10856152e-01 -1.19319983e-01 -1.63281095e+00
-3.89950484e-01 -3.34891319e-01 -1.06798029e+00 3.14384043e-01
3.50975215e-01 -4.01799560e-01 -3.29682082e-01 1.96399820e+00
3.07840616e-01 4.62547064e-01 1.74874052e-01 1.13741910e+00
8.66012096e-01 5.88082373e-01 1.89883813e-01 -6.23850584e-01
1.42872417e+00 -6.30805075e-01 -1.04943657e+00 2.05759421e-01
6.43087029e-01 -6.65107250e-01 5.98036587e-01 4.79709655e-01
-8.49875510e-01 -7.21839964e-01 -1.02974558e+00 3.05586725e-01
-2.16531456e-01 6.51575089e-01 5.54020107e-01 4.91442740e-01
-8.58297348e-01 3.77697885e-01 -6.30567789e-01 -2.02931881e-01
6.46254182e-01 4.66898769e-01 -8.89176726e-01 2.24506155e-01
-1.10980248e+00 6.50476754e-01 6.00065887e-02 6.28888905e-01
-6.86931491e-01 -4.91816372e-01 -8.02745700e-01 -1.62486270e-01
1.56143885e-02 -2.00638816e-01 9.68856037e-01 -2.04712296e+00
-1.71842980e+00 9.31697249e-01 -4.69675004e-01 -8.14830512e-02
2.41441295e-01 -7.34253451e-02 -6.57356858e-01 6.17904723e-01
-1.30793542e-01 4.47722554e-01 1.00724983e+00 -9.34618413e-01
-4.80950683e-01 -4.98004019e-01 -1.35294367e-02 1.85370841e-03
-6.09846473e-01 3.91728967e-01 -3.22946489e-01 -5.92144966e-01
-1.68171763e-01 -7.65094459e-01 3.54589671e-01 2.68769592e-01
-1.90817967e-01 -1.22755967e-01 1.11056471e+00 -4.62649196e-01
1.12132001e+00 -2.40157437e+00 -4.55116630e-02 1.12788580e-01
2.35339895e-01 5.58823287e-01 -1.37612849e-01 9.05568451e-02
-3.24477375e-01 -1.63123786e-01 8.87784958e-02 -2.54102975e-01
-3.83062392e-01 5.41687533e-02 3.82959247e-02 7.80981421e-01
3.97798687e-01 8.18419397e-01 -6.75218940e-01 -5.82862735e-01
1.69744059e-01 7.28559732e-01 -3.00110906e-01 3.22168559e-01
3.68894756e-01 4.38649774e-01 -5.17357588e-01 9.24291909e-01
7.46418118e-01 1.78701207e-02 -1.24326721e-02 -5.98263621e-01
-1.13056682e-01 -5.85863531e-01 -9.60412204e-01 1.44771361e+00
-4.40474063e-01 1.04662633e+00 3.19286019e-01 -1.10462701e+00
1.07985163e+00 6.22300863e-01 9.06250834e-01 -7.64518678e-01
5.75046599e-01 2.68340439e-01 -5.65669872e-02 -8.58570278e-01
2.94148356e-01 -1.47127688e-01 3.04006785e-01 1.41813025e-01
2.06087217e-01 5.27184069e-01 1.70047238e-01 -2.33701989e-01
7.39304721e-01 1.29262730e-02 4.66099203e-01 -5.14111184e-02
1.01574910e+00 -4.26926583e-01 6.54869437e-01 2.41811395e-01
-6.48572326e-01 2.82988459e-01 6.46447122e-01 -6.96640491e-01
-6.68034017e-01 -4.48498636e-01 -1.22846119e-01 8.77082646e-01
-4.24308442e-02 -4.41843301e-01 -8.06266189e-01 -5.05155504e-01
-2.50947654e-01 1.58322230e-02 -7.34819651e-01 -1.63127273e-01
-4.84807402e-01 -6.13152206e-01 8.47333312e-01 4.18739557e-01
9.03820574e-01 -1.23945069e+00 -8.30933809e-01 1.94325686e-01
-2.22112685e-01 -1.41774440e+00 -2.58164644e-01 -3.31071526e-01
-4.84799385e-01 -1.12038267e+00 -6.66426539e-01 -6.22678101e-01
4.65073854e-01 -1.49117336e-02 6.28963649e-01 1.74204782e-01
-4.92139667e-01 1.88957021e-01 -2.58714765e-01 -1.61092460e-01
-1.11502722e-01 -3.48429620e-01 -5.02881221e-02 8.71416688e-01
7.08347082e-01 -5.29668927e-01 -5.92019439e-01 1.87564686e-01
-6.78238690e-01 -1.22883052e-01 5.19753635e-01 9.32511866e-01
4.69787329e-01 -1.48240149e-01 3.04572165e-01 -2.30180696e-01
5.88259339e-01 -3.60756189e-01 -3.80786389e-01 -2.92446204e-02
-1.53861657e-01 -4.94132847e-01 6.47986770e-01 -5.41947365e-01
-1.06425273e+00 1.64058790e-01 -2.31050208e-01 -5.69662988e-01
-3.92613769e-01 5.32962084e-01 -1.69063732e-01 -2.21963555e-01
3.08110416e-01 3.72853577e-01 3.76255244e-01 4.09752876e-02
-5.62423840e-02 8.28673303e-01 4.91162419e-01 -2.96435595e-01
3.21140140e-01 6.38074517e-01 2.41301432e-01 -1.37415445e+00
-6.29946291e-01 -3.63040745e-01 -6.30473614e-01 -6.63239300e-01
9.62585688e-01 -9.04602706e-01 -1.18718743e+00 9.61523831e-01
-1.28690124e+00 2.18484439e-02 1.37043759e-01 5.47130346e-01
-5.03201663e-01 3.71204078e-01 -6.26944184e-01 -9.46995854e-01
-4.05311465e-01 -1.29830992e+00 1.03431904e+00 3.62478077e-01
-3.29074204e-01 -8.93767238e-01 -5.52236922e-02 1.26365468e-01
4.58088875e-01 6.81295097e-01 3.99853349e-01 -3.26004803e-01
-2.77798891e-01 -3.87760311e-01 -3.14479172e-01 5.07657230e-01
1.37849882e-01 3.63832772e-01 -1.26794600e+00 -3.69606391e-02
-4.88864705e-02 -4.59272355e-01 5.06431699e-01 3.99786949e-01
1.16128290e+00 -3.02382141e-01 4.30631563e-02 8.57766688e-01
1.21054471e+00 3.67729723e-01 6.68216348e-01 3.61480445e-01
4.44041550e-01 6.91066861e-01 5.70701003e-01 7.98935592e-01
3.78955975e-02 1.02468979e+00 3.54366541e-01 -3.63079131e-01
7.76522607e-02 8.63681883e-02 4.90752846e-01 4.59628463e-01
-4.29511279e-01 -1.69677846e-02 -4.59255427e-01 1.87352210e-01
-1.64471197e+00 -1.26974761e+00 6.44082427e-02 1.71013546e+00
6.33204222e-01 -3.55054945e-01 1.40458047e-02 2.14718938e-01
4.89657044e-01 4.26098019e-01 -2.53520519e-01 -5.08391380e-01
-4.43104535e-01 2.53564566e-01 -4.03670855e-02 2.66439766e-01
-1.24055326e+00 7.85321116e-01 5.24158859e+00 6.93867505e-01
-1.72182548e+00 3.98213454e-02 7.93298960e-01 -6.84922710e-02
4.54355717e-01 -4.73918617e-01 -5.74911177e-01 4.03718144e-01
9.09522414e-01 2.17322186e-02 9.80725139e-02 8.53522658e-01
6.09570801e-01 8.76514532e-04 -7.13494122e-01 1.42911947e+00
3.43130141e-01 -1.12487745e+00 -1.79657951e-01 1.13381132e-01
3.45386267e-01 -4.44473803e-01 7.19629899e-02 4.46806997e-02
-6.06318593e-01 -1.10676420e+00 5.13961434e-01 6.27676725e-01
9.18322384e-01 -8.96029294e-01 9.84556854e-01 1.99012943e-02
-1.21020317e+00 -1.04012728e-01 -8.59487131e-02 -1.81686860e-02
1.25645429e-01 4.28183377e-01 -4.01075274e-01 4.43891078e-01
6.36949182e-01 1.08012664e+00 -2.52254099e-01 5.72156131e-01
-1.21189766e-01 5.45482755e-01 -9.16957930e-02 2.89638969e-03
2.95509219e-01 -2.21876130e-01 4.96417940e-01 1.46048200e+00
3.58848602e-01 2.71328330e-01 -1.33590177e-01 7.58476138e-01
-5.22844531e-02 3.88229668e-01 -6.05608463e-01 -1.63287595e-01
-7.20465556e-02 1.63028014e+00 -3.03130686e-01 -2.50953078e-01
-5.17889738e-01 1.05137837e+00 1.43923059e-01 5.38523197e-01
-8.09497535e-01 -4.47560221e-01 1.07083547e+00 -1.97123304e-01
1.41900361e-01 -1.16038337e-01 2.00922012e-01 -1.17600083e+00
1.40821397e-01 -8.40836227e-01 1.34101957e-01 -8.71593297e-01
-8.92553270e-01 9.20499623e-01 -2.05887869e-01 -1.13679826e+00
-4.97113883e-01 -6.98586702e-01 -5.96535206e-01 6.81377828e-01
-1.59641945e+00 -1.06273758e+00 -8.39738190e-01 8.29706490e-01
3.12522441e-01 -3.82941633e-01 9.26932812e-01 5.02914786e-01
-7.81215072e-01 8.50282967e-01 -2.09385991e-01 5.27476132e-01
5.16530216e-01 -6.61969125e-01 -5.35717309e-01 7.46408224e-01
-5.97018301e-02 3.93452138e-01 4.54649955e-01 -1.17407031e-01
-1.39302075e+00 -8.08260500e-01 9.38892901e-01 9.97022986e-02
5.53793788e-01 -1.37321725e-01 -7.63060153e-01 3.69921237e-01
1.11087702e-01 5.67633927e-01 8.42483759e-01 -3.57306540e-01
-1.73922390e-01 -4.94095564e-01 -1.06314242e+00 3.00261557e-01
7.14299142e-01 -5.91225982e-01 -9.44971964e-02 -6.66588992e-02
-5.86335473e-02 -2.27360860e-01 -9.42334116e-01 3.92897159e-01
9.98156428e-01 -1.26570785e+00 7.36135542e-01 -5.69139540e-01
7.06267416e-01 -1.73198715e-01 -2.32721135e-01 -1.00119114e+00
1.01761475e-01 -6.79864168e-01 -1.61773354e-01 1.24909806e+00
-1.09361731e-01 -4.71118927e-01 7.47671366e-01 3.65154833e-01
2.58281708e-01 -9.62944150e-01 -1.04346883e+00 -5.02985001e-01
-4.74519253e-01 -4.33774650e-01 4.23256099e-01 9.29471910e-01
-1.50064379e-01 9.02624652e-02 -5.58074594e-01 -1.74750507e-01
3.95901322e-01 8.14526603e-02 9.14624929e-01 -9.89986539e-01
7.06847012e-02 -5.04247665e-01 -9.24236536e-01 -7.25476146e-01
6.09944880e-01 -6.32687986e-01 -1.68923184e-01 -6.53047442e-01
6.27141967e-02 1.23820290e-01 -3.26751083e-01 4.47666526e-01
1.71724215e-01 4.88016069e-01 1.96909964e-01 1.47884041e-01
-3.51778924e-01 7.72582412e-01 1.29922736e+00 -7.98692852e-02
-1.84403136e-01 -1.79064482e-01 -4.10132200e-01 7.67966926e-01
7.10488796e-01 -1.12731703e-01 -2.52361447e-01 4.31383364e-02
-3.47086847e-01 3.33574891e-01 5.71874022e-01 -8.60660970e-01
5.66547215e-02 -1.50664330e-01 6.09179437e-01 -2.26972818e-01
6.65198505e-01 -8.46777916e-01 4.78639118e-02 3.74629617e-01
-1.49164304e-01 1.08693212e-01 2.19242007e-01 2.43894860e-01
-7.66779602e-01 -4.24154885e-02 9.18655157e-01 9.86440852e-02
-1.08054936e+00 5.17520487e-01 -4.79875594e-01 -1.13886096e-01
1.13564754e+00 -5.17707646e-01 -1.02957621e-01 -5.15755773e-01
-6.64923549e-01 -4.16643739e-01 3.46846208e-02 3.81623924e-01
8.00766349e-01 -1.42775869e+00 -6.16439044e-01 4.74990129e-01
2.62087405e-01 -6.07564449e-01 3.01425457e-01 1.21770072e+00
-4.22545403e-01 2.35521242e-01 -5.94943821e-01 -6.71750963e-01
-1.64262533e+00 7.20910728e-02 7.29877532e-01 1.13253102e-01
-4.25097823e-01 5.81706285e-01 1.12416014e-01 2.84113824e-01
4.61027026e-02 8.04443583e-02 -5.60907423e-01 3.40167314e-01
8.88817430e-01 2.41075158e-01 -1.36313468e-01 -1.38584054e+00
-4.98777092e-01 8.35667312e-01 1.52530581e-01 1.41014814e-01
1.25816023e+00 -6.09340519e-02 -3.44814807e-01 1.66046277e-01
1.88913918e+00 -1.03407845e-01 -1.18958986e+00 -6.77347332e-02
-2.23373070e-01 -6.63946927e-01 1.36635348e-01 -4.70780551e-01
-1.53404701e+00 9.78006184e-01 8.14701378e-01 -2.72721529e-01
1.50588775e+00 -4.17158753e-01 6.51340544e-01 2.00775012e-01
4.51709516e-02 -9.85265195e-01 3.57763052e-01 2.24631697e-01
8.96049678e-01 -1.09227657e+00 -2.72600770e-01 -1.30191311e-01
-6.30509675e-01 1.53436494e+00 7.45979786e-01 4.10454199e-02
7.91332960e-01 2.69336313e-01 3.19882363e-01 -3.60011727e-01
-5.43730497e-01 -2.88471207e-02 2.18153194e-01 3.88398498e-01
5.25893033e-01 -1.75811142e-01 -1.46458670e-01 5.53986073e-01
-8.52089599e-02 5.23436248e-01 3.64366412e-01 7.11880565e-01
-2.57536799e-01 -5.68470597e-01 -1.72244325e-01 2.00513229e-02
-8.39365959e-01 3.47490907e-01 -3.17468077e-01 9.89995122e-01
2.12443560e-01 8.58614266e-01 8.14550966e-02 -4.13734496e-01
3.10888886e-01 1.29710957e-01 3.33290637e-01 1.19820200e-01
-4.04177606e-01 2.65189912e-02 -1.04713440e-03 -9.37065303e-01
-9.31401551e-01 -4.65961576e-01 -1.12163687e+00 -4.09565121e-01
2.32227407e-02 1.66232809e-02 5.99288762e-01 8.62083316e-01
3.70448351e-01 2.27965310e-01 8.02420139e-01 -8.82680118e-01
-1.16393536e-01 -8.29083443e-01 -7.41990030e-01 8.08627069e-01
6.35724425e-01 -7.88708389e-01 -4.51063067e-01 1.49961114e-01] | [13.629281997680664, 1.8062363862991333] |
0ccfd72d-92a7-413e-b33e-5af8da929db4 | swamp-swapped-assignment-of-multi-modal-pairs | 2111.05814 | null | https://arxiv.org/abs/2111.05814v2 | https://arxiv.org/pdf/2111.05814v2.pdf | SwAMP: Swapped Assignment of Multi-Modal Pairs for Cross-Modal Retrieval | We tackle the cross-modal retrieval problem, where learning is only supervised by relevant multi-modal pairs in the data. Although the contrastive learning is the most popular approach for this task, it makes potentially wrong assumption that the instances in different pairs are automatically irrelevant. To address the issue, we propose a novel loss function that is based on self-labeling of the unknown semantic classes. Specifically, we aim to predict class labels of the data instances in each modality, and assign those labels to the corresponding instances in the other modality (i.e., swapping the pseudo labels). With these swapped labels, we learn the data embedding for each modality using the supervised cross-entropy loss. This way, cross-modal instances from different pairs that are semantically related can be aligned to each other by the class predictor. We tested our approach on several real-world cross-modal retrieval problems, including text-based video retrieval, sketch-based image retrieval, and image-text retrieval. For all these tasks our method achieves significant performance improvement over the contrastive learning. | ['Minyoung Kim'] | 2021-11-10 | null | null | null | null | ['sketch-based-image-retrieval'] | ['computer-vision'] | [ 3.69250149e-01 -8.51097256e-02 -4.19733942e-01 -4.77231741e-01
-1.45528913e+00 -6.41920686e-01 6.70592546e-01 3.62368256e-01
-2.83722579e-01 6.68213189e-01 5.96201159e-02 3.79813612e-01
-3.39698076e-01 -5.84996700e-01 -7.29565620e-01 -8.93969536e-01
2.19000876e-01 8.00104380e-01 1.42806306e-01 4.50135358e-02
2.85553098e-01 1.99434921e-01 -1.65271783e+00 5.62913835e-01
5.34950256e-01 1.25321901e+00 7.84376636e-02 -2.57240981e-02
-2.54532039e-01 5.88689029e-01 -2.45247066e-01 -4.71778899e-01
2.38253340e-01 -2.48428136e-01 -7.77527869e-01 -1.07183799e-01
8.76772702e-01 -7.36679211e-02 -3.51083875e-01 1.07616901e+00
4.99377549e-01 7.56435394e-02 1.00155735e+00 -1.64743280e+00
-5.82861543e-01 2.82714695e-01 -7.86959410e-01 -2.66522139e-01
3.29057127e-01 -4.38058943e-01 1.33182454e+00 -1.03955650e+00
7.16320813e-01 1.45661473e+00 3.22163939e-01 7.20667660e-01
-1.26611090e+00 -8.67104828e-01 1.64180696e-01 4.44061220e-01
-1.51722848e+00 -4.11160380e-01 1.17602468e+00 -4.02612329e-01
1.95691645e-01 1.22336939e-01 2.63307214e-01 1.07776606e+00
-7.50242695e-02 1.17001998e+00 1.03729379e+00 -5.32375634e-01
-6.51645893e-03 4.26705927e-01 1.39675573e-01 5.56203544e-01
-2.99983442e-01 -1.08733192e-01 -6.70412898e-01 -2.78976977e-01
1.81415617e-01 3.14412624e-01 -2.46176928e-01 -1.01681113e+00
-1.43050575e+00 8.07568133e-01 4.17270005e-01 2.90924907e-01
-5.71408793e-02 7.87900761e-03 6.33329451e-01 3.81843776e-01
5.59628308e-01 2.08780825e-01 -5.75212419e-01 4.25888777e-01
-7.96192527e-01 1.29987746e-01 4.63285476e-01 9.53298986e-01
8.84829879e-01 -6.80162489e-01 -2.91841149e-01 1.15857744e+00
4.62247878e-01 5.66351593e-01 5.51464617e-01 -6.54388905e-01
5.71833551e-01 6.24422669e-01 -3.13470177e-02 -1.30831277e+00
-2.06572674e-02 1.24846272e-01 -7.49295294e-01 -1.62832305e-01
2.46569827e-01 3.17911178e-01 -7.95662522e-01 2.11796737e+00
2.34915808e-01 1.77550107e-01 2.52125651e-01 8.66175473e-01
9.45801675e-01 6.89065695e-01 2.05698684e-01 -2.14545533e-01
1.11231661e+00 -1.19875014e+00 -7.13371456e-01 -8.83256420e-02
4.62819576e-01 -9.54676807e-01 1.05100811e+00 5.37315682e-02
-8.49374771e-01 -4.57898736e-01 -9.13303018e-01 -1.50977045e-01
-5.28758645e-01 1.73458114e-01 2.30250016e-01 4.42870194e-03
-6.98680282e-01 5.47317624e-01 -3.38469923e-01 -3.10040653e-01
2.87326753e-01 1.28634632e-01 -7.34755039e-01 -3.29730541e-01
-1.33776295e+00 6.97123528e-01 4.98544365e-01 -3.12811702e-01
-7.04594135e-01 -6.03489161e-01 -8.44776213e-01 1.35703191e-01
4.09591347e-01 -4.57460761e-01 6.88183486e-01 -1.31505191e+00
-8.88083935e-01 1.37285519e+00 -1.36920035e-01 1.01082884e-01
4.61774737e-01 -1.37589127e-01 -4.88957018e-01 3.44710290e-01
5.51860929e-01 1.00794959e+00 1.27979267e+00 -1.64567089e+00
-5.31265497e-01 -5.99188745e-01 -4.01553921e-02 4.40010309e-01
-5.93165517e-01 -2.65622556e-01 -7.19175398e-01 -7.06674278e-01
3.66544515e-01 -1.05804610e+00 4.75845695e-01 4.36396480e-01
-3.06574047e-01 -5.47740877e-01 1.10114062e+00 -4.83262867e-01
9.39618766e-01 -2.38336611e+00 4.81679738e-01 1.54084414e-01
-7.64158592e-02 -1.29296575e-02 -4.20601487e-01 2.66528845e-01
-2.49754176e-01 -3.55156474e-02 -2.46990800e-01 -3.77230078e-01
3.93805802e-02 1.82630092e-01 -3.88350874e-01 2.39423618e-01
1.02472074e-01 7.30324030e-01 -1.01583028e+00 -9.45457637e-01
2.21279308e-01 3.21776181e-01 -2.96259612e-01 3.25152427e-01
-8.57640132e-02 3.85141850e-01 -4.66821492e-01 6.75663710e-01
6.95681870e-01 -3.05120558e-01 6.15845770e-02 -6.81806505e-01
5.15726805e-01 -2.30851591e-01 -1.04859555e+00 1.99651444e+00
-5.40392280e-01 6.74140751e-01 -2.80124575e-01 -1.26070547e+00
7.22184956e-01 2.83926100e-01 5.65740764e-01 -9.18109655e-01
-1.64345115e-01 3.78503412e-01 -6.14018857e-01 -7.24230349e-01
4.31446910e-01 -2.81257212e-01 -2.02150196e-01 4.77308780e-01
1.61083087e-01 -1.65464684e-01 1.10463044e-02 1.76079482e-01
4.56751436e-01 2.33201131e-01 -4.48019467e-02 -8.94587412e-02
7.79023826e-01 -3.01305264e-01 3.51964295e-01 6.01899207e-01
-1.41250223e-01 7.51403689e-01 3.46230656e-01 -4.12517697e-01
-1.01105189e+00 -1.12676287e+00 -2.76102960e-01 1.08926570e+00
7.12711751e-01 -3.37949276e-01 -2.85274327e-01 -1.09354651e+00
1.46539003e-01 3.64300430e-01 -7.50843406e-01 -5.23730278e-01
-2.63314992e-01 -3.24295670e-01 2.11769700e-01 1.63331568e-01
4.83911812e-01 -9.70802009e-01 -2.09201146e-02 -1.11800566e-01
-5.84320784e-01 -8.83940399e-01 -6.44255757e-01 3.04814409e-02
-6.37025058e-01 -1.25993240e+00 -8.77732992e-01 -1.25630760e+00
6.83093250e-01 3.39959532e-01 1.13591778e+00 -2.87057310e-02
-1.72321826e-01 6.84102297e-01 -2.99959779e-01 9.45540220e-02
-1.37464955e-01 1.05214752e-02 -5.21969423e-02 3.81995976e-01
4.47124422e-01 -1.47162423e-01 -6.57107651e-01 4.60595518e-01
-1.10273612e+00 -1.38989285e-01 2.87256926e-01 1.26613986e+00
9.62188244e-01 -1.05555570e-02 6.19810581e-01 -7.67651141e-01
3.86622578e-01 -6.14751458e-01 -1.63277715e-01 8.98183346e-01
-4.11942929e-01 3.07615221e-01 4.45154011e-01 -6.95192814e-01
-8.27549398e-01 6.55358434e-02 3.43990982e-01 -7.32588887e-01
1.37131726e-02 3.85247082e-01 -3.17327321e-01 1.89194139e-02
2.88077980e-01 2.34543920e-01 -6.67916089e-02 -3.66826802e-01
3.10678095e-01 6.30024135e-01 2.08725184e-01 -7.59591877e-01
6.47196412e-01 4.99889404e-01 1.93627831e-02 -6.49316609e-01
-1.12297535e+00 -5.29449701e-01 -4.76159483e-01 -2.14745864e-01
7.07790375e-01 -9.24883246e-01 -3.52502167e-01 3.64784420e-01
-1.04514682e+00 2.91326344e-01 -1.86663955e-01 3.16289932e-01
-5.49905956e-01 5.97446799e-01 -2.19722480e-01 -4.86313432e-01
-1.55720666e-01 -1.17153907e+00 1.54066956e+00 1.89807981e-01
1.82726420e-03 -1.08123171e+00 5.30817248e-02 4.04036492e-01
6.39201254e-02 -1.61894262e-02 1.38962257e+00 -8.95184338e-01
-4.85908836e-01 -3.52142453e-01 -4.97415155e-01 3.87526304e-01
2.36148775e-01 -1.39438570e-01 -1.01019788e+00 -4.95687693e-01
-2.63319671e-01 -8.05150449e-01 9.19720650e-01 7.24819154e-02
1.30638468e+00 -1.22959487e-01 -6.66860282e-01 1.69650525e-01
1.41526294e+00 -4.90565076e-02 4.26979184e-01 1.53734729e-01
6.36025071e-01 8.69978786e-01 9.20979857e-01 1.41951412e-01
2.23723799e-01 8.49768341e-01 4.63913739e-01 -2.74718497e-02
7.49614462e-02 -5.25244415e-01 -6.04363754e-02 6.47916436e-01
5.74246407e-01 -4.15122807e-01 -5.77157497e-01 7.48529673e-01
-1.98510826e+00 -7.88164735e-01 2.13783562e-01 2.46399093e+00
8.57797861e-01 -1.31169870e-01 -3.07553619e-01 -1.57794669e-01
9.76904511e-01 2.52807885e-01 -5.59850335e-01 2.24159762e-01
-2.40351006e-01 4.06193174e-03 -1.36218490e-02 3.61514866e-01
-1.49069703e+00 7.74378181e-01 5.29965401e+00 1.13757181e+00
-1.03433013e+00 4.69649918e-02 6.28704309e-01 -6.81528524e-02
-5.71504653e-01 7.44133368e-02 -4.86569256e-01 5.52395403e-01
2.54857719e-01 -5.13500115e-03 1.65311113e-01 5.34572482e-01
-3.77779573e-01 -4.40560170e-02 -1.51065779e+00 1.30219126e+00
5.68289876e-01 -9.11531568e-01 5.22965968e-01 -4.35000271e-01
8.27102184e-01 -3.78293186e-01 2.13552877e-01 4.17644173e-01
-3.60538602e-01 -5.57871163e-01 5.80540597e-01 7.59945214e-01
8.31359327e-01 -6.42352402e-01 6.63287938e-01 2.02370554e-01
-9.90665972e-01 4.90496717e-02 -2.50050813e-01 6.22423470e-01
-5.21354638e-02 3.68357360e-01 -1.29073828e-01 5.44817030e-01
7.12399840e-01 1.12073171e+00 -5.30056536e-01 8.94937217e-01
-2.21189335e-02 -8.12841672e-03 -2.00494155e-01 2.48868287e-01
4.15196605e-02 -2.16792017e-01 6.05762541e-01 7.98022032e-01
2.14693591e-01 -2.39117712e-01 2.06741706e-01 6.18789315e-01
-3.16170990e-01 2.03685760e-01 -7.07314014e-01 9.23537835e-02
5.44139802e-01 1.16889822e+00 -3.32370073e-01 -4.28931803e-01
-4.40582544e-01 1.16683555e+00 4.21643436e-01 3.43013018e-01
-6.88952208e-01 -4.31619138e-01 4.20107484e-01 -1.54652342e-01
2.69794632e-02 4.65504616e-01 1.69332191e-01 -1.32191145e+00
4.01995361e-01 -5.69181263e-01 6.57800615e-01 -9.64309216e-01
-1.89138448e+00 4.24581647e-01 2.69268863e-02 -1.62687707e+00
-8.93802494e-02 -4.52859342e-01 -4.30146158e-02 6.96046531e-01
-1.82183075e+00 -1.28508496e+00 -2.43029416e-01 7.43684530e-01
4.26698953e-01 -2.44700983e-01 9.64957356e-01 7.15088189e-01
-1.16865851e-01 7.60342956e-01 4.80256498e-01 7.80111179e-03
1.50270724e+00 -9.75807309e-01 -6.16667688e-01 7.30402321e-02
9.92422998e-02 4.31491166e-01 3.58864993e-01 -3.86403441e-01
-8.95184517e-01 -9.50643361e-01 1.07624769e+00 -2.65716106e-01
6.32395387e-01 -2.27239192e-01 -1.02703559e+00 5.17473280e-01
2.23860487e-01 2.20614389e-01 6.76680326e-01 -1.76034078e-01
-7.04482615e-01 -3.72337878e-01 -1.13189638e+00 5.24511218e-01
6.21876121e-01 -9.18352127e-01 -7.79234707e-01 6.89820111e-01
6.17815435e-01 -1.33486137e-01 -7.60367095e-01 5.29118419e-01
7.64826894e-01 -6.93615615e-01 1.17187643e+00 -7.94186056e-01
8.27370763e-01 -2.30750158e-01 -4.59623545e-01 -1.27354574e+00
1.24531761e-01 1.18524045e-01 2.93012142e-01 1.45384252e+00
4.03666258e-01 -4.31517422e-01 5.56165576e-01 6.15328372e-01
2.21871659e-01 -4.88859892e-01 -1.00434661e+00 -6.80624008e-01
2.67429173e-01 9.19183046e-02 3.07621628e-01 1.24570084e+00
2.19864268e-02 4.35670555e-01 -5.54414153e-01 -1.22739747e-01
8.07869613e-01 5.84596992e-01 4.46120083e-01 -1.33035350e+00
-8.29725489e-02 -2.41150200e-01 -3.52780014e-01 -9.49136853e-01
7.87279427e-01 -1.08667016e+00 8.50306600e-02 -1.10229635e+00
7.50741363e-01 -7.35941708e-01 -5.95995903e-01 5.31475782e-01
-2.99407244e-01 4.68611330e-01 3.19294035e-01 4.79598343e-01
-9.12873089e-01 7.73885846e-01 1.16034901e+00 -7.07030892e-01
1.20016828e-01 -2.35855371e-01 -2.58856684e-01 6.29124165e-01
5.36726713e-01 -6.48706257e-01 -3.32645327e-01 -5.24699688e-01
2.76308537e-01 2.95125574e-01 6.10346615e-01 -7.10246086e-01
2.43512198e-01 9.70035908e-04 1.84547111e-01 -6.90924883e-01
5.79011023e-01 -1.42446733e+00 -6.10955395e-02 9.12842378e-02
-9.01096106e-01 -2.40445122e-01 6.81465790e-02 7.71263301e-01
-7.17660129e-01 -3.88381392e-01 7.51774371e-01 4.47449572e-02
-6.99535191e-01 4.56721693e-01 2.34938100e-01 2.93233663e-01
9.95431066e-01 1.20320782e-01 -3.01063150e-01 -4.97943580e-01
-9.38289225e-01 3.85018647e-01 4.56984639e-01 6.88814104e-01
6.81365669e-01 -1.80606401e+00 -5.65029144e-01 1.96113586e-01
8.62307787e-01 -2.56968200e-01 4.25681442e-01 5.42228222e-01
1.06608413e-01 4.28647101e-01 -2.00206622e-01 -7.97931373e-01
-1.45800018e+00 8.04861724e-01 2.86885500e-01 -3.10135812e-01
-1.20821357e-01 6.23021841e-01 4.69233274e-01 -6.99250817e-01
2.72294790e-01 4.51228201e-01 -2.43640140e-01 4.48705077e-01
2.20875099e-01 1.04655780e-01 -1.07581802e-01 -7.37516999e-01
-3.86402488e-01 9.48530912e-01 -3.48768890e-01 -2.37684250e-02
1.09884226e+00 -2.79041052e-01 -3.67885321e-01 8.04154992e-01
1.88516188e+00 -2.19339356e-01 -1.03198385e+00 -7.59755909e-01
-5.64688332e-02 -5.55073977e-01 -1.66364342e-01 -7.21691132e-01
-1.24500895e+00 9.61755872e-01 9.58741069e-01 -9.41828117e-02
1.15005648e+00 2.77212441e-01 6.60437524e-01 3.36408317e-01
2.82088131e-01 -1.12434888e+00 3.48969638e-01 3.30024272e-01
8.20575774e-01 -1.71864891e+00 -1.10645965e-01 -2.25014597e-01
-4.45464879e-01 9.50935602e-01 6.07040048e-01 -3.02400813e-02
6.98721826e-01 -4.66826767e-01 -1.29680857e-01 -2.36038417e-01
-5.96643984e-01 8.42875093e-02 4.57000792e-01 3.69679242e-01
2.98436880e-01 -1.26619013e-02 -4.03200358e-01 6.53415546e-02
5.46215475e-01 -3.28086495e-01 -1.88327163e-01 7.87171960e-01
4.07722546e-03 -1.16913617e+00 -2.24959910e-01 4.09270376e-01
-1.61854938e-01 -1.78633958e-01 -5.02113760e-01 6.90088809e-01
-1.44972160e-01 7.04233646e-01 9.93220955e-02 -2.02252686e-01
2.33621567e-01 3.95663172e-01 5.09987772e-01 -2.85486430e-01
-1.54731497e-01 4.36366238e-02 -3.24567020e-01 -2.69353807e-01
-7.90192485e-01 -5.86032212e-01 -8.78790081e-01 4.65530194e-02
-4.32794034e-01 2.04156011e-01 4.56785351e-01 9.17327046e-01
2.90265322e-01 1.85520262e-01 8.77985239e-01 -6.25154853e-01
-4.20020461e-01 -6.10377014e-01 -6.34041071e-01 1.09319019e+00
4.31667894e-01 -9.21549022e-01 -6.96289361e-01 1.58469640e-02] | [11.11589241027832, 1.1065013408660889] |
deb1b04e-0929-49a9-a233-2b3d4fda5da5 | gradient-imitation-reinforcement-learning-for-1 | 2211.06014 | null | https://arxiv.org/abs/2211.06014v2 | https://arxiv.org/pdf/2211.06014v2.pdf | Gradient Imitation Reinforcement Learning for General Low-Resource Information Extraction | Information Extraction (IE) aims to extract structured information from heterogeneous sources. IE from natural language texts include sub-tasks such as Named Entity Recognition (NER), Relation Extraction (RE), and Event Extraction (EE). Most IE systems require comprehensive understandings of sentence structure, implied semantics, and domain knowledge to perform well; thus, IE tasks always need adequate external resources and annotations. However, it takes time and effort to obtain more human annotations. Low-Resource Information Extraction (LRIE) strives to use unsupervised data, reducing the required resources and human annotation. In practice, existing systems either utilize self-training schemes to generate pseudo labels that will cause the gradual drift problem, or leverage consistency regularization methods which inevitably possess confirmation bias. To alleviate confirmation bias due to the lack of feedback loops in existing LRIE learning paradigms, we develop a Gradient Imitation Reinforcement Learning (GIRL) method to encourage pseudo-labeled data to imitate the gradient descent direction on labeled data, which can force pseudo-labeled data to achieve better optimization capabilities similar to labeled data. Based on how well the pseudo-labeled data imitates the instructive gradient descent direction obtained from labeled data, we design a reward to quantify the imitation process and bootstrap the optimization capability of pseudo-labeled data through trial and error. In addition to learning paradigms, GIRL is not limited to specific sub-tasks, and we leverage GIRL to solve all IE sub-tasks (named entity recognition, relation extraction, and event extraction) in low-resource settings (semi-supervised IE and few-shot IE). | ['Philip S. Yu', 'Irwin King', 'Lijie Wen', 'Xiangli Yang', 'Chenwei Zhang', 'Shiao Meng', 'Xuming Hu'] | 2022-11-11 | null | null | null | null | ['event-extraction'] | ['natural-language-processing'] | [ 1.76557049e-01 4.57395285e-01 -5.41752398e-01 -4.70573187e-01
-8.52824509e-01 -6.19704366e-01 4.53954965e-01 9.60601270e-02
-8.92543375e-01 1.01559973e+00 2.45078817e-01 -1.39290467e-01
1.72252178e-01 -6.33816540e-01 -7.71978855e-01 -1.35129884e-01
9.88328010e-02 4.02960271e-01 -1.65721536e-01 1.12086786e-02
-6.33838624e-02 2.62348741e-01 -9.56367850e-01 8.12151283e-02
1.23254859e+00 7.03858137e-01 1.57406330e-01 4.08991843e-01
-2.53507555e-01 1.15254164e+00 -5.39622605e-01 -4.32702661e-01
1.07170708e-01 -5.13594508e-01 -1.11758304e+00 3.78544256e-02
-3.73241335e-01 -3.23742151e-01 -1.08874403e-01 9.47606742e-01
4.76604074e-01 1.57774314e-01 6.10411704e-01 -1.19770682e+00
-7.56528318e-01 8.92867446e-01 -4.00017112e-01 2.11961970e-01
3.75294119e-01 2.56285340e-01 1.07533991e+00 -1.08924174e+00
1.04428279e+00 7.99714029e-01 6.73361361e-01 7.01450586e-01
-1.14390945e+00 -7.39689231e-01 2.18093842e-01 -9.90026966e-02
-1.17105222e+00 -5.25793076e-01 7.30657041e-01 -1.81046486e-01
9.44235682e-01 -1.36105418e-01 3.68130893e-01 1.57611346e+00
-1.64006218e-01 1.19263279e+00 9.24724996e-01 -4.37387288e-01
5.00824392e-01 4.06952977e-01 2.46690243e-01 6.44361258e-01
3.30720693e-01 2.75197458e-02 -7.33663380e-01 9.53596905e-02
4.69182551e-01 4.71306918e-03 -2.44994730e-01 -1.69444740e-01
-1.27918375e+00 5.59597671e-01 3.45262021e-01 3.20689768e-01
-6.09361827e-01 -2.08038762e-01 5.03785610e-01 2.62721688e-01
3.96890849e-01 1.07981563e+00 -9.38896060e-01 -2.52847046e-01
-7.58549750e-01 -3.73962335e-02 1.00006020e+00 1.13974142e+00
9.41144347e-01 -3.98492105e-02 -2.47022510e-01 8.19670498e-01
1.00935936e-01 3.36951911e-01 6.68468297e-01 -7.13877439e-01
9.08059597e-01 9.30695534e-01 3.49210709e-01 -6.36818886e-01
-5.20533085e-01 -2.35363007e-01 -7.76638865e-01 -3.98292482e-01
5.25311768e-01 -7.38729477e-01 -7.62837589e-01 2.02045798e+00
4.55492616e-01 1.14218846e-01 3.41976136e-01 9.39095438e-01
8.76997650e-01 3.89272392e-01 3.43888521e-01 -3.56306970e-01
1.27887142e+00 -7.94586658e-01 -7.88760304e-01 -5.76360703e-01
1.02642488e+00 -1.71049938e-01 1.29601955e+00 -3.44291702e-02
-7.54058182e-01 -2.46296078e-01 -9.35483932e-01 -3.85300703e-02
-2.86034763e-01 3.72811943e-01 5.79892457e-01 2.72055864e-01
-3.58422488e-01 5.63927174e-01 -9.34177041e-01 -2.06166476e-01
4.97359067e-01 2.88902521e-01 -4.74153250e-01 3.05129662e-02
-1.57954991e+00 9.04604852e-01 5.56578696e-01 2.69964725e-01
-8.22541416e-01 -7.61814237e-01 -1.02127421e+00 8.55518058e-02
8.89162302e-01 -4.16540176e-01 1.27159178e+00 -9.43671227e-01
-1.51838326e+00 8.05244923e-01 -7.46044219e-02 -3.31835628e-01
3.39333653e-01 -4.54777002e-01 -3.30578774e-01 2.96936799e-02
2.67017961e-01 6.80300653e-01 5.71174622e-01 -1.08102071e+00
-4.13895994e-01 -1.30615473e-01 -4.80282381e-02 2.41150916e-01
-5.33387899e-01 2.91361809e-02 -3.32723022e-01 -5.14203727e-01
-2.32919499e-01 -8.87944877e-01 -1.84394300e-01 -5.57996929e-01
-5.79842806e-01 -5.18994510e-01 4.66531962e-01 -5.53580105e-01
1.28126991e+00 -2.05890012e+00 5.31005859e-03 -5.15807122e-02
3.39209825e-01 3.05249661e-01 -1.89463332e-01 2.14855894e-01
3.73959914e-02 3.52252990e-01 -1.45317629e-01 -3.11805159e-01
-5.71630448e-02 2.33444884e-01 -2.04026386e-01 1.25034172e-02
7.11012900e-01 1.09313822e+00 -1.26684725e+00 -6.66312933e-01
-3.72484714e-01 -3.65671031e-02 -5.28453887e-01 3.99422854e-01
-2.94351608e-01 6.97914243e-01 -8.31237912e-01 5.66121161e-01
1.96853533e-01 -7.05546200e-01 2.72563607e-01 -3.93557996e-01
1.30550131e-01 5.48821747e-01 -1.05986047e+00 1.63823533e+00
-6.08183503e-01 3.87508720e-01 -3.32375728e-02 -1.13306499e+00
8.70012820e-01 4.30272371e-01 4.13381934e-01 -5.71345508e-01
1.93132862e-01 1.81811720e-01 -1.35676667e-01 -7.99836636e-01
3.91137600e-01 -1.01483325e-02 -2.73204893e-01 7.28800595e-01
3.07071745e-01 -1.58731844e-02 3.25075418e-01 4.35234517e-01
1.34834945e+00 4.10916001e-01 4.08465952e-01 2.38808796e-01
1.86783195e-01 2.23508045e-01 1.10942841e+00 7.18515277e-01
-3.30502808e-01 9.90531594e-02 4.61640120e-01 -1.45013168e-01
-1.03612375e+00 -7.33635545e-01 2.62962263e-02 1.15064442e+00
1.00140564e-01 -4.67440844e-01 -7.11059153e-01 -1.24572039e+00
-3.35306853e-01 8.81616414e-01 -4.54505742e-01 -2.26919919e-01
-4.79374766e-01 -8.16016793e-01 6.76259816e-01 6.69486821e-01
7.12746322e-01 -1.58606458e+00 -5.30008256e-01 4.27775651e-01
-3.83598000e-01 -1.35819876e+00 -6.53425097e-01 4.49718922e-01
-7.00922489e-01 -1.04573429e+00 -3.45939785e-01 -6.18869901e-01
8.70065451e-01 -1.40126333e-01 1.15091288e+00 -1.73083633e-01
-5.28022312e-02 2.53916740e-01 -4.61247504e-01 -2.97156960e-01
-4.08900201e-01 4.08165365e-01 2.07760558e-01 -1.83476463e-01
7.39234447e-01 -3.70776951e-01 -4.25389916e-01 3.16756397e-01
-7.64600992e-01 1.52770609e-01 7.76078641e-01 1.14711487e+00
4.50169384e-01 -1.35759234e-01 9.68328238e-01 -1.19541693e+00
8.30798388e-01 -6.31897509e-01 -3.52552772e-01 5.35542786e-01
-9.04662132e-01 3.46271694e-01 8.51272285e-01 -7.79284596e-01
-1.35840178e+00 9.85685885e-02 2.91122198e-01 -4.71468180e-01
-1.53744712e-01 8.35787237e-01 -2.57120162e-01 4.71817255e-01
8.83334994e-01 6.82416484e-02 -3.02577317e-01 -2.49265358e-01
4.53386277e-01 8.30580294e-01 3.63563240e-01 -8.71161520e-01
5.10366261e-01 8.58253986e-02 -6.27286732e-01 -2.83477455e-01
-1.50046408e+00 -2.58147240e-01 -5.81227541e-01 9.96385049e-03
7.83521056e-01 -9.14282084e-01 -6.73214197e-01 1.01632953e-01
-1.03628469e+00 -6.08927190e-01 -4.07542408e-01 6.86196744e-01
-3.59925151e-01 1.15645230e-01 -7.44597256e-01 -6.80219710e-01
-5.76634049e-01 -8.75218630e-01 9.00966585e-01 5.09275973e-01
-4.61295456e-01 -7.97865748e-01 -2.03515328e-02 2.09170252e-01
1.02425434e-01 -8.22875500e-02 7.01144755e-01 -1.09015524e+00
-3.62341762e-01 -1.82176054e-01 -2.31777266e-01 1.91429451e-01
2.38290519e-01 -2.99364120e-01 -7.19820440e-01 -2.71628108e-02
-8.38371459e-03 -9.79131281e-01 4.60629225e-01 -1.71433717e-01
9.70485330e-01 -5.31831205e-01 -3.12736452e-01 4.33132648e-01
9.12015021e-01 2.94573512e-02 8.95897597e-02 4.10971642e-01
7.68770933e-01 7.31040597e-01 6.70662522e-01 4.11126286e-01
6.58990681e-01 3.31093907e-01 -2.02416897e-01 9.50463787e-02
1.18730433e-01 -7.09601760e-01 3.40154737e-01 7.89923966e-01
1.69766843e-01 -1.23175487e-01 -9.55057561e-01 5.06487191e-01
-1.87206066e+00 -8.02405775e-01 4.50622559e-01 2.01120281e+00
1.56005025e+00 2.87528664e-01 -1.40563801e-01 -2.86199123e-01
6.61511779e-01 -1.36469901e-01 -1.07473814e+00 6.64044842e-02
4.74112667e-02 -1.38906002e-01 4.06125128e-01 4.03433200e-03
-9.68494296e-01 1.14113545e+00 5.87377310e+00 6.10243797e-01
-1.02783680e+00 8.30725208e-02 7.47298837e-01 6.86710561e-03
-3.63138139e-01 3.56601761e-03 -1.01934445e+00 4.79999781e-01
1.11453617e+00 -1.32818595e-01 5.70395947e-01 8.58957708e-01
1.39384851e-01 2.13501416e-03 -1.33715975e+00 7.86377847e-01
-2.95857757e-01 -1.15857244e+00 -3.55789125e-01 -2.36376166e-01
6.92796767e-01 2.55472064e-01 -4.11086649e-01 8.19665253e-01
9.06743407e-01 -7.44392514e-01 4.41597760e-01 3.84397388e-01
8.47758055e-01 -3.32505494e-01 7.16735780e-01 7.28395045e-01
-9.82755840e-01 -1.56778425e-01 -1.14150062e-01 2.52358675e-01
2.48566493e-01 4.87879694e-01 -1.09819126e+00 4.98552531e-01
5.43338597e-01 7.43141651e-01 -3.86371851e-01 3.75855207e-01
-7.70083368e-01 8.61778140e-01 -3.67045581e-01 -2.33482748e-01
8.14084858e-02 -1.44118935e-01 3.41775239e-01 1.13003910e+00
-7.90397078e-02 2.14252815e-01 4.08360094e-01 1.01898658e+00
-6.81061685e-01 1.77498564e-01 -4.90861624e-01 -6.36623204e-01
9.21571732e-01 1.43224251e+00 -6.14142120e-01 -5.10474205e-01
-4.55598563e-01 7.82082260e-01 8.67914140e-01 5.71535945e-01
-5.89554608e-01 -4.46244538e-01 9.42950696e-02 -2.84751326e-01
1.62322387e-01 -6.53389916e-02 -5.37079990e-01 -1.51557589e+00
-4.29945774e-02 -1.01272070e+00 4.08966452e-01 -7.66002893e-01
-1.49368322e+00 5.71950674e-01 -1.87823251e-02 -1.10480976e+00
-5.81286252e-01 -2.07814127e-01 -4.74579573e-01 6.08562589e-01
-1.34722829e+00 -8.08239996e-01 -1.82688698e-01 4.26882744e-01
5.67206681e-01 -1.09577157e-01 6.93025708e-01 1.80591822e-01
-1.11171150e+00 7.42740810e-01 -3.75301957e-01 5.81252396e-01
9.14844394e-01 -1.16526008e+00 2.25443050e-01 7.60968626e-01
1.40965670e-01 8.90147150e-01 3.86007130e-01 -9.00802970e-01
-1.34330249e+00 -1.02576268e+00 9.21882749e-01 -4.54013199e-01
8.07792068e-01 -2.91457415e-01 -9.97642875e-01 8.12666714e-01
-1.51879176e-01 1.74009398e-01 7.88446188e-01 3.24977964e-01
-4.40743595e-01 1.87789779e-02 -1.06881583e+00 7.30575204e-01
1.19308794e+00 -6.82401419e-01 -8.02140176e-01 2.92350292e-01
8.91508341e-01 -3.80826771e-01 -9.77687955e-01 4.82044280e-01
1.11190505e-01 -2.08126336e-01 6.77366793e-01 -9.72226083e-01
5.04852831e-01 -1.48433805e-01 3.43296677e-01 -1.32375634e+00
1.00185700e-01 -6.63410187e-01 -3.39617223e-01 1.49765348e+00
8.99799407e-01 -3.97982121e-01 7.56033778e-01 1.27364469e+00
4.24262136e-02 -7.52612650e-01 -5.20464063e-01 -6.58040047e-01
-3.22752982e-01 -3.69835019e-01 5.52611470e-01 1.24412262e+00
4.02634025e-01 7.99835920e-01 -3.75790417e-01 3.21969427e-02
4.44421887e-01 6.07253090e-02 6.97952807e-01 -9.26283658e-01
-1.88535780e-01 -1.84381884e-02 4.29171890e-01 -9.53191459e-01
5.24404407e-01 -9.45537329e-01 3.39310855e-01 -1.36122465e+00
1.17100030e-01 -7.23902225e-01 -4.03531432e-01 1.00997031e+00
-6.89317942e-01 -2.25689292e-01 4.11132984e-02 5.54699421e-01
-9.89233851e-01 6.70527279e-01 1.18751621e+00 -3.23083624e-02
-6.30031705e-01 -8.20942223e-02 -7.49924064e-01 7.07949936e-01
7.35347688e-01 -7.08985686e-01 -3.51074308e-01 -3.13035548e-01
4.45311666e-01 2.22160384e-01 2.24421117e-02 -4.66299415e-01
4.34240371e-01 -2.15621606e-01 3.84643853e-01 -1.12628646e-01
-1.67762533e-01 -5.53004801e-01 -3.23133022e-01 9.13980976e-02
-6.39936626e-01 -1.46332800e-01 -1.53406054e-01 5.79612374e-01
-2.73819506e-01 -5.64854562e-01 3.05118263e-01 -2.37883940e-01
-7.55689561e-01 3.26094568e-01 -2.49844655e-01 6.24616683e-01
7.85612464e-01 7.99577683e-02 -1.74905777e-01 -3.11311930e-01
-7.62454569e-01 6.13100588e-01 2.29090437e-01 3.84254128e-01
4.32752639e-01 -1.01947176e+00 -6.30861819e-01 8.41351412e-03
1.82332709e-01 2.60312587e-01 -4.44663577e-02 8.70399594e-01
7.34261349e-02 1.12829678e-01 1.23914741e-01 -2.71413893e-01
-7.24554718e-01 6.62444472e-01 7.91112706e-02 -7.45842397e-01
-6.08265281e-01 6.94076419e-01 -1.07919909e-01 -7.14882076e-01
2.47107297e-01 -6.32040650e-02 -3.01178396e-01 1.48944438e-01
3.95904392e-01 3.10232788e-02 -9.06434730e-02 7.84945041e-02
-3.05926710e-01 -4.11807150e-02 -5.01921535e-01 -1.45432234e-01
1.31445146e+00 -1.50254220e-01 2.74880767e-01 3.67145747e-01
8.21364224e-01 2.56865062e-02 -1.50451899e+00 -5.43391347e-01
5.88063955e-01 4.61342596e-02 -7.69760758e-02 -9.58850503e-01
-6.97259426e-01 5.50575376e-01 -4.31690477e-02 -2.06803307e-02
9.44378793e-01 1.24478461e-02 8.33301187e-01 8.21348250e-01
4.43206072e-01 -1.46326876e+00 2.92587310e-01 5.17416656e-01
5.71325958e-01 -1.56140757e+00 -9.25774723e-02 -2.39690930e-01
-1.21227574e+00 8.34245145e-01 9.58530068e-01 1.07350387e-01
4.57866430e-01 3.76340985e-01 -3.73779386e-02 -9.75732058e-02
-8.25869977e-01 -1.85525611e-01 1.28402635e-01 3.91835570e-01
4.86302108e-01 -2.24173918e-01 -3.21706980e-01 9.59196627e-01
5.45407422e-02 3.48244399e-01 2.35146344e-01 9.45123971e-01
-9.15010273e-02 -1.07791162e+00 5.80235608e-02 5.91626585e-01
-5.44687033e-01 -2.13187501e-01 -2.66514301e-01 5.57895899e-01
-1.22352235e-01 9.64018047e-01 -2.47215882e-01 -2.65129715e-01
2.99041271e-01 2.27888733e-01 9.62265655e-02 -8.14467013e-01
-6.94900870e-01 -1.59159929e-01 4.17586029e-01 -4.22748953e-01
-4.69553500e-01 -3.59946966e-01 -1.62386119e+00 5.31767346e-02
-6.61799490e-01 4.36171770e-01 5.52797854e-01 1.36059833e+00
5.81228495e-01 5.18848002e-01 7.50517666e-01 -5.50291717e-01
-7.58704007e-01 -1.09468043e+00 -1.76570475e-01 7.31923044e-01
-9.69031826e-02 -5.67995369e-01 -3.82450104e-01 2.55754683e-02] | [10.021318435668945, 8.562751770019531] |
ead9e257-c783-4d70-a72e-77339748ab18 | unsupervised-part-of-speech-tagging-with | null | null | https://aclanthology.org/Q16-1018 | https://aclanthology.org/Q16-1018.pdf | Unsupervised Part-Of-Speech Tagging with Anchor Hidden Markov Models | We tackle unsupervised part-of-speech (POS) tagging by learning hidden Markov models (HMMs) that are particularly well-suited for the problem. These HMMs, which we call anchor HMMs, assume that each tag is associated with at least one word that can have no other tag, which is a relatively benign condition for POS tagging (e.g., {``}the{''} is a word that appears only under the determiner tag). We exploit this assumption and extend the non-negative matrix factorization framework of Arora et al. (2013) to design a consistent estimator for anchor HMMs. In experiments, our algorithm is competitive with strong baselines such as the clustering method of Brown et al. (1992) and the log-linear model of Berg-Kirkpatrick et al. (2010). Furthermore, it produces an interpretable model in which hidden states are automatically lexicalized by words. | ['Daniel Hsu', 'Michael Collins', 'Karl Stratos'] | 2016-01-01 | null | null | null | tacl-2016-1 | ['unsupervised-part-of-speech-tagging'] | ['natural-language-processing'] | [ 6.58853212e-03 4.86094594e-01 -3.98466825e-01 -3.10056299e-01
-7.89466918e-01 -9.23362195e-01 5.09095013e-01 1.42002106e-01
-4.16118860e-01 6.12924159e-01 2.95846373e-01 -7.04778790e-01
3.01753879e-01 -4.95568573e-01 -6.79172456e-01 -8.21283162e-01
-1.86895430e-01 6.29592717e-01 3.61736745e-01 6.56492859e-02
-2.09649414e-01 3.11961502e-01 -1.24051797e+00 3.27813089e-01
4.00382847e-01 4.87126231e-01 6.29818082e-01 5.97131371e-01
-2.37071857e-01 7.50300705e-01 -3.52858722e-01 -7.14197218e-01
-9.67187285e-02 -2.50311971e-01 -9.36411202e-01 -1.38660213e-02
1.53738055e-02 1.81527615e-01 -1.79942638e-01 1.09800780e+00
4.65526171e-02 2.51171887e-01 7.97613144e-01 -1.01983285e+00
-3.82530361e-01 9.92574215e-01 -4.88074005e-01 1.86870135e-02
9.45252553e-02 -3.27091008e-01 1.46873212e+00 -8.93436730e-01
5.11435688e-01 1.23882699e+00 6.52531147e-01 5.94387829e-01
-1.21917903e+00 -4.17744786e-01 3.91147584e-01 1.28009692e-01
-1.41996324e+00 -3.61496985e-01 2.25045562e-01 -4.71853405e-01
1.11072457e+00 4.11582053e-01 4.79840845e-01 1.17174292e+00
5.54354303e-02 1.09223902e+00 9.31284964e-01 -9.70613003e-01
2.84663767e-01 -1.80991814e-01 3.28425705e-01 4.68512595e-01
1.22678444e-01 -8.98348242e-02 -4.57330018e-01 -2.64565349e-01
5.63563228e-01 -3.64388861e-02 3.49583000e-01 -4.16358709e-01
-1.10747147e+00 8.84959280e-01 -2.19536141e-01 5.20123065e-01
-5.41649103e-01 6.83351383e-02 2.17113182e-01 -1.99652717e-01
2.61508256e-01 -3.39813977e-02 -8.02690148e-01 -4.50631022e-01
-8.51722300e-01 -4.88110669e-02 9.11238849e-01 1.15174508e+00
9.11326826e-01 -3.05835724e-01 7.20988438e-02 9.36781287e-01
6.97299421e-01 7.18177259e-01 4.70395625e-01 -1.01249707e+00
2.02519059e-01 -1.06655926e-01 3.16366881e-01 -3.50979745e-01
-4.47431892e-01 -2.61484474e-01 -5.62998176e-01 -4.12195683e-01
3.91468138e-01 -1.62193567e-01 -9.76062179e-01 1.86336279e+00
3.38556230e-01 4.24431562e-01 2.48122349e-01 4.33663756e-01
4.01759744e-01 7.89839923e-01 5.22684932e-01 -6.21737540e-01
1.78436875e+00 -9.41432059e-01 -9.29488182e-01 -5.78783274e-01
8.42432022e-01 -9.04697299e-01 7.59709597e-01 3.05422276e-01
-7.63757467e-01 -4.37089115e-01 -6.30718291e-01 2.48499185e-01
-1.65073708e-01 -2.55389437e-02 9.21075225e-01 7.53133953e-01
-9.95083809e-01 2.27393910e-01 -1.44320083e+00 -5.69253922e-01
-3.07328343e-01 2.11732417e-01 -3.78833115e-01 1.96925178e-01
-1.37848246e+00 9.06032920e-01 6.03875935e-01 6.79634660e-02
-6.88599169e-01 5.18018790e-02 -8.94288778e-01 -8.85516852e-02
4.92376924e-01 -2.88288981e-01 1.71230888e+00 -1.01604569e+00
-1.41813648e+00 8.74575257e-01 -6.03482842e-01 -5.91013432e-01
-1.30040735e-01 -2.67466038e-01 -5.95868647e-01 8.69325846e-02
2.07044795e-01 6.55781984e-01 5.72919488e-01 -1.24920511e+00
-8.04826260e-01 -1.31251544e-01 -6.78985417e-02 -2.20973580e-03
-2.08426684e-01 6.07369900e-01 -5.79819620e-01 -5.77485025e-01
2.77045190e-01 -1.28161383e+00 -4.45053309e-01 -8.03456247e-01
-5.23007751e-01 -3.03786784e-01 2.41190493e-01 -8.65938187e-01
1.54088163e+00 -2.20276499e+00 -4.83881459e-02 2.62567580e-01
-7.66688287e-02 3.77131432e-01 -2.60222447e-03 8.42901468e-01
-2.42661983e-01 1.36934906e-01 -1.75979078e-01 -4.42672074e-01
2.54076064e-01 9.39632058e-01 -2.92216182e-01 4.64211345e-01
1.30010964e-02 7.78592825e-01 -1.01833451e+00 -6.86054587e-01
3.06496620e-02 3.60557169e-01 -2.63550460e-01 2.34134480e-01
-2.26366594e-01 1.28001079e-01 -1.53280273e-01 3.99154991e-01
4.57377523e-01 -1.53738543e-01 9.43870008e-01 1.53619960e-01
-5.31251490e-01 7.81331003e-01 -1.34672415e+00 1.30922103e+00
-4.94220763e-01 2.88063049e-01 1.24686837e-01 -8.61941755e-01
5.81363261e-01 6.52661920e-01 3.07883739e-01 -3.89987417e-02
-1.69629067e-01 6.75360709e-02 1.37924939e-01 -2.72314310e-01
4.58648354e-01 -3.27556491e-01 -3.03669065e-01 4.27716881e-01
1.42744169e-01 4.94598299e-01 4.15388703e-01 2.60855615e-01
1.20962811e+00 2.31251672e-01 7.02560663e-01 -2.54043043e-01
1.28025621e-01 -1.99562699e-01 1.16164720e+00 8.73414516e-01
-1.46454915e-01 3.23353201e-01 5.14563918e-01 1.91716000e-03
-9.01077211e-01 -1.15813518e+00 -2.77550101e-01 1.45678568e+00
-3.85193646e-01 -7.16288626e-01 -8.40394795e-01 -7.46021569e-01
-3.46007645e-01 7.76982725e-01 -5.31160355e-01 1.91607267e-01
-5.61376691e-01 -3.86990428e-01 5.69516778e-01 6.74410641e-01
-2.13034406e-01 -1.11378694e+00 -2.66379356e-01 4.21913058e-01
-4.20098066e-01 -1.16070867e+00 -4.02686685e-01 9.20543253e-01
-5.79269350e-01 -7.17731595e-01 -4.77934062e-01 -1.04298580e+00
5.51416874e-01 1.17625833e-01 1.13655615e+00 -1.97888285e-01
5.00335574e-01 2.20261171e-01 -9.29099619e-01 -3.48505676e-01
-7.50493467e-01 3.05101305e-01 2.18665957e-01 5.25533780e-02
6.90132320e-01 -4.81738687e-01 -4.73285168e-02 3.45375776e-01
-9.74967897e-01 7.47841671e-02 9.35891747e-01 7.48026669e-01
7.47635663e-01 7.76564777e-02 1.52882844e-01 -1.12524581e+00
-2.05108017e-01 -5.40479898e-01 -4.70180750e-01 2.86815882e-01
-3.79719287e-01 1.01152755e-01 3.90173227e-01 -4.43795741e-01
-9.64160264e-01 5.96263707e-01 -5.47264636e-01 -1.48814321e-01
-6.68390512e-01 6.55942619e-01 -6.59126818e-01 4.30546790e-01
1.56289190e-01 1.96231127e-01 -4.07314211e-01 -7.96999216e-01
5.24418473e-01 9.61378753e-01 4.57677931e-01 -5.13965309e-01
8.81327391e-01 2.68888623e-01 -1.66667461e-01 -7.90028811e-01
-1.10793388e+00 -9.66971636e-01 -1.06810749e+00 3.24164927e-02
1.07272148e+00 -1.14694548e+00 -4.48362499e-01 2.37565935e-01
-1.25298548e+00 -4.71874475e-01 4.51108366e-02 6.67285979e-01
-5.16763806e-01 7.45053172e-01 -9.66451287e-01 -1.23796272e+00
2.30314314e-01 -7.65468180e-01 1.14310122e+00 -9.61091071e-02
-5.66221654e-01 -1.12558329e+00 3.27150941e-01 1.42337978e-01
-2.41811335e-01 -1.16371363e-01 9.04389679e-01 -9.77422953e-01
9.70485136e-02 -1.02330856e-02 3.25244427e-01 4.26134259e-01
1.42105408e-02 1.73609659e-01 -8.97563815e-01 -1.53880477e-01
-3.24782014e-01 8.56228992e-02 7.81887591e-01 2.74909019e-01
5.85822821e-01 -5.68278611e-01 -4.59526688e-01 -2.78711710e-02
9.58486438e-01 4.62078601e-01 6.51663899e-01 3.17583412e-01
7.46612966e-01 5.93312919e-01 7.65744865e-01 2.75147259e-01
6.11426532e-01 8.40915561e-01 1.63122952e-01 2.62697656e-02
2.97453076e-01 -5.07729948e-01 8.83160591e-01 1.34375656e+00
1.34759113e-01 -3.08518410e-01 -1.27922308e+00 7.25807607e-01
-2.13614655e+00 -1.01011968e+00 -5.26421905e-01 2.14855003e+00
9.87793744e-01 1.12607956e-01 3.45453918e-01 6.48076087e-02
1.01305473e+00 5.22316471e-02 3.13124359e-01 -5.95564604e-01
-1.19212084e-01 3.51342678e-01 6.55381203e-01 6.65122449e-01
-1.48247409e+00 1.36405349e+00 6.06986189e+00 7.67106831e-01
-4.53272343e-01 2.46247619e-01 6.55228198e-02 4.88189936e-01
-2.99498886e-01 5.21900594e-01 -1.20444703e+00 6.02989912e-01
1.56867623e+00 4.42966431e-01 1.33857191e-01 8.52413058e-01
9.51127931e-02 -2.28485182e-01 -8.59006584e-01 5.54901004e-01
-9.36290324e-02 -6.70450926e-01 -1.87982574e-01 3.47101808e-01
6.15631878e-01 7.98899680e-02 -8.92892033e-02 3.83275598e-01
9.70250726e-01 -6.68257952e-01 9.62895811e-01 2.39585206e-01
4.37563509e-01 -6.87464535e-01 8.17771673e-01 5.77631652e-01
-1.15240562e+00 1.79075345e-01 -4.12423551e-01 -1.17923796e-01
6.01529777e-01 8.34351778e-01 -7.63835967e-01 4.92874801e-01
5.05800843e-01 4.19442326e-01 -2.59703070e-01 7.27371037e-01
-7.66722679e-01 1.36747301e+00 -3.40333223e-01 3.75853330e-02
4.25157756e-01 -1.71379149e-01 3.45035583e-01 1.54801977e+00
1.96810454e-01 5.47059700e-02 4.32009429e-01 1.35053350e-02
2.56116122e-01 2.42133677e-01 -3.42760175e-01 -3.70190740e-01
6.47183657e-01 1.27444422e+00 -1.06989014e+00 -3.77219826e-01
-7.44681299e-01 8.88360560e-01 2.80685872e-01 2.45322391e-01
-6.70273006e-01 -1.05128065e-01 6.79877043e-01 -4.67506377e-03
9.07547235e-01 -4.90830421e-01 9.55968425e-02 -1.14078259e+00
-2.72527754e-01 -7.25554764e-01 5.31406701e-01 -6.44162774e-01
-1.24761140e+00 2.88626790e-01 6.04669228e-02 -1.03811514e+00
-6.88203454e-01 -6.09881461e-01 -1.71694651e-01 7.53080726e-01
-1.35002959e+00 -1.31714940e+00 6.44165039e-01 2.99678534e-01
8.54874477e-02 3.21086675e-01 1.21507525e+00 1.78398371e-01
-4.59002554e-01 2.76727825e-01 2.32279941e-01 6.43433630e-01
5.78694344e-01 -1.44839382e+00 6.64364278e-01 1.11993563e+00
8.79928112e-01 9.39951479e-01 8.37469339e-01 -7.90651917e-01
-1.19461513e+00 -1.01297605e+00 1.67196083e+00 -6.84874058e-01
9.25525606e-01 -7.38835931e-01 -8.57028127e-01 1.29442120e+00
1.78926587e-01 -1.69545531e-01 1.25931704e+00 6.25745893e-01
-4.76390094e-01 3.12746167e-01 -5.02641916e-01 4.44151968e-01
9.40687835e-01 -5.86321890e-01 -7.60790110e-01 3.79519403e-01
6.66276276e-01 -2.63394237e-01 -7.59588242e-01 1.26424283e-01
4.68684942e-01 -8.32587123e-01 4.89862531e-01 -7.54758060e-01
-1.19082205e-01 -3.71448010e-01 -3.73951167e-01 -1.22549486e+00
-7.30141997e-01 -7.49761105e-01 -2.31907129e-01 1.63200927e+00
6.74217939e-01 -2.72799134e-01 5.49252510e-01 3.23292315e-01
-4.37759250e-01 -2.75004029e-01 -1.09004378e+00 -1.18180215e+00
7.67327473e-02 -7.55242348e-01 3.85547519e-01 9.70118165e-01
1.67915970e-01 4.96488482e-01 -4.39809889e-01 4.89219427e-01
3.83566827e-01 -1.33233622e-01 5.11083841e-01 -1.40334976e+00
-7.04948664e-01 -6.36747330e-02 -4.27320778e-01 -1.20257950e+00
6.55421436e-01 -8.61518145e-01 5.74184954e-01 -1.45091796e+00
3.21064115e-01 -4.35135603e-01 -5.48821151e-01 9.46749687e-01
-1.87116086e-01 1.36821061e-01 1.64318070e-01 1.88519567e-01
-9.82796431e-01 6.01304621e-02 4.92869437e-01 3.50480229e-01
6.04584105e-02 7.23337680e-02 -5.12654901e-01 7.82101631e-01
6.71286047e-01 -8.92751634e-01 9.46775600e-02 -2.62777358e-01
5.94269872e-01 1.02066897e-01 -2.92080045e-02 -4.50451016e-01
8.42067376e-02 -2.82634646e-01 -3.68631780e-02 -3.33701462e-01
1.41063884e-01 -8.42897177e-01 3.34758133e-01 2.52788723e-01
-1.60796732e-01 2.15929359e-01 -2.63516214e-02 5.45843124e-01
-2.42004663e-01 -4.88916099e-01 5.84090889e-01 -8.86803344e-02
-8.09851289e-01 2.71915155e-03 -9.17221606e-01 7.24558458e-02
8.61476600e-01 8.56910497e-02 2.10296467e-01 -5.01508236e-01
-9.41578805e-01 -4.85284552e-02 4.19516921e-01 1.75299972e-01
1.78960916e-02 -1.31620300e+00 -2.99813062e-01 -8.99289250e-02
1.84728205e-01 -2.70229250e-01 -6.02853112e-02 9.57423747e-01
-6.42900988e-02 5.71655095e-01 3.16193581e-01 -3.87819439e-01
-1.22732782e+00 6.23019397e-01 -2.41116315e-01 -5.20192444e-01
-3.39928180e-01 7.29460657e-01 1.72399446e-01 -3.42034012e-01
6.85273409e-02 -1.74148872e-01 -1.87485233e-01 1.85237557e-01
4.78516638e-01 6.00289851e-02 6.46946505e-02 -1.24962044e+00
-6.64307714e-01 1.07867546e-01 -9.88564044e-02 -5.29106975e-01
1.22899961e+00 -2.87082464e-01 -2.73434192e-01 1.00996697e+00
9.47827160e-01 3.79788518e-01 -1.05066407e+00 -2.70615250e-01
5.92689812e-01 2.75243726e-02 -1.31564662e-01 -6.33473754e-01
-5.21146417e-01 7.37131000e-01 1.72820717e-01 4.01035666e-01
5.06691456e-01 3.64902526e-01 8.93247545e-01 3.52778733e-01
5.41254580e-01 -1.21318758e+00 -5.31019688e-01 9.39994156e-01
-2.02004276e-02 -8.46001327e-01 -4.60518360e-01 -5.12751281e-01
-7.94502497e-01 7.58889556e-01 3.97793762e-02 3.34441811e-01
6.49157941e-01 5.18250823e-01 3.60206395e-01 1.24138288e-01
-8.47233415e-01 -7.91013956e-01 4.75020707e-02 7.39300728e-01
5.22787929e-01 5.09773314e-01 -4.18249160e-01 8.46548498e-01
-2.42006525e-01 -4.92001563e-01 3.41110826e-01 1.12307060e+00
-5.87916553e-01 -1.56757367e+00 -3.55506897e-01 1.19170189e-01
-7.34663367e-01 -4.83462393e-01 -3.57781231e-01 6.30536854e-01
1.21895052e-01 1.26539791e+00 1.56008855e-01 -3.11076522e-01
1.14492550e-01 6.79369807e-01 2.09613249e-01 -1.23271167e+00
-3.83440197e-01 6.08508408e-01 2.57646203e-01 -1.85836360e-01
-5.75785577e-01 -1.24275935e+00 -1.41766107e+00 7.86426663e-02
-4.20655876e-01 7.27699816e-01 4.52693492e-01 1.19993412e+00
3.93775627e-02 9.72437710e-02 4.91876215e-01 -2.74119556e-01
-3.58097106e-01 -1.09034669e+00 -9.99951065e-01 1.24959201e-01
-1.46862045e-01 -7.18761981e-01 -3.45733166e-01 4.13695455e-01] | [10.337764739990234, 9.731013298034668] |
4cc6b953-0e5a-41e5-b0db-65ca50f7b10b | deepco3-deep-instance-co-segmentation-by-co | null | null | http://openaccess.thecvf.com/content_CVPR_2019/html/Hsu_DeepCO3_Deep_Instance_Co-Segmentation_by_Co-Peak_Search_and_Co-Saliency_Detection_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Hsu_DeepCO3_Deep_Instance_Co-Segmentation_by_Co-Peak_Search_and_Co-Saliency_Detection_CVPR_2019_paper.pdf | DeepCO3: Deep Instance Co-Segmentation by Co-Peak Search and Co-Saliency Detection | In this paper, we address a new task called instance co-segmentation. Given a set of images jointly covering object instances of a specific category, instance co-segmentation aims to identify all of these instances and segment each of them, i.e. generating one mask for each instance. This task is important since instance-level segmentation is preferable for humans and many vision applications. It is also challenging because no pixel-wise annotated training data are available and the number of instances in each image is unknown. We solve this task by dividing it into two sub-tasks, co-peak search and instance mask segmentation. In the former sub-task, we develop a CNN-based network to detect the co-peaks as well as co-saliency maps for a pair of images. A co-peak has two endpoints, one in each image, that are local maxima in the response maps and similar to each other. Thereby, the two endpoints are potentially covered by a pair of instances of the same category. In the latter subtask, we design a ranking function that takes the detected co-peaks and co-saliency maps as inputs and can select the object proposals to produce the final results. Our method for instance co-segmentation and its variant for object colocalization are evaluated on four datasets, and achieve favorable performance against the state-of-the-art methods. The source codes and the collected datasets are available at https://github.com/KuangJuiHsu/DeepCO3/
| [' Yung-Yu Chuang', ' Yen-Yu Lin', 'Kuang-Jui Hsu'] | 2019-06-01 | null | null | null | cvpr-2019-6 | ['co-saliency-detection'] | ['computer-vision'] | [ 5.28810501e-01 1.51715234e-01 -2.73093998e-01 -2.61199355e-01
-8.81602466e-01 -3.72292608e-01 3.88855100e-01 3.72959942e-01
-3.42352152e-01 4.57337528e-01 -3.43765765e-01 2.06085160e-01
1.84028428e-02 -6.07399583e-01 -8.69034410e-01 -7.42685854e-01
-2.46862005e-02 4.79981989e-01 8.28731179e-01 2.24765748e-01
3.62530112e-01 2.96721041e-01 -1.68016350e+00 5.99592686e-01
1.09778631e+00 1.16659808e+00 7.87749231e-01 1.67461321e-01
-2.14339107e-01 9.64385569e-02 -5.39046168e-01 2.80151330e-02
2.31753901e-01 -5.40274560e-01 -6.82514548e-01 4.22002077e-01
3.83637577e-01 2.08139673e-01 1.95727274e-01 1.37323153e+00
2.72047251e-01 1.21732734e-01 5.92815578e-01 -1.25985134e+00
-4.19162601e-01 4.56318557e-01 -1.11451221e+00 4.67135489e-01
-2.07167700e-01 1.91281512e-01 1.04771948e+00 -9.94668663e-01
3.99574459e-01 9.89185154e-01 1.62478253e-01 4.23694521e-01
-1.26235604e+00 -7.10366786e-01 3.36010098e-01 1.11091964e-01
-1.37652707e+00 -2.14929387e-01 8.27388704e-01 -3.36288363e-01
4.16099697e-01 2.94321299e-01 7.07669556e-01 4.07442003e-01
-2.14964435e-01 1.16009331e+00 1.19503224e+00 -1.86842039e-01
3.10586482e-01 1.90101117e-01 1.02332897e-01 4.20275927e-01
1.48941293e-01 -3.25369090e-01 -2.73852974e-01 1.48410499e-01
8.05286825e-01 3.07918906e-01 -3.18569094e-01 -2.92191148e-01
-1.37177730e+00 4.68834668e-01 9.69248593e-01 6.04209602e-01
-6.11309826e-01 6.06856830e-02 2.20700856e-02 -2.92729497e-01
3.75817865e-01 4.90467280e-01 -4.35913175e-01 5.86593151e-01
-1.04735088e+00 3.42573971e-01 1.33092836e-01 7.01876402e-01
1.07534575e+00 -7.06770599e-01 -3.73417765e-01 9.55310404e-01
2.22257107e-01 1.07887395e-01 4.82141495e-01 -7.06276119e-01
6.34986937e-01 1.03457415e+00 1.31017238e-01 -9.28562760e-01
-3.27235430e-01 -7.75866449e-01 -5.65097570e-01 9.79626626e-02
5.76167941e-01 7.52976313e-02 -1.10840774e+00 1.58666837e+00
6.04248226e-01 6.41259253e-01 -2.26482585e-01 1.23554325e+00
1.04030526e+00 7.34531701e-01 1.52470738e-01 -2.57312387e-01
1.65563059e+00 -1.20113766e+00 -4.23933238e-01 -6.05726659e-01
1.13274239e-01 -7.40633190e-01 1.04250503e+00 3.90160419e-02
-1.09911585e+00 -5.76664805e-01 -9.33425903e-01 1.88820660e-01
-3.17508757e-01 4.73416120e-01 4.76250708e-01 -1.77691549e-01
-6.85276508e-01 4.01181191e-01 -6.16251588e-01 -1.90514609e-01
6.65177882e-01 3.78745675e-01 -2.31751218e-01 1.98701378e-02
-9.91407275e-01 4.43435371e-01 5.80420554e-01 2.43286833e-01
-8.08680296e-01 -5.41433930e-01 -5.72251976e-01 1.39082313e-01
5.74393809e-01 -2.63922751e-01 1.06540322e+00 -1.35592735e+00
-7.15763807e-01 1.22912717e+00 -1.98663503e-01 -2.76618242e-01
3.15305173e-01 3.94686162e-02 -2.03529745e-01 8.92001018e-02
5.60215294e-01 1.01684344e+00 8.37349057e-01 -1.55060768e+00
-1.18171000e+00 -5.39939165e-01 -9.41337794e-02 5.02742350e-01
-9.93660763e-02 1.48025915e-01 -1.04062331e+00 -3.84613127e-01
4.64363158e-01 -6.56825721e-01 -2.95898348e-01 -1.96130425e-01
-7.54463196e-01 -4.11880761e-01 8.80133390e-01 -4.38738108e-01
1.06391644e+00 -2.09694266e+00 -4.26122062e-02 6.62285686e-02
2.98678786e-01 4.10948157e-01 -1.95571020e-01 -3.05452123e-02
-1.80559337e-01 1.37654215e-01 -5.40583849e-01 -2.45066419e-01
-4.62080300e-01 -1.36092484e-01 -1.60159230e-01 3.33689779e-01
5.06078243e-01 9.31178749e-01 -1.04788911e+00 -5.69799900e-01
1.91615939e-01 1.66719243e-01 -5.32650016e-02 2.32446223e-01
-4.31564540e-01 7.39229023e-01 -5.95161557e-01 5.76508999e-01
7.94278145e-01 -3.82235497e-01 -1.52305281e-03 -4.52837557e-01
-3.04257959e-01 5.00531979e-02 -1.34023428e+00 1.37270331e+00
-6.75619915e-02 3.45850050e-01 -7.14165196e-02 -1.20940197e+00
8.34998071e-01 3.16334404e-02 5.11030495e-01 -7.45674014e-01
7.93506578e-02 6.15969479e-01 8.87749419e-02 -5.24798870e-01
1.87661380e-01 1.39692888e-01 8.90032724e-02 2.24739745e-01
4.95567583e-02 2.10528374e-02 3.78470510e-01 -8.17601532e-02
6.20658934e-01 -1.46363437e-01 2.31345132e-01 -1.90917134e-01
5.65536320e-01 -1.22425370e-02 8.12209785e-01 6.18424118e-01
-2.17761546e-01 9.52697575e-01 5.07661760e-01 -2.01485008e-01
-8.17772388e-01 -9.00422573e-01 -2.10317031e-01 7.53047168e-01
8.58968258e-01 8.15771595e-02 -8.75267029e-01 -7.06582010e-01
-9.80612915e-03 3.30464572e-01 -7.60272801e-01 1.24674708e-01
-5.09052694e-01 -7.46546268e-01 -3.91823463e-02 3.80594015e-01
7.21827269e-01 -1.34217048e+00 -9.07852292e-01 1.51922986e-01
-3.32727045e-01 -9.53445017e-01 -6.32127464e-01 2.86725640e-01
-8.02738786e-01 -1.21161962e+00 -1.02061045e+00 -1.21149266e+00
9.53227222e-01 5.53673267e-01 9.84818518e-01 2.21266091e-01
-3.20826530e-01 -4.35219109e-01 -1.68490842e-01 -4.28020239e-01
2.25471050e-01 -7.98298270e-02 -4.57669318e-01 4.74595010e-01
2.90086508e-01 -3.33185613e-01 -9.59313452e-01 7.03145862e-01
-9.49886143e-01 2.92878091e-01 7.95410395e-01 7.63222933e-01
1.30536139e+00 1.96266800e-01 6.39380753e-01 -8.86399448e-01
3.59070629e-01 -5.02044797e-01 -6.50327265e-01 2.60468394e-01
-2.07137391e-01 -3.37666810e-01 2.71988839e-01 -4.26019967e-01
-6.85617805e-01 4.18831110e-01 3.58253300e-01 -5.70406973e-01
-3.42607439e-01 5.42582512e-01 -3.29301983e-01 2.51971692e-01
4.60185647e-01 2.16394126e-01 -1.40942216e-01 -3.27900946e-01
1.94262102e-01 6.02418423e-01 8.03915799e-01 -1.82506889e-01
4.67786402e-01 4.78286594e-01 -3.27166170e-01 -5.47172666e-01
-1.10381961e+00 -9.29046631e-01 -5.76687694e-01 -3.56031924e-01
1.03648460e+00 -7.15187550e-01 -2.77192414e-01 4.79755551e-01
-1.11164379e+00 -2.74889767e-01 -2.24315643e-01 3.09961259e-01
-3.81909192e-01 3.62068303e-02 4.76779975e-03 -7.38368094e-01
-2.39784822e-01 -1.35509825e+00 1.18730009e+00 9.02076006e-01
2.31177628e-01 -5.10032177e-01 -1.05444230e-01 3.83632511e-01
-5.31909354e-02 2.67446518e-01 6.20108843e-01 -9.50829327e-01
-7.42147624e-01 -2.85554111e-01 -4.76038069e-01 9.65368971e-02
1.81961372e-01 -6.65934980e-02 -1.02132416e+00 -7.11022392e-02
-2.68479288e-01 -1.66286185e-01 1.02372241e+00 7.32672453e-01
1.36573267e+00 -1.93390563e-01 -7.15963840e-01 3.80941629e-01
1.31432772e+00 3.40950578e-01 5.27627289e-01 1.76513046e-01
5.28054476e-01 8.64508569e-01 9.46527362e-01 3.29326955e-03
9.73419696e-02 9.98116851e-01 6.88182354e-01 -5.89877903e-01
-1.28631875e-01 -9.89664346e-02 -1.84776306e-01 9.81970727e-02
3.92916612e-02 -1.86766073e-01 -9.10351574e-01 9.97845709e-01
-2.12380672e+00 -8.48869324e-01 -4.10406858e-01 2.36749625e+00
7.02562511e-01 8.22132602e-02 3.94058347e-01 -1.03539579e-01
1.38245046e+00 5.74692674e-02 -8.73848438e-01 3.36368710e-01
-6.78576082e-02 1.21546470e-01 2.32732266e-01 1.72938570e-01
-1.44442570e+00 9.93518054e-01 4.78589773e+00 1.01230788e+00
-1.08101809e+00 2.36856312e-01 1.21069872e+00 1.40006859e-02
-5.67027628e-02 5.29979169e-02 -7.09385931e-01 7.32830465e-01
1.84104323e-01 1.38520971e-01 1.54989839e-01 6.61538780e-01
2.51325548e-01 -5.24021626e-01 -9.30995822e-01 8.01503956e-01
-1.82922080e-01 -1.23637927e+00 -2.35813260e-01 -1.44725800e-01
9.16270196e-01 7.43406415e-02 1.91635594e-01 -5.10124974e-02
-1.17631450e-01 -1.05250812e+00 8.07668209e-01 1.97594091e-01
6.01751924e-01 -6.54488504e-01 7.23510146e-01 4.52648968e-01
-1.37055814e+00 -8.47103819e-02 -4.21299636e-01 3.96459043e-01
-6.21380433e-02 9.16958332e-01 -5.27116001e-01 3.74868780e-01
7.99380600e-01 6.73706055e-01 -5.19064724e-01 1.65031374e+00
-3.95100802e-01 4.71130431e-01 -3.27463955e-01 7.40128830e-02
3.15300345e-01 -1.87179759e-01 5.14149725e-01 1.10767937e+00
1.75019428e-01 -6.25672340e-02 4.23428893e-01 1.45728672e+00
-3.73308621e-02 1.54048845e-01 -1.29721075e-01 1.87062219e-01
5.68473577e-01 1.74202764e+00 -1.32988060e+00 -3.90847623e-01
-1.72526002e-01 9.81750369e-01 3.64026338e-01 4.20843154e-01
-8.83282840e-01 -5.65649271e-01 3.83318990e-01 1.99969798e-01
3.56644541e-01 3.03211570e-01 -5.33197880e-01 -9.04898107e-01
2.21851349e-01 -5.23392618e-01 4.33678716e-01 -7.71729410e-01
-1.12514186e+00 5.50652087e-01 -3.80508266e-02 -1.33883321e+00
1.65029809e-01 -2.66389847e-01 -1.08245170e+00 1.12117445e+00
-1.51868618e+00 -1.13552773e+00 -6.02455378e-01 3.97363275e-01
6.14945650e-01 2.62761652e-01 1.83451176e-01 2.80000240e-01
-7.17872143e-01 3.87518972e-01 -9.90673527e-02 8.77139792e-02
3.40455264e-01 -1.15083778e+00 3.07607621e-01 9.55695629e-01
2.90514916e-01 4.62420523e-01 5.17132938e-01 -7.69491553e-01
-6.07963264e-01 -1.26456654e+00 6.84695244e-01 -5.42736687e-02
3.73137236e-01 -2.84480065e-01 -1.10336447e+00 2.27318317e-01
-4.96332757e-02 4.47514772e-01 1.19432047e-01 -3.26018214e-01
5.79400435e-02 2.33733343e-04 -1.02692211e+00 4.52873886e-01
8.85687351e-01 -1.73379377e-01 -3.17960739e-01 5.11620700e-01
5.78986347e-01 -4.34517652e-01 -3.30433279e-01 4.70324963e-01
1.68392822e-01 -1.04683459e+00 9.27610695e-01 -2.56148249e-01
5.99481702e-01 -9.10107374e-01 2.15802893e-01 -1.27202082e+00
-2.88617790e-01 -1.68840483e-01 1.66553378e-01 1.29028606e+00
6.31368995e-01 -4.48420972e-01 6.81332588e-01 2.43604764e-01
-3.00544351e-01 -1.24259448e+00 -7.24959314e-01 -6.50656879e-01
-2.87694573e-01 -1.11126125e-01 6.06635332e-01 7.59651780e-01
-2.33696237e-01 2.32146218e-01 1.03664607e-01 3.53340864e-01
6.57324553e-01 5.67553937e-01 5.08559108e-01 -1.10128152e+00
-2.37557560e-01 -5.95299244e-01 -3.71255815e-01 -1.00768435e+00
-1.37625664e-01 -1.02296484e+00 4.43878174e-01 -1.80967236e+00
5.65348685e-01 -7.18884110e-01 -5.70593834e-01 7.20739543e-01
-6.04345858e-01 5.75761378e-01 1.48667380e-01 5.89648426e-01
-6.66159272e-01 2.54458666e-01 1.39481258e+00 -2.95292288e-01
-5.13955235e-01 2.32214823e-01 -7.58745611e-01 6.35083973e-01
7.47683227e-01 -4.06606883e-01 -2.93356180e-01 -1.74573898e-01
-3.14023763e-01 -8.36995617e-02 6.23603523e-01 -1.13109756e+00
1.87601194e-01 -2.56049812e-01 4.82966483e-01 -7.51475990e-01
3.56932469e-02 -5.93500316e-01 1.27413750e-01 3.73105913e-01
-5.46582699e-01 -3.86783987e-01 6.41957298e-02 5.06020606e-01
-3.37858170e-01 -3.21492910e-01 1.11963809e+00 -2.76679337e-01
-9.39924538e-01 4.08107162e-01 4.42599952e-02 1.18136257e-01
1.37512064e+00 -3.14991981e-01 -3.55679423e-01 1.01003041e-02
-6.49980724e-01 5.19609988e-01 3.52298975e-01 3.08966964e-01
6.49442375e-01 -1.16026974e+00 -5.64967036e-01 1.58157349e-01
3.82550567e-01 5.38780510e-01 3.37990791e-01 1.13639927e+00
-6.54788390e-02 1.94657281e-01 -2.05856040e-01 -8.68495345e-01
-1.23505902e+00 4.49300379e-01 5.15389621e-01 -1.61458001e-01
-3.96817535e-01 1.08926654e+00 9.63154614e-01 -2.74505615e-01
8.75085518e-02 -2.90451616e-01 -6.28342986e-01 -6.48848712e-02
5.39727688e-01 -1.06046528e-01 -4.65119816e-03 -7.55379438e-01
-4.41614956e-01 5.43449998e-01 -1.63578466e-01 4.57102619e-02
1.16484034e+00 1.13369815e-01 -3.03402305e-01 3.48144293e-01
1.14763498e+00 -2.85496056e-01 -1.43791521e+00 -3.19484949e-01
1.00539431e-01 -5.00916541e-01 -5.12210093e-02 -8.99075329e-01
-1.44050753e+00 7.75405169e-01 7.44600058e-01 1.14410177e-01
1.17086148e+00 5.21372020e-01 6.47611499e-01 -2.13882521e-01
5.91077060e-02 -9.48423564e-01 3.07370961e-01 2.11552367e-01
7.47171938e-01 -1.18773150e+00 -2.18001768e-01 -6.86222076e-01
-7.80569673e-01 6.91304266e-01 1.02315354e+00 -2.17434883e-01
3.19954664e-01 -2.11523607e-01 1.04968706e-02 -4.62506652e-01
-3.87540728e-01 -5.62649548e-01 6.92854345e-01 4.61410463e-01
2.76699156e-01 2.16353536e-01 -4.69701558e-01 6.56910837e-01
3.24745774e-01 -1.45707548e-01 8.72678310e-02 6.53410256e-01
-8.45965922e-01 -7.76436865e-01 -4.18754637e-01 7.76668727e-01
-3.84736210e-01 9.13217291e-02 -4.39743429e-01 4.37690347e-01
4.81102705e-01 8.74624729e-01 3.72544825e-01 -4.51075613e-01
2.33440951e-01 -2.38240972e-01 1.24297716e-01 -8.23573351e-01
-5.53907573e-01 4.90236998e-01 -2.60218024e-01 -4.60419595e-01
-4.95716482e-01 -6.44011796e-01 -1.71277881e+00 4.87015665e-01
-5.61389029e-01 1.85318425e-01 5.04333317e-01 8.16706955e-01
3.35608602e-01 6.13342941e-01 6.79100931e-01 -1.04296780e+00
1.06573524e-02 -7.40337610e-01 -6.54196143e-01 6.18317485e-01
8.45012218e-02 -8.43797505e-01 -1.81011081e-01 4.98959199e-02] | [9.809576034545898, -0.07218162715435028] |
3d0675d3-ff2d-4ea2-8f32-b27351225cdf | cost-effective-training-in-low-resource-1 | 2201.05700 | null | https://arxiv.org/abs/2201.05700v1 | https://arxiv.org/pdf/2201.05700v1.pdf | Cost-Effective Training in Low-Resource Neural Machine Translation | While Active Learning (AL) techniques are explored in Neural Machine Translation (NMT), only a few works focus on tackling low annotation budgets where a limited number of sentences can get translated. Such situations are especially challenging and can occur for endangered languages with few human annotators or having cost constraints to label large amounts of data. Although AL is shown to be helpful with large budgets, it is not enough to build high-quality translation systems in these low-resource conditions. In this work, we propose a cost-effective training procedure to increase the performance of NMT models utilizing a small number of annotated sentences and dictionary entries. Our method leverages monolingual data with self-supervised objectives and a small-scale, inexpensive dictionary for additional supervision to initialize the NMT model before applying AL. We show that improving the model using a combination of these knowledge sources is essential to exploit AL strategies and increase gains in low-resource conditions. We also present a novel AL strategy inspired by domain adaptation for NMT and show that it is effective for low budgets. We propose a new hybrid data-driven approach, which samples sentences that are diverse from the labelled data and also most similar to unlabelled data. Finally, we show that initializing the NMT model and further using our AL strategy can achieve gains of up to $13$ BLEU compared to conventional AL methods. | ['Jan Niehues', 'Danni Liu', 'Sai Koneru'] | 2022-01-14 | null | null | null | null | ['low-resource-neural-machine-translation'] | ['natural-language-processing'] | [ 4.02797610e-01 3.38778734e-01 -6.05315030e-01 -4.89531249e-01
-1.36600089e+00 -6.08278811e-01 5.92393935e-01 2.04091221e-01
-8.98276687e-01 1.26802349e+00 -8.16240557e-04 -4.03592885e-01
2.66147852e-01 -4.78037447e-01 -1.05542254e+00 -3.64262819e-01
3.15062374e-01 1.16354990e+00 4.76227626e-02 -5.21116197e-01
3.20814624e-02 2.97260702e-01 -1.05161679e+00 2.30942950e-01
1.32719409e+00 4.69954789e-01 5.44198155e-01 3.78886938e-01
-4.47901458e-01 5.36912024e-01 -6.74006402e-01 -4.93560344e-01
6.08290553e-01 -6.02785289e-01 -8.76377225e-01 7.04535376e-03
4.20906752e-01 -7.44954720e-02 3.61028373e-01 8.19478452e-01
8.73645604e-01 7.02267811e-02 3.26155782e-01 -8.49688411e-01
-4.12078559e-01 9.17476416e-01 -3.63613367e-01 6.07405342e-02
1.68128356e-01 1.34673968e-01 9.02866423e-01 -1.05047858e+00
8.46185148e-01 9.40005064e-01 5.58169901e-01 8.72918069e-01
-1.37702930e+00 -7.77357817e-01 1.45180181e-01 6.20534047e-02
-9.42613244e-01 -1.02412224e+00 6.26736581e-01 -4.27111685e-02
1.28821599e+00 2.02294573e-01 5.19790232e-01 1.22812593e+00
-1.29351959e-01 9.48060393e-01 1.27327204e+00 -1.10219717e+00
2.56358206e-01 5.05110860e-01 -2.05957368e-01 5.23121238e-01
1.13477055e-02 -2.73736030e-01 -7.41129279e-01 -1.03683650e-01
2.96598047e-01 -3.40303272e-01 -1.19707920e-01 -3.98954004e-01
-1.45812595e+00 8.15076649e-01 7.77184069e-02 4.21544343e-01
-4.00303304e-01 -1.97756156e-01 3.57108742e-01 7.26697505e-01
9.85060215e-01 9.49242353e-01 -7.26964712e-01 -2.98117429e-01
-1.22123873e+00 -1.69089124e-01 9.39552128e-01 1.15979731e+00
9.94918287e-01 -7.22407326e-02 -2.93133166e-02 1.17837358e+00
-1.15514118e-02 6.70559585e-01 3.43258411e-01 -6.57708943e-01
1.00696349e+00 6.35913134e-01 1.34383440e-01 -4.18882877e-01
-5.42010926e-02 -4.14409548e-01 -5.34873009e-01 -1.37856171e-01
4.02932256e-01 -2.49469504e-01 -9.08741355e-01 1.84252644e+00
1.62987456e-01 -2.04216093e-01 5.76096252e-02 7.39538074e-01
4.14776504e-01 6.84801280e-01 -8.16186741e-02 -6.68073118e-01
7.56306469e-01 -1.16779292e+00 -7.40720689e-01 -6.80666864e-01
9.95266676e-01 -9.17264402e-01 1.30100286e+00 2.34691948e-01
-1.28821230e+00 -2.42272034e-01 -9.72597659e-01 -7.58244321e-02
-2.96435177e-01 2.22263083e-01 6.05998755e-01 6.66461408e-01
-1.01382315e+00 4.08918738e-01 -8.91604900e-01 -5.92722416e-01
5.51343143e-01 8.28100145e-01 -4.53520030e-01 -2.04234034e-01
-1.23483741e+00 1.28447330e+00 2.76987731e-01 7.86131322e-02
-8.11691105e-01 -3.18820328e-01 -6.64265692e-01 -2.22464964e-01
6.57712758e-01 -4.61881459e-01 1.16247165e+00 -1.46302104e+00
-1.67174268e+00 7.84029424e-01 -1.63010776e-01 -5.89604497e-01
6.38981283e-01 -1.51603609e-01 -1.16830252e-01 -5.43064438e-02
9.75334272e-02 9.09830213e-01 6.27872050e-01 -9.23599482e-01
-3.83068174e-01 -1.21367477e-01 1.72290817e-01 5.39138377e-01
-9.91949260e-01 2.53251344e-01 -4.23294008e-01 -3.16499680e-01
-8.49698707e-02 -1.14978671e+00 -4.33731884e-01 -1.53933689e-01
-1.22069269e-01 5.51049970e-02 4.44197267e-01 -7.38876581e-01
9.29860532e-01 -1.73217559e+00 2.85553604e-01 -4.67179492e-02
-3.04757264e-02 4.85885113e-01 -3.29972744e-01 4.81879354e-01
3.55692625e-01 1.84338480e-01 -4.22828496e-01 -6.15324378e-01
-1.03251725e-01 4.76161599e-01 -7.44739100e-02 1.46542862e-01
5.14524341e-01 1.00118482e+00 -9.57961261e-01 -7.20104575e-01
-1.74556762e-01 8.46154168e-02 -5.11303663e-01 2.50158608e-01
-4.48335439e-01 6.62012577e-01 -1.00119703e-01 5.44673145e-01
2.87885696e-01 -6.47720918e-02 4.48512673e-01 2.71478742e-01
-9.51296836e-02 5.94620883e-01 -7.66652167e-01 2.02179885e+00
-8.50953519e-01 6.13863647e-01 1.93736125e-02 -1.27526093e+00
1.11641943e+00 4.39726830e-01 2.25259498e-01 -8.81973267e-01
6.06558397e-02 7.75152743e-01 1.21583112e-01 -4.23268497e-01
5.21867514e-01 -1.83382452e-01 -8.68637413e-02 6.04795635e-01
3.33772838e-01 -1.48675755e-01 4.66136754e-01 1.32877395e-01
1.14969218e+00 2.36014768e-01 2.94032186e-01 -3.32779795e-01
2.70479977e-01 4.46553200e-01 8.40979099e-01 6.65464580e-01
-6.81138039e-02 3.18189234e-01 1.33916736e-01 -2.01954663e-01
-1.45071626e+00 -4.08927470e-01 1.25172973e-01 1.22037423e+00
-2.51991987e-01 -2.40557089e-01 -8.37604940e-01 -9.67040360e-01
-5.22328854e-01 6.47263050e-01 -2.38680005e-01 -6.74732821e-03
-9.37872350e-01 -1.11579859e+00 5.04672766e-01 2.70555764e-01
5.98534167e-01 -8.70246470e-01 -2.54423231e-01 2.99696892e-01
-4.52373534e-01 -1.02598274e+00 -3.45790565e-01 7.10710347e-01
-9.92120564e-01 -2.92566061e-01 -7.96864033e-01 -9.33852255e-01
8.82240176e-01 -1.02960408e-01 1.28459835e+00 -2.15625599e-01
1.12865940e-01 -1.66710019e-01 -5.78295171e-01 -6.58241868e-01
-8.43178272e-01 7.13313460e-01 2.07933560e-01 -2.16329172e-01
3.98061693e-01 -3.83400679e-01 -1.12883359e-01 2.97363222e-01
-5.90381563e-01 2.87633508e-01 9.81269240e-01 1.07705629e+00
5.40858924e-01 -3.45644563e-01 8.46208632e-01 -1.18895864e+00
5.67209840e-01 -3.46084356e-01 -4.85016406e-01 4.13152725e-01
-8.32961023e-01 9.33550000e-02 8.21164966e-01 -8.22682023e-01
-8.91971409e-01 2.31307730e-01 -1.81754500e-01 -2.10558549e-01
1.69892371e-01 4.59935188e-01 -3.70261967e-01 -1.09581612e-01
1.01014364e+00 8.39111768e-03 6.16779178e-02 -5.18004060e-01
2.05334172e-01 1.01077807e+00 -5.43736182e-02 -6.53631568e-01
7.33706236e-01 1.14837132e-01 -2.65071034e-01 -4.58433181e-01
-8.98148656e-01 -3.27342987e-01 -9.72168326e-01 -3.09344437e-02
2.89265215e-01 -9.29256737e-01 7.28075579e-02 2.68273409e-02
-1.00205314e+00 -6.52548611e-01 -3.79549265e-01 6.76211596e-01
-5.76724827e-01 1.59101203e-01 -4.26440239e-01 -7.77384818e-01
-5.84263444e-01 -1.29695535e+00 8.86265993e-01 -1.22092731e-01
-3.59040946e-01 -1.01546824e+00 1.46303460e-01 7.43099332e-01
6.11340225e-01 -2.37643421e-01 7.76951551e-01 -1.05282450e+00
-5.53879559e-01 -1.34850852e-02 1.06203400e-01 4.92495149e-01
9.90058258e-02 -4.17552769e-01 -8.30602050e-01 -5.37241995e-01
-3.05026816e-03 -7.13616371e-01 4.91076559e-01 -1.02390245e-01
6.26884103e-01 -3.57764542e-01 -1.78589284e-01 2.60953397e-01
1.27738750e+00 2.07604021e-01 4.28067118e-01 5.04598498e-01
5.03756166e-01 7.11373508e-01 8.99138331e-01 -1.95332661e-01
1.25452712e-01 9.21680391e-01 -4.41124029e-02 -4.99897301e-01
-1.01062991e-01 -1.29817612e-02 6.02459669e-01 1.53866374e+00
-1.03891857e-01 -3.35259527e-01 -1.30323625e+00 7.65556037e-01
-1.87603319e+00 -7.00618029e-01 7.16345087e-02 2.38747096e+00
1.53856301e+00 3.73763829e-01 -2.45035924e-02 5.47944084e-02
6.11450553e-01 -3.74156475e-01 -5.30747533e-01 -5.98456979e-01
-4.28040951e-01 4.60841566e-01 6.85453236e-01 5.14080048e-01
-8.21598589e-01 1.24300325e+00 6.27763081e+00 1.02879012e+00
-1.18730891e+00 6.52770638e-01 5.47130108e-01 -3.35100472e-01
-2.25640163e-01 3.22946548e-01 -9.18475926e-01 4.26202536e-01
1.37346506e+00 -1.76848117e-02 3.44873011e-01 7.64941752e-01
2.78338104e-01 6.02225447e-03 -1.12458980e+00 7.07811892e-01
3.03225964e-01 -1.26933134e+00 -1.09265096e-01 1.59287646e-01
8.34228933e-01 2.79968530e-01 -2.83773184e-01 4.15136278e-01
4.16571289e-01 -7.30219305e-01 6.59181654e-01 2.43114635e-01
9.36283231e-01 -6.36339903e-01 8.62038374e-01 9.00427759e-01
-5.24215341e-01 3.26359435e-03 -5.58535814e-01 -1.34549320e-01
5.96380569e-02 7.15062082e-01 -1.26564789e+00 5.51010370e-01
3.07965755e-01 5.79895377e-01 -5.51352680e-01 7.34513879e-01
-1.98801994e-01 9.62480962e-01 -4.27799404e-01 -2.60018885e-01
2.06370160e-01 -1.16167575e-01 4.49952036e-01 1.19509637e+00
2.73992240e-01 -3.48706096e-01 3.05671901e-01 6.01576209e-01
-3.82996380e-01 6.06391668e-01 -6.67913377e-01 -3.00005853e-01
4.49558765e-01 1.09370267e+00 -6.16461158e-01 -4.76682365e-01
-4.73205835e-01 1.20715094e+00 6.67005241e-01 5.23705184e-02
-6.28622234e-01 -1.59834772e-01 3.88765847e-03 1.76476479e-01
-1.42969862e-01 -2.32798621e-01 -2.57415086e-01 -1.41279602e+00
1.78908229e-01 -1.26618755e+00 -5.59716821e-02 -5.05351305e-01
-1.06485426e+00 7.03555524e-01 -1.21753417e-01 -1.35151684e+00
-4.06928658e-01 -3.21448654e-01 -4.44071665e-02 1.08391845e+00
-1.45872903e+00 -1.28927350e+00 1.61720514e-01 3.69538009e-01
8.00520360e-01 -4.51409638e-01 1.11983573e+00 6.40202701e-01
-6.23775899e-01 7.54298031e-01 3.67616713e-01 9.81921330e-02
1.03688240e+00 -1.15948927e+00 5.52539110e-01 9.85107064e-01
3.87798280e-01 8.96482050e-01 5.52913845e-01 -7.25948453e-01
-1.43884909e+00 -1.00503540e+00 1.42256904e+00 -4.92477745e-01
4.55974907e-01 -9.83911335e-01 -6.43984795e-01 6.63363099e-01
3.02681595e-01 -2.43019849e-01 7.59879708e-01 3.82301331e-01
-3.59918061e-03 -1.71671689e-01 -1.05650580e+00 5.57845950e-01
1.07305837e+00 -4.34610933e-01 -6.48531497e-01 6.11933112e-01
5.85641623e-01 -3.27149600e-01 -6.39102936e-01 4.12087679e-01
2.39522651e-01 -2.60661185e-01 5.80863476e-01 -6.19106114e-01
3.02250594e-01 1.61116440e-02 -1.25012383e-01 -1.48014963e+00
1.17119793e-02 -7.97624946e-01 2.35334828e-01 1.37089729e+00
1.13038707e+00 -5.95406592e-01 9.08455729e-01 4.55688477e-01
-2.41881371e-01 -7.26637125e-01 -9.55693185e-01 -1.03187501e+00
2.44271830e-01 -3.12159985e-01 3.84755939e-01 1.25505650e+00
5.23509160e-02 8.20586205e-01 -6.26759768e-01 -4.08905715e-01
3.61074150e-01 -9.59462672e-02 9.16723907e-01 -1.06907034e+00
-3.44525993e-01 1.21646829e-01 -6.07722588e-02 -9.91658509e-01
2.36125976e-01 -1.10485518e+00 3.24083388e-01 -1.58535278e+00
3.61873865e-01 -9.27443027e-01 -6.19499981e-02 8.51632416e-01
-9.15866718e-02 4.66241300e-01 5.71864694e-02 5.39791405e-01
-5.32926500e-01 2.46549040e-01 1.07582676e+00 -3.00820559e-01
-2.27160379e-01 -2.17549920e-01 -4.77915525e-01 3.78333867e-01
7.90132344e-01 -7.57942855e-01 -5.07350743e-01 -8.20405066e-01
5.40918887e-01 -1.30599529e-01 -2.25087523e-01 -7.41724908e-01
2.65839577e-01 -1.62709296e-01 5.90343960e-02 -2.83344716e-01
3.34509134e-01 -8.46874952e-01 5.27511612e-02 1.59361944e-01
-5.89214087e-01 -5.88504039e-02 9.31381956e-02 3.03216964e-01
-1.93424746e-01 -4.38257992e-01 7.61952341e-01 -4.50905293e-01
-3.83756459e-01 9.89312306e-02 -2.68542826e-01 1.50051296e-01
6.94691896e-01 -1.27653316e-01 -1.40706450e-01 -2.10869864e-01
-5.40525913e-01 1.88537404e-01 6.66531980e-01 2.85620779e-01
4.31064405e-02 -1.15370595e+00 -8.94192994e-01 2.15241853e-02
1.94979325e-01 1.10156134e-01 -2.86639154e-01 1.10666335e+00
-5.33299387e-01 4.29947436e-01 -3.21909994e-01 -5.88180184e-01
-1.27463126e+00 3.99050772e-01 4.94331382e-02 -5.05752861e-01
-3.31959277e-01 8.78253460e-01 -3.85618865e-01 -8.44506025e-01
4.13116701e-02 -7.50578148e-03 -2.80697420e-02 1.97487250e-01
1.48722216e-01 -6.00876696e-02 5.45720160e-01 -4.64193761e-01
-1.99262053e-01 1.32340714e-01 -3.25757205e-01 -4.53349441e-01
1.52116227e+00 -1.84834182e-01 -1.71884567e-01 7.16845572e-01
9.37126338e-01 2.97632784e-01 -8.69506478e-01 -6.03426933e-01
2.34053880e-01 -3.38595718e-01 -3.32071027e-03 -1.01793778e+00
-7.46492028e-01 9.28320348e-01 5.17250359e-01 -2.01982424e-01
1.19231868e+00 -2.27302998e-01 7.83385992e-01 7.17212200e-01
7.72048891e-01 -1.39758193e+00 9.20899957e-03 4.30857062e-01
4.09076869e-01 -1.48861897e+00 -1.18972566e-02 -3.76301289e-01
-5.19449413e-01 7.56769776e-01 3.89609277e-01 3.13507199e-01
-1.58804189e-02 5.25797665e-01 3.19669336e-01 2.50213504e-01
-9.80153680e-01 -2.02172711e-01 3.29742208e-02 3.79070312e-01
5.51731288e-01 -8.84301811e-02 -7.40479529e-01 1.80441514e-01
-1.73503742e-01 -8.90544131e-02 4.68543947e-01 1.06210279e+00
-3.96968335e-01 -1.82588208e+00 -1.78065211e-01 4.34209943e-01
-6.16756499e-01 -4.83686000e-01 -6.89714015e-01 5.67944765e-01
2.24316031e-01 9.37498450e-01 -1.49917871e-01 -2.54813403e-01
1.10812537e-01 4.55602676e-01 4.94520336e-01 -1.13325334e+00
-8.28208506e-01 1.92580953e-01 6.80875897e-01 -1.97247952e-01
-7.13496149e-01 -6.75813615e-01 -7.87737668e-01 -1.02385543e-01
-7.41006136e-01 3.87619942e-01 9.42078829e-01 1.06257868e+00
3.23481113e-01 -2.28028875e-02 6.75461829e-01 -5.00088453e-01
-3.91567051e-01 -1.26780665e+00 -5.99749666e-03 2.19629735e-01
-8.29226077e-02 -4.41160202e-01 -3.13298494e-01 1.65535510e-01] | [11.523652076721191, 10.256119728088379] |
ea287f43-e018-4c26-8aae-88d9f597b0e1 | learning-graph-neural-networks-for-image | 2207.11681 | null | https://arxiv.org/abs/2207.11681v2 | https://arxiv.org/pdf/2207.11681v2.pdf | Learning Graph Neural Networks for Image Style Transfer | State-of-the-art parametric and non-parametric style transfer approaches are prone to either distorted local style patterns due to global statistics alignment, or unpleasing artifacts resulting from patch mismatching. In this paper, we study a novel semi-parametric neural style transfer framework that alleviates the deficiency of both parametric and non-parametric stylization. The core idea of our approach is to establish accurate and fine-grained content-style correspondences using graph neural networks (GNNs). To this end, we develop an elaborated GNN model with content and style local patches as the graph vertices. The style transfer procedure is then modeled as the attention-based heterogeneous message passing between the style and content nodes in a learnable manner, leading to adaptive many-to-one style-content correlations at the local patch level. In addition, an elaborated deformable graph convolutional operation is introduced for cross-scale style-content matching. Experimental results demonstrate that the proposed semi-parametric image stylization approach yields encouraging results on the challenging style patterns, preserving both global appearance and exquisite details. Furthermore, by controlling the number of edges at the inference stage, the proposed method also triggers novel functionalities like diversified patch-based stylization with a single model. | ['DaCheng Tao', 'Xinchao Wang', 'Mingli Song', 'Yibing Zhan', 'Yiding Yang', 'Yining Mao', 'Yongcheng Jing'] | 2022-07-24 | null | null | null | null | ['image-stylization'] | ['computer-vision'] | [ 3.05733293e-01 -1.50879724e-02 4.59735915e-02 -4.79475051e-01
-5.39247394e-01 -5.35235763e-01 7.06793308e-01 -2.60253072e-01
3.28085646e-02 6.40722394e-01 -1.25063062e-02 2.59578675e-01
-8.25594738e-02 -9.81289566e-01 -1.04797137e+00 -7.52852380e-01
5.19486368e-01 6.33627713e-01 1.85771987e-01 -3.88756603e-01
2.11476266e-01 7.06546366e-01 -1.23543787e+00 2.08566889e-01
1.01921678e+00 1.05336642e+00 8.96425769e-02 6.15974545e-01
-4.45889682e-01 4.01754081e-01 -4.73212212e-01 -6.63490832e-01
2.13699907e-01 -6.90804482e-01 -4.46396023e-01 3.19600165e-01
9.13263023e-01 -1.91983372e-01 -3.09423864e-01 1.22153628e+00
5.78496695e-01 2.56465450e-02 7.26953268e-01 -1.21000850e+00
-1.12127495e+00 3.91199887e-01 -8.05420458e-01 -2.66207933e-01
1.41386583e-01 2.27081954e-01 8.88552845e-01 -7.61855185e-01
7.25973248e-01 1.64160967e+00 7.52515078e-01 3.45801145e-01
-1.62832987e+00 -6.74736440e-01 1.04892187e-01 -9.88474931e-04
-1.36166024e+00 -1.31334111e-01 1.53056526e+00 -3.75086069e-01
2.94050127e-01 7.72957653e-02 8.16472054e-01 1.34825873e+00
4.45496202e-01 5.20494819e-01 1.27628803e+00 -2.86984652e-01
4.37171087e-02 1.31259382e-01 -4.61176991e-01 8.22220206e-01
1.12852812e-01 2.02233970e-01 -6.40789807e-01 -4.51207794e-02
1.48820078e+00 -1.17016509e-01 -1.78755641e-01 -7.32352376e-01
-1.00004423e+00 7.67534614e-01 6.96519554e-01 3.35815370e-01
-2.84122825e-01 2.54025608e-01 3.65602106e-01 3.68392825e-01
7.74847209e-01 3.73765558e-01 -1.35256499e-01 2.33637169e-01
-1.08534324e+00 2.83711076e-01 6.89648151e-01 1.17928410e+00
1.18525445e+00 4.14160818e-01 -5.72158217e-01 8.27786505e-01
1.81263894e-01 5.55002451e-01 2.41161540e-01 -7.22083986e-01
6.52924180e-02 5.77285707e-01 -1.10617533e-01 -1.43657839e+00
-1.46979019e-01 -7.58350432e-01 -1.22616053e+00 4.18049514e-01
3.44943315e-01 1.07143775e-01 -7.71538794e-01 1.89841819e+00
3.04770768e-01 2.55545765e-01 -5.56241512e-01 7.95322478e-01
7.67285585e-01 3.75954211e-01 2.69395262e-01 2.91572064e-01
1.27883267e+00 -1.01668608e+00 -7.48809576e-01 3.41078788e-02
1.43253863e-01 -9.21059489e-01 1.28775787e+00 -9.41561162e-02
-1.14695847e+00 -8.15213263e-01 -8.49515557e-01 -3.07964414e-01
-2.35480562e-01 -3.40190344e-02 2.82462001e-01 3.53851348e-01
-1.29375684e+00 6.83640063e-01 -4.04112637e-01 -4.14539844e-01
5.28771043e-01 2.24876180e-01 -1.83579162e-01 5.41842245e-02
-1.00585127e+00 5.95589399e-01 1.82600453e-01 9.97913256e-02
-5.68076551e-01 -1.11975121e+00 -9.19976175e-01 1.69960782e-01
8.60257223e-02 -1.32597423e+00 5.76739669e-01 -1.53375530e+00
-1.98634875e+00 1.13476133e+00 3.18602547e-02 6.74967421e-03
8.04799914e-01 -1.16757102e-01 -2.81310409e-01 -8.04189295e-02
5.33038490e-02 7.35455453e-01 1.37666070e+00 -1.56482899e+00
-1.65909499e-01 -2.95622110e-01 -1.81178316e-01 2.61876881e-01
-1.66360915e-01 -1.64907053e-01 -5.55327177e-01 -1.29239762e+00
-2.04059407e-01 -7.23260224e-01 5.71701527e-02 5.76059937e-01
-4.22261208e-01 1.12140045e-01 9.78298426e-01 -6.07856989e-01
8.91299903e-01 -2.09151030e+00 6.36856019e-01 3.32694471e-01
3.08699608e-01 1.22529820e-01 -4.02896583e-01 2.73154229e-01
-1.98012982e-02 -1.61867350e-01 -4.34824049e-01 -5.43064356e-01
4.53785323e-02 1.51018590e-01 -6.61918297e-02 3.81185591e-01
4.45148557e-01 1.29440069e+00 -8.30051839e-01 -5.45832694e-01
3.82520854e-01 7.02922821e-01 -6.64405406e-01 5.33943295e-01
-3.88800472e-01 8.89079332e-01 -4.93062615e-01 5.56080878e-01
9.06575322e-01 -5.97016178e-02 -1.78819165e-01 -6.14880204e-01
1.57493562e-01 -3.47027153e-01 -1.04492617e+00 2.17263889e+00
-5.79275608e-01 4.47851121e-01 3.74359846e-01 -8.99343789e-01
1.24342310e+00 -1.28486767e-01 2.07087576e-01 -5.90904176e-01
3.03121090e-01 1.94419995e-01 -4.88835365e-01 -8.93028155e-02
4.18079495e-01 -3.74150664e-01 -4.91798222e-02 2.64326245e-01
3.14529806e-01 -4.30427492e-01 -2.74910301e-01 -7.96654634e-03
5.34808040e-01 5.49343705e-01 5.75808212e-02 -5.71234167e-01
7.18296468e-01 -3.93190593e-01 2.83812225e-01 5.15371025e-01
2.22114071e-01 7.94956684e-01 3.51107985e-01 -4.98183548e-01
-1.23258889e+00 -1.12027597e+00 1.35952860e-01 1.07778680e+00
2.61905283e-01 5.00868782e-02 -9.99310911e-01 -5.03118694e-01
1.80220362e-02 4.83491421e-01 -8.25008869e-01 -3.33061069e-01
-6.36862755e-01 -1.73870370e-01 4.66365188e-01 3.18154573e-01
7.90937960e-01 -1.31359506e+00 1.37089148e-01 2.81560898e-01
1.10065594e-01 -1.04766417e+00 -1.07683682e+00 -3.44788790e-01
-7.99755156e-01 -7.73983359e-01 -1.07709754e+00 -8.72031689e-01
7.17794180e-01 3.09271049e-02 1.30997348e+00 1.89717009e-03
-1.58591956e-01 3.26275796e-01 -7.59313926e-02 -3.75768244e-02
-6.33909941e-01 1.22993834e-01 -2.88839579e-01 6.32918656e-01
-1.30642429e-01 -1.12466931e+00 -6.60939693e-01 2.65143931e-01
-1.02130222e+00 1.11711979e-01 7.26598203e-01 1.02380013e+00
8.02387714e-01 -4.14725780e-01 5.08011937e-01 -1.07875335e+00
6.23438299e-01 -1.15650177e-01 -7.00097561e-01 3.65640640e-01
-5.09335339e-01 3.06004226e-01 7.94474423e-01 -5.69520056e-01
-1.32639158e+00 -9.27520394e-02 2.54773331e-04 -8.42049539e-01
-2.77673036e-01 -6.53653359e-03 -5.16438723e-01 -6.08725965e-01
6.07521653e-01 3.41823220e-01 1.74333319e-01 -5.47034502e-01
8.54372144e-01 1.18298836e-01 6.01504803e-01 -9.33769643e-01
1.29847419e+00 5.22389412e-01 2.80563444e-01 -7.31513441e-01
-6.42344713e-01 4.16720398e-02 -8.65522921e-01 -2.44314820e-01
8.86493444e-01 -8.02362263e-01 -4.11742389e-01 8.86156917e-01
-1.21418750e+00 -5.93329787e-01 -5.92619717e-01 -1.04461946e-01
-7.93327987e-01 5.65904200e-01 -6.56265676e-01 -1.82315648e-01
-5.42702377e-01 -9.96548414e-01 1.37391543e+00 1.52187869e-01
-2.09427640e-01 -1.31126714e+00 8.56441408e-02 1.43340789e-02
7.76196361e-01 5.65092921e-01 1.07135129e+00 -1.09752186e-01
-6.49427712e-01 9.36167687e-02 -5.84472418e-01 2.62920201e-01
3.30521882e-01 -7.59073496e-02 -8.23220611e-01 -3.89734149e-01
-1.70557067e-01 -5.04343100e-02 6.04028165e-01 4.07591134e-01
1.24487114e+00 -2.26945624e-01 -3.07661723e-02 1.24011600e+00
1.49733710e+00 -2.26257175e-01 5.17675877e-01 1.25038803e-01
1.34534943e+00 6.77497745e-01 1.02691539e-01 2.85187125e-01
1.64749488e-01 8.49580884e-01 2.69935399e-01 -6.10017776e-01
-6.98491275e-01 -4.68242913e-01 5.96347935e-02 6.01160586e-01
-1.49622941e-02 -1.88418999e-01 -1.27163649e-01 3.66366863e-01
-1.60926473e+00 -7.66150713e-01 -5.22448048e-02 1.93746912e+00
8.50787938e-01 -8.47119093e-02 -1.19538475e-02 -3.95846337e-01
9.86338019e-01 3.11425567e-01 -5.84434807e-01 -5.07863879e-01
-4.73211259e-01 5.32059312e-01 1.98022276e-01 5.48614621e-01
-9.46572661e-01 1.29179335e+00 5.27432871e+00 1.19551802e+00
-1.28896761e+00 6.73911795e-02 6.48598790e-01 1.40083194e-01
-6.72235668e-01 -2.90005088e-01 -3.41900289e-01 4.75862622e-01
2.61922866e-01 -5.57647459e-02 6.97047710e-01 6.83105528e-01
8.33418518e-02 3.73496950e-01 -1.04699290e+00 1.09629583e+00
1.47157416e-01 -1.45452666e+00 6.88355803e-01 -1.77190483e-01
9.16844010e-01 -4.64390755e-01 2.11006433e-01 -4.02237140e-02
1.60153359e-01 -8.93225253e-01 9.88381982e-01 7.47306585e-01
1.29727674e+00 -7.23423541e-01 3.06039065e-01 -1.55754879e-01
-1.28637266e+00 3.57370794e-01 -2.27635399e-01 2.92729378e-01
2.15554401e-01 5.99065125e-01 -5.67813106e-02 7.90783882e-01
5.80897212e-01 8.21541786e-01 -5.99596560e-01 8.00313830e-01
-2.23847687e-01 2.09412858e-01 -9.38029662e-02 1.82214797e-01
2.47268006e-01 -6.36458516e-01 7.83991873e-01 1.28706169e+00
4.01238292e-01 -2.36513615e-01 -1.55728295e-01 1.52306437e+00
-2.10055873e-01 1.24850728e-01 -7.09422827e-01 4.23882753e-01
2.36338302e-01 1.34275877e+00 -5.77160954e-01 -2.67933697e-01
-1.97937638e-01 1.44462299e+00 5.18253565e-01 5.38486898e-01
-6.65520966e-01 -2.52359658e-01 6.50496900e-01 2.07425460e-01
3.21346551e-01 1.50942728e-02 -4.80713546e-01 -1.13056624e+00
-1.00661017e-01 -7.46124029e-01 -5.64390495e-02 -8.42327476e-01
-1.67698336e+00 6.89095378e-01 -1.67917833e-01 -1.03878450e+00
2.23590136e-02 -4.32815313e-01 -9.14883971e-01 1.15103912e+00
-1.51707983e+00 -1.81328452e+00 -6.33146524e-01 8.33336353e-01
3.75820667e-01 -5.90925366e-02 6.85680151e-01 3.73269290e-01
-2.28812948e-01 1.00488913e+00 1.06688216e-02 -3.00950208e-03
1.07837009e+00 -1.21567488e+00 4.15633619e-01 6.90117180e-01
-9.90680233e-02 3.49063128e-01 6.64331138e-01 -6.75574422e-01
-1.25521004e+00 -1.37806022e+00 3.76458317e-01 -2.46947616e-01
5.61289907e-01 -4.89587009e-01 -1.10137141e+00 4.34298903e-01
3.62915844e-01 5.57572879e-02 1.96661532e-01 -1.67375892e-01
-3.97480249e-01 -2.37458557e-01 -1.17328990e+00 7.25341082e-01
1.18307388e+00 -7.08277404e-01 -2.28190377e-01 6.58942163e-02
6.92607760e-01 -3.68521005e-01 -9.26947653e-01 2.27114618e-01
5.35542965e-01 -9.58763301e-01 9.41129923e-01 -3.67485136e-01
4.34608698e-01 -2.22499564e-01 1.18103638e-01 -1.48430407e+00
-7.55174816e-01 -9.01368260e-01 8.92015100e-02 1.58537626e+00
8.73565301e-02 -4.57687080e-01 7.86324561e-01 2.96529889e-01
-2.36404672e-01 -5.72527230e-01 -7.22864091e-01 -5.23184657e-01
1.33389950e-01 1.36189908e-01 7.49512792e-01 1.00152361e+00
-8.47259760e-01 3.43406886e-01 -6.11387849e-01 -1.57834105e-02
9.13846612e-01 2.56700128e-01 9.13058579e-01 -1.12902892e+00
-5.13315856e-01 -7.62535810e-01 -3.25707436e-01 -1.00301135e+00
3.38791490e-01 -8.31825674e-01 -2.01406226e-01 -1.14823627e+00
5.78681417e-02 -3.80743384e-01 -1.05577357e-01 -5.93665317e-02
-2.87687480e-01 4.83947098e-01 9.49402153e-02 -3.05561628e-02
-2.47235328e-01 9.75787163e-01 1.88683116e+00 -1.80808142e-01
-1.01849191e-01 -1.76764041e-01 -5.61243951e-01 5.11078596e-01
6.17019236e-01 -1.85961366e-01 -3.43053222e-01 -5.17967343e-01
-2.71516666e-02 -2.74719261e-02 6.11393392e-01 -7.59867013e-01
7.01686461e-03 8.16078335e-02 4.41820294e-01 -2.42679462e-01
2.91274786e-01 -8.56522381e-01 4.32093799e-01 9.33012813e-02
-3.02230656e-01 8.98689870e-03 2.56371439e-01 8.27658415e-01
-2.20257550e-01 2.72262812e-01 1.22861362e+00 -4.38687094e-02
-4.27016169e-01 6.52665079e-01 6.84463456e-02 9.20275971e-02
7.87348628e-01 -2.90412396e-01 2.52096225e-02 -4.22969073e-01
-6.76687479e-01 -2.00392976e-01 7.60687172e-01 4.68548417e-01
3.57694924e-01 -1.64324033e+00 -6.61730945e-01 4.87724036e-01
1.47687584e-01 1.26859635e-01 5.20152748e-01 5.93992949e-01
-4.98830885e-01 3.36831361e-02 -6.45753264e-01 -5.95734239e-01
-9.92176771e-01 5.76817930e-01 5.16652465e-01 -2.69760132e-01
-5.64234674e-01 9.41626370e-01 7.67404974e-01 -5.79165459e-01
1.65841375e-02 -2.15852544e-01 1.40520334e-01 -1.15581304e-01
4.15726043e-02 2.39172891e-01 -2.19561487e-01 -7.83186972e-01
1.01471029e-01 1.17066133e+00 1.38720781e-01 9.08361748e-02
1.15726125e+00 -2.46678621e-01 -3.92177373e-01 3.11953694e-01
1.32043839e+00 -1.35472879e-01 -1.55578470e+00 -6.21100724e-01
-4.78733957e-01 -4.16584671e-01 -4.44597490e-02 -5.96012652e-01
-1.25787258e+00 8.67652118e-01 6.05052471e-01 -1.10424601e-01
1.12472928e+00 -6.85091764e-02 8.84322405e-01 -2.30044961e-01
2.78893888e-01 -7.39726067e-01 1.99190646e-01 2.30078802e-01
1.22604918e+00 -1.03063178e+00 -3.11541528e-01 -5.39384663e-01
-3.48760873e-01 1.02109694e+00 6.46705031e-01 -5.81184983e-01
6.20876253e-01 8.90449956e-02 -5.71095459e-02 -2.95044810e-01
-1.54295817e-01 2.26876765e-01 6.81663275e-01 6.56023979e-01
2.28319526e-01 1.09564684e-01 -6.03886768e-02 2.73234665e-01
-1.73628286e-01 -7.33885542e-02 -3.90370376e-03 3.43183428e-01
-2.14336812e-02 -1.27348936e+00 -3.40699732e-01 1.35969087e-01
-1.78928614e-01 -2.34164789e-01 -5.13547719e-01 8.27435732e-01
-3.95786986e-02 4.18810606e-01 1.25530675e-01 -2.19330594e-01
4.18741763e-01 -1.16310984e-01 6.51241481e-01 -3.40658695e-01
-6.96192086e-01 2.86297262e-01 -2.56689310e-01 -8.33317220e-01
-3.28413665e-01 -2.77879208e-01 -6.36597335e-01 -4.91186589e-01
4.51448224e-02 -2.56125331e-01 2.71540701e-01 7.66296029e-01
5.81835210e-01 7.44790792e-01 6.10752225e-01 -1.28302562e+00
-3.19775760e-01 -9.67476308e-01 -7.33720899e-01 8.01136732e-01
2.02659279e-01 -7.06690848e-01 -1.87972099e-01 5.95450252e-02] | [11.58558177947998, -0.6166903972625732] |
cfd3aad7-92e3-4ee5-bd53-e46f05f3f615 | adversarial-speaker-disentanglement-using | 2305.09167 | null | https://arxiv.org/abs/2305.09167v1 | https://arxiv.org/pdf/2305.09167v1.pdf | Adversarial Speaker Disentanglement Using Unannotated External Data for Self-supervised Representation Based Voice Conversion | Nowadays, recognition-synthesis-based methods have been quite popular with voice conversion (VC). By introducing linguistics features with good disentangling characters extracted from an automatic speech recognition (ASR) model, the VC performance achieved considerable breakthroughs. Recently, self-supervised learning (SSL) methods trained with a large-scale unannotated speech corpus have been applied to downstream tasks focusing on the content information, which is suitable for VC tasks. However, a huge amount of speaker information in SSL representations degrades timbre similarity and the quality of converted speech significantly. To address this problem, we proposed a high-similarity any-to-one voice conversion method with the input of SSL representations. We incorporated adversarial training mechanisms in the synthesis module using external unannotated corpora. Two auxiliary discriminators were trained to distinguish whether a sequence of mel-spectrograms has been converted by the acoustic model and whether a sequence of content embeddings contains speaker information from external corpora. Experimental results show that our proposed method achieves comparable similarity and higher naturalness than the supervised method, which needs a huge amount of annotated corpora for training and is applicable to improve similarity for VC methods with other SSL representations as input. | ['Helen Meng', 'Zhiyong Wu', 'Yang Chao', 'Shuai Wang', 'Xintao Zhao'] | 2023-05-16 | null | null | null | null | ['voice-conversion', 'voice-conversion'] | ['audio', 'speech'] | [ 3.74375701e-01 1.23796232e-01 -2.81270109e-02 -2.72037297e-01
-8.98121178e-01 -5.83898067e-01 7.32894897e-01 -2.08185673e-01
-4.31685418e-01 6.29079044e-01 5.34935355e-01 -2.95801491e-01
5.48821747e-01 -7.31627941e-01 -4.36992317e-01 -6.99542403e-01
2.24984378e-01 6.14382438e-02 8.50930437e-02 -4.82021242e-01
-7.23176897e-02 4.20609444e-01 -1.59185696e+00 3.17349464e-01
7.57758141e-01 9.07207072e-01 4.19951975e-01 9.31368828e-01
-3.02381605e-01 6.31753623e-01 -9.74967420e-01 -3.82800639e-01
1.86121345e-01 -8.66521180e-01 -6.25368834e-01 -2.65100926e-01
1.05753183e-01 -6.54554367e-03 -6.25602305e-01 1.08926868e+00
9.25628364e-01 3.32929850e-01 7.29845762e-01 -9.54703450e-01
-9.15022790e-01 8.47850442e-01 2.75986344e-01 3.37051660e-01
4.54570532e-01 2.23965034e-01 1.01680481e+00 -9.93116677e-01
3.96261275e-01 1.43543422e+00 3.82533818e-01 9.00211394e-01
-1.13196027e+00 -7.28261769e-01 -2.34589949e-01 3.59397471e-01
-1.15095556e+00 -1.02855003e+00 1.06005609e+00 -1.68036789e-01
1.01116586e+00 6.65024519e-01 3.27378899e-01 1.45721054e+00
-1.94999933e-01 7.26280391e-01 1.00047910e+00 -6.97606802e-01
1.29855365e-01 4.84219849e-01 -2.53123134e-01 2.91203797e-01
-5.00641763e-01 5.55194020e-01 -4.61136550e-01 8.68565664e-02
5.33159792e-01 -4.92764980e-01 -6.09414697e-01 1.76444113e-01
-1.13104391e+00 9.36688662e-01 2.55884290e-01 5.74422419e-01
6.40408993e-02 -2.44107068e-01 7.04983175e-01 6.80431664e-01
2.79215753e-01 6.13973022e-01 -2.53275931e-01 -2.34728545e-01
-7.79411614e-01 -4.17617679e-01 7.58924186e-01 9.37885165e-01
3.43615621e-01 9.24985826e-01 -2.63945937e-01 1.16987348e+00
2.38709468e-02 7.40180016e-01 1.40971661e+00 -4.92833525e-01
6.25570178e-01 1.32242426e-01 -3.52642268e-01 -8.36847603e-01
8.37295130e-02 -4.08103287e-01 -9.55044925e-01 6.43926766e-03
-4.65273671e-02 -1.19568177e-01 -6.44948304e-01 1.74866354e+00
7.57300574e-03 1.93223849e-01 6.24072433e-01 8.20175529e-01
9.90516603e-01 1.28752697e+00 -2.30917573e-01 -4.43903476e-01
9.81296301e-01 -1.28423536e+00 -1.17842448e+00 -1.80853873e-01
3.61548513e-01 -1.17431009e+00 1.43600702e+00 2.16296896e-01
-8.32908452e-01 -1.26890302e+00 -1.22056150e+00 5.95309101e-02
-5.07120609e-01 2.22154155e-01 -1.26705453e-01 9.37310219e-01
-8.13150167e-01 7.71499038e-01 -3.27312440e-01 -7.99321309e-02
3.62332561e-03 1.45623714e-01 -4.44226325e-01 3.27497721e-01
-1.69905329e+00 9.44011271e-01 4.63054866e-01 1.94076635e-02
-1.00674963e+00 -3.24572295e-01 -1.00771034e+00 3.44650187e-02
-2.91114673e-02 8.43795538e-02 1.17313170e+00 -1.18843627e+00
-2.18421602e+00 5.56699157e-01 1.62370250e-01 -4.56384599e-01
4.03103292e-01 -1.30131051e-01 -1.27721667e+00 1.52306512e-01
-2.29140207e-01 2.43517414e-01 1.35222650e+00 -8.68620038e-01
-2.33330235e-01 5.20040728e-02 -6.10272765e-01 1.38539717e-01
-6.64607465e-01 8.96966979e-02 1.06650189e-01 -9.91837025e-01
-1.35190859e-01 -6.52700424e-01 8.92638713e-02 -4.11585718e-01
-4.25106674e-01 -2.13392705e-01 7.06349194e-01 -9.30984080e-01
1.33682919e+00 -2.39704561e+00 1.30772099e-01 -1.32123590e-01
-5.90976775e-01 9.13753390e-01 -4.96682256e-01 6.17064297e-01
-9.40932482e-02 9.64530185e-02 -1.47778317e-01 -3.31456423e-01
-1.73908696e-02 1.05030626e-01 -7.16101587e-01 3.04547340e-01
4.60332543e-01 5.78341842e-01 -8.89964044e-01 -3.72650266e-01
2.98227996e-01 5.37302911e-01 -4.59386438e-01 8.66292536e-01
1.13020159e-01 4.59058464e-01 6.38482347e-02 1.70520052e-01
5.42333305e-01 7.24014044e-01 -1.01855442e-01 -1.82168141e-01
1.02283515e-03 7.48385549e-01 -9.27653968e-01 1.58478844e+00
-8.62394035e-01 7.60763049e-01 -1.31984083e-02 -1.00664377e+00
1.42948377e+00 7.53252149e-01 -2.36516267e-01 -6.15254581e-01
2.26525933e-01 3.94148409e-01 2.26064414e-01 -5.44157743e-01
4.23880160e-01 -5.16667426e-01 -9.57661942e-02 1.11243039e-01
4.82259482e-01 -7.01708436e-01 -3.34191740e-01 -2.81696916e-01
8.10378492e-01 2.03313325e-02 3.02924335e-01 1.23923346e-01
9.60577965e-01 -3.25331271e-01 4.43546534e-01 4.02552962e-01
-2.23999903e-01 8.16167891e-01 -1.48720518e-01 2.06447631e-01
-9.95065093e-01 -1.10076034e+00 -1.22827999e-01 8.82679999e-01
-6.48578554e-02 -2.45945573e-01 -7.89993882e-01 -3.51931900e-01
-3.02268952e-01 1.05361569e+00 -1.91087857e-01 -6.99209213e-01
-6.91983163e-01 -8.03453699e-02 1.04727304e+00 4.79929119e-01
2.01359853e-01 -1.34200549e+00 1.60310298e-01 5.02849758e-01
-1.46914989e-01 -1.24622869e+00 -8.27632427e-01 2.73264498e-01
-6.46713078e-01 -4.26471740e-01 -8.32900345e-01 -1.09304523e+00
2.24240378e-01 2.05325298e-02 6.43594027e-01 -4.21197087e-01
-1.34533271e-01 -3.01524457e-02 -6.54144347e-01 -5.91465905e-02
-1.24903560e+00 -1.93422213e-02 5.67675889e-01 3.17903697e-01
1.92262203e-01 -5.88018894e-01 -1.26459286e-01 3.51278186e-01
-8.03876698e-01 -1.95069000e-01 4.25837457e-01 9.68374431e-01
2.35104546e-01 -1.03847288e-01 1.04304528e+00 -5.42099476e-01
8.42839003e-01 -1.79100379e-01 -2.46179879e-01 6.13023564e-02
-2.02055380e-01 2.31963530e-01 1.43277478e+00 -9.51867342e-01
-1.04583573e+00 -3.09739947e-01 -5.87973118e-01 -6.49451315e-01
-2.09517375e-01 1.64139539e-01 -7.00186968e-01 4.76502120e-01
8.10216904e-01 6.24861777e-01 2.46121526e-01 -6.22643530e-01
6.37405097e-01 1.46519101e+00 8.09594035e-01 -2.93131500e-01
8.88885677e-01 -2.41758078e-01 -6.63335264e-01 -1.04447806e+00
-3.22432101e-01 -3.25058877e-01 -5.71055770e-01 8.98092538e-02
5.82759321e-01 -8.30232680e-01 -3.20577562e-01 3.97336751e-01
-1.39283884e+00 -1.13387823e-01 -6.09238803e-01 9.36048031e-01
-5.25895059e-01 4.33069676e-01 -6.50116801e-01 -8.47684681e-01
-3.49685341e-01 -1.13382077e+00 4.08833444e-01 -6.42113984e-02
-2.07687438e-01 -7.69129217e-01 3.25906336e-01 3.00687551e-01
6.96489930e-01 -2.51376212e-01 8.52731586e-01 -1.14287090e+00
-1.24164633e-02 -2.56644696e-01 4.01325673e-01 1.33776212e+00
5.56555748e-01 -1.78311080e-01 -1.46695662e+00 -2.27031827e-01
3.79058152e-01 -3.11669707e-01 6.09642982e-01 -3.81155401e-01
9.27916229e-01 -6.64440989e-01 1.39457583e-01 4.48414683e-01
1.01743233e+00 4.72847432e-01 6.22527599e-01 -1.81348160e-01
5.85607111e-01 5.79054296e-01 4.45897788e-01 1.81411337e-02
-3.77506316e-01 7.50719965e-01 1.20574594e-01 -5.96539490e-02
-6.82394683e-01 -6.21186852e-01 9.72624898e-01 1.94880497e+00
1.30357713e-01 -2.97861993e-01 -3.40810210e-01 4.61703449e-01
-1.15556264e+00 -1.12021589e+00 2.40842894e-01 2.17699862e+00
1.35134315e+00 2.23643929e-01 -1.16835333e-01 6.06669843e-01
1.08833408e+00 3.67000163e-01 -4.90254611e-01 -8.90902460e-01
-1.22520931e-01 5.02820611e-01 -2.89646275e-02 6.22854590e-01
-8.34703028e-01 1.11549187e+00 5.56339550e+00 1.39927733e+00
-1.31227815e+00 2.47432694e-01 1.44255117e-01 2.12267727e-01
-3.96358758e-01 -3.62244844e-01 -6.43719196e-01 4.77246732e-01
1.31399238e+00 -7.04180896e-02 5.92918575e-01 9.10401225e-01
1.46179646e-01 6.55382395e-01 -1.28206289e+00 1.03618467e+00
3.51798326e-01 -9.47896719e-01 3.50084752e-01 -3.37654620e-01
4.73127902e-01 -2.17723206e-01 1.49722964e-01 5.95104814e-01
-1.03285156e-01 -1.14995682e+00 6.78884745e-01 1.81968644e-01
1.24470174e+00 -7.82213330e-01 6.66853487e-01 3.85100424e-01
-1.15497673e+00 1.30011156e-01 -5.73491573e-01 -9.36625972e-02
3.32951769e-02 2.79482454e-01 -1.18136668e+00 5.00507236e-01
-1.31984763e-02 3.18158001e-01 -2.71373987e-01 6.13545239e-01
-3.96347731e-01 9.25387323e-01 4.00680751e-02 -2.80419916e-01
4.39559110e-02 -1.28029525e-01 6.36421680e-01 1.36349463e+00
3.66713375e-01 -1.61985397e-01 -2.36585066e-01 7.67358541e-01
-1.81072548e-01 4.68527019e-01 -7.21194804e-01 -5.68730652e-01
6.76549494e-01 9.27647054e-01 -8.81817341e-02 -2.17934713e-01
-2.48562470e-01 1.17477715e+00 1.32195756e-01 2.08444834e-01
-6.37964666e-01 -8.73756707e-01 6.18265390e-01 -7.62747303e-02
2.38056898e-01 -1.70572996e-01 2.07694575e-01 -1.13148260e+00
-2.90878087e-01 -1.01847732e+00 -1.92699105e-01 -7.63069570e-01
-1.52597785e+00 1.38966084e+00 -6.07245743e-01 -1.61841726e+00
-5.18345356e-01 -5.43739438e-01 -7.26873279e-01 1.01189411e+00
-1.55952787e+00 -9.14423525e-01 1.95806071e-01 6.76497936e-01
9.43221331e-01 -9.57658827e-01 1.28009224e+00 8.48419592e-02
-4.36360866e-01 8.91919553e-01 2.85368502e-01 3.82331163e-01
8.37974608e-01 -9.17405784e-01 5.21503687e-01 9.60634768e-01
3.67063850e-01 3.52628320e-01 4.74883318e-01 -2.91089952e-01
-1.15047634e+00 -1.12312305e+00 8.94105017e-01 -8.02899674e-02
6.31282389e-01 -5.13899386e-01 -9.25585091e-01 2.60963291e-01
4.73576128e-01 1.45831808e-01 9.58550453e-01 -3.66438091e-01
-3.76494437e-01 -2.81925291e-01 -1.01208413e+00 5.71432173e-01
8.66316497e-01 -1.12971580e+00 -1.28324854e+00 -1.01067156e-01
1.30761015e+00 -2.57572904e-02 -6.20699525e-01 2.40054429e-01
1.47467196e-01 -6.23201311e-01 9.07191157e-01 -6.73182309e-01
3.86217713e-01 -2.16313809e-01 -5.56166589e-01 -1.73204553e+00
-5.99078424e-02 -8.07617724e-01 1.59899890e-01 1.72762644e+00
3.99295479e-01 -5.74237347e-01 -7.50877410e-02 -3.96643765e-02
-4.44448978e-01 -2.13372976e-01 -9.77233887e-01 -1.21018398e+00
2.32354432e-01 -4.02754039e-01 7.39356816e-01 8.81574333e-01
2.22330362e-01 6.55634820e-01 -4.03345615e-01 1.07641451e-01
8.70034620e-02 -7.08363727e-02 6.55424774e-01 -9.21823263e-01
-3.30903649e-01 -2.49655560e-01 -4.38071281e-01 -9.46926594e-01
4.96423244e-01 -1.27861285e+00 2.37821937e-01 -9.86867726e-01
-5.76558948e-01 -2.60022759e-01 -3.93117726e-01 6.48555234e-02
-1.00878321e-01 8.46189633e-02 4.03840661e-01 1.06746659e-01
1.74584404e-01 1.04993165e+00 1.30965972e+00 -2.99316943e-01
-3.05622101e-01 3.32350999e-01 -2.22457394e-01 4.27864641e-01
8.92020106e-01 -5.41337550e-01 -4.67378646e-01 -7.09355772e-02
-5.48633456e-01 3.33721191e-01 -1.58879645e-02 -1.19987094e+00
1.90747589e-01 1.18227929e-01 8.06573853e-02 -3.47087145e-01
6.73618972e-01 -7.03864992e-01 -6.86684474e-02 4.68766123e-01
-7.26943612e-01 -4.44701582e-01 1.42020389e-01 4.03548896e-01
-6.22927785e-01 -6.04182959e-01 9.23468828e-01 -5.29763326e-02
-4.53809232e-01 -8.24772418e-02 -5.33991098e-01 2.29775622e-01
6.21107280e-01 -3.49920504e-02 1.80201344e-02 -3.54836076e-01
-7.85891891e-01 -5.41422188e-01 4.69260029e-02 7.14178801e-01
7.29912102e-01 -1.70073533e+00 -8.20243776e-01 7.53653944e-01
7.46613592e-02 -3.92696977e-01 -3.41045037e-02 1.56158507e-01
-7.59621263e-02 3.88777643e-01 -2.62523472e-01 -2.75909722e-01
-1.08892810e+00 7.32786238e-01 1.70828179e-01 2.49317467e-01
-4.09475535e-01 7.93785155e-01 -7.68082440e-02 -8.41476023e-01
4.26152796e-01 -3.17869097e-01 -2.88448304e-01 2.39480123e-01
5.19630134e-01 4.00060564e-01 1.94826394e-01 -9.33735907e-01
-2.50081450e-01 3.22544992e-01 1.85683265e-01 -4.44046259e-01
1.17303693e+00 -5.39702140e-02 2.98176885e-01 6.65624380e-01
1.77873254e+00 4.89713460e-01 -8.61943543e-01 -3.75448406e-01
-3.38130414e-01 -3.68109286e-01 6.90360814e-02 -5.42529106e-01
-9.74394202e-01 1.38940656e+00 6.14114702e-01 3.59764993e-01
8.35872769e-01 -2.49563813e-01 1.01028931e+00 2.03649461e-01
2.39559680e-01 -1.27482879e+00 3.06800604e-01 4.70747143e-01
1.25545442e+00 -1.12714541e+00 -6.32063448e-01 -2.07122892e-01
-9.07902181e-01 1.31866038e+00 4.35637802e-01 -1.78272977e-01
6.83123708e-01 3.39368194e-01 2.45678410e-01 6.69749260e-01
-4.66224194e-01 -2.43588209e-01 2.28244469e-01 8.13999832e-01
2.83814073e-01 1.25937730e-01 -7.41632804e-02 9.87689853e-01
-7.34355271e-01 -4.32732821e-01 5.61068058e-01 3.18203330e-01
-4.52052802e-01 -1.27091157e+00 -5.57565033e-01 -4.22955900e-02
-1.87519610e-01 -3.42572033e-01 -2.74438798e-01 3.10464621e-01
-8.73318315e-02 1.27555418e+00 9.09407735e-02 -7.30274081e-01
4.16422278e-01 3.08670074e-01 1.83275208e-01 -7.90832341e-01
-6.27033472e-01 1.29562765e-01 1.04410611e-01 -1.67471796e-01
-2.15228036e-01 -3.28735322e-01 -1.24511015e+00 1.40759349e-01
-6.76159024e-01 4.48806942e-01 6.51829302e-01 7.62514353e-01
8.24007392e-02 7.07069576e-01 1.42843902e+00 -7.96149611e-01
-1.04215741e+00 -1.27501512e+00 -5.95597565e-01 4.26170766e-01
5.63188374e-01 -2.25316525e-01 -8.21408451e-01 1.81310341e-01] | [14.821115493774414, 6.583869457244873] |
7c36d378-cdc0-42b5-81e3-6ed2095251c1 | layer-wise-cross-view-decoding-for-sequence | 2005.08081 | null | https://arxiv.org/abs/2005.08081v7 | https://arxiv.org/pdf/2005.08081v7.pdf | Rethinking and Improving Natural Language Generation with Layer-Wise Multi-View Decoding | In sequence-to-sequence learning, e.g., natural language generation, the decoder relies on the attention mechanism to efficiently extract information from the encoder. While it is common practice to draw information from only the last encoder layer, recent work has proposed to use representations from different encoder layers for diversified levels of information. Nonetheless, the decoder still obtains only a single view of the source sequences, which might lead to insufficient training of the encoder layer stack due to the hierarchy bypassing problem. In this work, we propose layer-wise multi-view decoding, where for each decoder layer, together with the representations from the last encoder layer, which serve as a global view, those from other encoder layers are supplemented for a stereoscopic view of the source sequences. Systematic experiments and analyses show that we successfully address the hierarchy bypassing problem, require almost negligible parameter increase, and substantially improve the performance of sequence-to-sequence learning with deep representations on five diverse tasks, i.e., machine translation, abstractive summarization, image captioning, video captioning, medical report generation, and paraphrase generation. In particular, our approach achieves new state-of-the-art results on ten benchmark datasets, including a low-resource machine translation dataset and two low-resource medical report generation datasets. | ['Xuewei Ma', 'Xu sun', 'Xian Wu', 'Chenyu You', 'Guangxiang Zhao', 'Xuancheng Ren', 'Fenglin Liu'] | 2020-05-16 | null | null | null | null | ['medical-report-generation'] | ['medical'] | [ 8.48657250e-01 4.20960188e-01 -1.47864580e-01 -1.88368604e-01
-1.31427562e+00 -4.98927623e-01 5.31310380e-01 -1.59406699e-02
7.86539074e-03 1.07137430e+00 7.96466410e-01 -2.81418353e-01
5.88611841e-01 -6.38326406e-01 -9.37602937e-01 -5.67777812e-01
5.19696534e-01 3.39796185e-01 -1.02789126e-01 -2.35643610e-01
1.14989810e-01 -1.13042496e-01 -1.05682826e+00 8.88032854e-01
7.39756823e-01 7.70471454e-01 4.39628273e-01 8.94368708e-01
-1.10188536e-01 1.15823257e+00 -6.81661367e-01 -7.09247231e-01
-1.80731013e-01 -1.00890005e+00 -9.21892226e-01 2.25874096e-01
4.00251925e-01 -8.20482492e-01 -6.50371075e-01 9.10786390e-01
7.68229783e-01 -4.48079467e-01 6.17716193e-01 -9.27917838e-01
-1.01672900e+00 5.71912050e-01 -6.27186596e-01 1.85808927e-01
6.83259070e-01 3.24107975e-01 9.81194139e-01 -8.31850886e-01
9.56913531e-01 1.09685695e+00 1.47633821e-01 8.96569848e-01
-1.08354688e+00 -5.06849468e-01 1.39173016e-01 1.17658414e-01
-1.05056131e+00 -6.46690607e-01 5.48323989e-01 -3.89287621e-01
9.37383711e-01 1.72113180e-01 4.36903000e-01 1.43745565e+00
3.51225674e-01 9.99959111e-01 7.02966750e-01 -5.49971946e-02
-9.19094384e-02 -1.50446445e-01 -3.16209316e-01 7.45076001e-01
2.32057080e-01 -7.05869943e-02 -6.56635463e-01 -1.36185335e-02
7.61902750e-01 -1.11610182e-01 -5.96362948e-01 -8.32693800e-02
-1.59094107e+00 8.29867005e-01 4.00810272e-01 1.82808992e-02
-5.59077561e-01 7.59819224e-02 6.34549260e-01 3.88721794e-01
4.98301595e-01 3.68873835e-01 -2.29135409e-01 -1.32062748e-01
-9.02528584e-01 2.05295712e-01 7.05948412e-01 1.27089715e+00
4.29048955e-01 4.67249230e-02 -6.40826523e-01 8.15133035e-01
-9.89344120e-02 4.68268991e-01 5.37533045e-01 -7.35335231e-01
1.28423297e+00 4.37241614e-01 1.17891379e-01 -7.52687395e-01
-6.98335171e-02 -4.17626917e-01 -1.31049311e+00 -4.47435349e-01
3.65356989e-02 -3.14652205e-01 -1.00846791e+00 1.74705172e+00
-2.75136139e-02 1.33593082e-01 4.73716497e-01 8.81329894e-01
1.16079164e+00 1.05486929e+00 -3.39383632e-01 -4.48355615e-01
1.33912778e+00 -1.36226714e+00 -6.13969147e-01 -3.81708831e-01
6.39861763e-01 -8.06586623e-01 6.80104196e-01 2.24349946e-01
-1.50531816e+00 -5.15224636e-01 -1.01301956e+00 -2.79999763e-01
2.48719290e-01 1.32906809e-01 2.81578690e-01 -1.30977169e-01
-1.14947283e+00 3.41285378e-01 -5.47938943e-01 -1.11076593e-01
4.12344128e-01 8.34614187e-02 -6.01436317e-01 -6.05585694e-01
-1.09042263e+00 7.95932233e-01 4.31122303e-01 4.18083696e-03
-9.38577890e-01 -6.45617247e-01 -1.00928235e+00 2.01123044e-01
2.01905698e-01 -1.42436600e+00 1.33031249e+00 -9.26418006e-01
-1.31171477e+00 8.68353546e-01 -3.76637191e-01 -5.40059090e-01
5.85858524e-01 -1.23478733e-01 -8.73207673e-02 5.63629687e-01
2.40581542e-01 9.85734403e-01 7.39349723e-01 -1.25453663e+00
-4.44416285e-01 -9.61551815e-02 1.44591779e-01 3.40061814e-01
-1.95596032e-02 -8.26889053e-02 -5.75475216e-01 -7.25735664e-01
-9.66519564e-02 -9.55406785e-01 -1.67186394e-01 -1.61785945e-01
-5.71080983e-01 8.91529173e-02 2.72534311e-01 -1.24076235e+00
1.22125113e+00 -2.03878689e+00 6.30614579e-01 -4.35504973e-01
1.29239753e-01 2.33262241e-01 -3.61350149e-01 5.75786173e-01
-2.01226145e-01 2.89279316e-03 -5.79776585e-01 -3.75328392e-01
-4.31505531e-01 1.73491210e-01 -5.80903172e-01 2.75240019e-02
4.51653421e-01 1.19766736e+00 -1.10448682e+00 -5.87305009e-01
-3.46619971e-02 2.82876104e-01 -6.58203542e-01 5.38963079e-01
-2.24055648e-01 6.09252036e-01 -3.11759979e-01 2.99438298e-01
4.00931001e-01 -6.73927605e-01 6.53429925e-02 -1.52886838e-01
2.70444453e-01 6.46996260e-01 -5.20683348e-01 2.23876333e+00
-8.13413620e-01 5.97542405e-01 -1.69319555e-01 -9.47693944e-01
6.86768591e-01 8.08195114e-01 3.47957939e-01 -6.90277576e-01
-2.27196023e-01 3.28326255e-01 3.43770231e-03 -5.50928116e-01
5.14742553e-01 -8.72369856e-02 -2.76415408e-01 6.71854377e-01
2.05832481e-01 -6.49811979e-03 2.47210264e-01 4.59228188e-01
1.17914629e+00 1.06930681e-01 5.67341387e-01 4.35670525e-01
5.16599953e-01 -6.36866093e-02 2.04801112e-01 6.64763629e-01
2.80782968e-01 1.18050516e+00 7.59008348e-01 -3.98019373e-01
-1.38652039e+00 -1.11095262e+00 3.90809596e-01 6.86708272e-01
4.74440530e-02 -4.98298019e-01 -8.89446437e-01 -8.19846749e-01
-4.62101012e-01 7.84160376e-01 -4.32922184e-01 -5.15355945e-01
-8.88176322e-01 -6.26830637e-01 5.04897416e-01 6.28318131e-01
4.79657859e-01 -1.07129335e+00 -4.37356174e-01 5.41438639e-01
-8.87408495e-01 -1.39280391e+00 -8.18289042e-01 -3.76085639e-01
-1.02945077e+00 -7.67254174e-01 -1.24543428e+00 -7.36035585e-01
8.51674795e-01 3.74059111e-01 1.40164316e+00 1.42032877e-01
-6.82730824e-02 6.45162687e-02 -3.55852455e-01 -7.82307759e-02
-9.04532433e-01 3.05979818e-01 -5.12609363e-01 9.05253645e-03
-2.55029172e-01 -5.15822649e-01 -8.17600548e-01 1.15051912e-03
-9.25781488e-01 9.03533161e-01 1.02912128e+00 1.07452416e+00
4.77776706e-01 -7.82119393e-01 8.31332207e-01 -6.53564215e-01
5.81859767e-01 -6.42682672e-01 -6.57451078e-02 3.35079432e-01
2.04896778e-02 3.01072061e-01 9.65397835e-01 -1.52133822e-01
-1.02295053e+00 -1.63985759e-01 -2.99244434e-01 -4.51464295e-01
-2.50908360e-02 4.15372103e-01 -9.18239504e-02 6.50781155e-01
5.13191223e-01 8.44172657e-01 8.20318907e-02 -2.28376493e-01
3.32052499e-01 7.62086570e-01 6.41902864e-01 -1.91616133e-01
5.35416365e-01 3.39960188e-01 -2.04557642e-01 -3.85273606e-01
-1.09235954e+00 -1.04603238e-01 -4.09545451e-01 6.21596090e-02
9.41516578e-01 -1.23435092e+00 -1.37314782e-01 2.01215863e-01
-1.76338971e+00 1.02686867e-01 -1.07498944e-01 2.54674494e-01
-8.84798765e-01 2.53117174e-01 -7.02244699e-01 -2.41013601e-01
-7.71397948e-01 -1.27145398e+00 1.47602379e+00 -5.99807911e-02
-2.53388464e-01 -7.39829540e-01 -1.31573796e-01 6.48871720e-01
1.31518200e-01 2.88212895e-01 1.05297232e+00 -4.35787946e-01
-8.24952066e-01 1.30766751e-02 -3.15126061e-01 4.58925098e-01
2.04151526e-01 -5.10200202e-01 -7.27686763e-01 -4.14548308e-01
-2.32177246e-02 -4.82199132e-01 1.16244757e+00 8.75299573e-02
1.28594470e+00 -7.02545524e-01 -2.44005084e-01 6.84694409e-01
1.24362564e+00 7.79855102e-02 8.96963298e-01 -5.33780418e-02
8.79607379e-01 4.54886377e-01 3.14167947e-01 3.20507437e-01
5.02578199e-01 6.64497733e-01 2.78079480e-01 -1.06587671e-01
-4.68926370e-01 -5.76384783e-01 4.88010734e-01 1.22459793e+00
4.84128185e-02 -5.69726884e-01 -5.31325161e-01 6.38763011e-01
-1.80206656e+00 -1.04484475e+00 3.18909168e-01 1.95700121e+00
1.02853990e+00 -8.02886561e-02 -1.42338991e-01 -1.58521742e-01
6.59401357e-01 4.73165333e-01 -6.29125714e-01 -4.63318437e-01
1.57420844e-01 -1.06038071e-01 2.30692059e-01 3.34616870e-01
-7.97689140e-01 7.30627596e-01 5.76199913e+00 6.21078730e-01
-1.04111755e+00 1.57154739e-01 8.08027089e-01 -3.07491869e-01
-5.00194728e-01 -3.04907620e-01 -5.28098166e-01 7.08688498e-01
1.03279519e+00 -2.98344940e-01 2.68525600e-01 5.46009898e-01
8.87507275e-02 1.54275000e-01 -1.43403447e+00 1.10591900e+00
4.09894615e-01 -1.76501918e+00 6.92358315e-01 8.65341276e-02
9.97903764e-01 7.59050017e-04 -9.99007747e-02 1.20057136e-01
3.41840014e-02 -1.16629410e+00 5.02494574e-01 4.35667127e-01
1.24324286e+00 -7.80879021e-01 8.98025572e-01 4.33718532e-01
-9.28026676e-01 4.17901166e-02 -2.47630134e-01 1.03559822e-01
6.31308317e-01 3.52873921e-01 -9.65497851e-01 9.92855489e-01
1.78001393e-02 8.62627268e-01 -3.36316586e-01 7.18140960e-01
-3.60178649e-01 4.25373018e-01 1.93790302e-01 2.17952669e-01
3.84518176e-01 1.49051070e-01 4.65845466e-01 1.16640711e+00
6.10704422e-01 7.94790387e-02 -3.44952047e-02 7.15358138e-01
-5.06490886e-01 -1.16632134e-01 -9.53428268e-01 -1.56255849e-02
1.06993124e-01 9.06597435e-01 -2.87121117e-01 -6.71737671e-01
-6.90360010e-01 1.35867381e+00 2.93541402e-01 2.74912655e-01
-9.10486341e-01 -1.74188063e-01 5.43176234e-01 9.39232856e-02
3.74886572e-01 -3.18027264e-03 -2.95415550e-01 -1.45289624e+00
3.11613053e-01 -1.10950506e+00 2.52418578e-01 -9.89500046e-01
-9.60236847e-01 7.06499398e-01 -3.96477804e-02 -1.45461655e+00
-1.01445878e+00 -1.69305772e-01 -5.43135524e-01 9.44192588e-01
-1.48857987e+00 -1.01024020e+00 -1.70583174e-01 4.23089504e-01
1.06738293e+00 -3.34031343e-01 7.84245968e-01 2.54917949e-01
-2.54443228e-01 5.82772493e-01 4.86751869e-02 2.02092931e-01
5.83632648e-01 -9.39961672e-01 8.23635995e-01 7.08842278e-01
1.70007452e-01 3.61216843e-01 4.65254813e-01 -4.93768603e-01
-1.29099190e+00 -1.21822989e+00 1.02022421e+00 -1.96727782e-01
3.34306151e-01 -3.59243542e-01 -8.67359102e-01 6.98310614e-01
5.43615043e-01 -1.90678522e-01 6.56592131e-01 -5.71325481e-01
-1.90599501e-01 7.24426955e-02 -7.27214158e-01 7.81836033e-01
1.25142813e+00 -3.97594690e-01 -6.41321003e-01 3.63448590e-01
1.06777465e+00 -6.45615697e-01 -6.67960048e-01 2.18304232e-01
4.59699541e-01 -8.79039884e-01 1.09790635e+00 -6.15750015e-01
1.47773290e+00 -8.58965665e-02 -9.69035085e-03 -1.52144182e+00
-1.13262914e-01 -6.22603416e-01 -3.51571977e-01 1.07639158e+00
5.14547408e-01 -3.80020380e-01 5.48182011e-01 -5.53633925e-03
-4.34994757e-01 -1.07693005e+00 -9.34508383e-01 -4.20592397e-01
4.35359105e-02 8.20189789e-02 6.48599088e-01 5.16732097e-01
-2.49189854e-01 8.32546413e-01 -7.98570395e-01 -5.22993207e-02
3.45099747e-01 3.63184690e-01 7.55184829e-01 -6.93216145e-01
-5.12770951e-01 -1.69780955e-01 -7.71483555e-02 -1.48731744e+00
1.92983046e-01 -1.13269293e+00 9.84371006e-02 -2.20603776e+00
6.29468679e-01 2.91180342e-01 -7.61724310e-03 2.04096496e-01
-4.38461274e-01 1.56722575e-01 4.51066613e-01 2.98611283e-01
-5.23467422e-01 5.63433945e-01 1.73285592e+00 -3.17906380e-01
1.70566961e-01 -1.10743918e-01 -1.12480938e+00 6.02777421e-01
6.63432896e-01 -4.96031910e-01 -6.01829886e-01 -9.51127291e-01
2.12754667e-01 7.81542182e-01 4.06554490e-01 -7.74010718e-01
-4.81574573e-02 -6.79012686e-02 5.05346775e-01 -5.91141999e-01
3.62224489e-01 -3.64888430e-01 7.15194792e-02 6.01966619e-01
-6.69630289e-01 1.70618683e-01 -3.17765214e-02 5.36015570e-01
-4.07174587e-01 -2.69070029e-01 6.02756560e-01 -4.89902943e-01
-3.67119998e-01 4.33535218e-01 -8.37039277e-02 2.33518764e-01
7.09670186e-01 -3.02326363e-02 -5.27191937e-01 -7.23898530e-01
-5.39370775e-01 3.01403940e-01 4.41645354e-01 6.27950728e-01
8.91410470e-01 -1.37567520e+00 -1.29763973e+00 1.06100477e-01
2.30843872e-01 3.55465889e-01 4.32449579e-01 6.42783523e-01
-4.19572741e-01 6.25753105e-01 -2.08696768e-01 -6.45061195e-01
-1.12471545e+00 5.59275925e-01 6.45473003e-02 -7.99876869e-01
-8.68847311e-01 6.45983458e-01 7.92069495e-01 -9.94278193e-02
-2.07388952e-01 -2.53140181e-01 -7.47822374e-02 -2.16539227e-03
7.58563638e-01 2.58736014e-02 1.48678096e-02 -6.01695359e-01
-1.61963761e-01 4.50785637e-01 -4.24283952e-01 -2.62643784e-01
1.06294000e+00 -1.67591289e-01 4.49689813e-02 2.76339561e-01
1.47103012e+00 -5.05126476e-01 -1.31982172e+00 -1.43078625e-01
-2.68076181e-01 -3.00883442e-01 -4.49245304e-01 -7.74560153e-01
-1.00091851e+00 1.10976410e+00 2.35798974e-02 -2.67882854e-01
1.17766750e+00 1.70523569e-01 1.28494871e+00 3.95766139e-01
3.84758234e-01 -7.72730112e-01 4.36723173e-01 4.86615479e-01
1.30744135e+00 -1.19319355e+00 -8.74015614e-02 -2.21304953e-01
-9.86906111e-01 8.84745896e-01 5.41647732e-01 6.25303239e-02
-1.76788449e-01 1.21486083e-01 -3.40873033e-01 2.11872458e-01
-1.32439780e+00 1.38628319e-01 1.13444105e-01 4.74859625e-01
5.85758805e-01 -1.47959217e-02 -1.18860632e-01 3.98806363e-01
-3.39207023e-01 1.88107818e-01 8.07583272e-01 7.32524514e-01
-2.59697467e-01 -9.05617118e-01 -1.98808804e-01 5.88193655e-01
-5.83658636e-01 -3.54762793e-01 -3.33548814e-01 4.09779042e-01
-2.01178044e-01 7.64495730e-01 1.66331276e-01 -1.65107101e-01
3.07770520e-01 8.17888454e-02 5.99462986e-01 -8.22028339e-01
-4.30590659e-01 -1.62755027e-01 3.61274332e-01 -4.76997137e-01
-1.86146662e-01 -4.92447227e-01 -1.05860972e+00 -1.69895634e-01
9.43437815e-02 -2.47099828e-02 4.41403449e-01 8.88138652e-01
6.77116752e-01 8.98972213e-01 6.50036514e-01 -7.17276871e-01
-6.22765899e-01 -1.05169332e+00 2.97997538e-02 4.85030323e-01
6.28243566e-01 -1.37202367e-01 4.80964072e-02 3.74242008e-01] | [10.958041191101074, 0.8731665015220642] |
7078c5b5-5e1c-4aa4-a82a-b812d3712512 | fast-moving-object-counting-with-an-event | 2212.08384 | null | https://arxiv.org/abs/2212.08384v1 | https://arxiv.org/pdf/2212.08384v1.pdf | Fast-moving object counting with an event camera | This paper proposes the use of an event camera as a component of a vision system that enables counting of fast-moving objects - in this case, falling corn grains. These type of cameras transmit information about the change in brightness of individual pixels and are characterised by low latency, no motion blur, correct operation in different lighting conditions, as well as very low power consumption. The proposed counting algorithm processes events in real time. The operation of the solution was demonstrated on a stand consisting of a chute with a vibrating feeder, which allowed the number of grains falling to be adjusted. The objective of the control system with a PID controller was to maintain a constant average number of falling objects. The proposed solution was subjected to a series of tests to determine the correctness of the developed method operation. On their basis, the validity of using an event camera to count small, fast-moving objects and the associated wide range of potential industrial applications can be confirmed. | ['Tomasz Kryjak', 'Krzysztof Blachut', 'Marcin Kowalczyk', 'Kamil Bialik'] | 2022-12-16 | null | null | null | null | ['object-counting'] | ['computer-vision'] | [ 4.23295468e-01 -3.35356921e-01 4.22082752e-01 2.94037014e-02
5.44050276e-01 -6.67681396e-01 5.55074930e-01 3.45795333e-01
-6.51703775e-01 4.52972084e-01 -8.03043425e-01 -2.66076446e-01
-1.57919854e-01 -8.97616088e-01 -4.55760360e-01 -6.99185014e-01
1.36126027e-01 3.42960387e-01 6.08246326e-01 1.73573121e-01
4.66698349e-01 8.57610106e-01 -1.90772069e+00 -2.23368198e-01
4.13210005e-01 8.01355004e-01 8.59313786e-01 1.07008255e+00
-8.74025468e-03 5.90737581e-01 -6.67882264e-01 1.25085562e-01
4.99788765e-03 -3.79455656e-01 -1.89345866e-01 4.77259576e-01
-5.95829003e-02 -4.94256526e-01 3.74339014e-01 1.09948421e+00
-1.90695152e-02 7.15017989e-02 6.89476967e-01 -8.39185297e-01
-4.22930986e-01 4.28582653e-02 -5.86785913e-01 4.61653858e-01
4.58599538e-01 2.76864409e-01 2.75251120e-01 -3.91069353e-01
4.64481145e-01 8.07857096e-01 2.14065567e-01 -5.18150181e-02
-1.20861030e+00 -4.31592643e-01 -2.00273007e-01 2.62274235e-01
-1.29115582e+00 -2.19683349e-01 3.40310961e-01 -4.24834251e-01
1.05698419e+00 2.28186756e-01 9.57362354e-01 2.03890428e-01
7.23300338e-01 -2.25777421e-02 1.16589499e+00 -8.87886286e-01
5.64689815e-01 3.65072638e-01 2.56306201e-01 5.53813219e-01
9.99166012e-01 -3.04083824e-02 -1.24005206e-01 3.44278306e-01
1.07148898e+00 5.01644425e-02 -2.12154672e-01 -2.93477416e-01
-1.08253694e+00 5.48515737e-01 -3.95696238e-02 6.19820952e-01
-6.18961930e-01 -7.07757697e-02 2.42661849e-01 -1.67686015e-01
1.57635193e-02 1.91436633e-01 -1.90892532e-01 -2.39704981e-01
-5.73931575e-01 -5.97172379e-02 1.06060934e+00 8.60798478e-01
8.32349584e-02 3.82691585e-02 2.79596984e-01 4.64294441e-02
3.00405651e-01 8.56854320e-01 3.16714585e-01 -5.76975524e-01
6.44788845e-03 5.44048369e-01 4.41047668e-01 -1.05239284e+00
-3.15318853e-01 1.92406267e-01 -2.95865923e-01 8.64772737e-01
5.49810410e-01 -1.05506979e-01 -7.27764726e-01 1.11321151e+00
4.23978984e-01 1.46541312e-01 2.74978913e-02 8.22325766e-01
3.50784421e-01 1.02143657e+00 7.70873018e-03 -5.47777772e-01
1.65455735e+00 -1.46104425e-01 -1.16870356e+00 2.11675838e-01
-5.23568466e-02 -1.01230657e+00 7.57290184e-01 6.93124473e-01
-9.25064743e-01 -7.00734377e-01 -1.28727698e+00 5.21315038e-01
-5.57363749e-01 5.64894319e-01 3.76170278e-01 7.86448896e-01
-5.73450267e-01 4.46076512e-01 -1.11291742e+00 -7.91415751e-01
-4.87912714e-01 2.81401575e-01 -7.34759718e-02 2.54346400e-01
-5.99303126e-01 1.12245452e+00 7.68822610e-01 5.91419876e-01
-4.60943818e-01 -4.07266356e-02 -5.81078291e-01 2.06929460e-01
1.94147155e-01 -2.50652701e-01 1.16422117e+00 -9.21167314e-01
-1.76413417e+00 8.35083008e-01 3.50149572e-02 -2.86421001e-01
5.39365768e-01 -3.28764051e-01 -5.57674587e-01 4.62018073e-01
-3.99609022e-02 -7.47200055e-03 7.16346681e-01 -7.58963645e-01
-1.02543831e+00 -3.40559989e-01 -1.75999254e-01 6.97710663e-02
-2.45626103e-02 -3.68721001e-02 -9.03080255e-02 7.55020157e-02
-8.36230665e-02 -8.04704785e-01 1.53778911e-01 -1.77160487e-01
-4.83710086e-03 1.66241005e-02 1.11439872e+00 -2.33975336e-01
8.54511321e-01 -1.92947447e+00 -3.77443373e-01 1.21829085e-01
-3.83248419e-01 4.37192082e-01 4.80758995e-01 6.44693449e-02
-7.87018836e-02 -6.56014204e-01 -3.33425589e-02 5.78373969e-01
-3.40392828e-01 -6.93154931e-02 1.42348528e-01 5.27264893e-01
3.97135675e-01 6.67650998e-02 -7.27237582e-01 -4.18072194e-01
1.06906164e+00 4.86820191e-01 1.81212574e-01 1.28909230e-01
-5.00266068e-03 9.91054848e-02 -4.17599499e-01 5.20341992e-01
5.31247377e-01 1.15168259e-01 1.28530368e-01 -2.75644511e-01
-7.98735261e-01 -4.27136838e-01 -1.78349829e+00 8.81603062e-01
-2.87077546e-01 7.17491388e-01 2.51866400e-01 -5.00819027e-01
1.15645051e+00 4.88324165e-01 2.73547113e-01 -6.76283002e-01
5.33607423e-01 2.05816567e-01 -1.72120109e-01 -8.13093185e-01
6.85964346e-01 9.68102515e-02 3.41665983e-01 7.57413954e-02
-3.15433651e-01 -3.15624148e-01 6.90002263e-01 -3.44798356e-01
8.35837424e-01 3.17690492e-01 7.05073297e-01 -2.30592340e-01
6.43993676e-01 2.19638601e-01 9.67006087e-02 3.49133372e-01
-9.22927558e-02 6.36114702e-02 1.57223269e-02 -2.21525818e-01
-9.90343451e-01 -8.46597672e-01 -3.27114373e-01 5.12413204e-01
6.54007137e-01 4.39136952e-01 -7.46542454e-01 4.22107428e-01
-2.78561085e-01 8.53257000e-01 -6.52793720e-02 1.04696322e-02
-4.35296923e-01 -9.05075848e-01 -2.16561571e-01 1.89120889e-01
4.23194319e-01 -1.20524192e+00 -1.76264870e+00 5.87392926e-01
4.31938976e-01 -1.17449856e+00 3.55949372e-01 4.18033153e-01
-1.09824204e+00 -1.35535467e+00 -4.63071495e-01 -9.39663708e-01
7.53826559e-01 3.51828009e-01 7.70403922e-01 -2.61298090e-01
-7.01718271e-01 3.10375899e-01 -4.56243783e-01 -7.49122143e-01
-4.30212110e-01 -3.62196475e-01 -2.87929922e-01 -1.28595829e-01
7.40884423e-01 -1.16052777e-02 -4.53664750e-01 1.33572832e-01
-9.56767857e-01 -2.36822233e-01 4.20725733e-01 1.20884500e-01
3.08348387e-01 5.03170431e-01 7.79107511e-02 -6.74411118e-01
6.22108996e-01 4.38759513e-02 -1.42458749e+00 1.48240358e-01
-5.20211995e-01 -3.15820396e-01 6.41367972e-01 -4.00057822e-01
-1.24525952e+00 3.16490203e-01 4.98353869e-01 1.51774138e-01
-6.10984266e-01 -7.20613897e-02 -5.53762987e-02 8.70146602e-02
3.03632975e-01 4.75723185e-02 2.44504996e-02 -7.82725662e-02
6.81458414e-02 6.15006089e-01 9.03501511e-01 -4.55828980e-02
4.97140110e-01 2.58924782e-01 3.07765603e-01 -1.05113482e+00
7.76157454e-02 -6.25767052e-01 -4.25830752e-01 -5.86540639e-01
1.20697689e+00 -7.18840122e-01 -1.20952952e+00 9.09932733e-01
-1.32491803e+00 8.15075561e-02 -7.28820711e-02 8.47540736e-01
-3.14964652e-01 1.12636447e-01 -7.62076795e-01 -1.19689178e+00
-4.52074558e-01 -9.25975144e-01 5.80825150e-01 7.83535123e-01
-1.01571575e-01 -6.88595891e-01 -1.83215275e-01 4.50784937e-02
1.10405453e-01 3.31447870e-01 6.20984614e-01 -3.64791974e-02
-5.97357392e-01 -6.50666416e-01 -9.53096822e-02 1.07273169e-01
4.79672730e-01 7.64244616e-01 -8.46492827e-01 -1.60251722e-01
2.92757899e-01 2.60987580e-01 4.43924218e-01 8.27317178e-01
5.51039100e-01 4.95345354e-01 -2.86183089e-01 1.66300908e-01
2.07905459e+00 9.99400914e-01 5.30823052e-01 4.37585056e-01
3.09018731e-01 2.83716381e-01 1.21100533e+00 6.47173882e-01
-4.01558936e-01 5.93563020e-01 7.33845413e-01 -1.44025043e-01
2.56530106e-01 4.67845440e-01 2.45500058e-01 5.11545718e-01
-4.26207036e-01 -5.60361147e-01 -4.16375607e-01 4.71909761e-01
-1.39207447e+00 -1.11100638e+00 -7.84435391e-01 2.42879152e+00
3.56500655e-01 3.89596403e-01 -4.72096689e-02 6.04825974e-01
1.29937518e+00 -3.78823936e-01 -1.32530347e-01 -9.05006826e-01
4.44358736e-01 4.47144270e-01 9.63780284e-01 4.50254619e-01
-1.07472050e+00 3.77311230e-01 5.91134501e+00 1.94160908e-01
-1.27927935e+00 -3.78280878e-01 -9.03326180e-03 -7.92253483e-03
4.56481844e-01 -1.20570213e-01 -7.87431419e-01 6.92398131e-01
8.10719311e-01 -5.52140996e-02 2.46660054e-01 7.01105356e-01
5.97625792e-01 -9.68568027e-01 -7.87834406e-01 5.99038720e-01
-4.62209694e-02 -7.06773043e-01 -1.60535693e-01 -2.88124979e-01
3.28559160e-01 -7.20930099e-01 -2.71726459e-01 -4.52016741e-01
-1.57887265e-02 -4.40452307e-01 8.65727127e-01 5.00402212e-01
4.98724043e-01 -7.76294827e-01 9.73498940e-01 1.22122772e-01
-1.27932870e+00 -1.02502987e-01 -4.06154841e-01 -4.17184085e-01
3.93547773e-01 6.23668134e-01 -1.03034818e+00 2.55817771e-01
4.16864574e-01 5.25276409e-03 -2.79065430e-01 1.10010505e+00
-3.05750035e-02 4.65134799e-01 -6.67985320e-01 -5.45096219e-01
1.20440073e-01 -6.58215821e-01 3.49179775e-01 1.33198190e+00
5.26656151e-01 2.15168640e-01 -2.56102860e-01 9.52835858e-01
7.83340871e-01 3.26392166e-02 -5.88907003e-01 -4.19709869e-02
5.67042887e-01 1.10418344e+00 -1.46658957e+00 -2.75753111e-01
-4.30889964e-01 8.70473862e-01 -4.43839133e-01 6.10309318e-02
-9.43350792e-01 -7.26881146e-01 8.76610428e-02 7.53663331e-02
6.20693982e-01 -2.37563089e-01 -1.03792310e-01 -4.66805339e-01
-6.14954643e-02 -2.59472787e-01 1.96621399e-02 -1.05615056e+00
-5.77559769e-01 8.67628008e-02 1.06827825e-01 -1.07999980e+00
-1.10157907e-01 -1.02599132e+00 -4.94020075e-01 8.59694660e-01
-8.77572238e-01 -7.24342406e-01 -5.91216028e-01 4.62055475e-01
7.21349478e-01 -2.10708052e-01 8.22477818e-01 -6.83661178e-03
-6.10220909e-01 -2.91246563e-01 2.85282880e-01 -2.86864698e-01
2.64442235e-01 -1.28005672e+00 -1.47058010e-01 1.25227880e+00
-1.84529513e-01 2.25127012e-01 1.07191646e+00 -7.11554587e-01
-1.25547659e+00 -6.29448295e-01 6.45948648e-01 6.11038543e-02
3.29933196e-01 -1.02331676e-01 -5.90483844e-01 3.56590599e-01
1.42830551e-01 -2.08757579e-01 9.57286060e-02 -5.76466143e-01
6.80094361e-01 -2.94727176e-01 -1.17653203e+00 1.80228040e-01
4.21489812e-02 -1.75082698e-01 -6.04587018e-01 6.01091869e-02
2.43759945e-01 -4.25225794e-01 -5.17706990e-01 -2.02642903e-01
6.40295982e-01 -1.02821362e+00 6.97773159e-01 2.82004148e-01
2.07494870e-01 -5.84892452e-01 4.69694197e-01 -6.90831006e-01
-3.44234377e-01 -2.86833853e-01 4.04572815e-01 1.31817245e+00
-4.79347967e-02 -6.13555014e-01 5.83382428e-01 5.22587240e-01
1.37385562e-01 2.61064619e-01 -5.99989951e-01 -5.02364814e-01
-9.23069358e-01 1.11155242e-01 4.20730300e-02 4.90573615e-01
8.65745619e-02 1.72923848e-01 1.40660420e-01 5.51482975e-01
5.94933450e-01 -1.93663854e-02 3.63855451e-01 -1.37426031e+00
-1.99705228e-01 -2.51963940e-02 -8.20858896e-01 -1.14112534e-01
-4.83370423e-01 -1.83774158e-01 1.81865454e-01 -1.33564889e+00
4.85763736e-02 1.71215877e-01 -3.69075648e-02 -1.10741295e-01
-2.70461917e-01 -1.28449455e-01 1.51430711e-01 -9.78779644e-02
-8.54128972e-02 -3.32619876e-01 8.89819682e-01 2.77321637e-01
-3.44362080e-01 2.79508531e-01 2.89702237e-01 6.64024055e-01
5.96160591e-01 -2.92649239e-01 -5.20371139e-01 -1.19274393e-01
2.57053096e-02 -1.28627848e-02 4.63334948e-01 -1.22890556e+00
1.55210555e-01 -9.46270376e-02 6.35373831e-01 -6.57701552e-01
7.60172009e-02 -1.64812076e+00 7.30496287e-01 9.43746448e-01
1.01239070e-01 3.15573573e-01 2.13432238e-01 6.69870019e-01
1.76845007e-02 -7.48280227e-01 7.09936678e-01 -2.41023600e-01
-1.11684418e+00 -6.02778554e-01 -7.79411733e-01 -7.49361634e-01
1.69001663e+00 -9.21086431e-01 -1.86398834e-01 1.71654344e-01
-5.31996489e-01 -2.28633776e-01 5.38920462e-01 -4.64633033e-02
2.28397042e-01 -8.42632115e-01 -2.05473363e-01 3.29690874e-01
-1.06348738e-01 -5.61137386e-02 8.09538364e-03 5.18639088e-01
-1.58914423e+00 4.22779918e-01 -7.30316162e-01 -6.44933224e-01
-1.58367312e+00 7.18492270e-01 1.70158669e-01 1.44362524e-01
-7.33758569e-01 3.89493048e-01 -2.45125502e-01 5.68428457e-01
1.38544753e-01 -7.86578834e-01 -4.04767722e-01 4.22541276e-02
5.26719630e-01 6.56879306e-01 1.47295743e-01 -2.46189777e-02
-2.34184712e-01 7.06093252e-01 2.08719790e-01 -5.65346666e-02
1.20595980e+00 -1.33554742e-01 3.82231884e-02 7.68238604e-01
4.07572776e-01 -4.76903319e-02 -1.36413991e+00 3.75345081e-01
-8.65308940e-02 -4.84392822e-01 1.12004712e-01 -6.13502502e-01
-7.40937591e-01 4.13650066e-01 1.10167599e+00 6.24764919e-01
1.37320137e+00 -6.46376073e-01 2.55607754e-01 1.73886031e-01
4.60731030e-01 -1.33850706e+00 -3.68767679e-01 5.37349358e-02
1.89616650e-01 -1.00809693e+00 2.74162024e-01 -5.82518637e-01
-2.90056378e-01 1.76214457e+00 4.05573428e-01 -1.80249467e-01
1.73082188e-01 8.37666392e-01 -1.47123411e-01 -2.11801544e-01
-5.52183688e-01 -3.23033392e-01 -4.63676602e-01 7.63275862e-01
4.87116724e-01 3.84207785e-01 -8.17196667e-01 -1.80795282e-01
2.51835108e-01 7.25722134e-01 1.01366615e+00 1.18307364e+00
-8.10607612e-01 -3.12488943e-01 -1.02097046e+00 3.04302752e-01
-5.11773169e-01 4.42392379e-01 6.26411438e-02 1.11080265e+00
4.90623832e-01 1.19921100e+00 6.01265728e-01 2.84018427e-01
6.28273547e-01 -1.55984685e-01 5.75881481e-01 -2.58697778e-01
-5.15065908e-01 1.41290545e-01 -1.18825614e-01 -3.19105119e-01
-5.07319868e-01 -8.68591368e-01 -1.27115107e+00 -2.61313558e-01
-6.77731216e-01 -1.07363567e-01 1.05577683e+00 8.60906839e-01
-4.54344004e-01 6.36934996e-01 3.70709777e-01 -4.72342193e-01
-4.01618421e-01 -9.15759385e-01 -1.14642274e+00 3.26023310e-01
9.74007174e-02 -4.93398756e-01 -4.58396763e-01 5.21911085e-01] | [9.149149894714355, -1.4198837280273438] |
37fa7c87-6a6f-4052-b3cf-0d9d72a68787 | joint-inference-for-fine-grained-opinion | null | null | https://aclanthology.org/P13-1161 | https://aclanthology.org/P13-1161.pdf | Joint Inference for Fine-grained Opinion Extraction | null | ['Bishan Yang', 'Claire Cardie'] | 2013-08-01 | null | null | null | acl-2013-8 | ['fine-grained-opinion-analysis'] | ['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] | [-7.308376789093018, 3.708082914352417] |
99337b04-4777-483f-b402-ca450d5b15c0 | reduce-communication-costs-and-preserve | 2208.12268 | null | https://arxiv.org/abs/2208.12268v3 | https://arxiv.org/pdf/2208.12268v3.pdf | FedPrompt: Communication-Efficient and Privacy Preserving Prompt Tuning in Federated Learning | Federated learning (FL) has enabled global model training on decentralized data in a privacy-preserving way by aggregating model updates. However, for many natural language processing (NLP) tasks that utilize pre-trained language models (PLMs) with large numbers of parameters, there are considerable communication costs associated with FL. Recently, prompt tuning, which tunes some soft prompts without modifying PLMs, has achieved excellent performance as a new learning paradigm. Therefore we want to combine the two methods and explore the effect of prompt tuning under FL. In this paper, we propose "FedPrompt" to study prompt tuning in a model split aggregation way using FL, and prove that split aggregation greatly reduces the communication cost, only 0.01% of the PLMs' parameters, with little decrease on accuracy both on IID and Non-IID data distribution. This improves the efficiency of FL method while also protecting the data privacy in prompt tuning. In addition, like PLMs, prompts are uploaded and downloaded between public platforms and personal users, so we try to figure out whether there is still a backdoor threat using only soft prompts in FL scenarios. We further conduct backdoor attacks by data poisoning on FedPrompt. Our experiments show that normal backdoor attack can not achieve a high attack success rate, proving the robustness of FedPrompt. We hope this work can promote the application of prompt in FL and raise the awareness of the possible security threats. | ['Gongshen Liu', 'Peixuan Li', 'Fangqi Li', 'Wei Du', 'Haodong Zhao'] | 2022-08-25 | null | null | null | null | ['data-poisoning'] | ['adversarial'] | [-1.82457626e-01 -3.45493294e-02 -2.89244831e-01 -2.62813121e-01
-6.98455632e-01 -1.19391608e+00 6.26730561e-01 2.22046927e-01
-6.52129531e-01 7.39427090e-01 -5.33027016e-02 -7.15805531e-01
1.03472829e-01 -8.90302777e-01 -8.85833025e-01 -7.64974535e-01
-2.07524151e-01 1.40198588e-01 3.47493589e-01 -5.37226573e-02
-1.24945834e-01 5.68133295e-01 -8.41660202e-01 2.93938160e-01
7.27629185e-01 7.58682907e-01 -2.38477737e-01 3.94559830e-01
-6.88191429e-02 5.20624876e-01 -6.73761189e-01 -7.71988273e-01
7.74970472e-01 6.69254139e-02 -8.10183525e-01 -5.86289227e-01
1.82417929e-01 -6.60345078e-01 -3.78318787e-01 1.23544800e+00
4.59723413e-01 -2.93411821e-01 5.54198958e-02 -1.69717801e+00
-2.02469721e-01 8.58707547e-01 -5.15083849e-01 -2.21632555e-01
1.66226879e-01 3.59229773e-01 6.85292900e-01 -2.98130453e-01
5.39889753e-01 1.18131483e+00 5.40784478e-01 6.16811216e-01
-1.26224005e+00 -1.34151375e+00 7.07293749e-02 -6.79983571e-02
-1.33263767e+00 -4.08948600e-01 5.10941505e-01 -1.00705422e-01
5.30049443e-01 8.03570807e-01 1.09497562e-01 1.23104346e+00
1.74426600e-01 7.42844582e-01 1.15107226e+00 -2.14496925e-01
3.74267638e-01 7.00600147e-01 2.77291507e-01 5.89904547e-01
6.10756099e-01 3.04417461e-01 -5.23969352e-01 -1.19697177e+00
3.46491039e-01 1.82662070e-01 -4.77960765e-01 -3.54261607e-01
-9.01815712e-01 9.75586772e-01 2.01835454e-01 9.28675830e-02
-8.85804882e-04 -3.32813300e-02 6.13503277e-01 6.61315620e-01
3.59557480e-01 3.67516965e-01 -9.20346677e-01 2.08356485e-01
-6.21630728e-01 2.62791783e-01 1.17085850e+00 7.18509853e-01
8.52275193e-01 -3.81683707e-01 1.85729824e-02 2.91861534e-01
2.04577506e-01 5.05485475e-01 5.07217884e-01 -6.69247091e-01
5.93584776e-01 4.31705922e-01 1.91109478e-01 -1.10089791e+00
-2.30340108e-01 -8.39912593e-02 -9.39838886e-01 2.83957291e-02
4.54558611e-01 -6.26733124e-01 -3.16692442e-01 2.12151575e+00
7.02511013e-01 1.06027052e-01 8.68799984e-02 6.02590084e-01
2.00465709e-01 5.93477011e-01 3.38568017e-02 -3.51062864e-01
1.51997781e+00 -6.48200810e-01 -6.72247350e-01 1.93027899e-01
1.05388534e+00 -5.26239693e-01 1.22489548e+00 4.07148093e-01
-5.27411580e-01 1.46423578e-01 -9.20616508e-01 1.49282828e-01
-5.28189600e-01 -2.32760593e-01 8.75904858e-01 1.14345849e+00
-7.74464488e-01 3.87366444e-01 -1.11457336e+00 -2.92089641e-01
6.41466379e-01 7.23203480e-01 -7.33396590e-01 1.11877777e-01
-1.47852647e+00 3.76660138e-01 3.97871137e-01 -5.43000102e-01
-8.82621765e-01 -8.37748468e-01 -5.04748523e-01 5.58496714e-02
6.35567725e-01 -6.26875758e-01 1.08231223e+00 -2.21121579e-01
-1.38624573e+00 5.78119338e-01 2.69035790e-02 -1.00049627e+00
7.89890468e-01 -1.52321056e-01 -4.73599523e-01 2.62118550e-03
-2.27913156e-01 2.69276679e-01 9.51847970e-01 -1.11309850e+00
-5.04939616e-01 -4.62600052e-01 2.55100429e-01 -2.65815467e-01
-1.04713261e+00 2.17493489e-01 -9.19222534e-02 -3.52763563e-01
-3.69693696e-01 -9.51343656e-01 -3.94779205e-01 1.28572583e-01
-6.14122093e-01 -7.04021603e-02 1.38679934e+00 -5.08867383e-01
1.41176033e+00 -2.24008393e+00 -7.35417545e-01 3.76397550e-01
2.28868991e-01 7.88373828e-01 9.31097474e-03 6.64882958e-01
3.24576586e-01 6.31758690e-01 -3.29098493e-01 -4.30802971e-01
1.61178589e-01 1.96803823e-01 -8.94248903e-01 5.54292679e-01
-5.39289951e-01 7.30206430e-01 -5.09321451e-01 -4.11301672e-01
-7.83621669e-02 2.07231432e-01 -7.33101785e-01 1.30078912e-01
-3.39480460e-01 3.46903950e-01 -6.09690666e-01 4.78848040e-01
1.15088201e+00 -2.57280588e-01 3.86572957e-01 -3.73114124e-02
1.24243163e-02 2.30373040e-01 -1.17592609e+00 1.31178188e+00
-2.29488701e-01 3.69286090e-02 4.53283608e-01 -4.71803397e-01
8.32760572e-01 4.50461954e-01 2.76710987e-01 -2.15236649e-01
5.93809374e-02 1.16895884e-01 -4.43431824e-01 -3.24358970e-01
2.37284392e-01 1.66761860e-01 -2.89962888e-01 9.56018925e-01
-2.96745390e-01 3.80692929e-01 -4.19249982e-01 4.79760319e-01
1.27469730e+00 -5.05037129e-01 2.49217510e-01 -3.01837087e-01
5.75227499e-01 -2.07052261e-01 6.90220475e-01 9.58966315e-01
-3.42290998e-01 -2.56578857e-03 5.45179844e-01 -4.20356274e-01
-5.25277972e-01 -7.80762851e-01 2.00949572e-02 1.02063465e+00
1.24487646e-01 -9.71093595e-01 -8.03965330e-01 -1.29984903e+00
2.85122603e-01 6.65804446e-01 -1.90401807e-01 -2.29000494e-01
-4.30636764e-01 -8.68381798e-01 1.19723523e+00 -1.25765815e-01
8.23817730e-01 -6.65884018e-01 -3.95757258e-01 -5.73117025e-02
-6.94810152e-02 -9.12181795e-01 -7.61224985e-01 1.22964904e-01
-6.12430751e-01 -1.25108242e+00 -6.56228215e-02 -4.66709852e-01
6.66049659e-01 3.45805675e-01 4.46768701e-01 2.94381175e-02
4.47460264e-02 1.91306174e-01 3.38480473e-02 -4.43994999e-01
-6.30053818e-01 3.40569407e-01 4.56654400e-01 2.65642881e-01
4.19524401e-01 -6.23396456e-01 -4.26546782e-01 3.21739167e-01
-1.27519107e+00 -3.15182000e-01 7.58314356e-02 6.28036022e-01
1.45507485e-01 2.18261868e-01 4.84222114e-01 -1.39059412e+00
8.69046271e-01 -5.06797194e-01 -9.09422159e-01 4.54627901e-01
-7.06302345e-01 1.70220762e-01 1.05568898e+00 -7.24812031e-01
-9.43659008e-01 1.72931835e-01 -1.93740800e-02 -5.13872266e-01
-1.25997365e-01 1.54425234e-01 -5.52936435e-01 -3.54204327e-01
8.01198006e-01 1.14592284e-01 2.45060906e-01 -8.00306737e-01
3.34102869e-01 1.00046325e+00 3.48984778e-01 -6.75841331e-01
1.06561637e+00 7.27255821e-01 -4.95080464e-02 -4.98613715e-01
-3.23162079e-01 -6.59486428e-02 1.89649403e-01 5.54216146e-01
4.09650087e-01 -8.37974548e-01 -1.17097294e+00 5.36478281e-01
-1.09474707e+00 -1.66038275e-01 7.06353486e-02 3.97944391e-01
-9.88528207e-02 6.94001138e-01 -6.16003871e-01 -7.45570838e-01
-7.66899884e-01 -9.42313373e-01 4.24555480e-01 -3.22946496e-02
-1.29097089e-01 -9.53888118e-01 5.89041784e-02 3.97309124e-01
5.57141125e-01 2.50102431e-01 8.83580387e-01 -1.26026320e+00
-5.11783302e-01 -4.12923396e-01 1.55367091e-01 3.09331298e-01
2.55734116e-01 -2.86915243e-01 -1.06542695e+00 -8.44051898e-01
4.48696971e-01 -2.71792531e-01 4.61365670e-01 -3.03428322e-01
1.36042821e+00 -1.22356546e+00 -4.32255387e-01 8.21830451e-01
1.40156841e+00 -3.62296253e-02 3.00988317e-01 2.56018549e-01
5.26861548e-01 3.70433807e-01 4.45879698e-01 6.62842512e-01
1.52105436e-01 4.09116894e-01 5.23043931e-01 3.19820130e-03
5.76789439e-01 -6.68966174e-01 5.60283840e-01 2.69406915e-01
7.80391574e-01 -2.03240037e-01 -6.72392666e-01 1.98533550e-01
-1.78146064e+00 -6.46525562e-01 1.98843256e-02 2.61528611e+00
1.17991567e+00 -8.09265897e-02 -1.88738834e-02 -1.55851632e-01
7.00176537e-01 1.64427295e-01 -6.24932051e-01 -4.77864265e-01
-5.64329736e-02 1.47513896e-01 1.11364901e+00 4.99846905e-01
-1.14368105e+00 9.50277805e-01 5.49991560e+00 1.00817430e+00
-1.37726259e+00 3.00628811e-01 5.87088525e-01 -2.35584244e-01
-3.41820568e-01 4.65556145e-01 -9.58014309e-01 7.63588727e-01
9.43294048e-01 -6.21604860e-01 6.96274042e-01 9.19457853e-01
-5.96103035e-02 2.88953394e-01 -9.91164267e-01 8.52508187e-01
-4.76234972e-01 -1.39276385e+00 -1.78978443e-02 4.58844066e-01
4.26541328e-01 2.01102525e-01 5.85308019e-03 2.53011942e-01
6.37630165e-01 -6.94413781e-01 1.49672881e-01 3.55048031e-02
7.34566867e-01 -9.04077291e-01 3.97186786e-01 1.05294824e+00
-7.38749623e-01 -2.93558240e-01 -5.04440844e-01 1.50301278e-01
-3.90580706e-02 5.05586267e-01 -9.73117828e-01 5.90541005e-01
6.81124508e-01 -1.74275130e-01 -5.40715396e-01 7.43802965e-01
-4.91925254e-02 8.14575791e-01 -9.22336280e-01 7.55783767e-02
9.76572279e-03 1.00642636e-01 6.55630410e-01 9.39093828e-01
5.90366609e-02 3.26150842e-02 2.41774589e-01 6.46100998e-01
-4.16020155e-01 1.36114791e-01 -7.68169105e-01 -8.90316740e-02
1.02606285e+00 1.22254205e+00 -4.44353484e-02 -2.51176566e-01
-9.39446688e-02 7.10382402e-01 1.45074785e-01 1.81400597e-01
-6.20232880e-01 -3.28117847e-01 9.26776588e-01 1.95539385e-01
-8.25462490e-02 -2.06135139e-01 -1.49157986e-01 -1.37937975e+00
6.76344633e-02 -1.11194706e+00 7.63011813e-01 4.55036089e-02
-1.33646274e+00 4.25080508e-01 2.22220179e-02 -1.00184715e+00
-3.60416062e-02 -2.57276325e-03 -6.40245438e-01 6.70278728e-01
-1.27570903e+00 -1.08774126e+00 2.79366285e-01 1.20342481e+00
-3.01802188e-01 -1.91122711e-01 1.04329062e+00 1.89908460e-01
-6.00862622e-01 1.38520229e+00 3.22528660e-01 8.72383639e-02
1.06321406e+00 -7.19578624e-01 3.58296007e-01 9.21309650e-01
1.96605071e-01 1.18651772e+00 6.32640004e-01 -6.45442963e-01
-1.53461039e+00 -1.31603777e+00 9.90885854e-01 -3.10703605e-01
4.16628540e-01 -8.13754797e-01 -1.08917356e+00 8.19863081e-01
1.43212780e-01 2.15787843e-01 8.05823565e-01 -1.95320621e-01
-6.57768846e-01 -3.12144667e-01 -1.93745649e+00 5.85625648e-01
5.88059366e-01 -6.61500394e-01 -1.82322815e-01 5.03034592e-01
1.27846551e+00 -1.41331583e-01 -7.70652473e-01 2.64723718e-01
2.40992814e-01 -8.23073387e-01 7.38934875e-01 -7.33570635e-01
-5.83563924e-01 -2.63285995e-01 -1.51818246e-01 -8.92416000e-01
1.41517639e-01 -1.38348734e+00 -5.09897947e-01 1.66369903e+00
1.68319717e-01 -1.44869673e+00 9.33838844e-01 1.03699374e+00
6.81348801e-01 -3.85880232e-01 -1.05998600e+00 -8.48168910e-01
1.73670575e-01 -2.84767240e-01 1.28995252e+00 1.24427938e+00
5.87545186e-02 -1.20816819e-01 -6.70862257e-01 4.92206365e-01
7.52550662e-01 -1.78221062e-01 1.15852308e+00 -1.00677645e+00
-4.10136968e-01 1.90820560e-01 1.27635762e-01 -7.56239533e-01
1.49501413e-01 -9.51107144e-01 -7.72448063e-01 -5.26585162e-01
3.44528109e-02 -5.65163791e-01 -3.78922492e-01 1.11137593e+00
8.84058625e-02 -7.02126771e-02 3.85004580e-01 3.35820735e-01
-3.60687435e-01 3.79196316e-01 7.28136957e-01 -4.78796437e-02
-3.77671987e-01 1.66332111e-01 -9.57722366e-01 4.70546842e-01
1.14658201e+00 -8.20144117e-01 -4.23290461e-01 -1.07545808e-01
3.82759459e-02 -8.46645832e-02 3.17441434e-01 -6.83019876e-01
5.13711870e-01 -2.29353443e-01 -1.54167712e-01 -3.59529495e-01
-5.67963645e-02 -1.14172292e+00 3.55880380e-01 6.58761203e-01
-2.87988514e-01 -2.05133200e-01 1.40618622e-01 6.08512461e-01
7.47233108e-02 1.36404321e-01 6.69320107e-01 -9.52396095e-02
-3.47736515e-02 6.23409152e-01 -3.83930542e-02 -1.98762983e-01
1.09521544e+00 2.49620125e-01 -6.44662738e-01 -3.08240116e-01
-3.72108787e-01 5.22138059e-01 7.86231637e-01 3.16656440e-01
1.06699355e-01 -1.03781986e+00 -4.04036731e-01 7.16213465e-01
-1.60198763e-01 -7.02451691e-02 1.57207578e-01 6.64271474e-01
-2.48462766e-01 4.62560266e-01 1.78270161e-01 -2.65897304e-01
-1.64735889e+00 9.16866422e-01 6.59036636e-02 -5.12187660e-01
-5.66086769e-01 7.13383555e-01 2.52343774e-01 -8.98363292e-01
4.94544446e-01 1.00816809e-01 4.09118742e-01 -1.87765509e-01
6.44850314e-01 3.49000663e-01 3.48001681e-02 -6.57059997e-02
-4.15311247e-01 -1.60276413e-01 -4.51594710e-01 -7.59528279e-02
1.01994658e+00 -1.26932517e-01 -3.56593907e-01 -2.42543489e-01
1.62312937e+00 4.91236120e-01 -1.24873900e+00 -5.25878787e-01
-8.88874158e-02 -6.85836434e-01 -1.32791385e-01 -7.74854481e-01
-1.06565559e+00 6.42383277e-01 6.35859549e-01 3.11464459e-01
1.02045572e+00 -1.97518647e-01 1.28543997e+00 5.37849307e-01
8.41098070e-01 -7.04320133e-01 -4.38831896e-01 1.78542018e-01
5.14025927e-01 -9.91977036e-01 -3.18039097e-02 -2.54334688e-01
-5.36758721e-01 6.50499046e-01 5.40256202e-01 2.34550342e-01
8.02114964e-01 4.56154138e-01 1.14911884e-01 1.73827201e-01
-1.03117967e+00 6.56917155e-01 -3.95465165e-01 4.67344850e-01
-4.09138173e-01 1.82172954e-01 -4.41671520e-01 1.04715562e+00
-3.22600901e-01 -2.22413428e-02 3.91012520e-01 9.80117142e-01
-1.95896760e-01 -1.72705412e+00 -6.48853540e-01 2.83001691e-01
-8.65940690e-01 8.34699646e-02 -5.82407176e-01 5.44931829e-01
-3.69786881e-02 9.59436715e-01 -3.73104125e-01 -6.40091598e-01
-2.63543800e-02 2.06805661e-01 -2.59112328e-01 -1.98312983e-01
-1.06924176e+00 -2.62458891e-01 -1.46574855e-01 -6.88901007e-01
3.79346907e-01 -3.26889306e-01 -1.16395092e+00 -8.12664986e-01
-5.23639619e-01 5.12570560e-01 6.97598338e-01 4.16472167e-01
7.14159191e-01 -4.39204454e-01 1.28180265e+00 -1.26849130e-01
-1.39974713e+00 -5.10694623e-01 -6.47112727e-01 1.64212376e-01
4.35178816e-01 -1.93224829e-02 -7.26402640e-01 -4.02467310e-01] | [5.819021224975586, 6.834571361541748] |
63a50dbd-b53f-4a8c-9d5a-818fc428431d | dual-skip-connections-minimize-the-false | 2110.13036 | null | https://arxiv.org/abs/2110.13036v1 | https://arxiv.org/pdf/2110.13036v1.pdf | Dual Skip Connections Minimize the False Positive Rate of Lung Nodule Detection in CT images | Pulmonary cancer is one of the most commonly diagnosed and fatal cancers and is often diagnosed by incidental findings on computed tomography. Automated pulmonary nodule detection is an essential part of computer-aided diagnosis, which is still facing great challenges and difficulties to quickly and accurately locate the exact nodules' positions. This paper proposes a dual skip connection upsampling strategy based on Dual Path network in a U-Net structure generating multiscale feature maps, which aims to minimize the ratio of false positives and maximize the sensitivity for lesion detection of nodules. The results show that our new upsampling strategy improves the performance by having 85.3% sensitivity at 4 FROC per image compared to 84.2% for the regular upsampling strategy or 81.2% for VGG16-based Faster-R-CNN. | ['Andreas Nürnberger', 'Tung Lung Liu', 'Philipp Ernst', 'Jiahua Xu'] | 2021-10-25 | null | null | null | null | ['lung-nodule-detection'] | ['medical'] | [ 1.29864946e-01 2.64421314e-01 -3.46154004e-01 1.73433408e-01
-6.49652541e-01 -2.63630971e-02 2.36121908e-01 -1.39915124e-01
-3.38508815e-01 5.41492939e-01 -4.22171466e-02 -6.11927211e-01
-1.13940343e-01 -9.66233730e-01 -2.12515175e-01 -5.22150695e-01
-1.46761566e-01 3.73297095e-01 6.30720019e-01 8.29424784e-02
-2.09383652e-01 9.02877629e-01 -9.70021367e-01 3.45815063e-01
5.94229221e-01 9.93423581e-01 3.28783244e-01 8.60883176e-01
-8.86511952e-02 8.65133464e-01 -2.16259196e-01 -4.28776890e-02
2.97658354e-01 -4.69900668e-01 -8.49748731e-01 9.80685204e-02
2.83912539e-01 -4.77331936e-01 -5.53020120e-01 8.44034195e-01
4.22319800e-01 -2.03182563e-01 7.57430434e-01 -6.24178767e-01
-3.15151870e-01 5.52467108e-01 -7.62983799e-01 6.26329839e-01
-3.10031712e-01 -3.00267059e-02 7.76837468e-01 -7.13795185e-01
2.36281812e-01 7.43469656e-01 9.18706536e-01 6.45310223e-01
-8.09123933e-01 -5.45027137e-01 -4.74975199e-01 -4.02985699e-02
-1.53889644e+00 3.92361823e-03 1.67577848e-01 -2.09015056e-01
8.30636203e-01 3.03699046e-01 9.28407431e-01 5.24538517e-01
6.19559467e-01 3.21810007e-01 5.92368960e-01 -1.39810979e-01
-2.97744155e-01 5.12645692e-02 -2.31517568e-01 1.30630088e+00
7.18195140e-01 1.33358657e-01 2.21923009e-01 -1.34048700e-01
1.72451997e+00 4.47486162e-01 -3.64497691e-01 -1.26329780e-01
-1.45367706e+00 7.85717309e-01 1.21077859e+00 6.86811388e-01
-3.44036609e-01 5.43791294e-01 1.17482096e-01 -1.66017383e-01
9.64580104e-03 6.38422489e-01 -6.84893429e-02 9.77538526e-02
-8.71064782e-01 -2.18683884e-01 4.43201244e-01 7.35697687e-01
3.09422791e-01 1.83005273e-01 -3.98535818e-01 6.73159540e-01
2.20697105e-01 4.95233148e-01 8.26330483e-01 -5.86761892e-01
1.42518878e-02 8.55449915e-01 -1.72668114e-01 -7.32213974e-01
-6.05077684e-01 -1.15802288e+00 -1.24969256e+00 -1.66970007e-02
3.83077115e-01 -1.27470657e-01 -1.08424616e+00 9.56671655e-01
1.02601953e-01 2.32814178e-01 -1.69349134e-01 6.49318755e-01
9.65381324e-01 3.79714519e-01 -6.86906129e-02 9.66438726e-02
1.65518463e+00 -1.10618567e+00 -3.99640977e-01 -3.97828128e-03
7.61089921e-01 -6.46710455e-01 6.41955793e-01 -1.42235801e-01
-8.97200644e-01 -5.23974478e-01 -9.82480884e-01 5.35310954e-02
1.25243561e-02 7.34329700e-01 5.38537860e-01 7.27110624e-01
-1.03608835e+00 5.83930731e-01 -1.04805076e+00 -3.71368855e-01
7.39271879e-01 6.98745847e-01 -1.12953037e-01 2.86242701e-02
-8.10817182e-01 7.97369480e-01 2.63927579e-01 -1.43370047e-01
-7.57634163e-01 -9.34669435e-01 -5.44005275e-01 2.44426414e-01
3.17094415e-01 -7.99963653e-01 1.48206651e+00 -5.28422356e-01
-1.09213722e+00 5.19064367e-01 -8.08399618e-02 -6.96258664e-01
5.60844064e-01 2.16223598e-01 -1.74386576e-01 3.65887672e-01
5.14540039e-02 8.44982147e-01 5.02077460e-01 -4.61267084e-01
-8.84584844e-01 -1.30842000e-01 -3.86770397e-01 2.65686035e-01
-1.91768140e-01 -4.16637480e-01 -4.71464217e-01 -5.17361283e-01
3.70285749e-01 -7.66300738e-01 -8.77912998e-01 5.56600511e-01
-2.95468330e-01 -1.12144994e-02 1.10598421e+00 -5.42964756e-01
9.18059349e-01 -1.90489018e+00 -5.12388051e-01 2.93520212e-01
5.84853232e-01 3.90232652e-01 2.46524706e-01 -3.74953955e-01
5.31352721e-02 2.26387322e-01 1.82141494e-02 3.69683683e-01
-8.35227132e-01 8.82894099e-02 3.27384472e-01 4.06591833e-01
3.95586282e-01 1.21646297e+00 -1.01163399e+00 -6.12283409e-01
3.75880390e-01 7.25491583e-01 -2.94491142e-01 -9.49549302e-02
2.00795263e-01 2.32111573e-01 -6.89329982e-01 6.76300526e-01
2.44243398e-01 -8.95155311e-01 1.19456552e-01 -3.69938493e-01
-5.19225113e-02 1.80584460e-01 -7.79465377e-01 1.06297231e+00
-5.53038299e-01 7.00408876e-01 -2.56380320e-01 -3.62967134e-01
8.30756783e-01 6.01818502e-01 6.72288835e-01 -3.07499200e-01
4.08735931e-01 4.59824860e-01 5.21623135e-01 -1.37750790e-01
1.81962922e-01 -2.10184067e-01 3.31967205e-01 1.45652533e-01
-3.02090853e-01 -1.81947872e-01 -9.91575792e-02 -1.06677320e-02
1.41002464e+00 -5.48824370e-01 8.91377747e-01 -3.93295825e-01
5.88855743e-01 1.50024638e-01 3.66281062e-01 6.82352722e-01
-2.49245122e-01 8.27551365e-01 4.12522733e-01 -4.84969169e-01
-9.25667167e-01 -1.06338382e+00 -2.77020067e-01 5.56860641e-02
-1.89425021e-01 5.16099147e-02 -3.45727921e-01 -9.43771660e-01
6.05017915e-02 9.45262089e-02 -7.15731680e-01 2.82085631e-02
-5.31946599e-01 -6.45995021e-01 7.17788398e-01 9.60573137e-01
9.54043448e-01 -7.26125360e-01 -8.25275004e-01 2.67993599e-01
-1.16909198e-01 -8.59298587e-01 -5.40861011e-01 3.87777835e-01
-1.31657696e+00 -1.41564977e+00 -1.04403615e+00 -9.93845642e-01
1.18299890e+00 6.48637056e-01 8.15641046e-01 2.82490820e-01
-9.55343425e-01 -5.10858111e-02 2.50340849e-02 -5.01556806e-02
-3.57484668e-01 3.78284425e-01 -3.36671859e-01 -4.31541115e-01
2.15219021e-01 -1.94595382e-01 -8.02668273e-01 3.76052260e-01
-6.60182297e-01 5.02666866e-04 1.21053827e+00 9.64945495e-01
9.10627723e-01 1.58480987e-01 3.62278193e-01 -1.16137087e+00
3.67563248e-01 -4.34228688e-01 -4.27572787e-01 -1.03648312e-01
-4.46033508e-01 -8.87067243e-02 6.59174979e-01 -3.25443029e-01
-7.04007924e-01 3.30286354e-01 -1.28607273e-01 -5.72131574e-01
6.75034970e-02 8.36432055e-02 5.14147520e-01 -6.53067887e-01
9.23702538e-01 1.08704135e-01 1.69365495e-01 4.63988632e-02
-7.16215819e-02 5.22218525e-01 4.47567493e-01 2.74721920e-01
8.02192807e-01 6.46350861e-01 3.69740248e-01 -1.08668089e+00
-8.12796414e-01 -7.65285730e-01 -5.00190914e-01 -3.12846124e-01
8.35206509e-01 -8.34173322e-01 -4.40409571e-01 -4.39072028e-02
-8.22931767e-01 -8.69698524e-02 -5.94342589e-01 7.46653438e-01
-2.35869870e-01 1.54120654e-01 -7.46490657e-01 -1.72083229e-01
-6.09231114e-01 -1.14751387e+00 6.83460891e-01 3.78701895e-01
-2.67673761e-01 -9.52762187e-01 -2.82287508e-01 -6.37419522e-03
8.65115523e-01 6.31294101e-02 8.26300383e-01 -3.71782601e-01
-6.83901966e-01 -3.30661207e-01 -7.17256486e-01 2.75896311e-01
7.53256559e-01 -4.11669927e-04 -7.20779896e-01 -1.23605743e-01
-1.50944352e-01 1.93496212e-01 8.57814550e-01 8.06155622e-01
1.28215969e+00 -4.91267778e-02 -8.87335300e-01 6.24751091e-01
1.54351473e+00 3.19987446e-01 5.66022158e-01 -9.66120809e-02
6.70914292e-01 -5.87520115e-02 2.06685185e-01 1.48562029e-01
-2.90055871e-01 3.14081390e-03 5.70253313e-01 -3.19332391e-01
-5.54831028e-01 -1.81652591e-01 -2.47985840e-01 5.32193244e-01
-2.25366294e-01 -1.32184297e-01 -1.01836872e+00 7.15612352e-01
-1.39197063e+00 -8.01587999e-01 -3.26410383e-01 1.98195863e+00
4.63275015e-01 1.15035631e-01 -1.52767673e-01 8.07653964e-02
8.24137390e-01 -6.45048395e-02 -3.32303673e-01 1.13956280e-01
4.55386609e-01 3.38076860e-01 1.00163174e+00 3.20337683e-01
-1.11664593e+00 4.41303819e-01 7.12962914e+00 7.72129595e-01
-1.24512887e+00 8.14089477e-02 9.50077593e-01 2.62463801e-02
-9.37948674e-02 -4.59603578e-01 -8.51459265e-01 -3.84364016e-02
6.06234431e-01 -8.71538147e-02 -1.94131106e-01 9.36382890e-01
1.17413394e-01 -1.69592589e-01 -8.39879692e-01 7.47641087e-01
-1.49410322e-01 -1.66270888e+00 1.10002615e-01 2.38057613e-01
7.77620852e-01 2.27741465e-01 1.54438585e-01 8.37421604e-03
2.19445229e-01 -1.16904056e+00 -2.17592612e-01 3.25779974e-01
1.20474994e+00 -7.77126551e-01 1.06618989e+00 2.63424724e-01
-1.43375313e+00 1.91009134e-01 -4.78175253e-01 3.19535941e-01
-1.68135017e-01 7.16196954e-01 -1.76829851e+00 1.19475976e-01
6.47507668e-01 6.99001789e-01 -5.29862106e-01 1.39779735e+00
-1.94137946e-01 6.11305892e-01 -5.00145376e-01 -4.77668703e-01
3.67459238e-01 1.97617605e-01 3.82463872e-01 1.18407393e+00
6.90033674e-01 -3.10960673e-02 4.29146141e-02 8.84920537e-01
-2.41068169e-01 7.69715104e-03 -7.17741013e-01 3.47821079e-02
2.30626509e-01 1.53019929e+00 -1.17625415e+00 -2.02746615e-01
-4.17411268e-01 6.93298638e-01 1.47837652e-02 -1.86845481e-01
-8.92742276e-01 -4.06398058e-01 1.57834530e-01 5.77376604e-01
4.66286719e-01 6.95891157e-02 -1.20959856e-01 -6.09004736e-01
-3.75206441e-01 -3.33006471e-01 3.03974420e-01 -4.61017877e-01
-9.25230205e-01 7.58547902e-01 -5.26382327e-01 -1.53022158e+00
-1.83180287e-01 -7.93967843e-01 -7.66397476e-01 7.64275491e-01
-1.52815044e+00 -9.21779752e-01 -6.64280474e-01 4.52165693e-01
4.56277639e-01 5.90251461e-02 8.03557694e-01 1.52982026e-01
-3.55430990e-01 4.45690036e-01 7.09405765e-02 3.50958914e-01
2.59872913e-01 -1.28815067e+00 2.61942655e-01 6.29412591e-01
-2.87273347e-01 2.06725076e-01 2.69766878e-02 -7.49818623e-01
-8.92382801e-01 -1.73039377e+00 6.90555751e-01 1.48435429e-01
4.71208125e-01 2.87776828e-01 -6.73545957e-01 5.59809566e-01
-1.10851288e-01 5.96336007e-01 5.42670310e-01 -5.07469296e-01
8.82697850e-02 -1.65830880e-01 -1.49660945e+00 6.72307909e-01
6.56662405e-01 -7.42100403e-02 1.04746094e-03 4.06598866e-01
4.42135066e-01 -5.13354838e-01 -9.57964480e-01 5.67942142e-01
4.49518472e-01 -8.01631808e-01 8.43537927e-01 -5.28147221e-02
4.40199733e-01 -2.09530875e-01 2.72064120e-01 -1.03233099e+00
-9.05922472e-01 -1.33773744e-01 1.33065730e-01 3.32165360e-01
7.16418922e-01 -6.01811409e-01 1.35664368e+00 -1.87245429e-01
-1.13250196e-01 -1.17990053e+00 -8.22285652e-01 -4.99269217e-01
-1.88124225e-01 -4.46696989e-02 1.14554875e-01 4.13919985e-01
-3.31719577e-01 1.24008752e-01 1.43905178e-01 1.05869353e-01
4.47355807e-01 -2.13380709e-01 6.77515417e-02 -1.19534206e+00
-2.02501938e-01 -5.69714904e-01 -5.99060714e-01 -9.50260818e-01
-7.50869513e-01 -8.99566829e-01 -3.48719358e-02 -1.78845179e+00
2.06642374e-01 -3.43608648e-01 -2.42046207e-01 3.65258485e-01
-2.70244642e-03 4.88113880e-01 -3.49174112e-01 1.47198632e-01
-1.34428158e-01 1.50986210e-01 1.86028600e+00 -1.18361615e-01
1.76906046e-02 5.47077119e-01 -6.45625234e-01 8.33937407e-01
9.96010244e-01 -4.49407488e-01 -4.01018411e-01 -1.50638372e-01
-1.74387455e-01 4.04824525e-01 3.27339798e-01 -1.33295953e+00
2.40096658e-01 -6.82825744e-02 8.82897079e-01 -8.82555544e-01
1.83908403e-01 -9.77073908e-01 2.24532951e-02 1.37430608e+00
-3.26818787e-02 -1.55238479e-01 2.45424673e-01 6.46812975e-01
-2.03688815e-01 -2.64271945e-01 1.17727423e+00 -6.29806936e-01
-3.95422250e-01 5.32716393e-01 -6.72094643e-01 -3.25097352e-01
1.33234954e+00 -2.94794470e-01 -1.04824118e-01 -2.19088078e-01
-5.60729563e-01 6.17587715e-02 1.46411895e-03 -1.18527748e-01
8.97835314e-01 -1.25960433e+00 -6.73546076e-01 2.86570400e-01
-2.42099866e-01 2.91886300e-01 1.42292097e-01 1.04749966e+00
-1.13189471e+00 7.22271204e-01 -1.03086144e-01 -6.69552505e-01
-1.36194921e+00 -6.58068582e-02 1.09444571e+00 -8.06185961e-01
-7.80748963e-01 1.14853549e+00 3.84700112e-02 -7.49814063e-02
1.65711775e-01 -6.67459309e-01 -3.51174623e-01 -4.76217926e-01
3.72425884e-01 5.12294590e-01 2.05158040e-01 -1.01946630e-01
-2.59512156e-01 3.80485535e-01 -4.82545316e-01 2.18162805e-01
8.60552728e-01 4.26972151e-01 -6.43993914e-02 1.86992027e-02
1.03066885e+00 -1.60542026e-01 -8.69306624e-01 -2.41476536e-01
-3.96286070e-01 -2.10954830e-01 4.13593173e-01 -5.82171857e-01
-1.48825359e+00 8.26295972e-01 7.44475245e-01 1.66757554e-01
8.51792037e-01 4.83341441e-02 7.46675611e-01 5.22491395e-01
6.49314374e-02 -3.81241977e-01 2.06012115e-01 3.42512518e-01
5.61430335e-01 -1.06320834e+00 2.59415507e-01 -8.30292642e-01
-2.83816725e-01 1.41186392e+00 9.43302691e-01 -4.56108004e-01
9.03945625e-01 4.23999995e-01 -7.51090944e-02 -2.19988361e-01
-6.39189482e-01 -1.90951273e-01 3.20404679e-01 4.68372136e-01
7.14834213e-01 3.20430338e-01 -8.17031637e-02 2.58570015e-01
-4.32834812e-02 -7.45775253e-02 4.86000538e-01 9.66749251e-01
-7.88320124e-01 -6.86123133e-01 -2.58049458e-01 1.28671288e+00
-6.66152298e-01 -9.42047536e-02 -2.23208845e-01 1.08824110e+00
6.14153743e-02 3.59951884e-01 4.07063037e-01 -2.41677180e-01
2.10025087e-01 -3.31618845e-01 4.37419713e-01 -8.86000156e-01
-7.02354670e-01 2.22773030e-01 -1.36237666e-01 -2.83739865e-01
-2.06547007e-01 -3.88591915e-01 -1.48888302e+00 -3.60946506e-02
-6.09821856e-01 4.38353866e-02 3.02973270e-01 5.08047581e-01
1.10414758e-01 1.29926026e+00 5.40644526e-01 -3.18395615e-01
-6.61545753e-01 -8.69309545e-01 -6.47050500e-01 -4.02983576e-01
3.40960860e-01 -3.32213402e-01 -3.30326617e-01 -2.32405454e-01] | [15.441780090332031, -2.1471214294433594] |
2057a0c4-5fbd-4af7-9ce6-ab6c3cc2f23e | lexicon-based-graph-convolutional-network-for | null | null | https://aclanthology.org/2021.findings-emnlp.248 | https://aclanthology.org/2021.findings-emnlp.248.pdf | Lexicon-Based Graph Convolutional Network for Chinese Word Segmentation | Precise information of word boundary can alleviate the problem of lexical ambiguity to improve the performance of natural language processing (NLP) tasks. Thus, Chinese word segmentation (CWS) is a fundamental task in NLP. Due to the development of pre-trained language models (PLM), pre-trained knowledge can help neural methods solve the main problems of the CWS in significant measure. Existing methods have already achieved high performance on several benchmarks (e.g., Bakeoff-2005). However, recent outstanding studies are limited by the small-scale annotated corpus. To further improve the performance of CWS methods based on fine-tuning the PLMs, we propose a novel neural framework, LBGCN, which incorporates a lexicon-based graph convolutional network into the Transformer encoder. Experimental results on five benchmarks and four cross-domain datasets show the lexicon-based graph convolutional network successfully captures the information of candidate words and helps to improve performance on the benchmarks (Bakeoff-2005 and CTB6) and the cross-domain datasets (SIGHAN-2010). Further experiments and analyses demonstrate that our proposed framework effectively models the lexicon to enhance the ability of basic neural frameworks and strengthens the robustness in the cross-domain scenario. | ['Degen Huang', 'Jingxiang Cao', 'Wei Liu', 'Junpeng Liu', 'Hao Yu', 'Kaiyu Huang'] | null | null | null | null | findings-emnlp-2021-11 | ['chinese-word-segmentation'] | ['natural-language-processing'] | [-2.33783070e-02 -1.71308368e-01 -3.17319781e-01 -3.53550106e-01
-7.40409911e-01 -5.78121960e-01 2.56476253e-01 1.33155473e-02
-6.20610535e-01 6.76568031e-01 1.53538167e-01 -5.59799373e-01
1.73632368e-01 -9.45784807e-01 -5.83708346e-01 -3.36254895e-01
2.43535578e-01 2.72368312e-01 6.23686433e-01 -4.98846471e-01
1.59732044e-01 2.72620898e-02 -9.72437739e-01 3.42713118e-01
1.42132521e+00 9.34047163e-01 6.37659729e-01 1.15168042e-01
-8.20394814e-01 4.18174177e-01 -6.77986562e-01 -4.95022893e-01
1.28682539e-01 -3.54875296e-01 -8.67442429e-01 -2.82107711e-01
-1.85089841e-01 1.53987063e-02 -1.92793664e-02 1.26165628e+00
4.87548709e-01 1.36175156e-01 2.92395413e-01 -7.16109633e-01
-1.02333069e+00 9.98704135e-01 -4.80050057e-01 1.02186806e-01
1.23956628e-01 4.49315049e-02 1.17875135e+00 -8.73817205e-01
6.24953151e-01 1.34734809e+00 6.60182834e-01 6.00116670e-01
-6.15574419e-01 -9.10054028e-01 5.59031427e-01 2.64991015e-01
-1.39496505e+00 -9.86147951e-03 6.81642592e-01 -5.02184741e-02
1.06460810e+00 -1.12662107e-01 4.81300086e-01 8.51533949e-01
-1.92177340e-01 1.26052105e+00 8.16359758e-01 -4.59197760e-01
-9.51664075e-02 -2.34874666e-01 4.62477207e-01 5.61408401e-01
4.50404584e-01 -3.54651034e-01 -1.84262559e-01 4.40229446e-01
7.29438424e-01 -3.90830934e-01 -3.74543786e-01 1.24418557e-01
-9.71060276e-01 9.82370019e-01 4.96788234e-01 6.63260937e-01
-8.49387608e-03 -2.66300559e-01 6.10771894e-01 2.01408770e-02
5.35967767e-01 5.45022845e-01 -7.68988311e-01 2.55045872e-02
-7.76421010e-01 4.74492460e-02 5.82506597e-01 1.22464573e+00
6.09338403e-01 2.07159504e-01 -3.15564483e-01 9.85392153e-01
3.04545075e-01 3.21036309e-01 8.72262597e-01 -2.51965791e-01
9.05148029e-01 1.07559419e+00 -1.83490202e-01 -8.82132888e-01
-4.72046018e-01 -4.70860630e-01 -8.17202985e-01 -6.74561977e-01
3.92953157e-01 -3.26499820e-01 -1.12842357e+00 1.72542512e+00
1.26248196e-01 2.19856992e-01 1.80469424e-01 9.84048128e-01
1.15413702e+00 1.04275846e+00 3.67975622e-01 -4.56435010e-02
1.38299417e+00 -1.10792768e+00 -9.49321389e-01 -5.69203615e-01
7.53653586e-01 -8.34953070e-01 1.40666008e+00 2.15093359e-01
-7.65142977e-01 -7.24096417e-01 -1.11219013e+00 -2.36660004e-01
-5.85717797e-01 4.74573910e-01 6.85113788e-01 5.35919428e-01
-8.00828636e-01 2.03224123e-01 -6.16450548e-01 -2.45868921e-01
4.41327125e-01 3.48860294e-01 -1.42651334e-01 -3.11260283e-01
-1.85264599e+00 6.90253377e-01 9.54838455e-01 4.23008949e-01
-4.63248491e-01 -5.58131039e-01 -1.02655864e+00 9.38672870e-02
5.36955357e-01 -3.50782841e-01 1.03831923e+00 -9.16687608e-01
-1.37503195e+00 7.42846251e-01 -1.40414843e-02 -5.24675131e-01
2.65110910e-01 -3.80868584e-01 -6.15863323e-01 6.79229796e-02
3.70040983e-01 7.54624307e-01 4.00964618e-01 -9.06715512e-01
-7.20709026e-01 -1.13641560e-01 -7.70723540e-03 2.26831719e-01
-3.70012313e-01 1.15940727e-01 -9.65370774e-01 -8.18360448e-01
3.43177803e-02 -6.15171254e-01 -2.44019166e-01 -7.41477966e-01
-5.25912285e-01 -5.71522415e-01 6.63135111e-01 -8.70777428e-01
1.49432707e+00 -2.18503523e+00 -6.62473887e-02 -1.58353597e-01
-1.97883055e-01 9.40389335e-01 -3.09750378e-01 2.45757848e-01
1.50245652e-01 4.61212188e-01 -3.40122938e-01 -1.51597053e-01
-1.89034179e-01 4.08482701e-01 -2.53975958e-01 7.94981346e-02
6.42534792e-01 1.25141120e+00 -8.05439115e-01 -5.97924411e-01
6.83304444e-02 2.12535635e-01 -3.42502892e-01 3.08383167e-01
-5.66397429e-01 3.30448568e-01 -5.94283819e-01 5.16185045e-01
9.06620026e-01 -2.07059458e-01 2.88786173e-01 -3.22365649e-02
-7.68520907e-02 5.45174241e-01 -1.24691463e+00 1.89657736e+00
-4.24200565e-01 3.34647924e-01 -5.92347421e-02 -1.14508522e+00
1.05400443e+00 1.57434285e-01 6.72080228e-03 -9.52264965e-01
2.47036621e-01 4.12707806e-01 9.00347754e-02 -4.32267249e-01
5.91699660e-01 9.31373313e-02 -3.38594347e-01 1.03664123e-01
1.04056381e-01 2.48722527e-02 4.37512279e-01 1.02964848e-01
6.63207471e-01 1.17621824e-01 3.34257662e-01 -4.08217251e-01
8.90580475e-01 3.13475758e-01 8.46706450e-01 4.08718079e-01
-2.78518736e-01 6.15911365e-01 4.57598418e-01 -2.64636457e-01
-6.28358960e-01 -7.12316334e-01 -8.92947838e-02 1.22461092e+00
4.73690063e-01 -3.79975110e-01 -1.01771986e+00 -7.96993315e-01
-2.16086581e-01 6.98021591e-01 -4.31390136e-01 -2.21213102e-01
-9.49437380e-01 -9.25976932e-01 7.99331367e-01 7.83920527e-01
9.25446630e-01 -1.22541451e+00 -1.04878228e-02 4.09489095e-01
-4.08030242e-01 -1.46331704e+00 -5.42797863e-01 9.15233269e-02
-7.34499931e-01 -1.02224338e+00 -6.84389651e-01 -1.41341317e+00
5.32290041e-01 3.09624195e-01 1.00406444e+00 2.84254640e-01
2.44279136e-03 -2.26098835e-01 -7.29283750e-01 -3.26184243e-01
-2.65006185e-01 5.78045785e-01 -4.00785714e-01 -2.35404998e-01
6.54194355e-01 4.30816077e-02 -4.76087123e-01 2.79449672e-01
-9.07461882e-01 1.63917422e-01 6.33093655e-01 9.55511808e-01
5.44629991e-01 6.45598918e-02 8.77526164e-01 -1.10889757e+00
8.10549617e-01 -3.70034754e-01 -8.11863005e-01 4.08035219e-01
-5.61874270e-01 9.70084742e-02 7.75298595e-01 -4.21667695e-01
-1.34600163e+00 -2.12222338e-01 -3.31162244e-01 1.21530950e-01
1.31754838e-02 8.25808883e-01 -6.89399481e-01 2.26609528e-01
1.43291354e-01 3.26648891e-01 -4.34818864e-01 -6.64871454e-01
3.89195830e-01 7.57504463e-01 4.56432164e-01 -6.62015378e-01
4.00194198e-01 -8.80185366e-02 -5.68571031e-01 -6.52465165e-01
-1.03685987e+00 -5.92282891e-01 -6.79167390e-01 2.32111067e-01
1.06645763e+00 -1.18892539e+00 -4.06450689e-01 6.10926032e-01
-1.46909881e+00 -2.51065344e-01 2.32030749e-01 2.64718294e-01
1.50075793e-01 4.83019143e-01 -7.98708498e-01 -5.23020267e-01
-5.05007803e-01 -1.10796714e+00 1.03423512e+00 5.39675415e-01
1.42359689e-01 -1.02552962e+00 -3.37342352e-01 3.55178446e-01
3.15523326e-01 -4.51543331e-02 1.14085448e+00 -8.99118721e-01
-5.98104060e-01 3.79687361e-02 -5.49620926e-01 4.30456221e-01
-4.64946451e-03 -2.23278269e-01 -8.13672900e-01 -4.53439988e-02
-1.67347163e-01 -1.42021671e-01 1.03378952e+00 3.64971489e-01
1.02114284e+00 1.74943488e-02 -3.52881640e-01 5.54597497e-01
1.49081349e+00 3.98072451e-01 5.59514642e-01 4.57152009e-01
8.44252050e-01 4.70209360e-01 8.46155941e-01 -3.82223679e-03
5.73001862e-01 3.12806070e-01 3.09343755e-01 -2.13603705e-01
-2.75412500e-01 -3.17609996e-01 2.87701041e-01 1.33082318e+00
2.21035331e-01 -4.29260403e-01 -1.03346670e+00 8.74240875e-01
-1.81577110e+00 -3.06870908e-01 -3.39920223e-01 1.67536008e+00
1.17856753e+00 3.94055009e-01 -3.48616272e-01 -1.74028613e-02
9.59586620e-01 3.04508060e-01 -4.34752196e-01 -3.31548423e-01
-4.16062802e-01 3.89946431e-01 3.85811687e-01 5.30630529e-01
-1.26138556e+00 1.81147408e+00 5.28458261e+00 1.25814652e+00
-1.12292194e+00 6.31258935e-02 5.86502194e-01 6.00229859e-01
-3.20477039e-01 -1.92922413e-01 -1.32634604e+00 5.02820134e-01
5.99561632e-01 -1.95306823e-01 3.26737106e-01 8.84867251e-01
-4.90045100e-02 2.90477034e-02 -7.30109930e-01 6.99672997e-01
-3.22358161e-02 -1.25346947e+00 1.47740051e-01 -3.90490383e-01
8.41345906e-01 1.46353036e-01 -7.82758147e-02 7.78530955e-01
3.29554737e-01 -9.65057135e-01 4.59567785e-01 -1.57776669e-01
9.11492050e-01 -8.38577628e-01 1.05836928e+00 4.51002359e-01
-1.48556685e+00 1.97075799e-01 -6.65897310e-01 4.70021321e-03
1.41246662e-01 5.84024072e-01 -8.75292301e-01 7.40772188e-01
5.20819664e-01 8.28072131e-01 -6.48696899e-01 7.58244574e-01
-7.81985581e-01 9.67803955e-01 -1.83896199e-01 -3.13099295e-01
6.89390242e-01 -2.74667680e-01 1.38635710e-01 1.55899143e+00
3.45887691e-02 1.17058463e-01 1.72873378e-01 8.44229221e-01
-2.27254286e-01 4.76171970e-01 -1.88206300e-01 -3.06729645e-01
4.60236013e-01 1.26095247e+00 -9.71007884e-01 -2.77337223e-01
-5.55547416e-01 7.54662156e-01 5.07676363e-01 3.93347323e-01
-7.90225625e-01 -5.86851180e-01 5.53556025e-01 -2.72575557e-01
5.60921431e-01 -3.23859274e-01 -5.30207038e-01 -1.23948300e+00
-3.63676660e-02 -9.14893329e-01 4.41966712e-01 -3.99708837e-01
-1.25194609e+00 7.46018231e-01 -2.72616059e-01 -9.47452545e-01
3.20842825e-02 -9.20232713e-01 -6.02566898e-01 1.01968789e+00
-1.93294680e+00 -1.29218650e+00 -6.61205202e-02 4.97462302e-01
7.88192511e-01 -1.17672861e-01 7.53707051e-01 3.46668333e-01
-8.73926878e-01 6.05609596e-01 3.86037794e-03 7.99481213e-01
6.75289333e-01 -1.21713829e+00 7.87533760e-01 1.20523512e+00
9.90442559e-02 8.20451200e-01 9.76290777e-02 -8.57131720e-01
-1.19026589e+00 -1.22690213e+00 9.67755258e-01 4.62704785e-02
6.22145355e-01 -5.73795199e-01 -1.16741467e+00 5.76386333e-01
3.02738190e-01 -1.10304736e-01 6.02463841e-01 5.68016954e-02
-8.91401023e-02 -6.60421653e-03 -6.95301235e-01 5.43120325e-01
9.41853702e-01 -3.77151698e-01 -9.44999099e-01 1.65043578e-01
1.29708672e+00 -5.74432015e-01 -5.05698979e-01 4.52157855e-01
6.52093440e-02 -6.14779890e-01 7.49570429e-01 -6.71166539e-01
3.57164055e-01 -2.61179894e-01 4.92510460e-02 -1.32415438e+00
-5.55299595e-02 -2.83377320e-01 3.25012863e-01 1.54090703e+00
5.64610898e-01 -5.06828308e-01 5.95943511e-01 3.34922791e-01
-3.44310284e-01 -7.86034346e-01 -7.67412841e-01 -6.68382704e-01
2.82526761e-01 -7.18679130e-01 9.25584316e-01 9.70506847e-01
-1.32380828e-01 5.01111805e-01 1.93568468e-02 1.61874875e-01
8.38832092e-03 2.37204388e-01 3.78307551e-01 -1.06296337e+00
5.33458479e-02 -5.23400784e-01 -5.64946681e-02 -1.50012839e+00
4.73495185e-01 -9.86220479e-01 1.19814217e-01 -1.72694540e+00
-5.86266592e-02 -5.13260603e-01 -3.99495989e-01 5.70695996e-01
-7.50131786e-01 1.09683524e-03 1.65337458e-01 -9.78098139e-02
-7.52120972e-01 5.56553781e-01 1.56320262e+00 -9.08081904e-02
-1.36549070e-01 -2.10859999e-01 -8.34399879e-01 6.40125215e-01
7.90601790e-01 -2.90654391e-01 -2.49365076e-01 -9.28154230e-01
2.53155231e-01 -3.15847546e-01 -2.67815411e-01 -5.63924134e-01
3.20567846e-01 -1.62555054e-01 2.84085155e-01 -7.01987028e-01
-2.28146330e-01 -5.33302307e-01 -6.19405329e-01 4.70608234e-01
-1.39219195e-01 3.20729241e-03 4.87968892e-01 4.68029082e-01
-5.67008853e-01 -3.39399129e-01 6.02572441e-01 -2.53857464e-01
-1.29075229e+00 2.21946016e-01 -1.03361465e-01 6.91120923e-01
9.23306406e-01 -1.47430757e-02 -3.45855296e-01 -3.17478590e-02
-1.75037026e-01 6.20734394e-01 1.98574532e-02 6.60333276e-01
5.30058444e-01 -1.17656994e+00 -5.37721276e-01 3.75487626e-01
-2.61052735e-02 6.70265853e-01 -2.93233944e-03 4.81702119e-01
-7.96578944e-01 6.35404706e-01 -8.58468190e-02 -4.61969048e-01
-7.29891598e-01 4.33310121e-01 2.17027023e-01 -5.97144604e-01
-3.12059283e-01 1.15786481e+00 3.87736410e-01 -5.49124897e-01
1.92774966e-01 -7.04523027e-01 -5.98257542e-01 1.06559679e-01
3.68821174e-01 9.31097418e-02 6.51495308e-02 -6.15234375e-01
-4.68147814e-01 5.57811141e-01 -3.00180763e-01 2.79402196e-01
1.32716131e+00 -1.97223634e-01 -1.54955089e-01 1.47487476e-01
1.05598307e+00 5.05132750e-02 -9.67544436e-01 -4.02010411e-01
4.92885351e-01 9.61457007e-03 -3.68046761e-02 -8.46910715e-01
-1.16600835e+00 1.24050462e+00 1.15363590e-01 -1.10069461e-01
1.17788780e+00 -3.20900589e-01 1.38916123e+00 2.17551053e-01
2.41634890e-01 -1.36684334e+00 -3.45433317e-02 9.44225609e-01
4.15070474e-01 -1.38708222e+00 -2.94892162e-01 -7.44576037e-01
-7.85870373e-01 1.17493379e+00 8.10031235e-01 -9.65043306e-02
4.84086365e-01 2.54915476e-01 2.58295536e-01 4.85436283e-02
-2.42437780e-01 -4.80822235e-01 4.58623737e-01 4.47051167e-01
5.80472648e-01 1.37274295e-01 -7.41195142e-01 1.41648674e+00
-5.97411320e-02 -2.25953951e-01 8.95400718e-02 5.54664493e-01
-5.09654820e-01 -1.31179523e+00 -2.72448547e-02 1.37755245e-01
-5.62436223e-01 -5.54165900e-01 -4.29956198e-01 1.08870447e+00
3.54316562e-01 9.08434451e-01 -1.36255160e-01 -1.53345466e-01
2.73503751e-01 9.12267417e-02 1.22455508e-02 -8.18383098e-01
-7.31887519e-01 2.29827419e-01 2.73503028e-02 -3.76731336e-01
-3.70221734e-01 -3.20114970e-01 -1.66200984e+00 7.13675693e-02
-5.64748585e-01 2.54805893e-01 3.80346984e-01 1.11865449e+00
2.31274143e-01 5.84761500e-01 1.98359862e-01 -1.57593980e-01
-3.62449586e-01 -1.03684235e+00 -5.10214269e-01 4.78023171e-01
-1.43403560e-01 -4.68761325e-01 -2.18503643e-02 -1.36039272e-01] | [9.96081256866455, 10.029912948608398] |
6beab1d3-47c1-4eac-bd55-d1ed2840bb35 | hier-metric-learning-beyond-class-labels-via | 2212.14258 | null | https://arxiv.org/abs/2212.14258v3 | https://arxiv.org/pdf/2212.14258v3.pdf | HIER: Metric Learning Beyond Class Labels via Hierarchical Regularization | Supervision for metric learning has long been given in the form of equivalence between human-labeled classes. Although this type of supervision has been a basis of metric learning for decades, we argue that it hinders further advances in the field. In this regard, we propose a new regularization method, dubbed HIER, to discover the latent semantic hierarchy of training data, and to deploy the hierarchy to provide richer and more fine-grained supervision than inter-class separability induced by common metric learning losses.HIER achieves this goal with no annotation for the semantic hierarchy but by learning hierarchical proxies in hyperbolic spaces. The hierarchical proxies are learnable parameters, and each of them is trained to serve as an ancestor of a group of data or other proxies to approximate the semantic hierarchy among them. HIER deals with the proxies along with data in hyperbolic space since the geometric properties of the space are well-suited to represent their hierarchical structure. The efficacy of HIER is evaluated on four standard benchmarks, where it consistently improved the performance of conventional methods when integrated with them, and consequently achieved the best records, surpassing even the existing hyperbolic metric learning technique, in almost all settings. | ['Boseung Jeong', 'Suha Kwak', 'Sungyeon Kim'] | 2022-12-29 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Kim_HIER_Metric_Learning_Beyond_Class_Labels_via_Hierarchical_Regularization_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Kim_HIER_Metric_Learning_Beyond_Class_Labels_via_Hierarchical_Regularization_CVPR_2023_paper.pdf | cvpr-2023-1 | ['metric-learning', 'metric-learning'] | ['computer-vision', 'methodology'] | [-1.10480838e-01 2.93047190e-01 -2.75310844e-01 -5.31848431e-01
-8.15564752e-01 -5.50527990e-01 5.73586285e-01 4.54716235e-01
-3.58070165e-01 5.93278825e-01 4.55583483e-01 -1.33361015e-02
-5.10375321e-01 -7.64495730e-01 -4.86607909e-01 -9.20870543e-01
-1.00190386e-01 7.45897889e-01 2.80699819e-01 -1.47348627e-01
2.02807814e-01 4.14771050e-01 -1.64039218e+00 2.07024217e-01
8.85702968e-01 1.15591681e+00 -1.91977113e-01 2.13845909e-01
3.16948533e-01 9.48403299e-01 -2.54003942e-01 -5.47550261e-01
3.21643949e-01 -4.19205844e-01 -9.78893518e-01 5.45677170e-02
4.34165955e-01 -2.09163368e-01 -4.75569367e-01 8.86980236e-01
1.49812475e-01 1.59399554e-01 9.92783010e-01 -1.51204586e+00
-6.83807135e-01 6.13579690e-01 -4.02755350e-01 -1.79836050e-01
1.33581638e-01 -4.36435997e-01 1.76967967e+00 -8.78397107e-01
4.93012607e-01 1.15219653e+00 8.97154093e-01 3.06896925e-01
-1.38463664e+00 -3.23760599e-01 -3.37011665e-02 4.43140239e-01
-1.45415580e+00 -1.49661690e-01 8.81945133e-01 -5.29011428e-01
3.00097674e-01 8.53761137e-02 3.93235445e-01 1.02742791e+00
-1.98270410e-01 8.07076931e-01 1.17667389e+00 -3.39731216e-01
5.16472638e-01 2.76400864e-01 3.81452531e-01 1.00687921e+00
1.15184195e-01 5.16885221e-02 -5.77153742e-01 -3.03667605e-01
4.69734460e-01 -7.79801682e-02 -1.48449734e-01 -9.31247771e-01
-9.82063890e-01 1.23643649e+00 7.51583338e-01 1.89332977e-01
-5.64666884e-03 -4.65842225e-02 3.44083786e-01 2.47864142e-01
4.95754242e-01 6.93172395e-01 -2.66870260e-01 1.97981998e-01
-1.07528675e+00 2.27982968e-01 8.28792393e-01 9.46435452e-01
1.00083828e+00 -4.73793685e-01 -1.55631542e-01 7.45163798e-01
1.59277543e-01 -2.88636815e-02 3.24987322e-01 -1.25485492e+00
3.58099222e-01 1.00823700e+00 5.45200557e-02 -1.10980392e+00
-6.33868277e-01 -2.99237609e-01 -7.00039446e-01 2.20591769e-01
8.08111548e-01 3.28464836e-01 -5.71500063e-01 2.03621268e+00
4.78448153e-01 5.99707700e-02 -1.55297801e-01 7.76496887e-01
1.71550095e-01 3.99963707e-01 -1.67345375e-01 2.12983459e-01
1.17136896e+00 -9.67603207e-01 -2.19630569e-01 1.88043907e-01
9.18517172e-01 -3.65345448e-01 1.48712122e+00 4.18476522e-01
-7.36827433e-01 -2.89793342e-01 -1.41702497e+00 -3.61401677e-01
-5.18913448e-01 -5.80196120e-02 6.46572948e-01 5.07392108e-01
-1.12339950e+00 8.82427216e-01 -7.72567630e-01 -2.22331613e-01
4.81586069e-01 1.12817287e-01 -3.35380822e-01 -2.01762497e-01
-1.22470951e+00 8.15214217e-01 2.86105573e-01 -3.07203621e-01
-8.21033716e-01 -9.99522090e-01 -9.74130154e-01 -4.66672964e-02
1.58100992e-01 -5.28327525e-01 1.01547766e+00 -3.46741647e-01
-1.08380842e+00 1.13019431e+00 2.63526142e-01 -5.16650796e-01
8.87050509e-01 -5.92490193e-03 -1.60878280e-03 2.26105943e-01
2.55692422e-01 6.86784863e-01 7.87116766e-01 -1.25684893e+00
-6.65989995e-01 -5.56623995e-01 3.70501161e-01 1.83064163e-01
-7.46752918e-01 -1.09231234e-01 -7.85752833e-02 -7.03578770e-01
4.08664644e-01 -1.08005047e+00 -9.30150151e-02 4.18870479e-01
-4.83764261e-01 -5.36555767e-01 7.98749208e-01 -4.60163981e-01
1.24098122e+00 -2.25705910e+00 3.26842278e-01 1.84785366e-01
4.53330934e-01 -2.61582106e-01 1.86317340e-01 4.35637385e-01
1.15999848e-01 -6.69122860e-02 -6.23905063e-01 -6.18197978e-01
5.28003216e-01 3.11269403e-01 -2.84064740e-01 9.55502212e-01
-2.44137064e-01 5.06859303e-01 -1.04882061e+00 -5.31245708e-01
4.15566787e-02 3.95122141e-01 -6.44226015e-01 3.75639886e-01
-1.89533979e-01 2.61559486e-01 -2.41672203e-01 4.26984429e-01
4.66001898e-01 -4.26204085e-01 -1.39129356e-01 -2.04422593e-01
2.08594486e-01 3.60040605e-01 -1.23221982e+00 1.69201970e+00
-3.54998529e-01 4.50227380e-01 -2.03669235e-01 -1.24581194e+00
1.12677348e+00 2.39280000e-01 8.64418149e-01 -4.90941793e-01
-6.06165081e-02 3.65022421e-01 -3.42573136e-01 -2.64264017e-01
2.18203485e-01 -1.60492912e-01 -3.17793101e-01 6.01909399e-01
-4.63283956e-02 -1.06305234e-01 2.50939041e-01 2.01713443e-01
1.11138070e+00 -4.26651649e-02 2.17304230e-01 -5.39824724e-01
6.38074398e-01 -1.46923602e-01 6.59193397e-01 5.20906091e-01
-2.44010329e-01 7.57869840e-01 6.47220671e-01 -4.07994837e-01
-1.33507931e+00 -1.42232382e+00 -5.55032551e-01 1.39949346e+00
3.11602086e-01 -5.27206361e-01 -8.78376126e-01 -9.62201059e-01
2.16367424e-01 6.44471943e-01 -9.50363576e-01 -5.30486584e-01
-4.84702468e-01 -7.19369948e-01 5.98554552e-01 6.72807813e-01
5.07960498e-01 -7.89547741e-01 -5.18506944e-01 2.00673006e-02
-9.15544629e-02 -1.14379358e+00 -6.19373500e-01 5.02646983e-01
-6.18120372e-01 -1.60303450e+00 -5.04495025e-01 -7.59743631e-01
5.98276615e-01 3.30862310e-03 1.14925218e+00 -3.61572311e-04
-2.65531301e-01 2.06777081e-01 -4.33027834e-01 4.15242277e-02
-2.34190881e-01 2.23450661e-01 2.52154261e-01 2.48222440e-01
2.66418844e-01 -8.16208482e-01 -6.30453050e-01 6.61104381e-01
-8.46504986e-01 -2.55096316e-01 3.06910053e-02 9.08583045e-01
3.74744952e-01 1.67735875e-01 4.70401376e-01 -1.00560069e+00
3.18924010e-01 -5.38501918e-01 -5.54445624e-01 2.00508282e-01
-8.41508031e-01 3.13749045e-01 8.55508447e-01 -1.08813792e-01
-6.92578137e-01 -1.11195214e-01 1.87495008e-01 -1.77567631e-01
-4.54745628e-02 3.37966532e-01 -3.44178557e-01 -6.91907154e-03
8.17880452e-01 -2.70609587e-01 -2.85035759e-01 -7.23057687e-01
4.04476553e-01 5.76185942e-01 5.88027179e-01 -9.58293796e-01
7.90437877e-01 7.14969993e-01 1.80481583e-01 -5.77554882e-01
-1.55618238e+00 -4.84858811e-01 -9.89035904e-01 2.32451931e-01
9.72455919e-01 -5.49286604e-01 -3.91320676e-01 2.22063124e-01
-7.89678216e-01 -4.20979321e-01 -5.75922072e-01 4.31364685e-01
-8.93236578e-01 1.96411729e-01 -7.44198859e-01 -3.86234105e-01
2.87647784e-01 -1.04273367e+00 1.00506365e+00 -2.54426122e-01
-3.34914565e-01 -1.37541211e+00 2.00187430e-01 4.34701860e-01
1.07453531e-02 4.56255138e-01 1.30951440e+00 -7.77867854e-01
-2.31332392e-01 -1.57604501e-01 -2.93896943e-01 5.27517736e-01
-2.48227119e-02 -1.15124360e-01 -1.05943751e+00 -4.12369967e-01
1.45943806e-01 -5.72752357e-01 7.82212496e-01 7.15448335e-02
1.34642243e+00 -2.49706805e-01 -1.42779455e-01 1.01227081e+00
1.19195580e+00 -4.40050334e-01 2.26303205e-01 4.58708704e-01
8.05458009e-01 9.15797353e-01 3.71585816e-01 5.41804373e-01
5.73938310e-01 9.20146465e-01 5.48525929e-01 -1.03998169e-01
-8.94746035e-02 -5.22505820e-01 1.28802508e-01 6.49107993e-01
2.99673676e-01 1.65955320e-01 -1.03052664e+00 2.32863918e-01
-1.87269139e+00 -5.02720237e-01 -7.03872368e-02 2.23521423e+00
8.24495494e-01 1.06553443e-01 5.18836379e-01 6.32894576e-01
5.47655523e-01 1.82092384e-01 -6.51345134e-01 -8.46729130e-02
-2.21808091e-01 -1.64662033e-01 3.29709768e-01 5.48972309e-01
-1.37811017e+00 6.70953810e-01 6.31777620e+00 6.47438169e-01
-6.33095562e-01 -1.17221836e-03 6.82387769e-01 2.35268384e-01
-2.41350234e-01 -1.86393745e-02 -8.22762966e-01 4.71871495e-01
6.90999389e-01 -2.76542276e-01 3.89435649e-01 9.93835270e-01
-1.37165952e-02 2.72814363e-01 -1.79475951e+00 7.84819365e-01
8.07652026e-02 -1.01358688e+00 -1.51379341e-02 1.72533885e-01
8.35284173e-01 -1.32867560e-01 2.26586536e-01 2.74981737e-01
4.34083700e-01 -1.13144171e+00 8.03545058e-01 1.43390164e-01
6.83923244e-01 -7.83891261e-01 6.36358440e-01 4.31493968e-01
-1.24303484e+00 -2.50816345e-01 -6.30453706e-01 2.00398099e-02
-2.05053434e-01 4.99313563e-01 -5.88370740e-01 2.61598647e-01
6.14667773e-01 9.63195860e-01 -7.93434560e-01 1.07112384e+00
-3.38449270e-01 6.15016162e-01 -2.43022710e-01 4.27546680e-01
5.86546302e-01 -4.40745026e-01 3.35820824e-01 1.03766727e+00
9.53446925e-02 -2.42054790e-01 4.71593648e-01 7.40953863e-01
-3.03915411e-01 2.14487836e-01 -5.43453813e-01 2.88279325e-01
5.77117741e-01 1.05687952e+00 -6.67961597e-01 -1.00529410e-01
-4.12418365e-01 8.05961847e-01 7.90135503e-01 2.93587685e-01
-1.02499676e+00 -2.91478783e-01 5.99401951e-01 3.90453011e-01
1.34244030e-02 -1.77527055e-01 -5.00771344e-01 -9.14073527e-01
2.67140061e-01 -6.41166687e-01 1.00573397e+00 -3.81440669e-01
-1.53820932e+00 4.03288722e-01 6.49342909e-02 -1.26721215e+00
-1.20958230e-02 -7.16735601e-01 -2.97205955e-01 5.18072665e-01
-1.26049531e+00 -1.07787418e+00 -4.30761129e-01 8.52025151e-01
9.59036052e-02 -4.72906455e-02 9.55452383e-01 3.69268954e-01
-3.82810265e-01 8.87991428e-01 3.73156011e-01 1.91389650e-01
7.46434271e-01 -1.83343971e+00 -7.50503391e-02 4.62058097e-01
4.28910494e-01 1.88544139e-01 5.82526267e-01 -1.66675508e-01
-7.24966705e-01 -1.07421649e+00 8.69548142e-01 -6.81479573e-01
9.97918367e-01 -7.49352038e-01 -1.06672025e+00 5.54409087e-01
-4.52052087e-01 3.08942050e-01 7.64618039e-01 1.76607102e-01
-1.06018353e+00 -4.32997286e-01 -1.08622324e+00 4.74897772e-01
1.13943374e+00 -7.69342005e-01 -6.55063510e-01 5.21693766e-01
8.29938710e-01 -3.45525853e-02 -9.16859150e-01 4.79468107e-01
2.65222371e-01 -1.20115030e+00 1.02903545e+00 -7.16996431e-01
4.26015943e-01 -3.08608353e-01 -5.91046274e-01 -1.19463372e+00
-2.47167751e-01 -5.03154159e-01 -2.62275875e-01 1.15447283e+00
3.97379249e-01 -3.31595957e-01 1.24542654e+00 5.86597383e-01
-3.31214041e-01 -9.77009237e-01 -9.91754949e-01 -1.00267136e+00
4.59839910e-01 -1.11118630e-01 4.61382568e-01 1.05908000e+00
1.57677770e-01 3.36845845e-01 -3.25917840e-01 3.60266268e-02
1.05398524e+00 1.22002900e-01 3.28090757e-01 -1.47022343e+00
-1.96900710e-01 -5.92509627e-01 -8.12261403e-01 -9.41274762e-01
4.53117430e-01 -1.39926600e+00 -3.39143649e-02 -1.23839796e+00
1.84564456e-01 -8.19870770e-01 -5.51232457e-01 4.04933065e-01
-1.20728739e-01 3.88600767e-01 -9.10647362e-02 3.00748706e-01
-8.29600632e-01 8.02303076e-01 9.83000934e-01 -1.34705767e-01
-7.32925981e-02 1.97817951e-01 -7.84642160e-01 8.77603114e-01
5.96036136e-01 -5.68608880e-01 -6.54936552e-01 -2.09329829e-01
1.21486075e-01 -2.67429292e-01 4.35074449e-01 -1.21705401e+00
3.64683867e-01 1.54941037e-01 2.26913556e-01 -1.61317468e-01
2.04157144e-01 -9.20039058e-01 -2.41727367e-01 1.60085410e-01
-7.40709126e-01 -4.27871607e-02 -5.14504492e-01 5.44624209e-01
-3.25402588e-01 -1.91559434e-01 1.21655893e+00 2.71215141e-01
-3.42734098e-01 5.47463179e-01 2.28425160e-01 7.06589103e-01
1.08942461e+00 -1.91026121e-01 -6.88704848e-02 -1.79912522e-01
-8.68035138e-01 3.06001931e-01 7.34599411e-01 2.63268977e-01
2.26032838e-01 -1.50975716e+00 -7.03591704e-01 -1.45266484e-02
4.30622160e-01 1.04216367e-01 -2.94739842e-01 7.38888144e-01
-3.90560091e-01 1.73683599e-01 -3.88686880e-02 -6.91179097e-01
-9.43770587e-01 7.07675517e-01 3.62560630e-01 -3.75137538e-01
-9.94099200e-01 8.13684404e-01 3.17536473e-01 -5.85960269e-01
8.84373963e-01 -2.13795975e-01 -7.08068535e-02 2.49324694e-01
3.09738815e-01 7.65778363e-01 8.70273858e-02 -5.76791048e-01
-2.53310412e-01 5.37320912e-01 -2.38219928e-02 -2.56825518e-02
1.37817657e+00 -1.09675713e-01 -1.68848410e-01 8.24580371e-01
1.59664893e+00 -2.91234046e-01 -1.67936885e+00 -5.03176570e-01
7.71077096e-01 -3.50307018e-01 -2.65780240e-01 -4.40823793e-01
-7.10803211e-01 9.45248246e-01 2.03688636e-01 3.93291146e-01
7.60787487e-01 3.43348205e-01 5.82768559e-01 4.69702810e-01
4.19035971e-01 -1.21479404e+00 4.47731942e-01 5.52538097e-01
7.66066670e-01 -1.20567846e+00 -5.70933595e-02 -3.79358828e-01
-4.34606165e-01 9.28626359e-01 4.68139648e-01 -4.03106242e-01
8.04218531e-01 7.64107192e-03 -5.76683171e-02 -1.42754719e-01
-3.72434169e-01 2.34464407e-02 4.59943235e-01 5.88817775e-01
3.69747549e-01 -1.57314986e-02 3.73580456e-02 5.10306716e-01
-6.98723912e-01 -5.43319046e-01 2.07080692e-01 5.17982244e-01
-4.42025065e-01 -9.40517545e-01 -9.66252461e-02 3.17823142e-01
-1.13553211e-01 1.05542086e-01 -3.79060298e-01 8.95794988e-01
1.78958625e-01 9.16040361e-01 -9.66635868e-02 -3.00122499e-01
4.24277753e-01 -2.14999355e-02 3.69565189e-01 -6.59388125e-01
-1.62424415e-01 -4.57661688e-01 -1.02230594e-01 -7.10794270e-01
1.39394235e-02 -7.21666813e-01 -1.17108941e+00 -2.31471226e-01
1.67586818e-01 5.63888073e-01 3.54239553e-01 8.88036132e-01
-2.40231961e-01 7.70687833e-02 1.05738604e+00 -5.34910679e-01
-1.19387150e+00 -6.64149225e-01 -1.04294157e+00 8.92058313e-01
3.88687581e-01 -1.01515532e+00 -7.66769946e-01 -4.35381085e-02] | [9.24476146697998, 3.155038356781006] |
a46e0427-47e4-4ca9-a9d0-421e362cb84d | supervised-and-unsupervised-speech | 1709.05362 | null | http://arxiv.org/abs/1709.05362v1 | http://arxiv.org/pdf/1709.05362v1.pdf | Supervised and Unsupervised Speech Enhancement Using Nonnegative Matrix Factorization | Reducing the interference noise in a monaural noisy speech signal has been a
challenging task for many years. Compared to traditional unsupervised speech
enhancement methods, e.g., Wiener filtering, supervised approaches, such as
algorithms based on hidden Markov models (HMM), lead to higher-quality enhanced
speech signals. However, the main practical difficulty of these approaches is
that for each noise type a model is required to be trained a priori. In this
paper, we investigate a new class of supervised speech denoising algorithms
using nonnegative matrix factorization (NMF). We propose a novel speech
enhancement method that is based on a Bayesian formulation of NMF (BNMF). To
circumvent the mismatch problem between the training and testing stages, we
propose two solutions. First, we use an HMM in combination with BNMF (BNMF-HMM)
to derive a minimum mean square error (MMSE) estimator for the speech signal
with no information about the underlying noise type. Second, we suggest a
scheme to learn the required noise BNMF model online, which is then used to
develop an unsupervised speech enhancement system. Extensive experiments are
carried out to investigate the performance of the proposed methods under
different conditions. Moreover, we compare the performance of the developed
algorithms with state-of-the-art speech enhancement schemes using various
objective measures. Our simulations show that the proposed BNMF-based methods
outperform the competing algorithms substantially. | ['Paris Smaragdis', 'Arne Leijon', 'Nasser Mohammadiha'] | 2017-09-15 | null | null | null | null | ['speech-denoising'] | ['speech'] | [ 4.03085023e-01 -2.98142582e-01 4.55645323e-01 -2.68417627e-01
-8.47137034e-01 -8.00774172e-02 3.00338686e-01 -1.05018236e-01
-5.45334339e-01 6.15294635e-01 3.13421309e-01 -2.52867609e-01
-3.27878177e-01 -4.51807439e-01 -3.79765451e-01 -1.15468633e+00
3.21358860e-01 -2.91365921e-01 1.21717170e-01 -3.02141637e-01
2.26957515e-01 2.19631895e-01 -1.73459816e+00 -1.79092333e-01
1.19366693e+00 9.50417340e-01 7.36389935e-01 7.08174527e-01
-9.80009809e-02 3.71619910e-01 -7.46414959e-01 -2.38782436e-01
1.18224345e-01 -6.78659558e-01 -2.66964048e-01 4.65777695e-01
-1.52369231e-01 -8.87581483e-02 -2.34382659e-01 1.52010548e+00
8.45039964e-01 7.03607142e-01 4.60537761e-01 -6.40738130e-01
-1.55608609e-01 3.14159840e-01 -3.78911108e-01 2.76969790e-01
5.47601357e-02 -1.49845123e-01 5.54600298e-01 -9.67686832e-01
9.50345248e-02 1.06010425e+00 6.62241220e-01 5.76480448e-01
-1.04931498e+00 -4.75769907e-01 1.15011139e-02 3.82845193e-01
-1.20407772e+00 -9.09687102e-01 9.90487278e-01 -2.27730498e-01
4.82069552e-01 2.31581330e-01 3.32727134e-01 6.26388669e-01
-6.34797588e-02 5.79301000e-01 1.48755479e+00 -7.37506926e-01
3.80763441e-01 1.35588184e-01 1.73960060e-01 3.56134653e-01
-6.55083656e-02 2.10315913e-01 -3.93547535e-01 -7.83091187e-02
4.43579137e-01 -2.10339427e-01 -5.92264593e-01 -5.94620369e-02
-7.87633002e-01 5.38804173e-01 -2.29441635e-02 5.46735585e-01
-6.78750932e-01 -2.09326908e-01 1.98450789e-01 2.12627664e-01
5.86300671e-01 -5.67902401e-02 -1.30414531e-01 -2.23595932e-01
-1.11970174e+00 -9.03286710e-02 6.37502789e-01 4.93439108e-01
5.75566769e-01 4.12054241e-01 5.14147207e-02 1.26659870e+00
6.30934894e-01 5.84128916e-01 4.22038913e-01 -6.48277819e-01
4.56175327e-01 -1.49006918e-01 2.18830347e-01 -9.63065207e-01
-1.72965378e-02 -7.91889369e-01 -1.01660466e+00 1.01557977e-01
5.09444326e-02 -2.99547613e-01 -8.63359869e-01 1.58769631e+00
3.93391609e-01 3.26841235e-01 3.52301240e-01 9.41749573e-01
4.85136509e-01 9.17944491e-01 -2.69589126e-02 -8.86359513e-01
8.83802950e-01 -8.13413441e-01 -1.32120299e+00 -1.43205002e-02
1.12606488e-01 -1.24594605e+00 6.60018742e-01 7.85113454e-01
-1.09455967e+00 -8.19823384e-01 -1.02355313e+00 4.31597948e-01
-4.90902103e-02 2.92306930e-01 2.00129002e-01 1.15920353e+00
-1.03069723e+00 6.15065455e-01 -7.72049785e-01 -2.08634898e-01
-2.00405374e-01 3.47255081e-01 -1.04410216e-01 6.10206909e-02
-1.12600565e+00 8.48819077e-01 2.56527543e-01 7.85020530e-01
-8.94904375e-01 -1.00386441e-01 -7.99706161e-01 1.22205973e-01
3.22422594e-01 -3.25726449e-01 1.36698711e+00 -9.87139761e-01
-1.87533343e+00 1.12895551e-03 -5.61725438e-01 -3.69457066e-01
2.17285320e-01 -2.51020163e-01 -7.38137603e-01 2.09164530e-01
-2.88214922e-01 -1.19627640e-02 1.28927743e+00 -1.45493162e+00
-5.01503348e-01 -1.62394509e-01 -2.14092076e-01 2.94329405e-01
-4.37506109e-01 2.38799229e-01 -2.66655475e-01 -8.67499769e-01
5.11873543e-01 -6.55252576e-01 -4.50135589e-01 -3.81438196e-01
-1.41912580e-01 4.44432013e-02 7.89670706e-01 -1.09675467e+00
1.30696142e+00 -2.45689654e+00 1.93820149e-01 2.89641500e-01
-2.11565048e-01 6.21348917e-01 1.17307700e-01 3.59206706e-01
-3.68269309e-02 -3.13547164e-01 -4.03990090e-01 -6.21366322e-01
-1.73526257e-01 3.52039993e-01 -3.56841944e-02 4.66491699e-01
-7.40267932e-02 -2.03961115e-02 -7.06666291e-01 -4.83360440e-01
5.77722192e-01 8.06894422e-01 -3.86014611e-01 4.01132464e-01
2.98452765e-01 6.17079973e-01 -5.26093990e-02 3.01556706e-01
8.70949090e-01 4.66920793e-01 3.57789770e-02 -3.39977682e-01
-3.63097936e-01 1.96966633e-01 -1.69462454e+00 1.39361906e+00
-6.86377585e-01 4.55103159e-01 6.76779449e-01 -1.28360188e+00
1.03427899e+00 8.03580642e-01 2.26667285e-01 -2.51271576e-01
2.55844504e-01 3.03944647e-01 1.11684859e-01 -5.41959703e-01
3.56375843e-01 -4.98492271e-01 5.58091879e-01 1.43387750e-01
2.87087828e-01 4.69707623e-02 1.79413751e-01 -9.71662849e-02
7.08030879e-01 -2.76817441e-01 2.69099206e-01 -1.26119509e-01
9.72873390e-01 -6.26434565e-01 6.77565396e-01 6.35316133e-01
-3.07728469e-01 3.94300878e-01 -1.91845313e-01 3.00657094e-01
-7.19164550e-01 -9.61085618e-01 -1.73968077e-01 6.80120409e-01
3.04181911e-02 -1.79399192e-01 -9.24948394e-01 -2.50897646e-01
-5.85359275e-01 6.01480126e-01 -1.59990508e-02 3.10471710e-02
-3.79999220e-01 -9.29953635e-01 2.18719155e-01 2.34458536e-01
6.64014697e-01 -8.12115550e-01 -7.06472620e-02 4.28299785e-01
-3.62761945e-01 -1.03001893e+00 -3.27892005e-01 3.89619052e-01
-9.67323422e-01 -5.18251061e-01 -9.18733120e-01 -7.98831224e-01
7.29485035e-01 5.53818047e-01 3.78170431e-01 -5.32728136e-02
2.36419648e-01 4.21077251e-01 -5.61775029e-01 -2.63406157e-01
-6.42939329e-01 -3.18892926e-01 3.92698348e-01 5.34090996e-01
2.76786033e-02 -7.21669614e-01 -4.75674659e-01 3.25642377e-01
-1.02322662e+00 -2.38289833e-01 9.01382506e-01 9.45723474e-01
5.58470190e-01 8.04639459e-01 7.94730783e-01 -4.67543364e-01
7.65324950e-01 -1.57803252e-01 -6.43621624e-01 7.10427240e-02
-5.01569629e-01 -7.42437840e-02 6.49394751e-01 -4.69680727e-01
-1.68525302e+00 1.65526077e-01 -7.41179466e-01 -2.07026660e-01
-3.75838071e-01 5.68783224e-01 -5.48707366e-01 -2.83379436e-01
3.35431546e-01 6.31036162e-01 1.90390134e-03 -9.11680818e-01
2.09356800e-01 1.17887342e+00 6.75914824e-01 -2.97253847e-01
9.37670827e-01 1.44216940e-01 -1.76518723e-01 -1.36673832e+00
-5.87386608e-01 -9.14800584e-01 -4.46724236e-01 -4.41975296e-01
7.22067237e-01 -7.69868255e-01 -3.77297938e-01 6.65715396e-01
-1.14823914e+00 1.84304357e-01 1.04110010e-01 9.67928350e-01
-4.67376709e-01 8.12756479e-01 -5.93812466e-01 -1.51563048e+00
-3.37408662e-01 -1.18536174e+00 6.22779191e-01 2.93935537e-01
1.84059367e-01 -9.11090672e-01 -5.30271418e-02 4.74651754e-01
4.57132906e-01 -3.16997617e-01 5.91980338e-01 -3.88006002e-01
-2.12713152e-01 -2.06682049e-02 1.23517685e-01 1.01267910e+00
3.78979981e-01 -3.24413300e-01 -1.01642871e+00 -3.28695536e-01
7.24173188e-01 2.37705380e-01 7.16570020e-01 5.77246249e-01
8.94373000e-01 -1.86721116e-01 4.92853150e-02 3.50177258e-01
1.35049105e+00 5.52746832e-01 8.59426439e-01 1.80051103e-01
2.58660257e-01 5.06687522e-01 7.60885179e-01 4.72817034e-01
-1.64767936e-01 5.02361655e-01 1.28409311e-01 -1.90846503e-01
-1.44125037e-02 1.50358034e-02 4.92821753e-01 1.55059218e+00
-1.06813967e-01 -1.56092063e-01 -4.35925305e-01 5.49076617e-01
-1.56810713e+00 -8.63069832e-01 -1.19892873e-01 2.29948092e+00
8.43975306e-01 1.56117454e-01 -8.53919536e-02 8.68019104e-01
1.03435135e+00 1.39168009e-01 -6.76415041e-02 -4.00084466e-01
-8.62198845e-02 3.64537090e-01 2.97755033e-01 6.44038677e-01
-1.17541838e+00 4.36634988e-01 5.63968801e+00 1.06444156e+00
-9.96529758e-01 2.98846453e-01 1.04228541e-01 3.68872792e-01
1.51211862e-02 -5.25492169e-02 -6.50743186e-01 5.15080690e-01
9.54175532e-01 4.60711978e-02 5.45006454e-01 5.61113358e-01
7.60764420e-01 -1.73885912e-01 -3.18239719e-01 1.00872755e+00
8.82265344e-02 -6.18984878e-01 -2.39731431e-01 -1.58292279e-01
6.26304507e-01 -4.96920854e-01 -3.10732070e-02 1.16469860e-01
-2.58318067e-01 -4.12653893e-01 4.73133266e-01 5.76013088e-01
3.07187259e-01 -9.08350646e-01 9.50096428e-01 5.40793240e-01
-1.10326481e+00 -4.68385555e-02 -4.42359537e-01 -1.94049757e-02
4.14563417e-01 1.03819311e+00 -4.74099815e-01 7.27643609e-01
5.86920857e-01 3.24746996e-01 -9.14733931e-02 1.50262606e+00
-4.42144156e-01 9.66872692e-01 -1.98736787e-01 4.71013039e-02
4.48724255e-02 -3.03750664e-01 8.89835477e-01 1.13862991e+00
5.12682736e-01 3.41941476e-01 -1.79502457e-01 4.24612343e-01
2.43765011e-01 3.52097601e-01 -2.50321388e-01 -1.23157375e-01
1.82813689e-01 1.20767152e+00 -5.97545207e-01 -2.28589088e-01
-4.55152452e-01 9.47986245e-01 -3.10012549e-01 5.18934906e-01
-5.48288643e-01 -6.34894252e-01 3.28901052e-01 -9.93833393e-02
4.49658304e-01 -4.27332431e-01 8.92138481e-02 -1.02873242e+00
5.23526892e-02 -9.85974610e-01 2.33502649e-02 -7.46043444e-01
-1.07637751e+00 6.26560390e-01 -2.03999504e-01 -1.43772650e+00
-1.73374817e-01 -3.06096137e-01 -5.15981019e-01 1.01252294e+00
-1.59114981e+00 -6.29112005e-01 -4.89727743e-02 6.58556104e-01
7.40574419e-01 -7.94038475e-02 6.27499700e-01 7.13536620e-01
-5.22440076e-01 3.93108219e-01 5.65459192e-01 -1.06966987e-01
5.80192924e-01 -1.04333389e+00 -1.13710873e-01 1.32960737e+00
1.30081967e-01 6.82816505e-01 1.08602250e+00 -5.82217753e-01
-1.04321921e+00 -7.88304448e-01 9.24206495e-01 4.85840768e-01
4.26935136e-01 -1.00879602e-01 -9.80998278e-01 1.14627473e-01
3.03091288e-01 -2.72854269e-01 7.16817319e-01 1.15859881e-02
2.91083753e-01 -2.94173688e-01 -1.00810266e+00 2.63050258e-01
5.89673877e-01 -4.85827148e-01 -7.58616447e-01 2.85253711e-02
5.17891526e-01 -2.23054245e-01 -6.73886538e-01 4.96863306e-01
2.30860069e-01 -9.39157665e-01 7.10830867e-01 -1.38834000e-01
-1.05980188e-01 -7.46227324e-01 -2.51178354e-01 -1.73853076e+00
-9.10353288e-02 -8.70044649e-01 -5.85813224e-02 1.49984431e+00
1.76939890e-01 -6.21881664e-01 4.63971049e-01 1.81025222e-01
-3.63631696e-01 -4.08559978e-01 -9.13916886e-01 -8.46333921e-01
-5.96432149e-01 -5.47814310e-01 1.25415668e-01 5.93365848e-01
-1.51121601e-01 1.90928698e-01 -7.09516108e-01 6.68037117e-01
8.35720241e-01 -1.68355554e-01 4.78637040e-01 -9.74549651e-01
-6.12966478e-01 -1.66669227e-02 -4.58940178e-01 -1.17625082e+00
2.26432327e-02 -2.76823640e-01 6.24177694e-01 -1.56068933e+00
-6.82537183e-02 -1.57562822e-01 -5.57790935e-01 -7.03122914e-02
-3.74017417e-01 -5.93649745e-02 8.96992981e-02 -7.46288076e-02
-2.51683116e-01 8.54900479e-01 9.73809779e-01 4.26284075e-02
-2.95526803e-01 5.24350286e-01 -4.31583405e-01 7.52643526e-01
7.03111231e-01 -5.76775253e-01 -3.59945208e-01 -2.74934232e-01
-3.31824183e-01 4.21740085e-01 1.90695837e-01 -1.24887908e+00
4.24600571e-01 1.74739927e-01 1.58632666e-01 -7.23623812e-01
5.25933683e-01 -9.25585687e-01 3.67006361e-02 2.43469328e-01
-1.71499923e-01 -3.40627432e-01 1.00939922e-01 7.69405961e-01
-6.98982894e-01 -6.24601841e-01 9.34757888e-01 1.31669611e-01
-5.75857818e-01 -2.56731004e-01 -8.15149844e-01 -5.36105216e-01
5.23860037e-01 -8.44902247e-02 6.43190071e-02 -7.19731927e-01
-1.07731628e+00 -2.54259765e-01 -1.97374955e-01 -6.68194797e-03
7.83975601e-01 -1.03486788e+00 -5.15110016e-01 1.37488559e-01
-3.02524030e-01 -3.98615032e-01 6.01934075e-01 1.09435296e+00
-8.46532434e-02 2.56405175e-01 9.21969637e-02 -4.50647235e-01
-1.49090934e+00 4.42114502e-01 2.80655235e-01 1.83653319e-03
-1.98118567e-01 6.01816535e-01 -4.48866524e-02 -2.17014790e-01
3.73949081e-01 -3.67974341e-02 -3.97126287e-01 -1.99108854e-01
5.66891909e-01 4.81012881e-01 3.87142181e-01 -8.40913951e-01
-1.25613948e-03 4.10350233e-01 2.37600237e-01 -5.16115665e-01
1.29764950e+00 -4.82431829e-01 -1.07776590e-01 2.47130260e-01
9.52575445e-01 2.69961804e-01 -8.63253236e-01 -2.97535509e-01
-1.51046097e-01 -6.25568748e-01 6.24188304e-01 -5.64023972e-01
-9.33662891e-01 1.00188196e+00 1.06344056e+00 2.39418000e-01
1.64728343e+00 -5.34666359e-01 9.17975664e-01 3.42837095e-01
2.81025112e-01 -1.17958069e+00 -1.71362251e-01 2.38748744e-01
6.06078446e-01 -1.10739779e+00 -2.63389766e-01 -5.23792922e-01
-2.66000509e-01 1.07550001e+00 3.23025882e-01 2.75578946e-01
7.76346385e-01 1.68840498e-01 2.26006866e-01 2.98453987e-01
-2.95877784e-01 -7.23157763e-01 2.27165639e-01 5.74511349e-01
2.82545626e-01 -1.24837913e-01 -6.76291108e-01 7.79753566e-01
1.27789760e-02 -2.14853864e-02 4.75293994e-01 9.17386770e-01
-8.25982213e-01 -1.25247419e+00 -9.06326890e-01 1.36466980e-01
-5.83258271e-01 -1.89148009e-01 6.64239749e-02 2.10482836e-01
8.64471346e-02 1.61370444e+00 -5.01425028e-01 -3.76008928e-01
3.93648356e-01 9.96799245e-02 3.31353128e-01 -5.00712097e-01
-3.51463586e-01 7.70038605e-01 4.63804742e-03 8.61554146e-02
-7.63437271e-01 -7.57118344e-01 -9.20604646e-01 6.44706637e-02
-8.94337058e-01 6.08770013e-01 9.96213853e-01 1.06747842e+00
-7.25327060e-02 5.18739522e-01 8.93775403e-01 -7.19550014e-01
-6.27608895e-01 -1.07832336e+00 -7.80171573e-01 2.34278247e-01
3.40354621e-01 -6.73844755e-01 -7.59579480e-01 1.61093026e-01] | [15.005529403686523, 5.820611953735352] |
44a3f073-c03f-4416-a048-c6691b6b6271 | extracting-temporal-and-causal-relations | 1604.08120 | null | http://arxiv.org/abs/1604.08120v1 | http://arxiv.org/pdf/1604.08120v1.pdf | Extracting Temporal and Causal Relations between Events | Structured information resulting from temporal information processing is
crucial for a variety of natural language processing tasks, for instance to
generate timeline summarization of events from news documents, or to answer
temporal/causal-related questions about some events. In this thesis we present
a framework for an integrated temporal and causal relation extraction system.
We first develop a robust extraction component for each type of relations, i.e.
temporal order and causality. We then combine the two extraction components
into an integrated relation extraction system, CATENA---CAusal and Temporal
relation Extraction from NAtural language texts---, by utilizing the
presumption about event precedence in causality, that causing events must
happened BEFORE resulting events. Several resources and techniques to improve
our relation extraction systems are also discussed, including word embeddings
and training data expansion. Finally, we report our adaptation efforts of
temporal information processing for languages other than English, namely
Italian and Indonesian. | ['Paramita Mirza'] | 2016-04-27 | extracting-temporal-and-causal-relations-1 | https://aclanthology.org/P14-3002 | https://aclanthology.org/P14-3002.pdf | acl-2014-6 | ['temporal-relation-extraction', 'timeline-summarization'] | ['natural-language-processing', 'natural-language-processing'] | [ 2.92834640e-01 3.53037357e-01 -4.45179760e-01 -5.31062484e-01
-4.26036745e-01 -7.73355782e-01 1.15464640e+00 8.76960933e-01
-4.06444609e-01 1.21196413e+00 1.21565974e+00 -4.57840055e-01
-5.23555636e-01 -8.79314423e-01 -2.24996641e-01 -3.68272454e-01
-8.29374135e-01 5.44472635e-01 3.18477303e-01 -3.02759200e-01
1.64032727e-01 6.09408081e-01 -1.05558801e+00 4.99758452e-01
4.54473287e-01 5.62139869e-01 -2.68269181e-01 9.53623772e-01
-3.01568598e-01 1.57452226e+00 -7.29618013e-01 -2.51457661e-01
-4.26380724e-01 -5.86607039e-01 -1.47561014e+00 -4.38774824e-01
-5.88517666e-01 -3.18478853e-01 -6.38122380e-01 3.44264328e-01
2.95689732e-01 5.68352103e-01 9.79318440e-01 -1.28006566e+00
-3.88604492e-01 1.36280930e+00 -4.05345976e-01 1.04808116e+00
9.61147308e-01 -2.43122831e-01 1.14505720e+00 -1.82333648e-01
1.10352731e+00 1.39909470e+00 4.46525991e-01 4.44718659e-01
-8.46916318e-01 -3.41380924e-01 4.03885722e-01 5.46272635e-01
-7.86131620e-01 -3.88196111e-01 7.26341963e-01 -3.16002369e-01
1.83362317e+00 3.74653578e-01 6.03879929e-01 1.35002232e+00
8.49728942e-01 7.08095253e-01 5.23321331e-01 -4.65664208e-01
-7.65534639e-02 -3.56730223e-01 4.75768536e-01 6.74042925e-02
1.94337934e-01 5.87272383e-02 -9.31127012e-01 -1.19390465e-01
5.86933970e-01 -6.79388344e-01 4.15719599e-02 7.35261858e-01
-1.36566532e+00 7.41813064e-01 -5.66970631e-02 6.63701534e-01
-6.38637424e-01 1.08462222e-01 8.54957044e-01 1.61487699e-01
8.23551893e-01 3.34403694e-01 -8.11582386e-01 -3.18073511e-01
-5.91064930e-01 4.97129440e-01 1.02648544e+00 8.29168797e-01
-9.36707556e-02 -1.23636134e-01 -2.87712991e-01 3.44162166e-01
3.00197333e-01 -2.20618695e-01 3.75223160e-01 -4.69128668e-01
8.46722543e-01 5.76787293e-01 2.93035626e-01 -1.10549021e+00
-6.65002644e-01 2.90826231e-01 -6.48574650e-01 -5.50965607e-01
3.12303662e-01 -6.06968343e-01 -6.18090868e-01 1.65875125e+00
6.72753751e-01 1.34460703e-01 4.64386791e-01 5.84599257e-01
1.02427530e+00 1.36230922e+00 3.60319287e-01 -1.13422716e+00
1.56806266e+00 -3.66365790e-01 -1.63766634e+00 -1.51592314e-01
5.73316813e-01 -6.58303559e-01 2.47989327e-01 8.15985724e-02
-1.01926446e+00 6.76912367e-02 -1.01469517e+00 -3.75113606e-01
-6.80595100e-01 -4.43765432e-01 1.12217796e+00 -3.89695494e-03
-4.24828112e-01 7.32652366e-01 -9.48269367e-01 -5.26669383e-01
-1.17674649e-01 2.90118247e-01 -4.10613596e-01 5.02532303e-01
-1.92623436e+00 1.19453716e+00 9.04228568e-01 3.10464613e-02
-5.75428426e-01 -7.52610505e-01 -9.72803354e-01 -2.00128108e-01
2.75512516e-01 -3.84101450e-01 1.47552145e+00 -5.64106703e-02
-1.02607310e+00 5.07784665e-01 -5.86364865e-01 -8.23466659e-01
1.97326854e-01 -6.77879035e-01 -1.10058033e+00 2.27261260e-01
3.38389784e-01 7.51427636e-02 1.36932954e-01 -7.57262528e-01
-8.45941782e-01 -1.27134442e-01 -1.43344656e-01 3.12645137e-01
-6.39539808e-02 9.45024669e-01 -2.95967519e-01 -7.11113691e-01
-2.46685043e-01 -3.90675187e-01 -3.48834276e-01 -8.46979916e-01
-5.33069134e-01 -9.88711178e-01 7.56636798e-01 -8.68340015e-01
2.11673355e+00 -1.70729327e+00 1.02589011e-01 -1.20416701e-01
-1.81814596e-01 -2.43166536e-01 1.35754675e-01 1.09972501e+00
-7.50718057e-01 3.23331058e-01 -7.68756121e-02 1.98223844e-01
-1.83925152e-01 5.39386511e-01 -7.11864352e-01 2.52651542e-01
7.34963179e-01 7.31353283e-01 -1.28340173e+00 -8.81985545e-01
1.98295176e-01 2.91699857e-01 1.80756569e-01 2.81044006e-01
-4.04646069e-01 3.24139774e-01 -6.09540403e-01 1.77057654e-01
-6.05122512e-03 3.13155055e-01 3.83424491e-01 -5.74132316e-02
-5.30568302e-01 1.22769880e+00 -1.24630368e+00 1.40333438e+00
-3.22186589e-01 8.60095441e-01 -6.96293712e-01 -7.79320717e-01
6.53311968e-01 1.14556348e+00 1.04580200e+00 -6.45512223e-01
3.05006802e-01 -2.63620168e-01 -2.09692910e-01 -9.09399569e-01
8.14642251e-01 -6.86345994e-02 -5.70711672e-01 7.81514466e-01
1.75586402e-01 1.54331438e-02 1.00915277e+00 4.86842930e-01
1.18710959e+00 2.87347704e-01 1.03558850e+00 -1.69179291e-01
3.34826648e-01 4.78351951e-01 8.72499585e-01 9.61902887e-02
-1.41203254e-01 2.12053344e-01 1.08427227e+00 -8.87758195e-01
-9.07830060e-01 -8.47252190e-01 -3.51787992e-02 9.02263761e-01
-2.81032503e-01 -7.59809494e-01 -9.65330005e-02 -7.29818046e-01
-2.90522516e-01 1.20974112e+00 -9.32775378e-01 1.42700002e-01
-1.10653412e+00 -1.18123293e+00 5.77352047e-01 6.90862060e-01
-7.42155081e-03 -1.22847581e+00 -7.80580521e-01 8.05982471e-01
-5.13150930e-01 -1.16544926e+00 -2.04310998e-01 4.69052583e-01
-6.46285534e-01 -1.33099890e+00 1.09577239e-01 -4.91021991e-01
-3.33663784e-02 -4.64009762e-01 1.16337025e+00 -7.44405627e-01
-2.47669518e-01 1.65799409e-02 -6.59539104e-01 -8.56050074e-01
-3.45797300e-01 -2.96762317e-01 3.08321625e-01 -2.64651954e-01
4.82626200e-01 -6.41332150e-01 -2.04480991e-01 -2.13668197e-01
-1.03075957e+00 -1.64781809e-01 2.02797614e-02 3.03350806e-01
1.54312402e-01 6.12470865e-01 5.33971906e-01 -8.40787947e-01
1.21654701e+00 -7.88790524e-01 -1.26456052e-01 3.15833092e-01
-3.52577388e-01 3.01255614e-01 4.34948683e-01 -5.79914391e-01
-1.71053064e+00 -9.09537151e-02 -1.93490051e-02 4.98839796e-01
-2.44743258e-01 8.77822280e-01 -1.47176385e-02 1.10692823e+00
8.25984657e-01 -1.35922223e-01 -7.64780998e-01 -6.15458144e-03
7.33707190e-01 3.03793430e-01 6.23963058e-01 -7.06310093e-01
4.04252321e-01 4.63952333e-01 -3.82703468e-02 -6.98659062e-01
-7.59250939e-01 -7.45790422e-01 -8.24331880e-01 -3.88910800e-01
1.15232086e+00 -6.51718616e-01 -4.65331018e-01 -1.26564100e-01
-2.00433660e+00 -1.12377159e-01 -5.04667759e-01 8.29398394e-01
-3.16574395e-01 -6.64445758e-03 -8.89178932e-01 -1.07870042e+00
-3.19284856e-01 -2.60528028e-01 6.73985124e-01 3.86780947e-01
-1.13568962e+00 -1.35351932e+00 5.03774285e-01 -2.68678725e-01
-3.10004562e-01 7.47802556e-01 8.05877686e-01 -9.58944380e-01
9.78247225e-02 -1.77532881e-01 -2.75266357e-02 -4.72732365e-01
5.35045922e-01 5.19274354e-01 -5.99742413e-01 5.60405552e-01
-2.23138407e-01 -1.96037702e-02 6.24182642e-01 3.32231700e-01
2.59584606e-01 -8.00041914e-01 -6.29298806e-01 1.08095501e-02
1.11780739e+00 8.75011086e-01 6.87428236e-01 2.40723282e-01
4.17008042e-01 1.18850863e+00 9.25477087e-01 7.12878764e-01
4.22560722e-01 2.50531286e-01 -1.68779939e-01 3.74593377e-01
-2.81811152e-02 -2.92804271e-01 3.77603889e-01 8.33698153e-01
-3.91003132e-01 -6.58659518e-01 -9.34302807e-01 1.18680108e+00
-1.98674548e+00 -1.51294625e+00 -9.78328407e-01 1.54730654e+00
1.45644391e+00 1.45093212e-02 4.64747474e-02 4.78634566e-01
3.47433567e-01 4.97171611e-01 8.30954127e-03 -9.18573260e-01
-4.16855998e-02 1.16890579e-01 4.03497756e-01 7.19282866e-01
-1.36416149e+00 1.02174914e+00 6.57631445e+00 4.83597726e-01
-6.80171967e-01 4.72569242e-02 2.21805468e-01 1.55873686e-01
-3.62531364e-01 2.43807584e-01 -8.52802217e-01 -9.88903642e-02
1.36935127e+00 -7.74157166e-01 -2.18945026e-01 2.23470718e-01
8.73728693e-01 -3.19719523e-01 -1.14821088e+00 3.92737895e-01
-2.84028232e-01 -1.52033937e+00 1.54201284e-01 -3.62993389e-01
6.17787540e-01 -5.30524015e-01 -6.57317638e-01 -5.27744554e-02
9.54985261e-01 -7.70366251e-01 6.95418835e-01 4.22647148e-01
3.78162652e-01 -7.72088945e-01 8.57130229e-01 -2.17478187e-03
-1.63135850e+00 1.29457876e-01 2.42502674e-01 -4.82664108e-01
9.06465113e-01 1.00249195e+00 -9.96587038e-01 1.27377069e+00
6.24671757e-01 7.12697327e-01 -7.89690986e-02 6.13068938e-01
-7.94635475e-01 6.87443316e-01 -4.16121095e-01 -2.99611568e-01
7.27630556e-02 9.42253992e-02 6.67238891e-01 1.71328998e+00
2.98220594e-03 6.51045740e-01 -3.81013185e-01 3.33155036e-01
4.61041063e-01 -2.18933392e-02 -9.39736605e-01 -3.23349506e-01
5.44883788e-01 9.08647060e-01 -9.02528346e-01 -4.78995353e-01
-1.64225921e-01 8.17768514e-01 1.37431817e-02 3.05936247e-01
-1.04551816e+00 -3.53785187e-01 3.85400951e-01 -2.27696672e-01
-3.98446679e-01 -5.89838147e-01 -4.42223310e-01 -9.61062551e-01
-1.27307707e-02 -4.73830134e-01 1.11278927e+00 -6.22922122e-01
-1.02988219e+00 5.24376690e-01 8.77090693e-01 -9.47548628e-01
-6.71195090e-01 -1.78076565e-01 -9.86421824e-01 6.66409373e-01
-9.69134629e-01 -8.97442281e-01 5.41828930e-01 4.93161261e-01
7.45425105e-01 3.55602831e-01 7.84466207e-01 1.74482584e-01
-7.30488241e-01 -2.36919880e-01 -7.73745537e-01 2.67758429e-01
6.97323859e-01 -1.39463151e+00 3.95241380e-01 1.38122213e+00
1.89401478e-01 6.92274749e-01 1.05431116e+00 -1.14653456e+00
-1.04217100e+00 -1.18358135e+00 2.33654046e+00 -4.05438781e-01
1.31275547e+00 -1.13265499e-01 -5.74095190e-01 8.21309328e-01
1.08653617e+00 -5.82373142e-01 7.40023434e-01 4.36699957e-01
-1.90603614e-01 -5.83030730e-02 -7.26008832e-01 9.27202225e-01
9.48978424e-01 -2.68976361e-01 -1.33756483e+00 5.81229627e-01
1.13196528e+00 -2.92669803e-01 -1.01987159e+00 3.65166217e-01
3.63834888e-01 -3.11178416e-01 7.63877332e-01 -9.40139830e-01
8.30804229e-01 -4.14991498e-01 1.46206543e-01 -1.05722749e+00
-3.67238611e-01 -1.08977008e+00 -3.03807735e-01 1.79275680e+00
8.11303318e-01 -3.31028074e-01 1.64095983e-01 5.74992061e-01
-9.16471258e-02 -1.97091088e-01 -9.90059078e-01 -5.42503655e-01
-2.00288534e-01 -1.08052957e+00 3.97376835e-01 1.36566067e+00
6.33554578e-01 8.13204646e-01 -2.86763787e-01 3.83737206e-01
1.54839873e-01 -7.34498054e-02 1.99872985e-01 -1.04393101e+00
2.15606272e-01 -2.76870906e-01 9.23635438e-02 -3.97350222e-01
1.04625970e-01 -2.61439502e-01 6.42934591e-02 -2.19136953e+00
-3.21355999e-01 1.52940467e-01 -8.42996612e-02 6.50596559e-01
-1.22382365e-01 -4.56718415e-01 -4.30986166e-01 -1.71268120e-01
-2.42396533e-01 3.31292003e-01 9.37094927e-01 -2.53041714e-01
-7.42646575e-01 -2.45098993e-01 -3.52177978e-01 5.69981635e-01
8.40164125e-01 -7.26058185e-01 -6.82293177e-01 -3.38082224e-01
7.07229972e-01 6.92874491e-01 9.56376940e-02 -4.48049068e-01
5.44376075e-01 -7.98137724e-01 1.93093479e-01 -8.81512642e-01
-2.01805338e-01 -5.78457057e-01 2.17798203e-01 1.78839639e-01
-5.82818031e-01 4.31047708e-01 5.56296289e-01 4.76427913e-01
-4.74875987e-01 3.62188630e-02 1.53615437e-02 2.59748045e-02
-9.80848968e-01 -9.47648138e-02 -8.11696231e-01 -1.02937065e-01
1.31381595e+00 2.06597582e-01 -5.18049419e-01 -2.62860835e-01
-8.46761763e-01 4.31066871e-01 -7.16700971e-01 6.88877106e-01
6.31034911e-01 -1.30618858e+00 -9.20620918e-01 -4.77455795e-01
-1.76959068e-01 4.33340706e-02 1.13871858e-01 8.97581279e-01
-5.55163264e-01 6.58252478e-01 1.18750222e-01 2.14425504e-01
-1.27508795e+00 7.82703578e-01 -2.84831405e-01 -8.89208674e-01
-5.26544034e-01 8.37922871e-01 -4.10640359e-01 2.17946023e-01
6.36200383e-02 -7.09664822e-01 -8.64437938e-01 1.00182068e+00
7.25971639e-01 1.85109839e-01 -7.94335455e-02 -4.41936880e-01
-7.68695414e-01 1.86200738e-02 3.16618942e-02 -8.41468453e-01
1.44206095e+00 -3.80170159e-02 -5.48651636e-01 8.75830472e-01
8.43205869e-01 1.09357052e-01 -6.38084471e-01 1.35053858e-01
6.07335269e-01 2.46090204e-01 -3.41404676e-01 -6.33309603e-01
-2.06744790e-01 3.10939103e-01 -2.77965754e-01 8.66658509e-01
1.20673645e+00 1.79607302e-01 5.31439364e-01 3.20649259e-02
-6.17795736e-02 -1.19403684e+00 -4.42691088e-01 7.42044389e-01
1.18450916e+00 -6.50584340e-01 4.48229760e-01 -3.90498608e-01
-6.52213454e-01 1.39283717e+00 2.38348499e-01 7.26038516e-02
6.76051140e-01 6.79349363e-01 -3.86874415e-02 -2.88223088e-01
-1.40452099e+00 -3.94528955e-01 3.46183389e-01 4.94933486e-01
1.05588591e+00 1.15440786e-01 -1.08504295e+00 7.01733291e-01
-2.96606213e-01 -1.55042663e-01 5.07239282e-01 1.00493288e+00
4.89714369e-03 -1.15051341e+00 -4.17912126e-01 1.09549142e-01
-7.99931288e-01 -1.38590097e-01 -9.77272570e-01 9.71525431e-01
1.07783295e-01 1.59501171e+00 2.16462642e-01 -4.04818147e-01
4.38342750e-01 1.25705868e-01 2.50761449e-01 -4.77638781e-01
-5.26695669e-01 1.30254552e-01 1.13362229e+00 -5.90704739e-01
-8.85280669e-01 -1.03016984e+00 -1.64932156e+00 -1.74408913e-01
6.23394810e-02 3.53497535e-01 2.67194808e-01 1.13408828e+00
-2.10446015e-01 1.17186511e+00 3.89641434e-01 -4.54938471e-01
3.18257600e-01 -1.11569893e+00 -1.62406236e-01 1.90057993e-01
1.30089432e-01 -2.01243684e-01 -8.32482949e-02 5.69117963e-01] | [9.065596580505371, 9.246320724487305] |
025fa920-3bc0-47cb-8f55-9d74f8e17db5 | citations-as-queries-source-attribution-using | 2306.17322 | null | https://arxiv.org/abs/2306.17322v1 | https://arxiv.org/pdf/2306.17322v1.pdf | Citations as Queries: Source Attribution Using Language Models as Rerankers | This paper explores new methods for locating the sources used to write a text, by fine-tuning a variety of language models to rerank candidate sources. After retrieving candidates sources using a baseline BM25 retrieval model, a variety of reranking methods are tested to see how effective they are at the task of source attribution. We conduct experiments on two datasets, English Wikipedia and medieval Arabic historical writing, and employ a variety of retrieval and generation based reranking models. In particular, we seek to understand how the degree of supervision required affects the performance of various reranking models. We find that semisupervised methods can be nearly as effective as fully supervised methods while avoiding potentially costly span-level annotation of the target and source documents. | ['David Smith', 'Ryan Muther'] | 2023-06-29 | null | null | null | null | ['retrieval'] | ['methodology'] | [ 8.58686939e-02 1.10578731e-01 -8.92860532e-01 -2.18449030e-02
-1.42149532e+00 -1.01246476e+00 1.45164955e+00 7.02060521e-01
-6.76704645e-01 8.19885790e-01 9.24007595e-01 -3.89084876e-01
-2.36763582e-01 -4.82344717e-01 -3.61304611e-01 -4.60573519e-03
3.31268698e-01 8.15863848e-01 4.70994055e-01 -3.54122788e-01
1.17524087e+00 3.35754901e-01 -1.08518088e+00 3.74743342e-01
9.35084641e-01 3.49418104e-01 -3.22386250e-02 6.38777137e-01
-5.12907922e-01 1.04570472e+00 -1.06074131e+00 -6.63200676e-01
-7.95111954e-02 -4.01498169e-01 -1.24002528e+00 -2.38992870e-01
5.08629799e-01 -9.75212604e-02 -3.80902827e-01 8.67876709e-01
3.54622662e-01 -5.26000671e-02 1.24736512e+00 -9.20233846e-01
-7.97812164e-01 1.39936161e+00 -4.37155247e-01 6.45663857e-01
6.49301112e-01 -6.15895152e-01 1.24709952e+00 -1.03126943e+00
9.21452761e-01 1.23244655e+00 3.72427106e-01 2.08070427e-01
-1.05866241e+00 -3.91602367e-01 -3.35894257e-01 5.94195798e-02
-1.33347714e+00 -8.19099247e-01 7.25429535e-01 -4.94460702e-01
5.86885989e-01 1.24189250e-01 -1.08264685e-01 7.36644983e-01
-1.86413955e-02 6.16338015e-01 1.06404340e+00 -8.48232567e-01
4.79859347e-03 4.06300366e-01 3.73491555e-01 6.93364561e-01
4.88902986e-01 -6.30196333e-01 -9.37035799e-01 -8.95675838e-01
3.77026200e-01 -4.30303335e-01 -1.62003785e-01 -7.79616972e-03
-1.56144190e+00 9.04153645e-01 2.31227830e-01 5.89512050e-01
-2.00587343e-02 -1.02547957e-02 4.03038383e-01 1.76891685e-01
4.52462763e-01 1.08907270e+00 -3.12197119e-01 1.23666994e-01
-1.22435319e+00 1.01572029e-01 1.10037458e+00 8.82964671e-01
7.54857481e-01 -4.10519242e-01 -2.73603976e-01 1.01698327e+00
4.12195981e-01 4.28162396e-01 8.32410574e-01 -8.97059143e-01
6.81269526e-01 6.60419047e-01 5.61970472e-01 -1.20638490e+00
-1.13825034e-02 -7.75227249e-02 -1.71534583e-01 -4.16390210e-01
5.35166204e-01 8.49313568e-03 -8.90523791e-01 1.03005457e+00
-1.67631730e-01 -6.13082290e-01 1.40276760e-01 4.86974895e-01
7.49971151e-01 7.76494563e-01 1.15117237e-01 -2.53197014e-01
1.10387456e+00 -1.02073097e+00 -5.36318064e-01 -2.62561053e-01
5.41511297e-01 -1.36236346e+00 9.38709080e-01 -2.42415108e-02
-1.14029741e+00 -2.54752547e-01 -9.29375589e-01 -2.27306828e-01
-5.97582400e-01 4.60765272e-01 2.49815062e-01 3.88841957e-01
-8.54993284e-01 6.31114006e-01 -5.71615040e-01 -4.27962333e-01
2.97251660e-02 -2.02083901e-01 8.92887264e-03 6.08554259e-02
-1.18345821e+00 1.00937617e+00 3.29359829e-01 -4.65363055e-01
-9.73098874e-01 -5.08909285e-01 -5.21062851e-01 -1.13863096e-01
1.57069519e-01 -2.51450390e-01 1.21141553e+00 -9.12746966e-01
-1.01705420e+00 1.01182187e+00 -3.35470974e-01 -1.57239273e-01
3.96988600e-01 -1.31757036e-01 -4.49036360e-01 4.54173744e-01
7.15034306e-01 3.96260500e-01 8.75420451e-01 -1.25442672e+00
-6.73359334e-01 -1.20090432e-01 -7.49638304e-03 3.67421955e-01
-7.42863476e-01 7.10963428e-01 -5.41418970e-01 -1.01816595e+00
-1.26463687e-02 -7.42947102e-01 1.24216184e-01 -4.25177604e-01
-4.59075630e-01 -6.13160312e-01 2.05038324e-01 -8.59440207e-01
1.46559572e+00 -1.66571760e+00 1.85445741e-01 3.89049292e-01
1.19487256e-01 -1.96207449e-01 -2.57113278e-01 7.48965979e-01
5.74237585e-01 7.08638132e-01 9.07772928e-02 5.74752316e-02
3.93716320e-02 -2.35905915e-01 -6.93804920e-01 1.21448122e-01
-1.68182045e-01 7.68312275e-01 -1.09738970e+00 -7.94012666e-01
-6.69996560e-01 2.22947687e-01 6.45115748e-02 -1.77696347e-02
-2.47998968e-01 -1.71660855e-01 -4.46127683e-01 8.94393742e-01
-1.65734023e-01 -5.47186077e-01 2.58780569e-01 7.16237649e-02
1.18824737e-02 8.53180230e-01 -8.63009453e-01 1.32920361e+00
-2.29254380e-01 9.43083405e-01 -2.12861806e-01 -1.52858555e-01
7.65845120e-01 1.60403386e-01 1.31751239e-01 -4.35236007e-01
-1.87282473e-01 6.37057304e-01 -2.50578433e-01 -9.54848081e-02
1.13146544e+00 1.66075289e-01 -1.70689315e-01 1.00549042e+00
-9.40931812e-02 -8.68428033e-03 7.27221370e-01 8.20041955e-01
9.55721438e-01 -5.51357009e-02 2.29395241e-01 -4.41488713e-01
2.69683838e-01 7.85001755e-01 4.63957861e-02 1.15565407e+00
2.65838593e-01 4.22803521e-01 2.17112169e-01 -2.36474872e-02
-1.06022739e+00 -9.02099192e-01 -6.16393164e-02 1.62212193e+00
2.76465237e-01 -6.09712303e-01 -3.45681012e-01 -7.70370662e-01
1.74653590e-01 7.49989212e-01 -4.08371925e-01 2.94199493e-02
-6.50265634e-01 -9.17167783e-01 9.80778396e-01 4.82392222e-01
-3.97617593e-02 -8.25104177e-01 -3.62936258e-01 1.31178126e-01
-6.15502477e-01 -6.04417801e-01 -6.03531539e-01 5.36719821e-02
-7.49792337e-01 -1.11702240e+00 -9.99830008e-01 -8.42441857e-01
8.21473479e-01 3.55700195e-01 1.45742965e+00 3.25603366e-01
2.29079008e-01 5.46144724e-01 -6.07349157e-01 -3.42485130e-01
-8.02064538e-01 7.56361544e-01 -6.43322244e-03 -5.20961404e-01
3.32318872e-01 9.65512693e-02 -1.43881574e-01 2.22003117e-01
-8.90560210e-01 -5.33424377e-01 5.99547207e-01 7.43250549e-01
1.89437062e-01 2.17359886e-02 6.65799677e-01 -9.87581193e-01
1.25745869e+00 -6.43057466e-01 -3.27613622e-01 7.84413934e-01
-1.12278318e+00 2.81393170e-01 2.95898020e-01 -4.15513217e-01
-1.12352073e+00 -5.04882574e-01 4.39507961e-01 2.00401902e-01
3.52034062e-01 9.15968716e-01 3.37265700e-01 -2.07213145e-02
1.07061899e+00 -2.87154200e-03 -5.37422776e-01 -6.63875043e-01
4.43022251e-01 8.30106616e-01 2.64270276e-01 -7.29616702e-01
1.04606247e+00 1.41658381e-01 -5.76184094e-01 -6.67114377e-01
-1.09117067e+00 -7.21580625e-01 -7.69411743e-01 -7.98989236e-02
4.55959260e-01 -9.12005901e-01 3.13590020e-01 2.16214299e-01
-1.10856724e+00 -2.15556949e-01 -5.26626594e-03 2.60413349e-01
-3.64999250e-02 4.38071787e-01 -8.46467972e-01 -3.55636865e-01
-2.59689927e-01 -7.05585122e-01 9.43291008e-01 1.76139131e-01
-5.01135588e-01 -1.15839636e+00 3.24287325e-01 5.61333239e-01
2.65840769e-01 -2.68322200e-01 1.26264012e+00 -1.07511938e+00
-4.52599436e-01 -1.59591511e-01 -3.69432986e-01 -3.29086721e-01
2.65176415e-01 1.34500653e-01 -5.82671404e-01 -7.61093050e-02
-5.95981419e-01 -5.95759630e-01 1.12566829e+00 -2.38909423e-01
4.02800232e-01 -6.71358347e-01 -5.55611789e-01 -6.94927126e-02
1.22632909e+00 -1.88368317e-02 4.06583667e-01 8.20535243e-01
6.87100351e-01 5.46628594e-01 5.42667568e-01 1.02538161e-01
4.59368378e-01 1.66185990e-01 -2.59775758e-01 3.50692689e-01
-8.34809430e-03 -5.00752509e-01 4.43189383e-01 8.61949861e-01
1.44681796e-01 -4.56263125e-01 -1.30039477e+00 8.03354979e-01
-1.42822599e+00 -9.70891297e-01 2.45915502e-01 2.09939957e+00
1.48375428e+00 2.52059937e-01 2.92947263e-01 -1.09959603e-03
6.19073868e-01 2.80260414e-01 -1.86157629e-01 -3.36428471e-02
-1.63953751e-01 -1.62716299e-01 6.03429675e-01 7.44819701e-01
-9.34705198e-01 1.01856625e+00 7.75975418e+00 7.47191191e-01
-8.09386134e-01 -1.66142642e-01 5.07786453e-01 1.74826249e-01
-5.67130864e-01 2.12977007e-01 -1.15010297e+00 3.74044925e-01
1.02063453e+00 -5.80794454e-01 3.05712223e-01 6.15004897e-01
-3.08775723e-01 -1.89370841e-01 -1.06616247e+00 4.00225401e-01
7.70080626e-01 -1.53080714e+00 4.63397622e-01 -3.32822114e-01
7.73024201e-01 1.56684741e-01 -3.68828416e-01 7.85489529e-02
6.99451625e-01 -7.70713627e-01 8.83275867e-01 5.37459791e-01
5.52215397e-01 -4.54079092e-01 4.79894519e-01 2.88902044e-01
-5.63944399e-01 4.31573056e-02 -4.48958188e-01 2.06011996e-01
7.59592727e-02 4.91337389e-01 -9.82699394e-01 1.76571280e-01
5.31743467e-01 4.63297337e-01 -1.18504524e+00 8.56199145e-01
-3.96577358e-01 7.39875376e-01 -1.25432938e-01 -4.37560737e-01
7.88093433e-02 3.57457280e-01 6.43870592e-01 1.37814140e+00
1.65192917e-01 -1.52541652e-01 2.21470997e-01 4.14952248e-01
-5.30574977e-01 3.73379081e-01 -2.57818848e-01 -5.45662463e-01
7.64302492e-01 1.07857168e+00 -1.14714217e+00 -6.46422207e-01
-2.31369939e-02 7.78349817e-01 3.63798052e-01 4.04462039e-01
-2.15078995e-01 -7.71726310e-01 -1.49563566e-01 2.69178301e-01
-5.41565903e-02 -4.03123677e-01 -3.12631018e-02 -1.45762801e+00
-1.96036398e-01 -1.01279473e+00 7.58601665e-01 -9.69947636e-01
-1.46720815e+00 6.29337490e-01 2.07394660e-01 -7.60294557e-01
-4.65378076e-01 -3.93954664e-01 -3.24896038e-01 7.11097300e-01
-1.65139484e+00 -8.19225669e-01 1.29586235e-01 1.32847428e-01
4.37965810e-01 -4.00192052e-01 4.53904927e-01 1.57833174e-01
-2.59962648e-01 3.60258132e-01 4.55800682e-01 5.73991776e-01
1.34055555e+00 -1.42601323e+00 8.95994753e-02 9.10712063e-01
6.54091597e-01 1.04396343e+00 6.33667350e-01 -1.04077172e+00
-1.07766902e+00 -8.13313484e-01 1.42538691e+00 -9.95977581e-01
1.15764225e+00 1.50721848e-01 -7.77791440e-01 6.92989707e-01
5.76038897e-01 -7.32668340e-01 7.28792131e-01 3.53287965e-01
-6.62010610e-01 1.62248343e-01 -7.60170579e-01 6.23791575e-01
6.44972801e-01 -8.39810908e-01 -1.13032615e+00 6.30630195e-01
4.69915271e-01 -9.28839967e-02 -7.79938996e-01 2.60882936e-02
3.49413007e-01 -2.45954797e-01 8.77785861e-01 -6.70757473e-01
7.33121097e-01 -2.85040975e-01 -2.73481440e-02 -1.17655718e+00
-3.28671515e-01 -4.78847951e-01 -1.35261908e-01 1.69928539e+00
9.35961306e-01 -3.35399717e-01 4.63848203e-01 6.24993205e-01
3.86544049e-01 -1.96884766e-01 -4.71849471e-01 -5.28281152e-01
2.09241405e-01 1.55152440e-01 2.91522413e-01 1.12246096e+00
2.98073649e-01 8.01468611e-01 -8.61274227e-02 -1.82923093e-01
5.24445415e-01 3.16712886e-01 4.94084865e-01 -1.26151025e+00
9.14461985e-02 -6.20730698e-01 2.02436849e-01 -8.37051153e-01
2.85762429e-01 -1.18646693e+00 1.06362276e-01 -1.66488767e+00
4.74967748e-01 -7.00988233e-01 -3.52863818e-01 5.96431971e-01
-3.79514813e-01 4.53452557e-01 4.99429705e-04 1.08929169e+00
-7.85209894e-01 5.65048084e-02 4.98069257e-01 -3.48420918e-01
-4.03044820e-01 -2.34376475e-01 -9.64630127e-01 8.88661385e-01
7.38076687e-01 -7.64339149e-01 -2.87272751e-01 -8.89123559e-01
7.19442010e-01 -1.12991445e-02 8.58291015e-02 -5.44744253e-01
5.83866298e-01 -3.58180374e-01 5.13265550e-01 -4.72130120e-01
-2.26206154e-01 -2.11096674e-01 -4.14743394e-01 4.26190011e-02
-1.14570725e+00 6.10288918e-01 -1.43689811e-01 4.19829518e-01
-2.31464967e-01 -8.28871131e-01 4.39013302e-01 -3.27281505e-01
-7.34808683e-01 -1.71230122e-01 -6.04224324e-01 6.55772984e-01
2.67294377e-01 2.79917896e-01 -5.66474855e-01 -2.26921514e-01
-3.46015990e-01 3.44204181e-03 7.81827390e-01 4.59290117e-01
3.38948607e-01 -1.17155278e+00 -7.64730394e-01 -4.40027863e-01
2.92381167e-01 -6.94180310e-01 -9.03172612e-01 3.51507396e-01
-5.59133530e-01 4.66581464e-01 -5.57896048e-02 -4.98913825e-02
-9.34343696e-01 1.62683770e-01 -3.43812555e-02 -3.85358334e-01
-1.12987630e-01 5.48062444e-01 -6.37753963e-01 -3.45017791e-01
2.30278745e-01 2.57098734e-01 -5.78944802e-01 6.56755447e-01
6.88972294e-01 7.80186057e-01 2.83857603e-02 -7.05222905e-01
-3.20721179e-01 2.73548603e-01 -2.56720513e-01 -7.26358235e-01
9.92780149e-01 -3.61788541e-01 -5.90333641e-01 6.27447188e-01
7.86508918e-01 7.28051245e-01 -4.11833614e-01 -4.60675001e-01
8.15468729e-01 -2.71024674e-01 -2.84038261e-02 -1.08305061e+00
-3.90756696e-01 4.48400885e-01 -1.13494746e-01 1.97045982e-01
6.82583272e-01 1.72742307e-01 4.90168571e-01 8.81555378e-01
4.04353738e-01 -1.35299444e+00 1.49979621e-01 6.90954864e-01
7.88785696e-01 -1.24073732e+00 4.52505857e-01 1.20425876e-02
-5.48452437e-01 1.15393937e+00 3.33457798e-01 2.16964573e-01
6.34339869e-01 3.15327048e-02 2.86444634e-01 -1.60135090e-01
-4.69426990e-01 -5.52883707e-02 6.88926578e-01 1.52174458e-01
7.13286817e-01 -4.05615121e-01 -4.69693244e-01 8.66202191e-02
-2.04514340e-01 -3.92939717e-01 8.38364720e-01 9.90235984e-01
-5.98454714e-01 -1.04626012e+00 -5.49582243e-01 6.79650784e-01
-9.19782817e-01 -3.45444083e-01 -1.02872801e+00 5.72781920e-01
-4.67242837e-01 1.12249756e+00 -1.61909699e-01 5.78063093e-02
-3.14827293e-01 1.95573404e-01 1.91528752e-01 -8.72007847e-01
-5.87556243e-01 2.94804484e-01 4.29408967e-01 2.08037607e-02
-7.72088528e-01 -7.65132368e-01 -1.08527100e+00 -1.24460906e-01
-4.92128849e-01 8.82632971e-01 4.24331784e-01 9.19177234e-01
2.33051091e-01 -1.16059929e-01 3.82870495e-01 -4.78694558e-01
-7.04976499e-01 -1.03775740e+00 -1.33749917e-01 3.18026900e-01
1.55942678e-01 -4.18571293e-01 -5.24803936e-01 3.36702108e-01] | [12.011384010314941, 8.581085205078125] |
df872e89-696a-460f-877e-6d7a8eee2c46 | scalable-3d-captioning-with-pretrained-models | 2306.07279 | null | https://arxiv.org/abs/2306.07279v2 | https://arxiv.org/pdf/2306.07279v2.pdf | Scalable 3D Captioning with Pretrained Models | We introduce Cap3D, an automatic approach for generating descriptive text for 3D objects. This approach utilizes pretrained models from image captioning, image-text alignment, and LLM to consolidate captions from multiple views of a 3D asset, completely side-stepping the time-consuming and costly process of manual annotation. We apply Cap3D to the recently introduced large-scale 3D dataset, Objaverse, resulting in 660k 3D-text pairs. Our evaluation, conducted using 41k human annotations from the same dataset, demonstrates that Cap3D surpasses human-authored descriptions in terms of quality, cost, and speed. Through effective prompt engineering, Cap3D rivals human performance in generating geometric descriptions on 17k collected annotations from the ABO dataset. Finally, we finetune Text-to-3D models on Cap3D and human captions, and show Cap3D outperforms; and benchmark the SOTA including Point-E, Shape-E, and DreamFusion. | ['Justin Johnson', 'Honglak Lee', 'Chris Rockwell', 'Tiange Luo'] | 2023-06-12 | null | null | null | null | ['image-captioning', 'text-to-3d', 'prompt-engineering'] | ['computer-vision', 'computer-vision', 'natural-language-processing'] | [ 1.44708473e-02 3.91071528e-01 2.31858954e-01 -4.76136595e-01
-1.43397510e+00 -9.69832599e-01 7.85987735e-01 -9.91983935e-02
-1.87076315e-01 3.09220225e-01 6.08103275e-01 -6.95320517e-02
5.84335029e-01 -1.32103816e-01 -1.08390892e+00 -2.63832379e-02
2.91187763e-01 1.06649101e+00 1.58230036e-01 -5.91125749e-02
1.54057190e-01 4.10880923e-01 -1.24100482e+00 3.19133013e-01
3.44718635e-01 9.70571637e-01 2.18250826e-01 7.55219340e-01
-2.10338131e-01 5.08754253e-01 -3.71480554e-01 -7.35182047e-01
3.59459728e-01 -2.91817427e-01 -7.73810863e-01 4.71240699e-01
1.22083616e+00 -6.09060585e-01 -2.21751705e-01 5.43353915e-01
5.11382222e-01 -3.14513087e-01 8.85939062e-01 -1.32775998e+00
-1.03404808e+00 1.54684946e-01 -7.20084548e-01 -1.73531547e-01
6.79577351e-01 3.63640845e-01 9.68366385e-01 -1.35594606e+00
8.69689524e-01 1.46759236e+00 8.61970603e-01 6.90742075e-01
-1.38202536e+00 -3.81339014e-01 -1.35908023e-01 -4.72978055e-01
-1.41860557e+00 -5.49285650e-01 4.86173421e-01 -6.55004382e-01
1.16591132e+00 4.17515375e-02 6.22708797e-01 1.28523827e+00
-7.53943101e-02 1.00660491e+00 8.76311421e-01 -3.01380038e-01
-3.57532687e-02 1.24967404e-01 -2.09556580e-01 6.07528269e-01
1.58566549e-01 -5.94319403e-02 -5.28635204e-01 4.72661294e-03
8.72770071e-01 -5.30089200e-01 -1.57327652e-01 -6.32054448e-01
-1.62690520e+00 4.90979731e-01 4.36297029e-01 -2.29951099e-01
-1.91530198e-01 4.64767665e-01 3.66450846e-01 -2.83002138e-01
6.84836507e-01 7.00157285e-01 -3.49826485e-01 -2.40455359e-01
-8.93071711e-01 5.12650132e-01 5.18999338e-01 1.74157846e+00
5.46440244e-01 -1.74500048e-01 -2.38704354e-01 4.54380453e-01
3.17757070e-01 1.26424885e+00 -4.46979068e-02 -1.14236009e+00
8.28404427e-01 5.83705783e-01 4.22778934e-01 -8.01220715e-01
-4.33732450e-01 -5.61470613e-02 -5.49154580e-01 -2.76087020e-02
2.39788130e-01 4.93306518e-02 -1.38909888e+00 1.34037519e+00
2.65207767e-01 -2.72730887e-01 3.57809067e-02 1.11556458e+00
1.14017653e+00 7.38097191e-01 1.65038750e-01 4.75342721e-01
1.22761655e+00 -1.25345743e+00 -3.30536038e-01 -4.35147047e-01
6.94538474e-01 -9.54583347e-01 1.08167291e+00 -6.72758222e-02
-1.34039402e+00 -5.87682247e-01 -7.33725667e-01 -5.93886554e-01
-2.72405416e-01 1.17552541e-01 2.56084293e-01 2.09580943e-01
-1.34704721e+00 5.50348572e-02 -4.52388018e-01 -5.45632899e-01
6.86006129e-01 1.28097802e-01 -8.89277160e-01 -1.40678644e-01
-5.93252063e-01 1.19661295e+00 3.35012734e-01 -1.52746484e-01
-1.15187633e+00 -9.61838782e-01 -1.08751631e+00 -2.17035696e-01
1.59370407e-01 -9.93668258e-01 1.56068921e+00 -4.43203777e-01
-8.99341166e-01 1.53306413e+00 3.09337471e-02 -4.11081195e-01
7.38630176e-01 -5.73106587e-01 -5.32592386e-02 3.82638454e-01
3.38794559e-01 1.57500672e+00 6.20046794e-01 -1.56383872e+00
-2.48069644e-01 -2.23091200e-01 -5.99914864e-02 4.60283458e-01
2.28104711e-01 -1.07594490e-01 -8.70067954e-01 -4.41682011e-01
5.10972068e-02 -1.15190148e+00 -1.69331893e-01 3.84750694e-01
-6.13400996e-01 -3.19487229e-02 7.88596094e-01 -7.02758729e-01
3.91427845e-01 -2.10980272e+00 1.96323246e-01 -1.49211794e-01
3.09467286e-01 1.02721177e-01 -3.82267833e-01 3.26459497e-01
1.20236287e-02 3.50801140e-01 -3.31640929e-01 -9.72291768e-01
1.75537527e-01 5.57693802e-02 -4.27143604e-01 2.85516232e-01
5.13068318e-01 1.60737586e+00 -8.68685842e-01 -8.92647445e-01
5.80129325e-01 4.98299658e-01 -6.24646664e-01 4.20992881e-01
-6.54143274e-01 3.46910655e-01 -3.43173623e-01 6.29663289e-01
6.09911740e-01 -7.20399737e-01 -2.79168397e-01 -5.05001068e-01
1.57485202e-01 -5.02866693e-02 -5.57453513e-01 2.27837467e+00
-4.51682657e-01 7.90704787e-01 -2.18447521e-01 -4.60103452e-01
9.63043094e-01 3.51388782e-01 4.08687472e-01 -5.94847500e-01
6.87973872e-02 2.15077326e-01 -8.24129939e-01 -6.90615892e-01
7.09168196e-01 1.31517157e-01 -5.05702734e-01 6.12267911e-01
3.26610684e-01 -1.18006217e+00 -9.25746262e-02 5.31804085e-01
8.40472221e-01 7.04899609e-01 6.58092573e-02 -2.52173729e-02
9.50078815e-02 4.34634835e-01 -3.46485645e-01 6.18763328e-01
-1.32395327e-01 1.36139870e+00 3.26596677e-01 -5.66162586e-01
-1.79546022e+00 -1.18987322e+00 1.45384986e-02 4.62391555e-01
2.89853752e-01 -4.35667276e-01 -7.37683475e-01 -9.80740547e-01
4.19947915e-02 8.88081968e-01 -7.36462355e-01 3.19294900e-01
-4.14367169e-01 -1.48625910e-01 5.67365468e-01 5.36847532e-01
4.15368170e-01 -9.07959938e-01 -4.13987756e-01 -8.73642489e-02
-3.55320096e-01 -1.66015685e+00 -9.17549551e-01 -1.26459867e-01
-6.22242332e-01 -8.94089460e-01 -1.17672658e+00 -8.07176530e-01
9.00311470e-01 4.01360303e-01 1.68733501e+00 -1.94390282e-01
-2.43364543e-01 7.92386353e-01 -4.28796232e-01 -4.31315154e-01
-6.98327482e-01 8.15902948e-02 -9.72990617e-02 -5.16621470e-01
2.24285483e-01 -4.47492421e-01 -3.70123476e-01 3.67683411e-01
-7.53250003e-01 7.11445451e-01 7.70292401e-01 4.24415171e-01
6.82842314e-01 -9.05302525e-01 1.70182750e-01 -4.16046560e-01
5.65587450e-03 -2.07859367e-01 -5.27275443e-01 3.20575058e-01
-2.71911502e-01 1.39374331e-01 1.18067130e-01 -2.00491130e-01
-8.12918782e-01 5.47488451e-01 -6.75199600e-03 -9.06024456e-01
-2.72403210e-01 -1.97760388e-01 -4.82086837e-02 9.40367430e-02
7.68327653e-01 3.59457016e-01 1.41173184e-01 -4.18309540e-01
9.83542383e-01 6.43104196e-01 8.72130394e-01 -5.42900860e-01
1.09167707e+00 4.98040617e-01 -1.72682643e-01 -4.90791380e-01
-1.24989021e+00 -2.91975945e-01 -8.75413716e-01 -2.42925063e-01
1.43588603e+00 -1.32837808e+00 -3.68408918e-01 2.72553146e-01
-1.72257769e+00 -2.13874415e-01 -2.56959766e-01 1.75754234e-01
-1.14079297e+00 7.30125606e-02 -3.79897773e-01 -3.80845487e-01
-6.57741964e-01 -9.80961323e-01 2.06608605e+00 -2.35080514e-02
-3.83075207e-01 -6.81243479e-01 -3.39980945e-02 8.23281944e-01
1.65268153e-01 4.93381858e-01 5.25021672e-01 -5.63222528e-01
-7.32108712e-01 -3.40370327e-01 -7.15819299e-01 3.20392609e-01
-9.59469080e-02 -2.74764866e-01 -9.60499763e-01 1.25097543e-01
-6.29992783e-01 -8.81968677e-01 3.73888999e-01 1.73371404e-01
8.60609353e-01 -1.89144343e-01 -3.07018697e-01 4.90511566e-01
1.23539853e+00 -2.36564726e-01 4.41072524e-01 1.64960966e-01
1.02459967e+00 4.21811074e-01 6.39681399e-01 1.96344301e-01
5.63268423e-01 7.81082749e-01 6.18445516e-01 -4.18327540e-01
-4.47427332e-01 -7.67830610e-01 8.16585869e-02 5.61907649e-01
1.66377157e-01 -5.66616476e-01 -1.15600014e+00 7.55765080e-01
-1.67333293e+00 -7.70332932e-01 -1.71206161e-01 1.57155859e+00
7.80602932e-01 1.53619453e-01 -7.66173378e-02 -4.90929037e-01
7.34088540e-01 6.77345172e-02 -5.75414956e-01 -8.56501050e-03
-2.17524514e-01 -2.06160963e-01 5.94430029e-01 4.16090012e-01
-9.89963412e-01 1.07950425e+00 6.90814447e+00 4.39402878e-01
-7.95536757e-01 -6.45335242e-02 8.26045930e-01 -3.14335674e-01
-3.95079195e-01 -1.44991934e-01 -8.75913382e-01 1.74736097e-01
8.13268960e-01 -2.34872133e-01 7.39464834e-02 1.05290580e+00
2.56521463e-01 7.08475560e-02 -1.54950094e+00 1.37085652e+00
6.05752766e-01 -1.68582630e+00 5.31355739e-01 1.30116820e-01
8.52278352e-01 4.25598294e-01 -1.06245331e-01 1.53483987e-01
3.86596173e-01 -1.14341545e+00 1.22784233e+00 3.98471892e-01
1.06864738e+00 -3.28916967e-01 5.98397374e-01 4.54613613e-03
-7.91845560e-01 5.26636243e-01 4.62107211e-02 4.10653591e-01
8.20698559e-01 3.67215723e-01 -1.32672489e+00 5.12370646e-01
8.52276266e-01 8.74797165e-01 -8.36371601e-01 7.53045142e-01
-8.97235498e-02 1.17001049e-02 -3.51498932e-01 3.08056939e-02
5.29962778e-01 2.49343723e-01 5.50149739e-01 1.23124433e+00
4.16192889e-01 1.47970572e-01 9.61494520e-02 9.90903258e-01
-2.70008951e-01 -2.00164005e-01 -8.82748365e-01 -2.17063501e-01
3.77683848e-01 1.18136525e+00 -6.23296082e-01 -5.54536045e-01
-2.16301829e-01 1.15370274e+00 1.14178136e-01 8.43351856e-02
-1.05064178e+00 -3.16815078e-02 3.54741573e-01 2.59233713e-01
3.77615869e-01 -4.44440812e-01 -2.27304310e-01 -9.67563510e-01
1.19479880e-01 -6.88712239e-01 -1.11464746e-01 -1.89038599e+00
-1.34041369e+00 8.98613691e-01 3.08349609e-01 -1.44817567e+00
-2.99124926e-01 -5.43836713e-01 7.69740939e-02 7.16562390e-01
-1.16248739e+00 -1.61644304e+00 -6.47252798e-01 2.87538290e-01
8.47616613e-01 1.83136508e-01 7.43373215e-01 1.55220866e-01
3.08981016e-02 3.03883165e-01 -3.51188302e-01 1.47488490e-01
9.18945372e-01 -1.17301619e+00 1.32982409e+00 3.77888262e-01
3.27516526e-01 -8.50792378e-02 6.91069901e-01 -6.08921885e-01
-1.20522571e+00 -1.28894365e+00 8.78490746e-01 -1.42543507e+00
5.89794636e-01 -7.94185698e-01 -5.04566967e-01 8.32396865e-01
2.61885673e-01 1.50157169e-01 3.42439532e-01 -4.07201737e-01
-5.24497747e-01 3.98912013e-01 -1.08670318e+00 6.62648320e-01
1.43399739e+00 -4.99005109e-01 -7.83191741e-01 6.76804304e-01
1.34582400e+00 -9.07127380e-01 -9.26706970e-01 1.31945640e-01
4.02590722e-01 -6.90019310e-01 1.12269187e+00 -4.18854982e-01
9.13552761e-01 -4.08987284e-01 -3.43043983e-01 -1.02907360e+00
1.68615896e-02 -6.53176606e-01 7.59210512e-02 1.16743493e+00
5.39974332e-01 1.85701519e-01 8.06537747e-01 7.77444601e-01
-4.97737020e-01 -5.32607913e-01 -7.69535780e-01 -7.29365945e-01
-1.25845000e-02 -4.79489088e-01 6.64307952e-01 7.86233306e-01
-4.41865593e-01 5.78214943e-01 -4.52463776e-01 -1.89553648e-02
7.77666211e-01 -1.38662094e-02 1.31609976e+00 -8.66787195e-01
-5.46785817e-03 -3.93184982e-02 -5.48708737e-01 -1.36273050e+00
7.55906478e-02 -9.64038491e-01 2.52215326e-01 -1.75894427e+00
2.93952107e-01 -4.35581088e-01 5.99903584e-01 7.37426043e-01
9.33366269e-03 8.08741570e-01 4.85662401e-01 3.03250700e-01
-9.14630949e-01 6.83070660e-01 1.55957806e+00 -2.74540514e-01
-4.91190031e-02 -6.52094722e-01 -6.46746874e-01 6.72387779e-01
3.18668932e-01 -3.37878108e-01 -3.24951530e-01 -9.38799441e-01
9.39322412e-02 7.70480707e-02 6.12506807e-01 -1.02264667e+00
-1.48697793e-01 6.36744499e-02 6.43351674e-01 -1.12824607e+00
8.01871121e-01 -8.13836396e-01 1.15104035e-01 -1.51446417e-01
-4.37546492e-01 3.94783407e-01 5.61680198e-01 5.12705684e-01
1.38457581e-01 1.03704706e-01 4.68343407e-01 -2.90947318e-01
-6.79207683e-01 4.09056455e-01 2.52492577e-01 4.45691258e-01
1.00101483e+00 -3.43098372e-01 -5.19398689e-01 -4.82108653e-01
-6.03469253e-01 5.15188277e-01 8.24451029e-01 8.86575878e-01
6.72336936e-01 -1.37995565e+00 -9.71764088e-01 -7.87868798e-02
6.03855610e-01 5.97211599e-01 2.27926970e-01 4.32966411e-01
-7.89442837e-01 6.21831059e-01 -8.85992497e-02 -1.20748997e+00
-1.10116279e+00 5.27332366e-01 2.59925187e-01 6.19666800e-02
-1.04609549e+00 8.81131589e-01 3.52973938e-01 -5.48186183e-01
8.55798870e-02 -3.08689326e-01 2.96347290e-01 -3.01127136e-01
3.30130458e-01 -2.70080686e-01 -9.65233371e-02 -8.91350210e-01
-4.09524381e-01 9.43001151e-01 -1.76373854e-01 -4.51036334e-01
1.37110496e+00 -3.33252579e-01 3.70172530e-01 2.05431268e-01
1.31770456e+00 -3.66976410e-01 -1.61404359e+00 -1.77808478e-02
-7.25393742e-02 -3.08510005e-01 -1.26906708e-01 -9.78132844e-01
-5.98095179e-01 6.92886651e-01 3.47897440e-01 -2.76636273e-01
6.43270373e-01 6.85104370e-01 9.89748478e-01 4.43677783e-01
3.72841537e-01 -7.21271396e-01 4.40352350e-01 3.75652730e-01
1.31473196e+00 -1.47123921e+00 1.46674305e-01 -3.62752795e-01
-9.11794782e-01 8.60374868e-01 8.39124858e-01 -6.29736036e-02
2.43815914e-01 5.18815927e-02 2.01783121e-01 -3.57456028e-01
-6.98381543e-01 1.38303950e-01 4.55276698e-01 6.80063784e-01
1.63511008e-01 -1.88577905e-01 5.59690893e-01 8.95940438e-02
-4.80999291e-01 -6.01380505e-02 6.95189834e-01 7.01330483e-01
-7.59898871e-02 -6.17145181e-01 -4.31821585e-01 1.37835354e-01
8.39901492e-02 -2.12346673e-01 -6.05839372e-01 1.05308270e+00
-1.51455775e-01 5.82882226e-01 3.87879342e-01 -2.02159360e-01
5.45391858e-01 1.11266859e-02 5.93584120e-01 -6.42592788e-01
-1.55461937e-01 -1.42023891e-01 3.62232000e-01 -6.30698025e-01
-5.99447489e-01 -4.49796557e-01 -1.16560388e+00 -4.12934870e-02
-1.20038085e-01 -2.15189960e-02 1.02157879e+00 8.58375669e-01
7.23580122e-01 9.56464112e-02 3.96546900e-01 -1.34572589e+00
-3.05803925e-01 -8.45845520e-01 -5.15208058e-02 6.41076446e-01
3.99592310e-01 -5.26629269e-01 -4.73530442e-01 5.43083608e-01] | [8.199196815490723, -3.289128303527832] |
bc8f4206-79cb-46be-b912-0655391ecefb | cst-yolo-a-novel-method-for-blood-cell | 2306.14590 | null | https://arxiv.org/abs/2306.14590v1 | https://arxiv.org/pdf/2306.14590v1.pdf | CST-YOLO: A Novel Method for Blood Cell Detection Based on Improved YOLOv7 and CNN-Swin Transformer | Blood cell detection is a typical small-scale object detection problem in computer vision. In this paper, we propose a CST-YOLO model for blood cell detection based on YOLOv7 architecture and enhance it with the CNN-Swin Transformer (CST), which is a new attempt at CNN-Transformer fusion. We also introduce three other useful modules: Weighted Efficient Layer Aggregation Networks (W-ELAN), Multiscale Channel Split (MCS), and Concatenate Convolutional Layers (CatConv) in our CST-YOLO to improve small-scale object detection precision. Experimental results show that the proposed CST-YOLO achieves 92.7, 95.6, and 91.1 mAP@0.5 respectively on three blood cell datasets, outperforming state-of-the-art object detectors, e.g., YOLOv5 and YOLOv7. Our code is available at https://github.com/mkang315/CST-YOLO. | ['Raphaël Phan', 'Fung Fung Ting', 'Chee-Ming Ting', 'Ming Kang'] | 2023-06-26 | null | null | null | null | ['cell-detection', 'blood-cell-detection'] | ['computer-vision', 'medical'] | [-6.03152871e-01 -4.54628468e-01 2.28784323e-01 -2.92238616e-03
-4.40690488e-01 -6.11759685e-02 3.70666414e-01 2.64410406e-01
-5.75562119e-01 5.49846292e-01 -6.62114471e-02 -1.57852117e-02
6.39298320e-01 -9.39644098e-01 -3.59466106e-01 -8.95168364e-01
3.51422280e-02 1.95073247e-01 7.60019124e-01 4.50882092e-02
-1.16779953e-01 6.57502592e-01 -9.64060247e-01 5.59627950e-01
7.41641402e-01 1.52223706e+00 -2.08186761e-01 1.17462814e+00
-2.55630046e-01 1.20448267e+00 -4.67213482e-01 -6.24089479e-01
5.74471392e-02 -4.92285788e-01 -2.01395810e-01 -1.43461466e-01
3.93670797e-01 -3.58900130e-01 -4.50463176e-01 8.97464991e-01
9.83702004e-01 -4.15596396e-01 5.83733857e-01 -1.25511205e+00
-6.54511988e-01 5.08476079e-01 -6.44195855e-01 8.45382690e-01
-1.36797115e-01 3.06696415e-01 3.87541264e-01 -1.20553935e+00
3.85584533e-01 1.40868878e+00 8.51102531e-01 3.85166258e-01
-9.08534825e-01 -8.44938815e-01 -2.21664160e-01 3.15556914e-01
-1.59598303e+00 -4.02268142e-01 2.60954738e-01 -4.70941007e-01
9.74522352e-01 1.90779999e-01 7.18118131e-01 5.57815373e-01
3.75983149e-01 1.30063474e+00 9.30451810e-01 -3.01267326e-01
-6.05226867e-03 1.61863282e-01 1.94165424e-01 1.26450205e+00
7.41294920e-01 -1.45158887e-01 -5.01414776e-01 9.79683846e-02
1.15824568e+00 2.79814273e-01 -8.12945664e-02 -1.85796216e-01
-1.03303969e+00 7.82333553e-01 8.46429229e-01 3.24927896e-01
-2.06942528e-01 3.70813340e-01 5.72437227e-01 1.19209573e-01
2.83916920e-01 2.22906806e-02 -1.43824205e-01 4.73767906e-01
-6.09849751e-01 2.18174741e-01 6.18300796e-01 1.12432361e+00
4.63252276e-01 3.62781435e-01 -7.33467937e-01 7.19377577e-01
2.68482327e-01 5.65653563e-01 2.92209923e-01 -5.30431390e-01
3.77810374e-02 9.85670030e-01 -2.60723084e-02 -4.59161073e-01
-5.91923952e-01 -8.48243952e-01 -1.03249395e+00 1.67540014e-01
6.43945992e-01 -2.48433828e-01 -1.05091274e+00 1.10267866e+00
4.03151482e-01 6.73041940e-01 6.15377491e-03 1.04963052e+00
1.67039955e+00 4.09770846e-01 3.20841134e-01 1.86494485e-01
1.85808694e+00 -1.23177242e+00 -3.37744534e-01 2.72550523e-01
8.00808072e-01 -7.77412176e-01 5.01246512e-01 2.02037126e-01
-1.07120788e+00 -8.24931324e-01 -9.86286402e-01 -2.52977908e-01
-5.33782840e-01 5.22683859e-01 6.42736018e-01 6.60470188e-01
-9.31513250e-01 3.75195965e-03 -9.56448555e-01 -5.92754960e-01
9.09324586e-01 1.85751081e-01 -7.90547132e-02 -8.58789831e-02
-8.88451457e-01 7.87072301e-01 1.07443117e-01 1.24550208e-01
-1.01018846e+00 -9.28103983e-01 -6.64454520e-01 1.64646983e-01
2.08917126e-01 -9.32489812e-01 1.15611815e+00 -3.28155905e-01
-9.80934799e-01 9.71267164e-01 -3.59984972e-02 -7.95455039e-01
6.57654762e-01 -1.56557471e-01 -3.13797683e-01 6.51787400e-01
1.02901153e-01 8.32257509e-01 2.99240261e-01 -7.97107041e-01
-9.34078395e-01 -1.78113386e-01 -2.65222520e-01 -2.89853364e-01
3.89816496e-03 2.44397253e-01 -6.65938675e-01 -7.73591578e-01
-4.20736849e-01 -4.35602844e-01 -2.41116226e-01 6.29018903e-01
-3.93231571e-01 -5.70151985e-01 6.38263285e-01 -2.79706717e-01
1.13074422e+00 -1.78675222e+00 -4.61976051e-01 -2.69369446e-02
7.11596370e-01 6.42712235e-01 -1.97592556e-01 9.89968255e-02
3.61270428e-01 -2.51877457e-01 5.36060706e-02 -4.16160315e-01
-2.21478999e-01 -2.15671092e-01 1.84501633e-01 5.64183652e-01
3.13966572e-01 1.38520646e+00 -8.11495364e-01 -1.06376171e+00
6.24332488e-01 9.30104256e-01 -3.08973432e-01 2.06139863e-01
2.45655328e-01 1.57256812e-01 -5.49953818e-01 1.38119578e+00
8.54552448e-01 -5.34652233e-01 -3.30715925e-01 -4.83158261e-01
-2.50529200e-01 -5.08809626e-01 -1.01998639e+00 1.15566945e+00
-8.89266655e-02 6.35714293e-01 -6.73841266e-03 -8.34830701e-01
9.31269705e-01 1.57748386e-01 4.89282370e-01 -6.45085096e-01
7.30172813e-01 3.46753336e-02 -5.50674200e-02 -5.04898131e-01
2.03352123e-02 -1.89604029e-01 3.24546129e-01 -8.97824317e-02
4.49120045e-01 5.46036780e-01 5.50190985e-01 4.42206323e-01
9.66075778e-01 -3.12830359e-01 3.97603631e-01 -4.11246091e-01
8.34109485e-01 -4.22770344e-02 7.96371937e-01 9.39457893e-01
-7.20998168e-01 6.60186708e-01 5.56238353e-01 -7.01409221e-01
-8.46231759e-01 -1.32327783e+00 -3.63682032e-01 8.49509239e-01
2.19169110e-01 -1.38622299e-01 -8.47924471e-01 -4.62528706e-01
4.04423714e-01 4.11404856e-02 -7.70971715e-01 1.97938934e-01
-7.39612699e-01 -1.12880576e+00 8.61540020e-01 9.86028910e-01
9.24728572e-01 -1.16938913e+00 -1.04882085e+00 4.79111403e-01
1.08340010e-01 -1.16800654e+00 -4.54081178e-01 -2.13332891e-01
-6.80133462e-01 -1.42694271e+00 -1.00093937e+00 -9.76842999e-01
5.88683367e-01 3.89360517e-01 1.09513009e+00 5.78297555e-01
-1.14019489e+00 4.50432077e-02 -4.29150492e-01 -6.28659368e-01
2.06221253e-01 -2.21652538e-01 -2.33563080e-01 1.50537163e-01
9.93415475e-01 1.18658589e-02 -1.23210847e+00 1.89365119e-01
-6.04194999e-01 -4.41672541e-02 7.67723322e-01 8.97001684e-01
5.70844769e-01 -5.46978116e-01 7.76913643e-01 -9.50515985e-01
1.80715889e-01 -1.60989776e-01 -6.42315865e-01 4.01276916e-01
-4.62309420e-01 -3.95346940e-01 5.10661185e-01 -3.35709155e-01
-9.12234366e-01 6.57571033e-02 -1.24765903e-01 -5.44948101e-01
-1.20587699e-01 -1.28840983e-01 3.77043933e-01 -3.84381026e-01
4.81047988e-01 3.92238498e-01 -1.31769791e-01 -5.50968945e-01
1.83558613e-01 5.88720024e-01 6.86937213e-01 -1.53408319e-01
2.94000745e-01 9.33814049e-01 -1.79746523e-02 -7.73536146e-01
-7.11064637e-01 -6.95669711e-01 -5.17093599e-01 -2.93781847e-01
9.83290792e-01 -1.29133046e+00 -9.65098023e-01 8.11262846e-01
-8.25364590e-01 -1.28231272e-01 7.04642683e-02 4.93733853e-01
-1.22478522e-01 2.04061985e-01 -1.32536495e+00 -6.98891819e-01
-1.12297833e+00 -1.04505408e+00 1.00500274e+00 8.26256931e-01
1.43426865e-01 -9.36056674e-01 -3.43284845e-01 2.83181190e-01
6.96627259e-01 5.79032063e-01 3.37116063e-01 -3.27992082e-01
-7.21794724e-01 -2.06040040e-01 -9.01052892e-01 8.82137716e-02
-2.58034140e-01 3.39883380e-02 -9.22616780e-01 -5.87620854e-01
-4.33088362e-01 -3.19105625e-01 1.58751988e+00 9.02752280e-01
9.47248280e-01 1.51353121e-01 -5.72130799e-01 9.71366942e-01
1.54769933e+00 2.41308406e-01 6.04741693e-01 2.41949812e-01
5.78449428e-01 8.35641176e-02 3.43681633e-01 5.79714715e-01
4.31361496e-01 2.92861879e-01 6.88530505e-02 -8.27441394e-01
-6.68812752e-01 7.31776580e-02 -7.93447196e-02 5.55147171e-01
-1.53177585e-02 -1.87972218e-01 -8.12219441e-01 5.00733376e-01
-1.78018034e+00 -7.46135831e-01 -3.76678079e-01 1.50014305e+00
3.91971618e-01 1.70489073e-01 3.47366571e-01 -3.33177179e-01
7.21399307e-01 -2.39083320e-01 -5.82446694e-01 -1.94383822e-02
-3.57539117e-01 4.13583517e-01 6.41289413e-01 4.58105564e-01
-1.46363986e+00 1.13691497e+00 5.23890972e+00 9.92012858e-01
-9.62510645e-01 2.57976204e-01 1.02131641e+00 -2.21096948e-01
6.30909085e-01 -5.37626326e-01 -1.15193927e+00 5.20916402e-01
2.22888082e-01 2.14294400e-02 -2.01610416e-01 7.86472976e-01
-1.16801955e-01 -3.68758619e-01 -8.86053264e-01 1.09087622e+00
1.72986656e-01 -1.65682232e+00 4.61760052e-02 -4.86142218e-01
5.06825566e-01 -5.15105249e-03 -7.87113234e-02 4.63528842e-01
1.64600179e-01 -9.84544635e-01 2.94717610e-01 5.48704147e-01
8.56228828e-01 -6.48754597e-01 1.21989977e+00 -1.51053980e-01
-1.76455021e+00 1.55553326e-01 -7.84414589e-01 2.00660929e-01
9.57809463e-02 5.52303970e-01 -6.90746129e-01 3.29838023e-02
1.02766013e+00 7.84643710e-01 -8.30822706e-01 1.64495456e+00
7.12505877e-02 4.64426935e-01 -2.22343013e-01 -3.35534275e-01
2.31225193e-01 2.09605485e-01 3.34399313e-01 1.92122447e+00
2.29193091e-01 3.47301036e-01 5.67357279e-02 7.14251220e-01
-1.84894279e-01 -8.12299475e-02 -1.02370955e-01 4.95495290e-01
5.76393723e-01 1.68463922e+00 -1.12171197e+00 -9.49800909e-01
-5.50664842e-01 6.79222643e-01 1.07630387e-01 2.41968781e-01
-7.73464799e-01 -6.25707924e-01 6.52662814e-01 -7.80573040e-02
6.97607994e-01 2.77128100e-01 -2.87303180e-01 -1.28575587e+00
-3.84184629e-01 -1.91426232e-01 8.32485974e-01 -5.84327936e-01
-1.47760367e+00 6.83366656e-01 -4.42348063e-01 -1.21648419e+00
5.25302231e-01 -8.81139994e-01 -8.64361703e-01 4.47728902e-01
-1.85833836e+00 -1.31511283e+00 -6.79555178e-01 6.30046785e-01
5.48977494e-01 -3.52175623e-01 3.20860922e-01 4.62779880e-01
-8.25893581e-01 8.64454091e-01 -1.79265574e-01 7.78538764e-01
6.26566648e-01 -1.15745389e+00 2.80637503e-01 8.30181181e-01
-2.66804069e-01 5.00082195e-01 2.43766487e-01 -4.17886138e-01
-1.24261796e+00 -1.38751733e+00 7.38532186e-01 -4.26706374e-01
4.52552199e-01 -5.36572218e-01 -7.51438797e-01 5.10933876e-01
3.47286791e-01 6.85304105e-01 6.51215076e-01 -6.65538371e-01
-3.05813581e-01 -4.07089800e-01 -1.36359143e+00 3.54119390e-01
8.49278271e-01 8.03983808e-02 -2.78328061e-01 1.96902514e-01
5.53715825e-01 -5.56515932e-01 -9.01295900e-01 3.12786490e-01
7.97302067e-01 -1.14326358e+00 1.34024215e+00 -4.44344252e-01
4.44225557e-02 -5.64049542e-01 1.63349569e-01 -7.16744363e-01
-5.14786005e-01 -7.90356621e-02 -4.06308085e-01 1.07665217e+00
4.12053205e-02 -5.94780207e-01 9.94125903e-01 -1.42486870e-01
-1.32511720e-01 -1.15101445e+00 -1.13734794e+00 -6.69957399e-01
1.32017568e-01 1.87834367e-01 5.44316828e-01 6.65297151e-01
-1.93254709e-01 2.98881769e-01 -3.18260103e-01 -3.08855046e-02
1.09692144e+00 3.35414320e-01 7.79306293e-01 -9.79466498e-01
-6.98101744e-02 -6.77620292e-01 -6.00983202e-01 -1.06858873e+00
-7.09570169e-01 -6.96347833e-01 -3.42273295e-01 -1.56150651e+00
6.45373285e-01 -3.07394117e-01 -8.80637586e-01 3.83364916e-01
-3.79816264e-01 8.68736207e-01 3.96985948e-01 -1.09919887e-02
-1.06336081e+00 3.45937818e-01 1.32616377e+00 -9.22118202e-02
5.42795770e-02 -3.32748443e-01 -6.82892025e-01 4.97155398e-01
1.01872516e+00 -3.74734074e-01 3.06895673e-01 -2.36468017e-01
-6.27913833e-01 1.58566646e-02 5.54439485e-01 -1.29545820e+00
5.93759239e-01 2.55206317e-01 1.18585837e+00 -8.80208552e-01
2.52241999e-01 -2.46225566e-01 -2.92884588e-01 1.37129653e+00
-4.66109850e-02 -5.64325452e-02 9.31348577e-02 2.95500010e-01
-2.27363303e-01 4.57832009e-01 1.28259444e+00 -3.10216010e-01
-8.03108275e-01 4.15614128e-01 -4.91323024e-01 -5.39627373e-02
1.25965130e+00 -3.08790535e-01 -8.88010025e-01 2.57277906e-01
-5.96765399e-01 9.96738076e-01 -6.99813059e-03 1.91872090e-01
9.82962728e-01 -1.36901438e+00 -1.13072741e+00 2.98453718e-01
4.06450093e-01 -2.87913829e-01 3.84993762e-01 1.24149013e+00
-1.00580728e+00 8.25269580e-01 -4.07138288e-01 -7.61178672e-01
-1.40678823e+00 4.18171585e-01 6.14050984e-01 -2.84706146e-01
-9.32213187e-01 1.39980161e+00 5.32137752e-01 2.50156187e-02
1.87870473e-01 -5.91747344e-01 -3.04712743e-01 -1.50079086e-01
8.35986614e-01 6.49091065e-01 -2.69538492e-01 -3.66683364e-01
-6.64564848e-01 6.30104959e-01 -1.01430528e-01 7.54936874e-01
1.06665850e+00 8.72420147e-02 4.25243787e-02 -7.63106644e-02
9.65973794e-01 -4.52113688e-01 -1.16477394e+00 -5.29584467e-01
-3.02775860e-01 -3.46387893e-01 -3.67192402e-02 -6.91460967e-01
-1.44887125e+00 1.10102618e+00 9.83653843e-01 -1.60128117e-01
1.08528578e+00 2.68082500e-01 1.20024002e+00 1.82807654e-01
1.48543209e-01 -7.19772518e-01 2.59188592e-01 1.10854156e-01
4.82185662e-01 -1.25975561e+00 1.26774475e-01 -5.38864493e-01
-5.75964510e-01 1.12932205e+00 1.14777553e+00 -4.32608336e-01
6.75012171e-01 5.62862515e-01 3.71847391e-01 -2.98541844e-01
-8.57096076e-01 -7.69608080e-01 7.13416487e-02 4.46196139e-01
6.08102441e-01 3.16523284e-01 -3.17852020e-01 4.05853182e-01
2.77058303e-01 2.39227027e-01 2.98202217e-01 8.60664427e-01
-9.79558527e-01 -5.35657883e-01 -3.48036289e-01 7.65566587e-01
-5.32355189e-01 -1.03299320e-01 -2.10298240e-01 8.45306396e-01
3.63929063e-01 8.27457905e-01 3.10778826e-01 -2.67674655e-01
2.23890170e-01 -3.98685783e-01 4.26638544e-01 -3.78388733e-01
-7.43622184e-01 5.73685825e-01 -2.56000578e-01 -5.71530163e-01
-1.55740172e-01 -4.42667067e-01 -1.35577095e+00 -4.31020945e-01
-2.73449987e-01 -1.10987127e-01 -1.20264262e-01 5.16026318e-01
1.91859052e-01 6.75311923e-01 1.37794673e-01 -3.44434381e-01
-4.13882837e-04 -1.11076438e+00 -5.77089727e-01 8.12892169e-02
5.57852030e-01 -8.19005549e-01 6.43925369e-02 2.45861914e-02] | [14.77100944519043, -3.100222587585449] |
cc6ce6ee-261c-4c16-b734-f55f888a21c2 | cross-lingual-speaker-identification-using | 2210.05780 | null | https://arxiv.org/abs/2210.05780v1 | https://arxiv.org/pdf/2210.05780v1.pdf | Cross-Lingual Speaker Identification Using Distant Supervision | Speaker identification, determining which character said each utterance in literary text, benefits many downstream tasks. Most existing approaches use expert-defined rules or rule-based features to directly approach this task, but these approaches come with significant drawbacks, such as lack of contextual reasoning and poor cross-lingual generalization. In this work, we propose a speaker identification framework that addresses these issues. We first extract large-scale distant supervision signals in English via general-purpose tools and heuristics, and then apply these weakly-labeled instances with a focus on encouraging contextual reasoning to train a cross-lingual language model. We show that the resulting model outperforms previous state-of-the-art methods on two English speaker identification benchmarks by up to 9% in accuracy and 5% with only distant supervision, as well as two Chinese speaker identification datasets by up to 4.7%. | ['Dan Roth', 'Dong Yu', 'Dian Yu', 'Ben Zhou'] | 2022-10-11 | null | null | null | null | ['speaker-identification'] | ['speech'] | [ 1.70674592e-01 -1.10996313e-01 -3.47410530e-01 -7.36231267e-01
-1.48642230e+00 -8.49612474e-01 6.73959792e-01 -1.90609723e-01
-4.12708163e-01 6.65779650e-01 3.00847828e-01 -5.19785285e-01
7.63513744e-02 -2.02064067e-01 -5.67127228e-01 -5.49112439e-01
3.38919371e-01 6.23971939e-01 1.72403976e-01 -2.47141734e-01
1.31599188e-01 2.02235207e-01 -1.40829766e+00 4.66415226e-01
9.64626074e-01 8.80245328e-01 -1.41914502e-01 5.17262399e-01
-2.94723481e-01 9.30417478e-01 -6.34288132e-01 -7.89218903e-01
-2.25734517e-01 -4.43762243e-01 -1.20530665e+00 -3.48539650e-02
4.19604957e-01 -1.13690712e-01 1.77208856e-01 1.09826517e+00
6.27263963e-01 -5.65696172e-02 5.27974486e-01 -8.22892547e-01
-6.58068895e-01 1.32424486e+00 -3.97944063e-01 2.42632866e-01
5.23526490e-01 -8.33526999e-02 1.07282555e+00 -9.25520539e-01
4.07455087e-01 1.26405597e+00 8.71432424e-01 9.58618343e-01
-1.37992334e+00 -6.91587687e-01 3.14823776e-01 2.88729936e-01
-1.51045537e+00 -1.13462806e+00 8.32967699e-01 -2.56654769e-01
1.03660953e+00 4.37721461e-01 -1.56637847e-01 1.44962525e+00
-6.81632280e-01 1.12825954e+00 1.35867798e+00 -7.50593662e-01
2.64745146e-01 5.34071624e-01 4.14372444e-01 5.94407499e-01
-5.11575162e-01 -1.28353566e-01 -8.06149602e-01 -5.32275951e-03
8.05832222e-02 -5.78805685e-01 -2.22158000e-01 1.95389360e-01
-1.06809807e+00 7.75713503e-01 4.20583673e-02 4.15176600e-01
3.82103771e-02 -5.06562710e-01 6.19365156e-01 2.69330740e-01
4.00939226e-01 3.59631389e-01 -7.30833054e-01 -3.17781866e-01
-9.60097313e-01 -2.00870067e-01 9.88332748e-01 8.95646572e-01
6.42008185e-01 1.16145141e-01 4.06139083e-02 1.31886137e+00
1.75619826e-01 5.81268072e-01 7.13910639e-01 -6.52249992e-01
6.67496562e-01 3.29257578e-01 -2.27045849e-01 -3.83448958e-01
-1.86639592e-01 -3.99887085e-01 -3.02658290e-01 -3.00268441e-01
5.61605215e-01 -3.19307417e-01 -6.72918797e-01 1.86125541e+00
2.46198490e-01 2.11908862e-01 3.75461251e-01 6.48062944e-01
8.85183156e-01 4.79743779e-01 1.01589181e-01 -7.73123503e-02
1.47969639e+00 -1.11447215e+00 -6.16853476e-01 -4.71033871e-01
6.38927877e-01 -7.58593619e-01 1.46498430e+00 2.76026070e-01
-9.53365088e-01 -4.55398649e-01 -9.22271729e-01 2.12718192e-02
-3.19663495e-01 3.22651535e-01 2.67137885e-01 1.07356358e+00
-1.00675714e+00 1.12812474e-01 -6.85895979e-01 -3.17070544e-01
1.84642792e-01 3.20523918e-01 -1.88324422e-01 5.28404936e-02
-1.20494783e+00 9.10802484e-01 6.47530258e-02 -7.12190121e-02
-7.47324526e-01 -7.96259701e-01 -8.29602301e-01 -1.17488652e-01
4.79335189e-01 -1.78696692e-01 1.54375446e+00 -1.03954613e+00
-2.05162215e+00 1.21188688e+00 -5.40245295e-01 -5.12097657e-01
5.05113244e-01 -2.66996294e-01 -7.32836008e-01 -1.62850246e-01
1.81661993e-01 4.24096882e-01 5.63263118e-01 -1.15409911e+00
-7.71620452e-01 -5.77766657e-01 4.38756905e-02 2.28925124e-01
-5.73908448e-01 6.90491140e-01 -7.22234130e-01 -4.29582238e-01
-6.68665841e-02 -8.97328436e-01 1.65329084e-01 -7.64551878e-01
-7.12350786e-01 -5.18446624e-01 7.74542212e-01 -8.47517610e-01
1.13540411e+00 -2.10199475e+00 1.53219521e-01 -1.65202860e-02
-2.75956362e-01 2.24448055e-01 7.88323507e-02 -2.05903072e-02
2.73617566e-01 2.15519369e-01 -2.02449471e-01 -7.20340371e-01
2.21665442e-01 1.22584306e-01 -3.45103860e-01 8.91255140e-02
3.03962022e-01 6.41876936e-01 -7.01392651e-01 -5.16440868e-01
3.65517028e-02 5.18656075e-01 -3.14048767e-01 1.39300093e-01
-5.24167120e-02 3.97413582e-01 -2.00837433e-01 7.40702033e-01
3.36215347e-01 5.65020703e-02 4.20021355e-01 2.32883364e-01
-8.55310559e-02 9.60936964e-01 -1.05411208e+00 1.58800983e+00
-7.03209162e-01 5.54721713e-01 3.69272172e-01 -1.02385545e+00
8.62651169e-01 3.64988238e-01 -1.28347710e-01 -4.70684588e-01
1.50771230e-01 4.23644543e-01 -5.65171428e-02 -3.64363730e-01
1.49427429e-01 -3.05433661e-01 -3.27817351e-01 3.70701730e-01
5.20131588e-02 1.30839005e-01 -1.32372066e-01 6.35506399e-03
8.47133040e-01 -6.39136285e-02 2.19744310e-01 -4.67234164e-01
1.18485558e+00 -7.49180615e-02 7.69289851e-01 6.79177642e-01
-3.78730118e-01 3.55726987e-01 3.43560636e-01 -5.29182740e-02
-4.46119726e-01 -8.32062662e-01 -3.34506780e-01 1.71595550e+00
-1.32728159e-01 -4.38303083e-01 -1.19204569e+00 -1.03602326e+00
-1.18709788e-01 1.09036851e+00 -4.95618612e-01 7.11936057e-02
-7.93627322e-01 -4.16510344e-01 1.05150890e+00 6.80711687e-01
4.55298185e-01 -1.10624540e+00 1.92586198e-01 1.39057100e-01
-2.92766690e-01 -1.56170130e+00 -5.27637839e-01 4.88959789e-01
-4.44939882e-01 -5.81088245e-01 -3.73554438e-01 -1.05767226e+00
3.52339268e-01 -1.16723120e-01 1.23513472e+00 -7.68131763e-02
7.37996623e-02 8.76059011e-02 -2.32732102e-01 -3.91403943e-01
-7.52229273e-01 5.18659711e-01 2.87072450e-01 1.90735087e-01
1.00421143e+00 -1.12625562e-01 1.51402857e-02 5.39138854e-01
-2.53003836e-01 -2.25903094e-01 3.58221889e-01 9.95276332e-01
2.47056067e-01 -8.41787234e-02 9.03447866e-01 -1.02907264e+00
4.40407008e-01 -1.59021422e-01 -5.46265244e-01 3.98797065e-01
-4.23527986e-01 1.75985113e-01 7.51865923e-01 -3.99198860e-01
-1.36614883e+00 2.32354030e-01 -4.81056571e-01 -1.58564776e-01
-5.88499367e-01 3.88009578e-01 -6.16098940e-01 1.99047640e-01
7.34739184e-01 4.04352695e-01 -1.18727274e-01 -5.34056664e-01
2.89059043e-01 1.21920216e+00 7.22540200e-01 -9.09540534e-01
5.65410614e-01 1.77617639e-01 -7.91770458e-01 -8.50307822e-01
-1.17958021e+00 -5.67650497e-01 -8.05612564e-01 7.90101439e-02
8.99134457e-01 -1.03707600e+00 -5.69891512e-01 3.91136050e-01
-9.17003036e-01 -4.43826079e-01 1.09334558e-01 3.68273169e-01
-3.83176595e-01 2.08082601e-01 -9.04375196e-01 -9.07305777e-01
-2.59285063e-01 -1.37605453e+00 1.12064290e+00 -1.14006221e-01
-3.83291870e-01 -1.03367448e+00 -2.76910782e-01 9.21959579e-01
3.21112782e-01 -5.00685334e-01 7.06746519e-01 -1.00468099e+00
-2.19892487e-01 3.03499103e-02 7.84566849e-02 4.66626018e-01
2.14549452e-01 -1.28471285e-01 -1.49104357e+00 -5.17210662e-02
-8.34755674e-02 -6.20496869e-01 6.16804302e-01 1.39914423e-01
9.67521906e-01 -1.99835896e-01 -3.72160196e-01 5.47150552e-01
9.11820233e-01 2.54561240e-03 2.28723824e-01 3.84678751e-01
6.57111645e-01 8.40298533e-01 2.88536996e-01 -9.34299603e-02
6.03619635e-01 9.28215265e-01 -2.57003814e-01 1.88719690e-01
-1.18834712e-01 -2.89037108e-01 7.67219961e-01 1.02397907e+00
1.66315228e-01 1.96356982e-01 -1.26150334e+00 6.16034508e-01
-1.71005869e+00 -9.54281509e-01 9.50407460e-02 1.97971106e+00
1.35300827e+00 2.53200114e-01 3.53225440e-01 1.70152768e-01
8.75935435e-01 -1.68228045e-01 -5.19564569e-01 -5.19912720e-01
-2.09009469e-01 -8.46696645e-02 2.59377629e-01 7.33560026e-01
-1.20660722e+00 1.50780690e+00 6.47973490e+00 9.08781826e-01
-1.33256137e+00 2.07661644e-01 7.07607985e-01 7.18474761e-02
-1.51561171e-01 -2.03212142e-01 -1.43390894e+00 3.80844384e-01
1.28419805e+00 1.05633453e-01 4.87266421e-01 1.05341816e+00
-8.28861743e-02 4.62901205e-01 -1.36853838e+00 1.02308631e+00
5.01074672e-01 -1.05538189e+00 -2.91065454e-01 -2.26013418e-02
7.20176935e-01 1.13100395e-01 9.19385105e-02 6.61491454e-01
5.70988059e-01 -9.88909781e-01 9.78957713e-01 -1.81896016e-01
7.16067255e-01 -7.95748770e-01 7.28700757e-01 5.55505931e-01
-8.91209245e-01 -3.42541225e-02 -1.91253591e-02 1.13352187e-01
6.68158531e-02 2.36867830e-01 -1.05908489e+00 2.91936278e-01
8.33460569e-01 4.88554835e-01 -5.48543692e-01 3.18040341e-01
-5.74473143e-01 1.26633155e+00 -3.55997026e-01 -2.62256205e-01
2.57610351e-01 8.92132372e-02 2.86466599e-01 1.51308620e+00
5.51604740e-02 -1.33351594e-01 2.47684568e-01 5.83467007e-01
-2.35042349e-01 3.73225838e-01 -3.58417124e-01 2.91668653e-01
7.01502204e-01 1.00700939e+00 -3.92935216e-01 -6.07668400e-01
-7.38597810e-01 1.04929042e+00 7.08761871e-01 2.81471729e-01
-7.77286410e-01 -9.95937288e-02 7.59664893e-01 -1.61517143e-01
3.75640869e-01 8.46312493e-02 -5.53543627e-01 -1.24930978e+00
1.51618078e-01 -1.41051805e+00 5.73738515e-01 -2.84157783e-01
-1.58828616e+00 8.86656642e-01 -3.71931523e-01 -7.41887808e-01
-6.09645784e-01 -7.71466196e-01 -5.33696055e-01 8.89458179e-01
-1.61911905e+00 -1.34716070e+00 1.20884225e-01 6.88643992e-01
8.66409659e-01 -6.10493302e-01 1.09417748e+00 2.86240339e-01
-8.20936620e-01 1.09118068e+00 4.19328064e-02 6.45830393e-01
9.36568081e-01 -1.45966530e+00 4.24999714e-01 7.95489550e-01
4.57107008e-01 8.28928530e-01 5.54974675e-01 -1.98426306e-01
-1.28712475e+00 -8.14230263e-01 1.28443599e+00 -9.52965021e-01
9.62970257e-01 -7.62804151e-01 -1.02572060e+00 9.44531381e-01
3.21607590e-01 -9.76495538e-03 9.82690156e-01 9.52192008e-01
-7.99217463e-01 -1.97204486e-01 -9.84969318e-01 6.15409374e-01
1.10082412e+00 -1.03175604e+00 -9.05537605e-01 2.39319459e-01
4.88420606e-01 -3.77043188e-01 -6.59679592e-01 1.55022860e-01
4.66878623e-01 -8.66535366e-01 8.67290854e-01 -7.91551888e-01
1.64425597e-01 -2.46921480e-01 -3.55836540e-01 -1.37056458e+00
-1.76866010e-01 -6.05559886e-01 2.38678768e-01 1.81069243e+00
1.14729595e+00 -5.67654192e-01 7.55768716e-01 5.79092681e-01
-2.82205701e-01 -3.25200200e-01 -9.88995492e-01 -1.01901150e+00
3.11845094e-01 -7.88179755e-01 7.53936172e-01 1.26326883e+00
3.71319443e-01 8.89597833e-01 -1.89766899e-01 4.01181072e-01
4.60472584e-01 1.11622110e-01 8.01339984e-01 -1.04906809e+00
-2.68576711e-01 -6.10545695e-01 -3.86474639e-01 -1.08257186e+00
9.31563020e-01 -9.98205125e-01 3.33943427e-01 -9.75885451e-01
1.67316765e-01 -4.23686117e-01 -3.74035358e-01 5.94118595e-01
-3.99378866e-01 1.94383606e-01 -1.53561383e-01 -1.29052124e-03
-6.93545938e-01 1.97866231e-01 4.02052104e-01 -2.90073007e-01
-2.97433406e-01 1.08593740e-01 -1.00994229e+00 8.16097379e-01
9.43113565e-01 -3.64748597e-01 -2.01108903e-01 -7.41258383e-01
-3.59823674e-01 -1.35267049e-01 -1.68148000e-02 -8.00844908e-01
3.77334863e-01 6.47189841e-02 1.88177511e-01 -3.85189086e-01
3.47465634e-01 -5.31010628e-01 -4.66667622e-01 8.03918317e-02
-6.76384807e-01 -2.11905345e-01 1.64356336e-01 1.96978375e-01
-5.06285369e-01 -2.40472555e-01 7.89729476e-01 1.18976064e-01
-7.77680516e-01 -1.49934471e-01 -2.86540478e-01 2.13571057e-01
8.67179215e-01 7.75427818e-02 -2.31009990e-01 -3.37402135e-01
-5.82092047e-01 1.43646121e-01 2.92579353e-01 6.13065362e-01
6.09147958e-02 -1.15677750e+00 -1.03019476e+00 3.14729184e-01
3.86494726e-01 -3.39092284e-01 5.19539341e-02 7.42659390e-01
-5.55288382e-02 6.04629576e-01 2.48681813e-01 -9.33983624e-01
-1.57586277e+00 4.39612508e-01 4.34093177e-01 -4.02077250e-02
-3.74387234e-01 1.16562188e+00 -4.84288894e-02 -9.98537779e-01
4.21438128e-01 -1.62152082e-01 -1.05481528e-01 -3.47090438e-02
7.52454996e-01 3.93146686e-02 2.38658756e-01 -1.04855788e+00
-8.56189728e-01 5.77028334e-01 -2.65702695e-01 -3.15931320e-01
1.02421558e+00 -2.53316134e-01 1.63494483e-01 5.35385132e-01
1.14558899e+00 5.41698635e-01 -1.03043759e+00 -6.73355043e-01
5.51304042e-01 -1.68487191e-01 1.48242638e-01 -1.08637071e+00
-6.91706061e-01 8.56798649e-01 1.68350071e-01 3.26286942e-01
7.92295575e-01 4.59031284e-01 6.11959934e-01 5.20005524e-01
2.01382682e-01 -1.37469220e+00 -2.76101708e-01 7.46161461e-01
6.84873462e-01 -1.62076330e+00 -4.14169312e-01 -5.56300759e-01
-9.73079801e-01 7.60359943e-01 6.54250562e-01 4.04815942e-01
6.39558315e-01 5.41272581e-01 6.84995770e-01 1.93520248e-01
-7.44798064e-01 -3.15753996e-01 3.57992172e-01 6.08786762e-01
8.21023405e-01 3.31155270e-01 2.23047882e-01 8.37091804e-01
-6.51491463e-01 -4.09691751e-01 -1.28451094e-01 7.16817498e-01
-2.05160946e-01 -1.26948869e+00 -4.21482593e-01 3.79730910e-02
-7.30517805e-01 -3.23999941e-01 -7.18198895e-01 5.86051404e-01
-1.12398490e-01 1.33258140e+00 -1.45832598e-01 -2.84386992e-01
2.56739259e-01 5.78387678e-01 1.53241500e-01 -6.44322813e-01
-7.00389385e-01 1.50278360e-01 5.11995196e-01 -1.64981097e-01
-3.54795039e-01 -1.21229720e+00 -1.20020270e+00 -1.87474340e-01
-4.06949401e-01 2.62336463e-01 7.75745034e-01 1.06308615e+00
2.91076034e-01 3.44980657e-01 6.77194238e-01 -3.80801678e-01
-9.01481330e-01 -1.14267123e+00 -3.21095079e-01 4.58214432e-01
3.14348489e-01 -2.90985495e-01 -6.44231796e-01 3.04020792e-01] | [14.179730415344238, 6.737485885620117] |
d2d8ec2c-6e60-4342-a557-79189f1ccfc8 | histgnn-hierarchical-spatio-temporal-graph | 2201.09101 | null | https://arxiv.org/abs/2201.09101v2 | https://arxiv.org/pdf/2201.09101v2.pdf | HiSTGNN: Hierarchical Spatio-temporal Graph Neural Networks for Weather Forecasting | Weather Forecasting is an attractive challengeable task due to its influence on human life and complexity in atmospheric motion. Supported by massive historical observed time series data, the task is suitable for data-driven approaches, especially deep neural networks. Recently, the Graph Neural Networks (GNNs) based methods have achieved excellent performance for spatio-temporal forecasting. However, the canonical GNNs-based methods only individually model the local graph of meteorological variables per station or the global graph of whole stations, lacking information interaction between meteorological variables in different stations. In this paper, we propose a novel Hierarchical Spatio-Temporal Graph Neural Network (HiSTGNN) to model cross-regional spatio-temporal correlations among meteorological variables in multiple stations. An adaptive graph learning layer and spatial graph convolution are employed to construct self-learning graph and study hidden dependency among nodes of variable-level and station-level graph. For capturing temporal pattern, the dilated inception as the backbone of gate temporal convolution is designed to model long and various meteorological trends. Moreover, a dynamic interaction learning is proposed to build bidirectional information passing in hierarchical graph. Experimental results on three real-world meteorological datasets demonstrate the superior performance of HiSTGNN beyond 7 baselines and it reduces the errors by 4.2% to 11.6% especially compared to state-of-the-art weather forecasting method. | ['Junbo Zhang', 'Shenggong Ji', 'Bin Wang', 'Tianrui Li', 'Fei Teng', 'Peng Xie', 'Minbo Ma'] | 2022-01-22 | null | null | null | null | ['self-learning', 'spatio-temporal-forecasting'] | ['natural-language-processing', 'time-series'] | [-6.14552021e-01 -3.09633315e-01 2.64915854e-01 -2.39324808e-01
5.42019963e-01 -3.71230006e-01 6.44332051e-01 2.60525793e-01
-7.70007372e-02 6.62230909e-01 1.10860363e-01 -8.73069763e-01
-3.96868110e-01 -1.40840864e+00 -5.52115798e-01 -9.28407371e-01
-9.16985512e-01 9.01826769e-02 5.04463375e-01 -7.59347260e-01
-3.02013993e-01 5.37781715e-01 -1.05331767e+00 -3.16376567e-01
8.43550920e-01 6.50604367e-01 2.13161513e-01 8.64417493e-01
-3.87313724e-01 8.65628302e-01 -3.75569999e-01 -1.15930118e-01
2.92733666e-02 -4.29240197e-01 -2.72569120e-01 -1.67068973e-01
2.98375338e-01 2.80623566e-02 -8.35128486e-01 9.55062211e-01
4.36657399e-01 5.43320537e-01 3.23585331e-01 -1.27449572e+00
-7.09116161e-01 7.61899352e-01 -5.45507193e-01 6.84243679e-01
-3.90322536e-01 7.44775683e-02 9.25615370e-01 -4.60188359e-01
1.53348267e-01 1.14410043e+00 8.77194524e-01 -3.44952419e-02
-7.69115567e-01 -7.68533111e-01 6.45742118e-01 1.69133484e-01
-1.28890622e+00 2.28880480e-01 8.24296832e-01 -5.82015634e-01
1.06788266e+00 2.36430287e-01 8.47979605e-01 3.26949298e-01
7.28927493e-01 -1.83789119e-01 7.66552031e-01 2.16595799e-01
-2.66133130e-01 -5.02803266e-01 3.09666783e-01 9.06933129e-01
2.20531702e-01 2.19962701e-01 -2.02999502e-01 -4.89890985e-02
7.62278497e-01 4.38987911e-01 -2.96260923e-01 3.90965909e-01
-8.45210552e-01 7.63424993e-01 1.42115462e+00 5.27511477e-01
-5.26934981e-01 2.82424986e-01 3.20259988e-01 3.15440387e-01
1.07115734e+00 -1.77260581e-02 -3.53128552e-01 5.04916072e-01
-8.19322824e-01 8.44489112e-02 6.38957322e-01 5.71263611e-01
8.12216163e-01 7.89402008e-01 1.40047505e-01 4.54114407e-01
4.05806631e-01 7.86153615e-01 4.92131338e-02 1.33462148e-02
5.90421081e-01 9.84169066e-01 -2.77156025e-01 -1.95893645e+00
-1.11875832e+00 -8.92093241e-01 -1.85112965e+00 -1.44737080e-01
-2.03744676e-02 -5.65856278e-01 -1.00540864e+00 1.36753452e+00
5.02380848e-01 9.12187934e-01 -2.36531779e-01 8.94756854e-01
1.24626911e+00 1.39461720e+00 1.82926178e-01 -8.50657895e-02
9.98150885e-01 -8.44835281e-01 -9.14121151e-01 3.85236032e-02
6.82697535e-01 -2.15980783e-01 6.83030427e-01 -2.15746105e-01
-5.69784105e-01 -8.17432821e-01 -8.29750776e-01 3.28497142e-01
-9.28919911e-01 -1.80843696e-01 6.80193722e-01 9.42525193e-02
-1.46182752e+00 5.87469041e-01 -7.84927905e-01 -4.95714024e-02
-2.65882518e-02 4.25146878e-01 -2.81283468e-01 2.17471734e-01
-1.93857467e+00 5.61545908e-01 4.71680433e-01 8.77643704e-01
-6.18543744e-01 -6.50892913e-01 -9.41854775e-01 3.67680222e-01
1.06884480e-01 -6.75528049e-01 3.93221140e-01 -4.12737072e-01
-9.69866931e-01 6.47585839e-02 -4.53071110e-02 -3.32264870e-01
8.21134672e-02 3.87002349e-01 -1.17195773e+00 -2.01239571e-01
-2.35559747e-01 -5.40542342e-02 4.51592654e-01 -1.06224370e+00
-6.09531820e-01 -3.51194501e-01 6.77025095e-02 1.15433432e-01
-4.25909162e-01 -3.18186373e-01 -4.27999765e-01 -6.87394202e-01
1.15891650e-01 -7.80339539e-01 -5.86422622e-01 -5.93431056e-01
-2.96191633e-01 -2.97292024e-01 9.61850226e-01 -9.09192920e-01
1.83660793e+00 -1.98823583e+00 7.65320957e-02 4.90273595e-01
7.77500808e-01 3.65093768e-01 -1.42790869e-01 7.54788518e-01
-4.77475822e-02 2.44274423e-01 -3.96055102e-01 -1.33296221e-01
-2.93027222e-01 5.41130304e-01 -2.82833219e-01 6.20299041e-01
-3.04588955e-02 1.03532636e+00 -9.53421354e-01 -1.96151689e-01
2.47609124e-01 7.47096121e-01 -5.07716350e-02 3.50100547e-01
-6.56240955e-02 7.06157386e-01 -4.21218008e-01 5.43051548e-02
9.01057124e-01 -5.25108039e-01 1.77716523e-01 1.29804807e-02
-3.64070296e-01 5.01185991e-02 -9.63561356e-01 9.54436004e-01
-3.64804327e-01 5.40310204e-01 3.57453264e-02 -9.68210876e-01
1.15321696e+00 4.60236013e-01 2.02450246e-01 -7.94863582e-01
-4.75393087e-02 -1.44720346e-01 1.81159005e-01 -5.32532573e-01
4.99164164e-01 -3.27170007e-02 1.41124278e-01 -2.58984383e-05
-2.09700599e-01 7.78532550e-02 7.81678557e-02 3.40383798e-01
6.87246501e-01 -3.17821056e-01 -5.98685220e-02 -4.95018840e-01
6.56172633e-01 -3.48537117e-02 4.44810271e-01 4.48406816e-01
2.01132715e-01 1.54123008e-01 5.28606951e-01 -1.10051835e+00
-7.04595923e-01 -4.92246002e-01 1.89131916e-01 1.13631272e+00
4.77750078e-02 -4.08674866e-01 -2.30727106e-01 -4.56629038e-01
1.07957810e-01 4.27608401e-01 -7.80470729e-01 1.17208980e-01
-6.10821664e-01 -1.11944342e+00 4.91405278e-01 3.21492136e-01
5.90296328e-01 -9.85080719e-01 2.52718389e-01 3.82560790e-01
1.07804827e-01 -1.05264676e+00 -1.71761632e-01 -8.77510980e-02
-9.11090136e-01 -8.30682278e-01 -4.21022207e-01 -5.95184088e-01
6.78271055e-01 5.38152039e-01 1.18182898e+00 6.68775260e-01
1.23512946e-01 -1.04465798e-01 -2.45285034e-01 -2.28322372e-01
-8.67130309e-02 3.69203329e-01 -1.74201548e-01 2.82775611e-01
-1.48319989e-01 -8.27223837e-01 -7.48116255e-01 2.79374808e-01
-9.93469894e-01 6.41375482e-02 1.44112661e-01 5.93275726e-01
1.88936502e-01 5.79157829e-01 3.04792941e-01 -9.08183157e-01
6.01919591e-01 -1.12903976e+00 -1.02946329e+00 1.31561190e-01
-6.49138868e-01 -3.27216208e-01 1.09714413e+00 7.32189219e-04
-8.24289024e-01 -2.78372854e-01 1.45054549e-01 -2.70281732e-01
1.14795752e-01 1.32353258e+00 2.91355133e-01 -1.56684861e-01
3.51138413e-01 4.16031241e-01 -3.23515952e-01 -3.02727848e-01
3.54376554e-01 2.00216874e-01 4.70640451e-01 5.66485003e-02
1.00628459e+00 3.86267722e-01 5.14385462e-01 -9.65432286e-01
-5.40858328e-01 -4.43735629e-01 -6.31845653e-01 -5.81055045e-01
1.08685744e+00 -1.06233895e+00 -8.69010866e-01 7.44429648e-01
-1.09992313e+00 -6.74119949e-01 5.82699537e-01 5.05429089e-01
5.51619709e-01 2.66612619e-01 -5.92984855e-01 -6.85242414e-01
-3.92700076e-01 -6.66886806e-01 7.14670360e-01 3.53067368e-01
4.86483097e-01 -2.01180339e+00 2.43725866e-01 -3.07445228e-01
7.97279060e-01 6.02980316e-01 8.21600020e-01 -3.16431016e-01
-6.18566096e-01 -1.23731881e-01 -5.52548647e-01 2.70825941e-02
2.87822634e-01 2.50790358e-01 -5.71628630e-01 -4.00861084e-01
-3.46611977e-01 4.14937854e-01 9.86492693e-01 5.53213179e-01
1.02216685e+00 -5.34360349e-01 -2.51994669e-01 9.70848262e-01
1.48088944e+00 1.04096688e-01 3.06592315e-01 -2.91041806e-02
1.60224795e+00 6.38086617e-01 -1.33091256e-01 3.43754083e-01
9.89048958e-01 1.09256245e-01 6.93668664e-01 -6.49904490e-01
1.17800027e-01 9.24676657e-03 -1.75360106e-02 1.45802319e+00
-6.22209370e-01 -7.93302894e-01 -1.48743558e+00 5.04281700e-01
-2.02810526e+00 -1.02434444e+00 -1.12135959e+00 1.78508627e+00
1.53349370e-01 2.46750731e-02 -5.95460534e-02 -1.46115214e-01
7.57325768e-01 8.55687797e-01 -3.19402844e-01 -1.72107801e-01
-3.61719906e-01 -4.88287285e-02 9.26977098e-01 8.62494469e-01
-1.20833588e+00 1.04696083e+00 5.40469170e+00 3.65241289e-01
-1.41295457e+00 1.02066077e-01 6.52596295e-01 3.05262595e-01
-3.82007360e-01 -1.97159693e-01 -8.48803699e-01 3.62126768e-01
1.13910258e+00 -3.67988013e-02 4.83137459e-01 2.92123646e-01
6.87126338e-01 3.50357533e-01 -1.29763201e-01 5.57792306e-01
-2.61493951e-01 -1.53698409e+00 3.03471655e-01 -9.62031167e-03
9.46291327e-01 5.56529641e-01 -1.77842721e-01 2.30442390e-01
6.31989598e-01 -1.13712049e+00 -1.41011134e-01 7.86946177e-01
3.77940744e-01 -7.39909708e-01 8.30908537e-01 2.61651397e-01
-2.06444454e+00 6.86589852e-02 -4.67798084e-01 -5.42409003e-01
1.98215753e-01 7.81391501e-01 -5.47551632e-01 1.19070053e+00
8.01170647e-01 1.29204643e+00 -4.69124407e-01 8.15184891e-01
-2.62436152e-01 1.17072010e+00 -1.62518576e-01 -2.54653785e-02
8.86881709e-01 -6.29415810e-01 4.11280304e-01 1.28208959e+00
3.05525750e-01 4.94497120e-01 3.98992538e-01 2.98102766e-01
4.59266715e-02 1.13177657e-01 -8.35100710e-01 6.91022575e-02
1.88782975e-01 1.41202331e+00 -7.03574538e-01 -4.21522290e-01
-5.35980463e-01 4.94699776e-01 4.60367858e-01 8.36047053e-01
-9.17099893e-01 -4.40045744e-01 5.63618302e-01 9.92941260e-02
2.09618151e-01 -8.40243638e-01 1.54434526e-02 -8.91024113e-01
-2.10316852e-01 -3.74625057e-01 6.57982290e-01 -8.11957598e-01
-1.30808485e+00 1.00917518e+00 -1.23647109e-01 -1.04808283e+00
1.81484610e-01 -4.39059317e-01 -1.21182871e+00 1.39107800e+00
-2.01856422e+00 -1.38687134e+00 -7.29662597e-01 9.63980913e-01
-3.10213063e-02 -1.07707098e-01 5.00416696e-01 3.69003475e-01
-7.64309108e-01 3.09512783e-02 1.68160275e-01 3.86442482e-01
1.64668098e-01 -1.18588686e+00 9.55209374e-01 1.05755675e+00
-1.50911063e-01 4.77267563e-01 3.57747167e-01 -9.76533592e-01
-1.36809742e+00 -1.69471073e+00 1.25870430e+00 9.12243724e-02
1.19109190e+00 -3.14896524e-01 -1.40619504e+00 6.36079907e-01
3.45615685e-01 4.89903420e-01 2.79992402e-01 2.02469602e-01
-1.96065426e-01 -4.35191780e-01 -3.21773827e-01 5.62452674e-01
9.18843448e-01 -3.66330802e-01 -5.79253249e-02 6.18625641e-01
1.05463302e+00 -4.85672683e-01 -9.28196728e-01 3.81980330e-01
5.47615439e-03 -7.86272466e-01 6.74139977e-01 -6.74592495e-01
2.70889848e-01 -5.69805324e-01 2.43292347e-01 -1.78810775e+00
-9.97628868e-01 -5.61299443e-01 1.03789531e-01 8.36721539e-01
5.14505446e-01 -1.08361733e+00 2.83652872e-01 2.59281993e-01
-4.15419310e-01 -4.32670444e-01 -7.69608200e-01 -5.51131666e-01
-9.79997292e-02 -2.27052510e-01 9.40124512e-01 1.52370715e+00
-4.54604834e-01 4.91119117e-01 -6.21486306e-01 9.17088926e-01
3.54324520e-01 1.37730151e-01 6.63827419e-01 -1.35064971e+00
-1.05056144e-01 -6.14983320e-01 -4.97738481e-01 -7.65894473e-01
1.29532889e-01 -9.27846968e-01 -4.29086328e-01 -1.93247819e+00
-5.00063896e-01 -4.46528435e-01 -6.21373117e-01 4.86210793e-01
-5.79940319e-01 6.31430745e-02 -8.82107168e-02 5.87394275e-02
-3.13694656e-01 6.04855001e-01 1.49040723e+00 -2.08319455e-01
-3.54696035e-01 -2.55383793e-02 2.23703608e-02 3.16059381e-01
9.44373250e-01 -4.19041187e-01 -6.15617752e-01 -7.53424108e-01
7.50845671e-01 2.68650234e-01 4.08605367e-01 -9.33011830e-01
5.90159178e-01 -2.68400311e-01 1.79719493e-01 -7.04848588e-01
-1.59992829e-01 -7.96814203e-01 5.06690264e-01 5.13394058e-01
9.86974239e-02 8.14182639e-01 4.75473702e-01 6.63176060e-01
-4.84989256e-01 6.63906872e-01 1.97343260e-01 -3.39468084e-02
-7.87934721e-01 1.08748603e+00 -3.20632160e-01 -1.62142277e-01
6.73021972e-01 1.79202214e-01 -5.39394319e-01 -4.54334915e-01
-8.22860718e-01 7.86012828e-01 -1.72445729e-01 3.28260303e-01
4.24286067e-01 -1.14247036e+00 -1.11058688e+00 2.20050976e-01
-1.39152547e-02 8.78566280e-02 8.27496231e-01 9.27427769e-01
-7.81354964e-01 3.19409579e-01 9.37345549e-02 -5.53307354e-01
-1.06587958e+00 4.11151111e-01 6.28937125e-01 -4.13036197e-01
-7.05272436e-01 8.66785645e-01 2.41485715e-01 -6.37569606e-01
-1.47836074e-01 -5.46497464e-01 -6.86157286e-01 1.92419544e-01
2.72942036e-01 3.12111795e-01 1.05198696e-01 -7.43850529e-01
-1.34135887e-01 5.04708529e-01 5.97271442e-01 2.80142784e-01
1.61160958e+00 -3.03617418e-01 -6.08116567e-01 6.61036491e-01
1.02157414e+00 -2.52508372e-01 -1.07469630e+00 -4.09042418e-01
-3.39406997e-01 -5.20521067e-02 5.13163626e-01 -3.76114458e-01
-1.74004269e+00 1.13410091e+00 2.03513101e-01 1.00137973e+00
1.13500643e+00 -4.87575948e-01 8.37830782e-01 2.57001698e-01
-1.81022197e-01 -4.54368591e-01 -6.09463751e-01 9.67158794e-01
7.98455834e-01 -1.07781255e+00 8.90178382e-02 -2.19728678e-01
-3.81357938e-01 1.23055124e+00 5.65136492e-01 -2.65265167e-01
1.37505996e+00 -6.25984073e-02 1.91565096e-01 -7.98392951e-01
-6.04932070e-01 -2.84957767e-01 6.99873984e-01 1.51205420e-01
5.10844886e-01 4.75928187e-01 5.83865196e-02 1.61530804e-02
-1.82171181e-01 -3.83494556e-01 1.12291507e-01 3.82689595e-01
-2.24684849e-01 -5.11473715e-01 -1.58566341e-01 4.65035379e-01
-1.60195425e-01 -5.11822164e-01 -5.48311435e-02 8.13565612e-01
-4.75490093e-02 1.00750995e+00 3.04833412e-01 -6.72180712e-01
2.26830527e-01 -2.39253685e-01 -2.64470190e-01 -2.94723272e-01
-1.04786432e+00 -1.07111186e-01 -1.17103122e-01 -3.02231520e-01
-5.44938385e-01 -1.28477082e-01 -1.39885318e+00 -7.72200882e-01
-7.14914277e-02 3.18114042e-01 5.83300471e-01 8.64225268e-01
5.83835602e-01 1.05219901e+00 9.51927125e-01 -8.63486528e-01
3.92464489e-01 -8.56445432e-01 -8.20795000e-01 2.09570989e-01
6.94912493e-01 -4.09258723e-01 -4.38310474e-01 -1.73881620e-01] | [6.608173847198486, 2.7861597537994385] |
c2008a86-19df-461e-9249-10f47cf33df8 | multi-attention-multi-class-constraint-for | 1806.05372 | null | http://arxiv.org/abs/1806.05372v1 | http://arxiv.org/pdf/1806.05372v1.pdf | Multi-Attention Multi-Class Constraint for Fine-grained Image Recognition | Attention-based learning for fine-grained image recognition remains a
challenging task, where most of the existing methods treat each object part in
isolation, while neglecting the correlations among them. In addition, the
multi-stage or multi-scale mechanisms involved make the existing methods less
efficient and hard to be trained end-to-end. In this paper, we propose a novel
attention-based convolutional neural network (CNN) which regulates multiple
object parts among different input images. Our method first learns multiple
attention region features of each input image through the one-squeeze
multi-excitation (OSME) module, and then apply the multi-attention multi-class
constraint (MAMC) in a metric learning framework. For each anchor feature, the
MAMC functions by pulling same-attention same-class features closer, while
pushing different-attention or different-class features away. Our method can be
easily trained end-to-end, and is highly efficient which requires only one
training stage. Moreover, we introduce Dogs-in-the-Wild, a comprehensive dog
species dataset that surpasses similar existing datasets by category coverage,
data volume and annotation quality. This dataset will be released upon
acceptance to facilitate the research of fine-grained image recognition.
Extensive experiments are conducted to show the substantial improvements of our
method on four benchmark datasets. | ['Feng Zhou', 'Yuchen Yuan', 'Errui Ding', 'Ming Sun'] | 2018-06-14 | multi-attention-multi-class-constraint-for-1 | http://openaccess.thecvf.com/content_ECCV_2018/html/Ming_Sun_Multi-Attention_Multi-Class_Constraint_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Ming_Sun_Multi-Attention_Multi-Class_Constraint_ECCV_2018_paper.pdf | eccv-2018-9 | ['fine-grained-image-recognition'] | ['computer-vision'] | [ 4.92001697e-03 -1.83362067e-01 -2.77763575e-01 -7.53209651e-01
-5.78601778e-01 -1.86666280e-01 2.10349157e-01 -1.95414349e-01
-4.40339267e-01 3.81534606e-01 9.06812623e-02 1.37329876e-01
-1.58251673e-01 -6.27248645e-01 -8.86902213e-01 -8.63455892e-01
1.85371250e-01 2.90976614e-01 2.62259603e-01 -4.02928516e-02
5.88138103e-02 5.66908956e-01 -1.64092314e+00 2.70524114e-01
6.70161188e-01 1.28777301e+00 5.16688824e-01 2.94713348e-01
8.86407506e-04 8.30769897e-01 -4.03727263e-01 -2.89484262e-01
2.36119345e-01 -1.48964301e-01 -6.95637167e-01 2.01891467e-01
7.22410083e-01 -4.95566666e-01 -5.75176001e-01 1.19087219e+00
4.11406338e-01 9.47796032e-02 4.63457197e-01 -1.25820053e+00
-1.23969388e+00 2.42885098e-01 -7.18330026e-01 3.64555866e-01
-2.40545571e-01 2.46712953e-01 9.56717670e-01 -9.83874798e-01
8.17850232e-02 1.37703037e+00 2.80913353e-01 5.61856270e-01
-1.21541774e+00 -7.73003221e-01 5.70974827e-01 3.88955504e-01
-1.42133379e+00 -1.99521273e-01 6.43238008e-01 -5.36993027e-01
9.45439458e-01 5.39343245e-02 3.81827891e-01 7.17034101e-01
-2.39087362e-02 9.49944377e-01 8.73190522e-01 -1.54337436e-01
-1.36730582e-01 -1.98693827e-01 2.24341720e-01 7.46524274e-01
2.18930557e-01 2.65876859e-01 -2.25652814e-01 1.53122112e-01
8.76319349e-01 5.51175654e-01 -2.32477605e-01 -5.05906701e-01
-1.42419541e+00 1.04113746e+00 9.10103381e-01 2.10712701e-01
-4.93784636e-01 1.73543170e-01 2.62200087e-01 2.30276823e-01
3.57448250e-01 1.25045717e-01 -6.78845823e-01 1.61734223e-01
-6.15931392e-01 2.39118382e-01 2.00077683e-01 1.04979289e+00
1.03154922e+00 -3.03525686e-01 -4.74303067e-01 1.05544710e+00
3.83735955e-01 2.85440326e-01 5.92673600e-01 -6.68684721e-01
4.03877109e-01 7.61551738e-01 -6.32608756e-02 -9.55953360e-01
-2.31468320e-01 -2.86087632e-01 -1.09101617e+00 9.78142098e-02
2.02762604e-01 2.06388846e-01 -1.09735513e+00 1.76709771e+00
4.15581852e-01 6.09133691e-02 -3.40317041e-01 1.22346675e+00
9.36133623e-01 4.50376511e-01 3.22408348e-01 2.32820898e-01
1.46517527e+00 -1.51582074e+00 -3.82891566e-01 -3.15092772e-01
4.03317362e-01 -6.27943575e-01 1.28191936e+00 1.74442381e-01
-7.35370159e-01 -8.17197144e-01 -8.88802767e-01 -3.63391250e-01
-5.38800657e-01 3.94890189e-01 6.71957731e-01 4.89721969e-02
-6.06765270e-01 2.95351505e-01 -7.67628729e-01 -1.86597243e-01
8.49994659e-01 4.01750147e-01 -6.09434605e-01 -2.90254891e-01
-9.26035881e-01 7.79448748e-01 2.77885765e-01 2.69824117e-01
-9.37308490e-01 -7.61240423e-01 -9.46089685e-01 1.78859904e-01
1.62014440e-01 -6.69742525e-01 1.10028660e+00 -8.91095519e-01
-1.38161159e+00 9.44879115e-01 -8.54519103e-03 -2.10681260e-01
2.59970307e-01 -2.41392687e-01 -1.09681934e-01 -4.90457304e-02
3.89595211e-01 1.01863682e+00 8.05103600e-01 -8.59800994e-01
-7.38799036e-01 -4.53923970e-01 1.77232787e-01 -9.20914412e-02
-1.37852177e-01 3.33345205e-01 -7.02504337e-01 -9.55005288e-01
-1.26846179e-01 -7.56817997e-01 -2.91528523e-01 2.49694988e-01
-2.00574428e-01 -5.49028158e-01 7.83860445e-01 -3.97332162e-01
1.23857355e+00 -2.32948494e+00 1.77139953e-01 -2.66646594e-01
2.80785829e-01 2.66269386e-01 -3.68958771e-01 6.72780424e-02
-2.16397613e-01 -1.36699095e-01 -2.87710160e-01 -1.87674165e-01
1.75621197e-01 2.54103303e-01 -1.58813223e-01 5.02830744e-01
5.69983125e-01 1.13566637e+00 -7.06328213e-01 -5.19756079e-01
4.76300269e-01 4.82179433e-01 -6.64700210e-01 5.09960473e-01
-2.59333812e-02 3.82814139e-01 -5.94067335e-01 8.78646553e-01
8.14016104e-01 -3.99255246e-01 -4.04617190e-01 -4.78267521e-01
4.01857235e-02 -2.23890692e-01 -8.93552899e-01 1.69293332e+00
-5.60854793e-01 3.30711335e-01 -9.82459709e-02 -1.26061714e+00
9.18484926e-01 -1.12772994e-01 2.33294487e-01 -8.95186543e-01
1.33550256e-01 9.43290070e-02 1.97824761e-01 -6.55041039e-01
1.79112628e-01 6.65064529e-03 -3.83227557e-01 1.80665284e-01
2.88656116e-01 3.67685825e-01 1.94353014e-01 -7.02280998e-02
8.02316487e-01 -9.79765132e-02 3.17383707e-01 -3.25544149e-01
5.42124152e-01 -1.84759170e-01 7.08149910e-01 5.30866206e-01
-5.09916484e-01 6.02734625e-01 1.01987280e-01 -6.56702101e-01
-9.84043300e-01 -7.27199197e-01 -2.58166939e-01 1.58119857e+00
3.92757505e-01 -4.93028648e-02 -7.20735431e-01 -1.02061605e+00
3.78019601e-01 2.76510030e-01 -9.87430871e-01 -1.97119698e-01
-5.33247113e-01 -5.55541635e-01 4.21790987e-01 9.02725518e-01
7.11801291e-01 -1.41547620e+00 -6.50345564e-01 2.76855648e-01
-2.71304343e-02 -9.24362063e-01 -8.59354794e-01 2.13460997e-01
-5.63125014e-01 -1.17995727e+00 -8.59223366e-01 -1.04479325e+00
8.02021861e-01 5.27581334e-01 9.89132345e-01 1.90220088e-01
-5.57112217e-01 -5.68621382e-02 -2.88331121e-01 -1.89323798e-01
4.02374238e-01 -3.28121148e-02 -1.19165428e-01 3.13047707e-01
8.40137482e-01 -2.88603663e-01 -6.87516093e-01 6.78501248e-01
-8.14361155e-01 -1.05651498e-01 8.91932011e-01 1.33778059e+00
9.56730247e-01 -1.20521262e-01 7.38642752e-01 -4.82546955e-01
3.56354505e-01 -5.27083993e-01 -6.85413361e-01 3.46908808e-01
-2.10956112e-01 1.48019254e-01 6.55056715e-01 -5.00508010e-01
-8.43174934e-01 1.76511124e-01 -8.00111592e-02 -8.69063497e-01
-4.58653957e-01 2.73628891e-01 -5.63794851e-01 -1.90996528e-01
2.23008737e-01 1.26210123e-01 -3.48088682e-01 -7.96364844e-01
4.30223495e-01 6.98691785e-01 6.95600033e-01 -4.64958638e-01
5.39978802e-01 2.73066401e-01 -4.52333599e-01 -4.33586121e-01
-1.08837688e+00 -5.44931293e-01 -8.43976915e-01 1.22827835e-01
1.04673684e+00 -1.03960085e+00 -8.09332192e-01 7.48065591e-01
-9.91882443e-01 -3.11538309e-01 -8.24773461e-02 5.35687387e-01
-3.78275484e-01 6.91447929e-02 -5.50434709e-01 -3.74990523e-01
-3.97979289e-01 -1.30610132e+00 1.25085568e+00 4.82767731e-01
1.88510090e-01 -6.94600880e-01 -1.01973817e-01 5.47303259e-01
5.05898833e-01 -1.39551964e-02 7.45546222e-01 -4.06521320e-01
-6.28047109e-01 -1.69696108e-01 -7.85436690e-01 3.15803885e-01
1.85234249e-01 -1.15524806e-01 -7.72373378e-01 -3.75336438e-01
-3.43651116e-01 -6.72291517e-01 1.14229536e+00 3.53668332e-01
1.83842981e+00 -2.54127830e-01 -3.26884508e-01 7.93722928e-01
1.44812036e+00 -3.44135333e-03 5.27438164e-01 5.03064275e-01
8.01437318e-01 4.90371406e-01 8.42303336e-01 3.00758600e-01
6.36367857e-01 8.27661335e-01 4.86051530e-01 -2.79912800e-01
-2.60197937e-01 -3.27589422e-01 3.49315852e-02 5.30238509e-01
1.59937978e-01 2.77216267e-02 -4.97915447e-01 8.08096409e-01
-1.89324498e+00 -1.07954872e+00 1.79790139e-01 1.86310077e+00
5.87708414e-01 -1.58338889e-01 1.63402170e-01 -2.03819543e-01
8.42881441e-01 5.10425195e-02 -8.64287138e-01 -2.24863395e-01
-1.23544343e-01 2.23475769e-01 4.84911621e-01 1.71068713e-01
-1.39792943e+00 1.02925146e+00 5.31648159e+00 9.19967115e-01
-1.23643661e+00 2.14896277e-01 6.80446863e-01 -8.32856223e-02
1.36426732e-01 -3.44277322e-01 -8.24554741e-01 7.19556928e-01
3.88896108e-01 6.58372939e-02 4.32994395e-01 1.32392442e+00
-5.38526960e-02 1.60801746e-02 -1.20274150e+00 1.13028264e+00
1.95572618e-02 -1.10444820e+00 4.19557355e-02 -1.91145074e-02
5.66483676e-01 3.25199723e-01 -4.91534024e-02 5.97518742e-01
3.02198678e-01 -1.11244357e+00 7.95355916e-01 4.85326946e-01
8.59266222e-01 -6.60568774e-01 6.70841932e-01 3.86276245e-01
-1.49125075e+00 -1.79120421e-01 -7.15645552e-01 1.18325204e-01
-1.45854652e-01 4.51639742e-01 -1.21776082e-01 3.19734186e-01
1.12103188e+00 6.02541327e-01 -5.82649171e-01 1.21297884e+00
-2.12819368e-01 3.53064060e-01 -2.50618935e-01 -1.27569623e-02
3.87933373e-01 8.51637870e-02 -2.82434020e-02 1.23768961e+00
5.82555383e-02 1.99909106e-01 4.12990063e-01 9.42701697e-01
-1.67012289e-01 -1.03134677e-01 -1.87754765e-01 7.53297880e-02
3.04796517e-01 1.44465268e+00 -3.87344599e-01 -3.74031156e-01
-6.37779891e-01 1.03374493e+00 7.20328271e-01 1.77734584e-01
-9.74816442e-01 -7.44324505e-01 9.32017207e-01 -1.49106756e-01
9.03168321e-01 1.12126529e-01 -4.30893935e-02 -1.36721480e+00
1.36462599e-01 -8.88852477e-01 5.25036991e-01 -5.84757149e-01
-1.76480591e+00 7.17288554e-01 -1.33388028e-01 -1.21423447e+00
1.11307494e-01 -5.44973493e-01 -5.97447395e-01 9.39740241e-01
-1.88962805e+00 -1.44395018e+00 -5.15636802e-01 9.12569523e-01
7.23131955e-01 -5.04964441e-02 7.23163903e-01 6.92886055e-01
-8.10847282e-01 1.03655756e+00 -9.20393020e-02 3.12164962e-01
7.00114310e-01 -1.00864065e+00 2.52251953e-01 6.45011783e-01
-6.90004155e-02 5.76005042e-01 1.60286352e-01 -3.01580101e-01
-1.18935645e+00 -1.51530957e+00 8.01637769e-01 -1.79318532e-01
6.19543076e-01 -4.24501479e-01 -1.03684056e+00 6.87183440e-01
-4.57036728e-03 7.69329429e-01 5.99606812e-01 6.72230497e-02
-7.16846108e-01 -2.61054844e-01 -1.16194046e+00 -6.83442364e-03
9.83056366e-01 -5.42682350e-01 -7.90055335e-01 1.37621686e-01
7.30401933e-01 -2.47060210e-01 -8.80863488e-01 5.24347007e-01
5.29905736e-01 -7.60905921e-01 8.08161199e-01 -7.86294341e-01
4.97334480e-01 -7.46174037e-01 -4.18365926e-01 -1.34786403e+00
-9.72186565e-01 -1.02128997e-01 -1.00825690e-01 1.19156671e+00
1.08864851e-01 -6.75107121e-01 5.90254128e-01 3.03316236e-01
-3.31221104e-01 -1.01151180e+00 -8.65317523e-01 -6.81021452e-01
2.19186231e-01 -7.31546506e-02 9.09675777e-01 8.09683263e-01
-3.35976750e-01 2.96314478e-01 -3.95813465e-01 2.86695868e-01
7.34851241e-01 6.68701708e-01 8.06642532e-01 -1.05824220e+00
-3.44256163e-01 -4.83280897e-01 -7.03803897e-01 -1.42279756e+00
2.27279942e-02 -7.74812222e-01 2.64441580e-01 -1.27639198e+00
5.15228033e-01 -6.54635131e-01 -5.64335048e-01 8.31925988e-01
-3.98338109e-01 4.19332683e-01 1.99667528e-01 2.74482429e-01
-7.72939265e-01 8.16035390e-01 1.29240656e+00 -4.81852531e-01
2.38632455e-01 -2.00735018e-01 -9.03419912e-01 6.61952615e-01
6.37851536e-01 -4.74767417e-01 -2.24223912e-01 -6.02669299e-01
-5.31370699e-01 -3.23422998e-01 5.85212767e-01 -7.90543914e-01
3.30393463e-01 -2.81865388e-01 6.15483403e-01 -7.49476373e-01
9.11209211e-02 -7.40863085e-01 -1.21292740e-01 2.79875457e-01
-2.21637234e-01 -1.75851196e-01 1.11939184e-01 4.62936342e-01
-4.84304726e-01 1.53401658e-01 1.12703466e+00 -1.01904549e-01
-8.57352078e-01 1.03658211e+00 1.94616184e-01 1.02797754e-01
1.20798624e+00 -4.11483496e-02 -2.99594283e-01 2.13117942e-01
-5.08996606e-01 4.16607082e-01 2.84187615e-01 6.93398118e-01
5.39769292e-01 -1.78572667e+00 -8.38160932e-01 4.96266186e-01
6.87148988e-01 3.17410797e-01 7.13171244e-01 6.28638327e-01
-1.44616947e-01 5.53991854e-01 -4.08208072e-01 -6.85264289e-01
-1.18726659e+00 9.75426674e-01 4.53522176e-01 -1.78799033e-01
-4.67772692e-01 1.18778026e+00 8.72995615e-01 -5.55865228e-01
3.32920164e-01 -4.03221548e-01 -6.76126480e-02 -1.23488441e-01
7.44533837e-01 -6.54618666e-02 -8.45493898e-02 -8.29486787e-01
-4.63026226e-01 8.11024010e-01 -3.84823442e-01 6.85182512e-01
1.45974457e+00 -1.21010818e-01 -2.10627262e-02 1.20607272e-01
1.49611986e+00 -4.01573420e-01 -1.46659708e+00 -3.96565765e-01
-3.09296519e-01 -7.59075403e-01 1.23374000e-01 -7.67455935e-01
-1.45804119e+00 1.09347296e+00 6.55360460e-01 8.24090913e-02
1.33274460e+00 3.53839487e-01 8.71511579e-01 2.27267161e-01
3.56135666e-01 -9.48526382e-01 6.54369518e-02 3.19247127e-01
1.19420266e+00 -1.47639704e+00 -2.61229604e-01 -2.01508820e-01
-5.53932607e-01 9.60187614e-01 1.15239275e+00 -3.29823911e-01
8.62448215e-01 7.14033470e-02 -2.10706126e-02 -1.87092453e-01
-8.33370864e-01 -2.91324019e-01 2.82486171e-01 5.24323940e-01
3.19409251e-01 2.52273560e-01 -8.30138251e-02 9.09446180e-01
5.88676222e-02 1.60413072e-01 -1.40173674e-01 6.49738789e-01
-4.50076133e-01 -8.68505061e-01 -1.66759968e-01 4.71957743e-01
-3.67450505e-01 -1.44906476e-01 -1.24147676e-01 5.33038974e-01
5.73100924e-01 7.98503041e-01 2.29520783e-01 -4.48973656e-01
5.05680919e-01 -5.16477406e-01 4.10310119e-01 -4.49594826e-01
-4.60159600e-01 -4.71237712e-02 -3.07715565e-01 -6.49598956e-01
-4.44961160e-01 -5.40273011e-01 -1.08062911e+00 -3.36273015e-01
-5.31188190e-01 -6.64588884e-02 3.49512875e-01 8.28080297e-01
6.87444746e-01 4.98738915e-01 8.26710165e-01 -9.21790481e-01
-5.94232619e-01 -1.12881279e+00 -4.40205842e-01 4.26688433e-01
3.71311516e-01 -1.03790808e+00 -2.12876022e-01 4.47409451e-02] | [9.59341049194336, 2.0033609867095947] |
72b5089a-4eaf-4ac5-a922-0763c7db50a3 | graph-constrained-data-representation | 2107.13362 | null | https://arxiv.org/abs/2107.13362v2 | https://arxiv.org/pdf/2107.13362v2.pdf | Graph Constrained Data Representation Learning for Human Motion Segmentation | Recently, transfer subspace learning based approaches have shown to be a valid alternative to unsupervised subspace clustering and temporal data clustering for human motion segmentation (HMS). These approaches leverage prior knowledge from a source domain to improve clustering performance on a target domain, and currently they represent the state of the art in HMS. Bucking this trend, in this paper, we propose a novel unsupervised model that learns a representation of the data and digs clustering information from the data itself. Our model is reminiscent of temporal subspace clustering, but presents two critical differences. First, we learn an auxiliary data matrix that can deviate from the initial data, hence confer more degrees of freedom to the coding matrix. Second, we introduce a regularization term for this auxiliary data matrix that preserves the local geometrical structure present in the high-dimensional space. The proposed model is efficiently optimized by using an original Alternating Direction Method of Multipliers (ADMM) formulation allowing to learn jointly the auxiliary data representation, a nonnegative dictionary and a coding matrix. Experimental results on four benchmark datasets for HMS demonstrate that our approach achieves significantly better clustering performance then state-of-the-art methods, including both unsupervised and more recent semi-supervised transfer learning approaches. | ['Herwig Wendt', 'Guillem Rodriguez-Corominas', 'Lluís Garrido', 'Mariella Dimiccoli'] | 2021-07-28 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Dimiccoli_Graph_Constrained_Data_Representation_Learning_for_Human_Motion_Segmentation_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Dimiccoli_Graph_Constrained_Data_Representation_Learning_for_Human_Motion_Segmentation_ICCV_2021_paper.pdf | iccv-2021-1 | ['motion-segmentation'] | ['computer-vision'] | [ 2.01623127e-01 -7.65819699e-02 -3.13453734e-01 -1.90042138e-01
-7.10695744e-01 -5.94229698e-01 4.53476191e-01 -2.70930618e-01
-4.90949690e-01 2.46001601e-01 4.06658262e-01 1.60717711e-01
-1.63602993e-01 -2.43346572e-01 -5.25705636e-01 -1.23308384e+00
-1.86704081e-02 6.90956533e-01 2.68989444e-01 6.71561109e-03
9.41931680e-02 2.05862060e-01 -1.33417571e+00 2.94259697e-01
1.00739872e+00 5.71467102e-01 1.79099560e-01 2.86532581e-01
2.14813039e-01 7.50776052e-01 -5.28876446e-02 2.80599505e-01
4.13478315e-01 -5.92197657e-01 -1.00431538e+00 6.68362319e-01
2.50524580e-01 3.32144536e-02 -5.40723443e-01 8.23056936e-01
3.11193496e-01 5.60229242e-01 9.81203556e-01 -1.28434086e+00
-2.54345328e-01 2.96089619e-01 -6.92206383e-01 -1.72845170e-01
4.99320589e-02 -1.48994559e-02 7.78495908e-01 -8.40959609e-01
1.12339985e+00 1.06184828e+00 5.84360480e-01 5.57217181e-01
-1.62243748e+00 -2.36120492e-01 1.17079474e-01 4.55858141e-01
-1.44277513e+00 -3.48149419e-01 1.16693783e+00 -9.93007839e-01
4.04734462e-01 2.08928481e-01 6.83383882e-01 1.02970982e+00
-3.26386571e-01 1.06862819e+00 1.05729222e+00 -4.79428977e-01
7.09988654e-01 4.50456962e-02 1.22080557e-01 4.19283956e-01
-1.69668302e-01 -1.05142310e-01 -5.23546338e-01 -2.31215388e-01
6.79057539e-01 -7.36091360e-02 -3.12835693e-01 -1.56714272e+00
-1.34925044e+00 1.14190531e+00 3.16923171e-01 5.76162398e-01
-3.17538232e-01 -1.17251202e-01 2.74394870e-01 -7.95516819e-02
3.59297633e-01 -4.89816256e-02 -2.31815830e-01 -1.11507826e-01
-1.24046087e+00 2.09414717e-02 6.28471017e-01 8.80114377e-01
8.63950193e-01 -1.03891559e-01 -3.58960144e-02 7.27838814e-01
3.46534401e-01 2.55395651e-01 6.45349920e-01 -1.28697777e+00
4.52344775e-01 6.43630624e-01 -1.68977827e-02 -9.96963739e-01
-4.18167949e-01 -1.83515474e-01 -1.10423458e+00 -1.53743476e-02
5.22820234e-01 1.86971836e-02 -8.44973266e-01 1.80354548e+00
4.91504341e-01 4.72213298e-01 4.88996431e-02 1.05765069e+00
1.58298969e-01 5.67368686e-01 -2.70857722e-01 -3.50291491e-01
6.78429365e-01 -1.03403449e+00 -6.66425109e-01 3.09790391e-02
7.60446370e-01 -5.17147422e-01 7.06125081e-01 3.76369566e-01
-8.07407200e-01 -5.03304183e-01 -9.21696067e-01 -5.22188582e-02
-2.32765287e-01 2.24077433e-01 4.39766973e-01 5.12597859e-01
-1.14861369e+00 6.60445392e-01 -1.31335068e+00 -7.11744726e-01
2.52874523e-01 4.22688514e-01 -4.15854216e-01 -6.90731546e-03
-7.39825547e-01 4.14942384e-01 5.63264430e-01 6.35801554e-02
-6.24767601e-01 -4.84119773e-01 -9.12083864e-01 -3.38566393e-01
4.04153347e-01 -6.74651623e-01 7.43435562e-01 -7.36422479e-01
-1.51938903e+00 8.15559804e-01 -3.06767553e-01 -3.78671706e-01
4.83560413e-01 -1.63554877e-01 -5.88973425e-02 5.39014816e-01
2.28835091e-01 7.34903514e-01 1.17519808e+00 -1.49399078e+00
-2.74491608e-01 -4.46929365e-01 -4.84947562e-01 2.83220977e-01
-7.18809128e-01 -3.76453102e-01 -7.54659235e-01 -7.98144341e-01
4.30941015e-01 -1.40913606e+00 -5.94169736e-01 -1.18360221e-01
-2.71705031e-01 2.23339442e-02 1.13371444e+00 -6.41483247e-01
1.34473443e+00 -2.27229333e+00 1.14371765e+00 4.78462726e-01
1.18595503e-01 1.44266218e-01 -1.47269499e-02 4.75563616e-01
-2.33421832e-01 -3.87566447e-01 -8.42400670e-01 -3.77959818e-01
-8.29338208e-02 4.82292265e-01 -1.16637396e-02 7.90778518e-01
6.92459382e-03 8.08494985e-01 -9.68088090e-01 -6.05760813e-01
4.53090876e-01 4.78074372e-01 -8.12237799e-01 6.59211725e-02
6.36986196e-02 1.11582804e+00 -3.46442282e-01 2.16599077e-01
5.99834859e-01 -2.92743176e-01 3.85628581e-01 -3.01554918e-01
-7.95059204e-02 -2.59610891e-01 -1.61096776e+00 2.44542694e+00
1.69921726e-01 4.39018786e-01 2.46156082e-01 -1.60826099e+00
4.65112567e-01 3.22956115e-01 1.18978012e+00 -1.89829633e-01
1.39617071e-01 8.83411691e-02 -1.72753140e-01 -4.43110555e-01
1.64744258e-01 1.17415737e-03 1.00585073e-01 2.48325109e-01
1.92383349e-01 4.83919643e-02 2.75846690e-01 4.69297558e-01
8.27643394e-01 3.70241910e-01 8.74664709e-02 -5.53434372e-01
7.85397172e-01 2.41169959e-01 8.22229922e-01 3.11443388e-01
-2.42931426e-01 8.15048933e-01 2.19749331e-01 -1.84606552e-01
-9.50179815e-01 -1.01974726e+00 -4.95906249e-02 7.48200238e-01
1.20543569e-01 -4.37204421e-01 -1.11491680e+00 -6.42334223e-01
-2.47472331e-01 3.14228505e-01 -7.65610695e-01 -1.09107450e-01
-8.53950679e-01 -6.94634736e-01 6.00792579e-02 6.26357555e-01
3.65525544e-01 -7.01319993e-01 -6.19708538e-01 6.12980090e-02
-5.97610593e-01 -1.28018188e+00 -6.69656515e-01 2.08779588e-01
-1.26167512e+00 -1.06890821e+00 -8.79494667e-01 -1.03081942e+00
8.69381428e-01 3.74569923e-01 6.50958240e-01 -1.57940194e-01
-2.81442672e-01 8.85418713e-01 -5.24580300e-01 2.04581201e-01
-1.89013273e-01 1.13015577e-01 2.59945601e-01 5.05559266e-01
3.21377724e-01 -7.67446578e-01 -5.59165955e-01 3.73485327e-01
-1.09254038e+00 8.73131678e-02 4.38054264e-01 9.76708710e-01
6.50329947e-01 9.86614674e-02 3.92667390e-02 -8.59060884e-01
-4.64769565e-02 -5.71336806e-01 -3.55788291e-01 -5.60976490e-02
-5.53520620e-01 1.13120981e-01 5.40317953e-01 -4.95100528e-01
-9.22108650e-01 8.13384891e-01 5.24216413e-01 -9.28155839e-01
-2.60152072e-01 4.38227683e-01 -2.91759908e-01 -8.92563537e-03
5.10795057e-01 4.41940486e-01 2.84923255e-01 -8.23534131e-01
6.88240230e-01 4.24469054e-01 6.80272281e-01 -6.34552777e-01
1.25808990e+00 1.16471982e+00 5.93538918e-02 -1.04378653e+00
-4.27447140e-01 -1.30197144e+00 -1.67098320e+00 -1.65383115e-01
1.20899427e+00 -9.09161866e-01 -2.10590452e-01 4.79569495e-01
-8.29014540e-01 -3.82529080e-01 -5.18066287e-01 6.31465971e-01
-8.88363361e-01 8.29652548e-01 -5.00839114e-01 -5.99027574e-01
3.24135661e-01 -9.25434172e-01 1.04638302e+00 -2.80875176e-01
-1.71345025e-01 -1.12525451e+00 2.92271823e-01 4.94952530e-01
-3.65603268e-02 5.84601521e-01 7.21177161e-01 -2.94731617e-01
-4.28360671e-01 1.43819824e-01 2.95321107e-01 3.06296736e-01
1.42329827e-01 -3.25046420e-01 -7.44807601e-01 -6.38203382e-01
3.32984746e-01 -1.15008364e-02 1.01993728e+00 5.21662533e-01
7.96697080e-01 -2.54871011e-01 -5.21479249e-01 7.00378358e-01
1.28328538e+00 1.57454669e-01 4.83504415e-01 3.01463038e-01
1.18510020e+00 8.94089222e-01 6.12848818e-01 4.49840099e-01
4.38832790e-01 9.02636945e-01 -1.52254794e-02 -1.18010640e-01
1.17902756e-02 -1.27085865e-01 4.77122158e-01 1.30968225e+00
-1.39294192e-01 4.58143055e-01 -1.08366156e+00 8.70946944e-01
-2.51375556e+00 -8.39041412e-01 -2.93311656e-01 2.22389412e+00
5.76283097e-01 -3.52691263e-01 6.04721785e-01 5.37691534e-01
6.62949741e-01 1.22300908e-01 -6.36953890e-01 2.76701927e-01
-1.74630374e-01 -7.93138593e-02 3.56211573e-01 3.65325689e-01
-1.42875671e+00 1.03830862e+00 5.78888559e+00 6.86303079e-01
-9.93844926e-01 1.53263345e-01 1.25097305e-01 1.21013023e-01
-1.87541991e-02 1.27285600e-01 -2.02375367e-01 3.53402257e-01
7.66126275e-01 3.35272253e-02 5.38396537e-01 6.91106677e-01
3.15930277e-01 8.00351128e-02 -1.26696932e+00 1.07826090e+00
2.85491318e-01 -1.14106810e+00 7.88028985e-02 3.89301747e-01
1.10832560e+00 -1.45231918e-01 1.39520615e-01 -1.41509157e-02
8.27807784e-02 -6.11764848e-01 7.07224548e-01 4.11162466e-01
3.54860574e-01 -5.75790346e-01 3.58651847e-01 4.76261914e-01
-1.25713503e+00 -1.72646716e-02 -2.37964153e-01 5.37849069e-02
2.75872797e-02 3.35686982e-01 -4.58595812e-01 6.52986526e-01
7.98115790e-01 1.24715698e+00 -6.60099268e-01 7.87996292e-01
1.07556589e-01 6.25900567e-01 -2.32302547e-01 5.98785043e-01
4.26995784e-01 -8.22523654e-01 8.09653640e-01 1.13263416e+00
-2.41467468e-02 2.37018242e-01 4.16063279e-01 7.44693577e-01
3.09200197e-01 2.25731745e-01 -5.65174341e-01 -2.79497150e-02
2.84572132e-03 1.20201278e+00 -9.15951550e-01 -3.53692293e-01
-4.07798886e-01 1.36637092e+00 1.92706898e-01 7.43645310e-01
-5.64496279e-01 2.62688845e-01 3.27075124e-01 6.15059212e-03
8.05283546e-01 -6.64236605e-01 -2.11272970e-01 -1.45512700e+00
1.83646560e-01 -8.10432911e-01 6.20387912e-01 -2.27462158e-01
-1.12811995e+00 1.30362839e-01 2.24472716e-01 -1.80671561e+00
-3.86392146e-01 -4.92452651e-01 -2.96793669e-01 3.04358363e-01
-1.29169488e+00 -1.28627861e+00 -1.12598099e-01 1.07838368e+00
4.76373851e-01 -2.58655697e-01 6.28107607e-01 3.04790169e-01
-5.69772720e-01 1.40339464e-01 6.73272491e-01 2.59856194e-01
7.14346826e-01 -1.48182416e+00 -8.86189565e-02 9.43945885e-01
4.46514785e-01 7.88043916e-01 5.16056359e-01 -4.76309717e-01
-1.65052915e+00 -1.10023284e+00 2.74156868e-01 -4.88345176e-01
6.65725470e-01 -4.61217642e-01 -9.30088282e-01 7.41648138e-01
6.01113215e-02 -5.06502837e-02 9.88850176e-01 -6.06835298e-02
-3.89071286e-01 1.85505643e-01 -8.69717777e-01 4.89975810e-01
9.81265664e-01 -5.22406638e-01 -5.15830994e-01 2.64737666e-01
3.85467708e-01 -2.47261778e-01 -8.53220701e-01 2.98456311e-01
3.54209900e-01 -9.70813990e-01 9.82326031e-01 -4.98996317e-01
1.77751973e-01 -7.12911665e-01 -2.50728756e-01 -1.29212439e+00
-5.01237750e-01 -8.65117371e-01 -4.37906176e-01 1.11154759e+00
6.18457608e-03 -7.50460476e-02 1.20120943e+00 5.58197975e-01
-9.49131250e-02 -4.23238069e-01 -1.06911576e+00 -9.70236897e-01
4.06790793e-01 -3.59301239e-01 -1.18235029e-01 1.40910780e+00
1.91660181e-01 3.76988441e-01 -4.65762436e-01 1.64861739e-01
1.14107203e+00 1.71436206e-01 8.05127800e-01 -1.32217789e+00
-2.96892643e-01 -2.63369232e-01 -5.23793221e-01 -1.16693592e+00
2.69907594e-01 -1.00436687e+00 -1.08017735e-02 -1.33931184e+00
3.49716961e-01 -9.98233631e-02 -2.47625560e-01 3.27850163e-01
8.01032558e-02 1.50138095e-01 3.26963365e-01 6.58700466e-01
-7.53490388e-01 8.27569783e-01 9.26270008e-01 -2.78727561e-01
-5.47144711e-01 -3.28993797e-01 -3.51995766e-01 8.70260835e-01
4.58047003e-01 -3.30977738e-01 -5.74060977e-01 -3.40872586e-01
-4.18507397e-01 -1.01502791e-01 3.51907790e-01 -1.12078726e+00
6.40482605e-01 -1.05145760e-01 2.28334635e-01 -6.66964471e-01
3.18468720e-01 -1.10504234e+00 6.60463125e-02 4.87212032e-01
-8.77769962e-02 -4.92741227e-01 -2.61871275e-02 9.54890370e-01
-4.22375679e-01 1.96112886e-01 8.83408010e-01 4.83652204e-02
-8.44608784e-01 1.69509158e-01 -4.80415463e-01 1.26472041e-01
9.67760682e-01 -5.35627842e-01 4.16894853e-01 -3.05106670e-01
-1.00983977e+00 2.55032569e-01 6.41208589e-01 5.05049825e-01
4.88108516e-01 -1.52523649e+00 -5.79353929e-01 9.38687250e-02
1.94257513e-01 1.48693845e-01 3.27396095e-01 1.31757593e+00
-1.63534626e-01 5.19097269e-01 -2.35641450e-01 -1.12275207e+00
-1.05602562e+00 8.49913001e-01 -6.60868213e-02 -1.04480669e-01
-8.81357729e-01 4.26159829e-01 3.98310125e-01 -4.61214483e-01
2.18326032e-01 -2.57149160e-01 -2.96782702e-01 2.32053831e-01
1.67003527e-01 8.01412523e-01 -2.42842570e-01 -1.21126437e+00
-4.10486281e-01 1.05841088e+00 1.15847424e-01 -3.17640334e-01
1.42044628e+00 -3.82419199e-01 -1.45964697e-01 7.35579073e-01
1.46230209e+00 -2.32869670e-01 -1.40110266e+00 -4.48716402e-01
4.11477566e-01 -4.06324774e-01 -1.10430159e-02 -8.24159682e-02
-1.07599556e+00 8.19209218e-01 5.84070683e-01 -9.94403139e-02
1.27172971e+00 -3.03993337e-02 7.84431875e-01 2.43935153e-01
3.79444808e-01 -1.53412402e+00 2.27283388e-01 3.31137210e-01
6.97439373e-01 -1.34036922e+00 -4.98094559e-02 -7.16683805e-01
-8.27886939e-01 8.97854149e-01 3.22590739e-01 -2.27488816e-01
7.89899230e-01 -1.58714801e-01 7.60300606e-02 -2.35227093e-01
-2.53826052e-01 -3.31090271e-01 6.41867638e-01 7.90092289e-01
1.28157273e-01 -9.25856680e-02 -3.28281313e-01 3.92897576e-01
4.64750938e-02 -4.84566800e-02 3.39451373e-01 9.93792534e-01
-8.43000412e-02 -1.28972328e+00 -5.39210498e-01 -5.49579971e-02
7.67280757e-02 2.03671947e-01 -5.66907406e-01 6.28353357e-01
1.37122348e-01 1.00707686e+00 -1.53595552e-01 -2.70147920e-01
5.91640957e-02 1.65671915e-01 4.28346306e-01 -5.99554896e-01
-3.13954763e-02 6.39068246e-01 -6.01456940e-01 -7.49669433e-01
-9.73997355e-01 -1.15062523e+00 -1.32838809e+00 2.62192935e-01
3.81075628e-02 4.35537726e-01 4.52436179e-01 8.79259586e-01
3.26538205e-01 2.09403127e-01 6.37256980e-01 -1.09708130e+00
-2.06175998e-01 -6.44479275e-01 -7.78399944e-01 9.05388713e-01
2.38119990e-01 -7.70072103e-01 -3.45988005e-01 7.61212409e-01] | [7.9792866706848145, 4.3353047370910645] |
dfbbb243-dc27-4471-a7e9-35750d529e71 | weakly-supervised-generative-network-for | 2008.05770 | null | https://arxiv.org/abs/2008.05770v1 | https://arxiv.org/pdf/2008.05770v1.pdf | Weakly Supervised Generative Network for Multiple 3D Human Pose Hypotheses | 3D human pose estimation from a single image is an inverse problem due to the inherent ambiguity of the missing depth. Several previous works addressed the inverse problem by generating multiple hypotheses. However, these works are strongly supervised and require ground truth 2D-to-3D correspondences which can be difficult to obtain. In this paper, we propose a weakly supervised deep generative network to address the inverse problem and circumvent the need for ground truth 2D-to-3D correspondences. To this end, we design our network to model a proposal distribution which we use to approximate the unknown multi-modal target posterior distribution. We achieve the approximation by minimizing the KL divergence between the proposal and target distributions, and this leads to a 2D reprojection error and a prior loss term that can be weakly supervised. Furthermore, we determine the most probable solution as the conditional mode of the samples using the mean-shift algorithm. We evaluate our method on three benchmark datasets -- Human3.6M, MPII and MPI-INF-3DHP. Experimental results show that our approach is capable of generating multiple feasible hypotheses and achieves state-of-the-art results compared to existing weakly supervised approaches. Our source code is available at the project website. | ['Chen Li', 'Gim Hee Lee'] | 2020-08-13 | null | null | null | null | ['multi-hypotheses-3d-human-pose-estimation'] | ['computer-vision'] | [ 6.10034503e-02 4.04629976e-01 2.02005845e-03 -4.55392867e-01
-1.00866497e+00 -3.05492043e-01 5.26008308e-01 -4.40332860e-01
-4.03072894e-01 7.90066719e-01 1.29637584e-01 2.59175211e-01
3.25956829e-02 -6.43506289e-01 -9.59520578e-01 -5.86402714e-01
3.79362494e-01 1.17161739e+00 1.88962519e-01 -4.10869718e-02
8.24646205e-02 2.00131312e-01 -1.44025254e+00 6.04572892e-02
6.98900640e-01 8.40265930e-01 2.13469476e-01 4.39241886e-01
1.86128780e-01 3.50635499e-01 -4.86066401e-01 -4.27658707e-01
3.95226657e-01 -6.01015270e-01 -6.17843688e-01 3.93893480e-01
4.71771806e-01 -5.14187098e-01 -5.46058975e-02 1.02863312e+00
6.91878498e-01 2.98221316e-02 9.80724275e-01 -1.36846662e+00
-3.80162209e-01 6.94441274e-02 -9.00558650e-01 -4.80486840e-01
5.27829885e-01 4.38950472e-02 7.64520347e-01 -1.06262589e+00
6.59325182e-01 1.46577716e+00 7.82519162e-01 5.76883554e-01
-1.40284348e+00 -3.96529108e-01 -9.31599438e-02 -2.60988399e-02
-1.64831078e+00 -2.90561229e-01 9.60600317e-01 -5.78135550e-01
4.41912353e-01 -2.13228345e-01 6.37189031e-01 1.33957899e+00
1.92697272e-01 8.39780748e-01 1.16438901e+00 -4.09693509e-01
2.31748983e-01 7.68333003e-02 -3.54176044e-01 8.73964131e-01
1.15589157e-01 1.02966338e-01 -6.53301537e-01 -1.14379421e-01
1.20203757e+00 -2.33531415e-01 -2.32682049e-01 -7.39363670e-01
-1.19459105e+00 8.36176634e-01 4.18776035e-01 -1.60943419e-01
-4.65606064e-01 2.02723250e-01 -1.57767609e-01 -2.29040116e-01
4.91676033e-01 8.30955803e-02 -2.39538923e-01 4.51944508e-02
-8.48793089e-01 5.05572796e-01 7.44473755e-01 9.26275432e-01
9.09631729e-01 -3.86044234e-01 7.06458166e-02 7.43430138e-01
8.23083639e-01 6.76618516e-01 9.89572033e-02 -1.24427414e+00
4.02385771e-01 2.88777292e-01 3.96300048e-01 -9.21160102e-01
-3.80148560e-01 -4.72876012e-01 -8.69377315e-01 2.88173616e-01
6.46442711e-01 -1.87429681e-01 -1.06047678e+00 1.84901261e+00
6.05336070e-01 1.01306491e-01 -5.70507571e-02 1.32867718e+00
5.77523112e-01 4.86961603e-01 -3.02993327e-01 8.66893977e-02
9.91602302e-01 -8.67868304e-01 -6.82196498e-01 -6.20213151e-01
2.27918327e-01 -8.09842050e-01 1.01316297e+00 3.10681373e-01
-1.13263965e+00 -4.81877774e-01 -8.04337204e-01 -5.11135347e-03
2.82031894e-01 3.41771007e-01 1.53779492e-01 2.00671092e-01
-7.89970875e-01 3.87396842e-01 -1.00442696e+00 -2.56155163e-01
2.36771300e-01 2.07414836e-01 -3.81027371e-01 -1.30840763e-01
-1.02908194e+00 9.59064543e-01 4.24090207e-01 2.03282163e-01
-8.30759764e-01 -5.44115543e-01 -9.96074915e-01 -3.88842493e-01
4.05415595e-01 -1.07350993e+00 1.20075107e+00 -6.22988284e-01
-1.57959843e+00 9.95950818e-01 -1.47550955e-01 -1.92383304e-01
8.55073929e-01 -3.95675153e-01 3.33020836e-01 1.49559855e-01
4.55401957e-01 9.17812288e-01 7.36887693e-01 -1.72629333e+00
-4.14942443e-01 -4.71014202e-01 -5.14849387e-02 4.54994529e-01
1.57164246e-01 -6.55421078e-01 -7.76258588e-01 -5.88322639e-01
5.05723596e-01 -1.20380247e+00 -3.59308630e-01 2.52535969e-01
-6.69926167e-01 -3.70254256e-02 4.85212088e-01 -6.92685366e-01
6.51118159e-01 -2.04932523e+00 5.70446134e-01 2.27382436e-01
9.12344158e-02 -3.27515870e-01 -5.39050810e-03 8.62993449e-02
3.65416914e-01 -2.92949438e-01 -5.12290418e-01 -8.52258742e-01
2.84748524e-01 2.97290772e-01 2.77971085e-02 7.64127612e-01
2.56252866e-02 8.37599158e-01 -8.32697153e-01 -6.60782874e-01
3.10806513e-01 7.56435454e-01 -6.49702728e-01 4.85377371e-01
-3.36196184e-01 6.80443823e-01 -2.30706483e-01 4.48634356e-01
7.58839071e-01 -4.14131820e-01 1.13768838e-01 -4.01643395e-01
2.36150682e-01 -3.11660077e-02 -1.36114371e+00 2.08354664e+00
-3.92933100e-01 1.82557806e-01 4.37975563e-02 -8.35606575e-01
9.61092889e-01 1.72369152e-01 5.01673043e-01 -2.74196148e-01
2.94388145e-01 3.02114904e-01 -2.61494905e-01 -2.92451620e-01
3.40883255e-01 -5.18236101e-01 -1.38037384e-01 3.58624071e-01
1.12314031e-01 -3.90244693e-01 -1.59344137e-01 1.09623447e-02
5.41162431e-01 7.26050913e-01 1.10016041e-01 -7.97760338e-02
2.90268213e-01 -9.58733484e-02 7.34004438e-01 5.26230633e-01
1.03411585e-01 1.15039921e+00 5.09499013e-01 -3.30900699e-01
-1.11338103e+00 -1.44674480e+00 -1.89504717e-02 3.29989702e-01
3.61713260e-01 -1.37943223e-01 -8.22409153e-01 -6.44078255e-01
-5.01825437e-02 6.30775690e-01 -5.66629350e-01 6.17999211e-03
-4.36692059e-01 -5.93949080e-01 3.14569980e-01 5.44415832e-01
5.30257761e-01 -6.88932598e-01 -5.55998623e-01 3.01973503e-02
-8.54441226e-01 -1.22747624e+00 -5.57851791e-01 -7.16437101e-02
-7.02338755e-01 -9.41329420e-01 -1.02223790e+00 -6.91767573e-01
9.54487503e-01 -1.60789102e-01 1.09027719e+00 -3.16192061e-01
-1.36646301e-01 4.68674630e-01 -3.66656594e-02 -2.90637761e-01
-3.01474929e-01 -3.84459272e-02 1.78204104e-01 1.16017736e-01
2.78992057e-01 -5.53188086e-01 -4.95295882e-01 5.74458718e-01
-6.44697607e-01 1.92216218e-01 5.87974429e-01 9.56169009e-01
1.09044433e+00 -6.33612126e-02 2.82430708e-01 -5.25110006e-01
2.09469527e-01 -3.05036873e-01 -6.71254873e-01 -5.80186136e-02
-1.71219736e-01 3.06272328e-01 1.41289487e-01 -5.35686791e-01
-1.16278541e+00 5.89990437e-01 -1.92885071e-01 -7.18653440e-01
-3.05998772e-01 4.14642453e-01 -3.54936659e-01 1.85923308e-01
6.87873781e-01 5.48709258e-02 3.49047750e-01 -5.22734225e-01
2.82215059e-01 4.13449407e-01 7.08141804e-01 -7.35231936e-01
9.65146124e-01 6.86737597e-01 2.76383609e-01 -6.87252879e-01
-1.31625962e+00 -2.85855204e-01 -8.32634270e-01 -3.43753189e-01
1.00517702e+00 -1.03763175e+00 -5.72742462e-01 5.63740671e-01
-1.38853633e+00 -4.22533929e-01 -2.26490945e-01 6.86504722e-01
-9.11966145e-01 5.75600147e-01 -4.38864499e-01 -7.02461839e-01
-1.29140625e-02 -1.14678419e+00 1.52965391e+00 -2.35028323e-02
-4.66269255e-01 -9.99303639e-01 1.50487468e-01 4.70967233e-01
-3.30297239e-02 6.15509033e-01 5.25144339e-01 -2.72925422e-02
-6.37240708e-01 -1.77932665e-01 -6.12460449e-03 3.13366950e-01
-2.58602425e-02 -4.55680043e-01 -7.13833332e-01 -3.02144140e-01
6.62283376e-02 -4.76084143e-01 6.42519534e-01 7.15721488e-01
8.90052855e-01 -1.28402621e-01 -3.23039591e-01 5.64046443e-01
1.21328807e+00 -3.81600738e-01 6.00265384e-01 2.95696765e-01
8.52043271e-01 8.31243932e-01 9.17081177e-01 6.57565713e-01
6.85510874e-01 9.07216191e-01 5.43909609e-01 5.16384020e-02
-1.95724696e-01 -6.74721301e-01 1.72713682e-01 5.43026745e-01
-1.27696633e-01 -2.90398031e-01 -9.33896482e-01 5.85338652e-01
-2.04046488e+00 -6.79280579e-01 -1.43305197e-01 2.28135872e+00
7.38095343e-01 1.42773181e-01 1.28099233e-01 -6.17174581e-02
6.61556005e-01 -1.37851626e-01 -4.70572412e-01 4.16642129e-01
1.10175595e-01 -8.55140314e-02 1.69434190e-01 8.14349651e-01
-1.02338779e+00 7.80809522e-01 5.74856186e+00 5.43887079e-01
-7.54935324e-01 3.64977047e-02 4.42579210e-01 -1.35562032e-01
-3.62924337e-01 -2.99012270e-02 -9.76435483e-01 4.66516078e-01
3.86295795e-01 2.00032055e-01 1.90245137e-01 6.89467490e-01
7.99971074e-02 -2.72389919e-01 -1.33134592e+00 1.24157846e+00
2.29878992e-01 -9.63885903e-01 -9.18200165e-02 3.54593545e-01
7.49979675e-01 -1.44929320e-01 1.96781196e-02 -5.72134480e-02
2.05508307e-01 -8.52626204e-01 1.05985141e+00 7.72447050e-01
8.00789118e-01 -9.07936990e-01 7.88113654e-01 7.83958018e-01
-9.59804833e-01 3.62876713e-01 -2.69569039e-01 9.21355411e-02
6.38315856e-01 8.88198197e-01 -8.31376553e-01 3.83685619e-01
7.91826665e-01 5.70098877e-01 -1.40384108e-01 7.98070788e-01
-5.40683091e-01 1.02188163e-01 -6.34995162e-01 1.59661472e-01
-7.11558834e-02 -2.61489064e-01 6.37521327e-01 7.89239168e-01
5.24089038e-01 -1.54974777e-02 4.65211183e-01 1.28736138e+00
2.15045437e-02 -2.88659185e-01 -3.87305140e-01 4.35832500e-01
3.04361492e-01 9.74979460e-01 -6.86885595e-01 -5.07065728e-02
1.42927235e-02 1.33073843e+00 3.17891985e-01 2.61627644e-01
-9.85692978e-01 -1.24963611e-01 3.67321312e-01 2.72970885e-01
1.57835320e-01 -3.30170065e-01 -1.82983518e-01 -1.19265842e+00
3.41195822e-01 -7.28719950e-01 1.24086842e-01 -8.04445505e-01
-1.49438846e+00 4.19260591e-01 3.33991170e-01 -1.23876965e+00
-5.71282446e-01 -5.26669443e-01 -1.82526931e-01 9.14153218e-01
-1.17828250e+00 -1.18072772e+00 -4.63182896e-01 3.37972611e-01
2.57301480e-01 5.03425822e-02 5.22393763e-01 2.76461065e-01
-1.17864616e-01 4.32052702e-01 -2.71002114e-01 3.81113067e-02
9.04569209e-01 -1.31604505e+00 3.00870597e-01 5.57424426e-01
-1.29242400e-02 2.91290522e-01 8.48780394e-01 -7.33824074e-01
-1.17538273e+00 -9.42982554e-01 7.72452056e-01 -6.96795464e-01
2.25131661e-01 -4.02223885e-01 -6.75312817e-01 7.10717499e-01
-1.84767753e-01 1.31849483e-01 4.14599329e-01 -1.02250673e-01
1.58817917e-02 1.64065272e-01 -1.31114721e+00 3.44214976e-01
1.07826650e+00 -2.48570517e-01 -4.78222936e-01 3.14796537e-01
4.41948593e-01 -8.72333586e-01 -7.87746310e-01 5.06727695e-01
5.56316376e-01 -1.04664302e+00 1.13014925e+00 7.78976977e-02
5.83503544e-01 -5.13763070e-01 -2.32777670e-01 -1.38427567e+00
-1.76488727e-01 -2.68229961e-01 -1.55098453e-01 1.04126608e+00
2.89910555e-01 -3.87656897e-01 1.16856015e+00 3.39032054e-01
5.51177636e-02 -8.33674431e-01 -1.00549221e+00 -7.29564607e-01
-5.64599819e-02 -3.00112098e-01 3.22726995e-01 5.95011413e-01
-5.58123350e-01 4.38532442e-01 -7.04504311e-01 4.85911489e-01
1.35905027e+00 3.75708714e-02 1.06625772e+00 -1.22145963e+00
-6.75114393e-01 7.30820224e-02 -3.52223635e-01 -1.58694148e+00
3.15114766e-01 -6.71643853e-01 5.83619952e-01 -1.69300091e+00
1.40293717e-01 -2.84802526e-01 3.44687611e-01 2.84722269e-01
-2.03807075e-02 4.06424791e-01 -4.11511064e-02 1.49569064e-01
-4.38061208e-01 8.84746492e-01 1.41620755e+00 2.08803434e-02
-1.71510667e-01 7.40049705e-02 -4.06871110e-01 1.02913213e+00
5.86314738e-01 -5.53595245e-01 -4.34389740e-01 -4.78974372e-01
1.09343193e-01 7.35062733e-02 7.72566438e-01 -1.01853776e+00
8.76828730e-02 -1.45357639e-01 6.01830900e-01 -1.03503251e+00
7.86767066e-01 -7.52038300e-01 2.41391405e-01 3.25252116e-01
-6.14819638e-02 -3.10684562e-01 -6.98524937e-02 6.79980218e-01
-1.24133348e-01 -2.26153314e-01 9.25123572e-01 -2.60703295e-01
-3.92743558e-01 3.68544012e-01 -9.14493203e-02 1.85303912e-01
8.70463610e-01 -5.75184897e-02 2.43980885e-01 -6.58428013e-01
-6.95350587e-01 2.55414248e-01 6.23555005e-01 3.72774184e-01
9.10371006e-01 -1.61102498e+00 -8.31604183e-01 1.56673238e-01
2.17074994e-02 6.08880460e-01 3.99699621e-02 7.24743783e-01
-5.50316036e-01 3.83348875e-02 -7.08122402e-02 -1.06410778e+00
-1.00977671e+00 2.10151955e-01 5.85101843e-01 -1.22584924e-01
-5.35331249e-01 7.30405152e-01 2.98715234e-01 -9.46409702e-01
2.95056254e-01 1.03028780e-02 2.62664735e-01 -2.82608151e-01
1.59082279e-01 3.60633284e-01 -4.11800206e-01 -8.76178920e-01
-3.75588477e-01 9.37045693e-01 2.97045738e-01 -4.86227512e-01
1.37782073e+00 -2.10705683e-01 -9.40682888e-02 3.88476282e-01
1.24310839e+00 -2.56725550e-01 -1.58717728e+00 -3.41054410e-01
-3.88579100e-01 -5.49652696e-01 -1.27495646e-01 -5.69960892e-01
-9.81248617e-01 7.86018312e-01 5.48857272e-01 -4.06853914e-01
7.64034152e-01 2.99478173e-01 7.52057672e-01 2.53886044e-01
4.86797780e-01 -1.00036049e+00 4.71028328e-01 4.92619216e-01
9.85396564e-01 -1.38977098e+00 1.04737721e-01 -5.22902489e-01
-5.25452375e-01 9.07199562e-01 9.74604785e-01 -8.84266868e-02
6.35117412e-01 2.53847271e-01 1.19737573e-01 -2.10498482e-01
-1.87318951e-01 -3.23790722e-02 5.63475192e-01 6.39473438e-01
3.70115399e-01 -7.19761550e-02 -1.31735131e-01 3.77890348e-01
-3.30263138e-01 -1.87477432e-04 3.40915859e-01 8.82926822e-01
-2.81873882e-01 -1.18054450e+00 -7.85354674e-01 4.72253226e-02
-1.09539762e-01 3.97146136e-01 -3.00792277e-01 6.79231107e-01
1.85210019e-01 6.57393754e-01 -5.50412666e-03 -2.95138806e-01
4.28647935e-01 -1.07738651e-01 8.68510664e-01 -5.71101546e-01
3.45306873e-01 4.41273272e-01 -1.38209552e-01 -4.75503892e-01
-5.85175514e-01 -6.56335175e-01 -1.34288836e+00 -2.98046544e-02
-2.44973823e-01 -3.63704115e-02 5.99325299e-01 9.35963631e-01
3.01867396e-01 2.21667945e-01 2.82354772e-01 -1.33458936e+00
-6.39353573e-01 -8.99901867e-01 -4.73807931e-01 4.95914966e-01
2.97445267e-01 -9.77726042e-01 -4.55139518e-01 4.94587608e-02] | [7.101036071777344, -1.11672842502594] |
fba84173-7a9e-4afc-a4f3-b2d8151dd9a8 | improving-cross-lingual-transfer-learning-for | 2006.05474 | null | https://arxiv.org/abs/2006.05474v2 | https://arxiv.org/pdf/2006.05474v2.pdf | Improving Cross-Lingual Transfer Learning for End-to-End Speech Recognition with Speech Translation | Transfer learning from high-resource languages is known to be an efficient way to improve end-to-end automatic speech recognition (ASR) for low-resource languages. Pre-trained or jointly trained encoder-decoder models, however, do not share the language modeling (decoder) for the same language, which is likely to be inefficient for distant target languages. We introduce speech-to-text translation (ST) as an auxiliary task to incorporate additional knowledge of the target language and enable transferring from that target language. Specifically, we first translate high-resource ASR transcripts into a target low-resource language, with which a ST model is trained. Both ST and target ASR share the same attention-based encoder-decoder architecture and vocabulary. The former task then provides a fully pre-trained model for the latter, bringing up to 24.6% word error rate (WER) reduction to the baseline (direct transfer from high-resource ASR). We show that training ST with human translations is not necessary. ST trained with machine translation (MT) pseudo-labels brings consistent gains. It can even outperform those using human labels when transferred to target ASR by leveraging only 500K MT examples. Even with pseudo-labels from low-resource MT (200K examples), ST-enhanced transfer brings up to 8.9% WER reduction to direct transfer. | ['Jiatao Gu', 'Juan Pino', 'Changhan Wang'] | 2020-06-09 | null | null | null | null | ['speech-to-text-translation'] | ['natural-language-processing'] | [ 5.32289803e-01 4.44640279e-01 -2.30361938e-01 -3.23441595e-01
-1.85483396e+00 -5.50455570e-01 5.75419664e-01 -4.93033797e-01
-5.57866037e-01 8.43451798e-01 4.20496076e-01 -8.59987140e-01
8.46148908e-01 -2.38670051e-01 -9.87289131e-01 -3.30911309e-01
4.48622048e-01 7.77778983e-01 -1.65420443e-01 -2.97098517e-01
-4.96261537e-01 1.31027132e-01 -6.18819058e-01 6.37831271e-01
8.23205113e-01 5.43205321e-01 5.79229832e-01 8.26800406e-01
-9.54385027e-02 6.40912831e-01 -7.15774894e-01 -5.29999137e-01
1.57041281e-01 -5.88952243e-01 -1.05629027e+00 8.01494718e-02
2.24227279e-01 -3.17961276e-01 -6.99507296e-01 6.44872844e-01
6.16440654e-01 1.55549515e-02 5.31369328e-01 -5.92009604e-01
-1.04622936e+00 8.37129772e-01 -2.98754603e-01 1.17500469e-01
2.95935631e-01 2.61571556e-01 8.82294595e-01 -1.34766746e+00
4.12278742e-01 1.33123684e+00 2.90460408e-01 1.04209411e+00
-1.18698108e+00 -5.29747486e-01 -7.22550899e-02 -1.79085508e-01
-1.44652045e+00 -1.15559077e+00 2.18732744e-01 -2.20782943e-02
1.67603636e+00 2.51471668e-01 3.04143950e-02 1.41540086e+00
-2.06114128e-01 8.97099555e-01 9.74539280e-01 -6.65356815e-01
-1.83400899e-01 3.09123427e-01 -3.56158823e-01 3.77364337e-01
-2.58435100e-01 1.57394856e-02 -7.31670916e-01 2.87689477e-01
4.04117793e-01 -4.44012702e-01 -3.59704226e-01 5.11425376e-01
-1.52606416e+00 6.10455215e-01 2.00749472e-01 3.74545336e-01
-2.22253844e-01 1.33415803e-01 5.99507689e-01 7.92100489e-01
8.35819066e-01 4.18305933e-01 -8.23935807e-01 -3.43968272e-01
-8.31047773e-01 -6.18209302e-01 6.78258598e-01 1.46388745e+00
7.94979632e-01 5.17207623e-01 -3.23924541e-01 1.22020006e+00
8.55247378e-02 1.17891121e+00 7.35378087e-01 -4.58793551e-01
1.16679966e+00 4.59677167e-02 -1.13628082e-01 8.47421810e-02
1.63722441e-01 -6.25825584e-01 -7.13550448e-01 -4.37494546e-01
1.89749777e-01 -4.64803159e-01 -1.17675173e+00 1.76875496e+00
-1.24540284e-01 -3.09747402e-02 6.20003939e-01 6.69691741e-01
5.68325579e-01 1.22964406e+00 -4.74496419e-03 -2.07584471e-01
1.08491385e+00 -1.45440137e+00 -6.45885170e-01 -7.89687753e-01
1.15072513e+00 -1.00976014e+00 1.40331256e+00 -1.81063667e-01
-1.29062974e+00 -5.12923956e-01 -7.75886536e-01 -2.26286486e-01
-9.65872332e-02 4.89858806e-01 -1.41682044e-01 4.78834957e-01
-1.44440746e+00 1.65724531e-01 -7.29735553e-01 -5.71996391e-01
1.18853986e-01 3.79369140e-01 -5.16118586e-01 -3.70224565e-01
-1.31671250e+00 1.34339571e+00 1.54742748e-01 -8.62257332e-02
-1.38966811e+00 -5.40733814e-01 -9.95245099e-01 -4.43459430e-04
2.45692611e-01 -5.15052438e-01 1.67208087e+00 -1.30891097e+00
-2.06123376e+00 9.92902637e-01 -5.80542386e-01 -4.46122527e-01
2.95409173e-01 -3.09591502e-01 -4.70651299e-01 -1.41506627e-01
-2.75601912e-02 6.04341149e-01 7.83793747e-01 -8.27985764e-01
-3.87804389e-01 4.35334854e-02 -2.83459544e-01 4.89492744e-01
-4.04490829e-01 5.80654263e-01 -5.58352411e-01 -6.07819140e-01
-2.78600037e-01 -1.03735244e+00 8.14182162e-02 -5.45719385e-01
-3.24783444e-01 -1.38220310e-01 4.75079417e-01 -1.26158845e+00
1.05958629e+00 -2.07436824e+00 2.96260685e-01 -2.43963748e-01
-3.54758352e-01 8.43908370e-01 -6.10188961e-01 6.54563189e-01
-2.45995633e-02 2.01428249e-01 -3.35286766e-01 -6.84466183e-01
-2.05587745e-01 3.03278446e-01 -4.99552935e-01 1.85201958e-01
6.14343107e-01 1.31965923e+00 -7.72325873e-01 -3.64591740e-02
1.32660881e-01 5.81160486e-01 -2.62270123e-01 5.14872789e-01
-7.52492398e-02 7.57452011e-01 -1.46669865e-01 3.31944764e-01
1.03086695e-01 -1.90866649e-01 9.34067294e-02 2.45618030e-01
1.96061924e-01 1.17978263e+00 -1.67416736e-01 1.80707002e+00
-1.21702218e+00 8.20944548e-01 -8.75551347e-03 -8.42212915e-01
1.06203687e+00 8.48313510e-01 -1.40043989e-01 -8.58935535e-01
-1.09129138e-01 7.12922513e-01 1.74818337e-01 -1.70212433e-01
1.77577034e-01 -3.81024837e-01 -1.07781485e-01 5.22846639e-01
3.43485773e-01 -1.59860268e-01 -4.27019954e-01 5.78382537e-02
1.35177827e+00 2.73650065e-02 1.44803196e-01 6.32562712e-02
5.90376318e-01 -8.02372992e-02 3.03402543e-01 3.21137398e-01
1.37660220e-01 5.92199445e-01 -1.19952671e-01 -2.66191233e-02
-1.41425157e+00 -1.04548156e+00 2.49161258e-01 1.31063473e+00
-5.61933160e-01 -2.77710199e-01 -7.95288205e-01 -8.14435244e-01
-3.90372574e-01 1.06591868e+00 -7.44380057e-02 -3.86008471e-01
-9.74120855e-01 -2.59578645e-01 1.01848435e+00 4.50907886e-01
2.49929547e-01 -1.08951259e+00 3.27710956e-01 3.26988071e-01
-5.27094007e-01 -1.52522612e+00 -1.02885246e+00 3.00655365e-01
-6.90406859e-01 -2.15744346e-01 -1.23811197e+00 -9.14882302e-01
7.27893949e-01 2.98999101e-01 1.09526110e+00 -5.24381511e-02
3.78153205e-01 1.52559280e-01 -4.33675766e-01 -8.96302611e-02
-1.24939144e+00 5.02164423e-01 1.25677973e-01 -1.79253763e-03
3.58614594e-01 -2.13568568e-01 -2.88578533e-02 4.30796355e-01
-4.10695523e-01 2.67132819e-01 8.78652215e-01 9.63756144e-01
3.38194758e-01 -8.28492761e-01 8.67507577e-01 -5.36890686e-01
2.49274194e-01 -3.28134924e-01 -3.19589704e-01 4.29912418e-01
-2.46952325e-01 1.72111660e-01 9.50198948e-01 -6.07154250e-01
-1.00410557e+00 5.60194300e-03 -4.64328468e-01 -4.59190845e-01
-2.62333211e-02 3.70382428e-01 -4.92229909e-01 2.88102686e-01
5.76297700e-01 5.26549757e-01 -1.55668125e-01 -6.27633452e-01
5.28635621e-01 1.40991497e+00 3.14131945e-01 -3.87280852e-01
8.29041362e-01 -8.84133354e-02 -6.00134432e-01 -7.87399411e-01
-9.05969262e-01 -4.50642437e-01 -6.27455354e-01 1.93405688e-01
8.72948766e-01 -1.32259429e+00 -1.83438770e-02 1.98881388e-01
-1.54894495e+00 -9.38303173e-01 -3.31537783e-01 7.41945565e-01
-6.68334544e-01 -7.67497811e-03 -8.73754919e-01 -6.71046495e-01
-6.38416469e-01 -1.17434013e+00 1.35309923e+00 -5.55002511e-01
-1.26549214e-01 -9.87962842e-01 -5.91042377e-02 5.21353483e-01
6.18864179e-01 -7.42081344e-01 7.09005535e-01 -9.59030092e-01
-4.05277133e-01 -7.90945590e-02 -1.81517556e-01 8.93085659e-01
3.06281686e-01 -5.94687104e-01 -1.13230765e+00 -5.57132006e-01
-1.32708699e-01 -5.03157616e-01 7.05223083e-01 -1.61045700e-01
5.73378861e-01 -5.26874185e-01 -2.00618073e-01 4.57884610e-01
9.63028312e-01 1.45308793e-01 6.23978913e-01 -1.49344623e-01
9.24687684e-01 6.11501873e-01 4.29386139e-01 -3.46663207e-01
2.76190102e-01 1.07231665e+00 -2.06468046e-01 -3.98679465e-01
-7.35860109e-01 -6.11030936e-01 1.30338418e+00 1.86761987e+00
3.64069164e-01 -4.78354216e-01 -1.17371833e+00 6.71049058e-01
-1.43930256e+00 -4.52903003e-01 6.36933520e-02 2.42104936e+00
1.27676439e+00 -4.12002169e-02 -1.08512454e-01 -3.47103298e-01
8.86715651e-01 -1.30477548e-01 -3.95027369e-01 -6.70567811e-01
-1.38281614e-01 3.09959412e-01 6.47591352e-01 1.00190115e+00
-5.88813484e-01 1.61420035e+00 6.19153547e+00 1.09433043e+00
-1.34625328e+00 8.29708636e-01 6.29682302e-01 -1.18223010e-02
-4.58564460e-01 5.86029142e-02 -9.71238792e-01 3.53926063e-01
1.92561293e+00 -3.35653275e-01 7.29633510e-01 6.49007499e-01
2.83461720e-01 5.70332289e-01 -1.34949577e+00 9.52286124e-01
2.27268398e-01 -8.93561184e-01 2.77470440e-01 1.02792598e-01
7.09534705e-01 6.48070931e-01 -3.32543887e-02 7.04985499e-01
4.96884942e-01 -1.25471973e+00 5.88141561e-01 -6.52756840e-02
1.72271132e+00 -6.53722525e-01 6.89154327e-01 5.55701911e-01
-9.76414084e-01 3.60869914e-01 -6.03023350e-01 6.49807379e-02
4.75701839e-01 3.06681544e-01 -1.37603784e+00 4.69038606e-01
9.55484957e-02 5.19051552e-01 -2.09278852e-01 2.96135992e-01
-4.68494177e-01 1.16746223e+00 -2.48516589e-01 1.75185204e-02
3.87783796e-01 1.24989592e-01 4.99603003e-01 1.56953335e+00
5.57595611e-01 -3.23009789e-01 4.67859693e-02 4.84767973e-01
-5.65178156e-01 4.44043756e-01 -8.70745480e-01 -2.02990919e-01
4.81776714e-01 8.64581645e-01 -5.08513525e-02 -6.96965218e-01
-7.06065297e-01 1.60230529e+00 6.20880544e-01 4.79655564e-01
-5.74265897e-01 -1.69618413e-01 6.64392769e-01 1.01797849e-01
1.59985229e-01 -4.42759424e-01 -5.49477749e-02 -1.27335835e+00
5.29079214e-02 -1.21397233e+00 -1.89805344e-01 -8.20861161e-01
-1.14361238e+00 1.05936074e+00 -4.88866150e-01 -1.17030466e+00
-4.96593386e-01 -4.54634845e-01 -2.25813642e-01 1.56454265e+00
-1.66393793e+00 -1.47624230e+00 3.38667184e-01 5.14091372e-01
1.23643231e+00 -5.21727800e-01 1.07104146e+00 6.44635797e-01
-4.02430713e-01 1.01407111e+00 2.16068998e-01 3.19323987e-01
9.63884592e-01 -1.07616961e+00 1.18330824e+00 8.68707538e-01
4.23592478e-01 4.20334905e-01 2.69617081e-01 -5.73336601e-01
-1.40777075e+00 -1.53629434e+00 1.50591552e+00 -7.52000690e-01
6.28978431e-01 -7.86637425e-01 -9.55907285e-01 1.12150681e+00
4.96896803e-01 -5.48169911e-02 5.91826320e-01 -2.45627165e-02
-5.04299819e-01 -1.78029761e-02 -6.30645871e-01 8.00828159e-01
1.17507744e+00 -1.13384211e+00 -5.91609478e-01 6.05068266e-01
1.38112664e+00 -3.07258099e-01 -6.74611747e-01 1.64417177e-01
1.12038732e-01 1.93460677e-02 6.42943561e-01 -7.47590661e-01
2.58007318e-01 -1.49046034e-01 -4.03705925e-01 -1.71336901e+00
-3.04775164e-02 -1.07104933e+00 1.11871645e-01 1.27107656e+00
1.09230304e+00 -6.64984524e-01 2.72689581e-01 2.83862382e-01
-7.04452932e-01 -5.13198256e-01 -1.17061782e+00 -1.28160477e+00
4.97775972e-01 -5.54442227e-01 4.57159489e-01 7.47891128e-01
-3.99506837e-03 9.73590076e-01 -5.72668850e-01 1.04347967e-01
5.78576103e-02 -5.56925774e-01 7.92538404e-01 -4.55259681e-01
-4.40220982e-01 -5.04102074e-02 1.97696555e-02 -1.58664787e+00
5.88550568e-01 -1.52475846e+00 3.19807976e-01 -1.49900067e+00
5.26119769e-02 -4.98881549e-01 -1.75873443e-01 6.97037339e-01
-1.69122338e-01 3.36364359e-01 2.27689266e-01 3.14674288e-01
-3.44543636e-01 8.04396808e-01 1.19045126e+00 -8.01352337e-02
-5.47183566e-02 -5.63407466e-02 -4.44059759e-01 3.43641967e-01
7.50713706e-01 -7.12328732e-01 -2.01584458e-01 -1.00983036e+00
-1.69054106e-01 3.12260419e-01 -1.78071871e-01 -6.66725934e-01
7.25465044e-02 7.31715560e-02 -9.21924338e-02 -1.32701948e-01
4.61942226e-01 -5.44899523e-01 -1.97219193e-01 2.74948239e-01
-6.21607721e-01 3.92705165e-02 2.51130044e-01 1.46327540e-01
-3.03490698e-01 -1.76684499e-01 7.61780441e-01 -2.28402484e-02
-1.49697721e-01 1.92652479e-01 -5.80648363e-01 2.87929773e-01
5.37820816e-01 1.64664671e-01 -1.78550035e-01 -8.48064661e-01
-6.81105614e-01 -1.23950668e-01 2.86388934e-01 6.25949860e-01
5.11088431e-01 -1.28025222e+00 -1.28184032e+00 2.19581872e-01
2.04143539e-01 -2.59314388e-01 -1.40838370e-01 7.19548643e-01
-8.92784148e-02 7.26273596e-01 2.94245720e-01 -5.20141721e-01
-1.13679636e+00 5.09371459e-01 3.71494919e-01 -3.10171396e-01
-5.05212903e-01 8.89917493e-01 3.64664793e-01 -7.54885137e-01
5.79142850e-03 -7.38102943e-02 5.10547400e-01 -5.88337660e-01
5.11675239e-01 1.49697810e-01 3.79515439e-01 -9.87132251e-01
-3.66432846e-01 4.10683483e-01 -1.57587960e-01 -7.23953307e-01
1.02385926e+00 -4.75557953e-01 2.26494610e-01 5.47585130e-01
1.42705858e+00 2.73061663e-01 -1.04332566e+00 -6.53903127e-01
1.07732274e-01 -7.64783546e-02 -5.89085519e-02 -9.75958109e-01
-7.51333177e-01 1.40989423e+00 3.30183804e-02 -4.02419984e-01
8.64026904e-01 2.62677372e-01 1.22994876e+00 6.02081716e-01
5.37466347e-01 -9.96650577e-01 9.89794582e-02 1.03446615e+00
1.18816459e+00 -1.20126772e+00 -7.72367060e-01 -3.60278964e-01
-9.08163667e-01 7.64228940e-01 3.82881522e-01 1.94942683e-01
2.46857986e-01 3.67343187e-01 3.93911988e-01 4.66793299e-01
-1.09069180e+00 -3.79694581e-01 4.48983818e-01 6.40428007e-01
6.70290709e-01 2.63644278e-01 1.19574934e-01 2.61291593e-01
-2.87002057e-01 -3.02838087e-01 2.62658179e-01 4.23079908e-01
-2.82034606e-01 -1.36091828e+00 -1.50370955e-01 1.63199931e-01
-5.23867905e-01 -7.13057220e-01 -4.05409217e-01 3.78109097e-01
-4.75356877e-01 1.27993596e+00 -7.42597785e-03 -4.59391534e-01
3.00066710e-01 4.31829691e-01 1.97905973e-01 -1.32755208e+00
-6.85010135e-01 2.12841645e-01 5.11587322e-01 -3.99364859e-01
2.12893441e-01 -3.20609808e-01 -1.05980277e+00 -1.21725224e-01
-3.84176791e-01 1.61163554e-01 8.86098623e-01 1.00663662e+00
4.29510504e-01 4.54664946e-01 8.15310001e-01 -4.67507094e-01
-7.53126860e-01 -1.36301506e+00 -5.13027683e-02 6.17310554e-02
4.69008088e-01 9.45877109e-04 -4.30112422e-01 1.52330175e-01] | [14.52535629272461, 7.185739040374756] |
6b2104ee-9105-465c-b504-4297a696316c | physics-constrained-unsupervised-deep | 2306.11014 | null | https://arxiv.org/abs/2306.11014v1 | https://arxiv.org/pdf/2306.11014v1.pdf | Physics Constrained Unsupervised Deep Learning for Rapid, High Resolution Scanning Coherent Diffraction Reconstruction | By circumventing the resolution limitations of optics, coherent diffractive imaging (CDI) and ptychography are making their way into scientific fields ranging from X-ray imaging to astronomy. Yet, the need for time consuming iterative phase recovery hampers real-time imaging. While supervised deep learning strategies have increased reconstruction speed, they sacrifice image quality. Furthermore, these methods' demand for extensive labeled training data is experimentally burdensome. Here, we propose an unsupervised physics-informed neural network reconstruction method, PtychoPINN, that retains the intrinsic speed of deep learning-based reconstruction while improving reconstruction quality by combining the diffraction forward map with real-space constraints from overlapping measurements. In particular, PtychoPINN significantly advances generalizability, accuracy (evidenced by a 10 dB PSNR increase), and linear resolution (with a 3- to 6-fold gain). This blend of performance and speed offers exciting prospects for high-resolution real-time imaging in high-throughput environments such as X-ray free electron lasers (XFELs) and diffraction-limited light sources. | ['Apurva Mehta', 'Aashwin Ananda Mishra', 'Oliver Hoidn'] | 2023-06-19 | null | null | null | null | ['astronomy'] | ['miscellaneous'] | [ 5.92276692e-01 -3.42174143e-01 1.60517752e-01 -4.42195147e-01
-9.92801726e-01 -2.22473606e-01 4.40788001e-01 -3.59402210e-01
-9.34266627e-01 8.68334889e-01 -1.27084740e-02 -3.00008237e-01
-4.35926169e-01 -5.12941599e-01 -5.12026906e-01 -1.07640648e+00
1.00660674e-01 6.64514840e-01 -1.90970138e-01 4.81875390e-01
4.61375475e-01 8.26795876e-01 -1.13153481e+00 2.54698265e-02
6.28978014e-01 1.03741837e+00 4.46428657e-01 3.00410032e-01
8.74484479e-02 6.86035633e-01 -2.33996496e-01 1.51832104e-01
8.61504972e-02 -5.26060820e-01 -4.84235913e-01 -1.50735155e-01
5.43058813e-01 -8.75790536e-01 -6.86122239e-01 8.77840161e-01
7.09124506e-01 1.61315426e-02 3.54917377e-01 -3.20067167e-01
-6.02503836e-01 1.40155882e-01 -6.81478024e-01 2.54221529e-01
-8.52290262e-03 7.54560471e-01 6.87689066e-01 -7.70959795e-01
8.91956627e-01 6.36414826e-01 5.25156736e-01 5.83345592e-01
-1.65196228e+00 -7.41355419e-01 -6.42275333e-01 1.96196094e-01
-8.86683941e-01 -8.18390369e-01 6.29766762e-01 -3.72873157e-01
9.60727274e-01 9.65970196e-03 6.12944126e-01 9.47501898e-01
4.64171290e-01 3.45522553e-01 1.63908219e+00 -3.76941681e-01
2.82220513e-01 -4.47734654e-01 -1.25559986e-01 6.28309608e-01
2.64418006e-01 4.47227687e-01 -7.36068010e-01 1.06894542e-02
1.25490880e+00 1.56522855e-01 -8.64840209e-01 -1.97854549e-01
-1.33049107e+00 3.73943001e-01 4.63265657e-01 2.98741728e-01
-4.37744826e-01 2.15269223e-01 2.91881226e-02 1.60612136e-01
2.73959428e-01 1.37337923e+00 2.36776341e-02 -7.59164914e-02
-1.18375564e+00 1.60288155e-01 3.59637648e-01 4.86740381e-01
7.43925154e-01 2.47016594e-01 2.16564640e-01 3.86105984e-01
1.55314893e-01 6.16879940e-01 2.24761188e-01 -1.37114203e+00
-9.15352181e-02 1.33530498e-01 2.41624549e-01 -4.63658512e-01
-5.30500412e-01 -6.32123530e-01 -9.37948465e-01 7.37316847e-01
6.26125276e-01 1.81686282e-02 -1.02615559e+00 1.53831589e+00
1.60038814e-01 -1.25528835e-02 -1.11100003e-01 1.26072693e+00
6.89189374e-01 6.35214925e-01 -3.29239279e-01 -6.42011523e-01
1.06690097e+00 -4.53574628e-01 -7.53310680e-01 -1.62255481e-01
1.98718458e-01 -7.57766724e-01 8.96767080e-01 6.52682364e-01
-1.29334307e+00 2.66396068e-02 -1.11588240e+00 -2.95502603e-01
5.77812433e-01 -6.23943135e-02 5.78959227e-01 -1.65530052e-02
-6.84314311e-01 1.09734821e+00 -1.16273129e+00 1.13414131e-01
7.32561648e-01 4.87542123e-01 -4.30673361e-01 -3.83347154e-01
-2.73399800e-01 6.53349161e-01 -9.36631579e-03 -2.51023442e-01
-5.33475459e-01 -1.26480186e+00 -1.92815706e-01 3.59899215e-02
1.87584281e-01 -6.99278951e-01 1.09944677e+00 -1.11549139e-01
-1.85995245e+00 9.93463993e-01 -1.42890528e-01 -4.52731818e-01
3.09623510e-01 -2.91176975e-01 -1.10538803e-01 6.85914338e-01
4.15320694e-02 2.92335391e-01 6.45052016e-01 -1.00441718e+00
-2.40422636e-01 -5.87680399e-01 -5.56211770e-01 -8.84706154e-02
1.10306203e-01 6.54666778e-03 -6.41910657e-02 4.73029912e-02
6.64527178e-01 -6.15922272e-01 -2.59324044e-01 5.26173353e-01
-2.76705146e-01 3.98159295e-01 1.03333831e+00 -3.91963333e-01
4.57389623e-01 -2.05563474e+00 -5.39973825e-02 -2.15239435e-01
7.87956536e-01 4.73486304e-01 8.24556947e-02 5.54258823e-01
1.57781914e-01 -4.41667825e-01 -2.69056499e-01 -2.50006914e-01
-5.13323069e-01 6.56427667e-02 -2.21140504e-01 8.92788053e-01
-2.82699638e-03 1.07745576e+00 -8.04770410e-01 -9.74835455e-02
3.29318315e-01 5.62804520e-01 -5.93090951e-01 2.82645404e-01
-2.73959368e-01 1.37519300e+00 -1.56538859e-01 6.51077926e-01
6.78817809e-01 -9.57360864e-01 3.75936747e-01 -3.90901923e-01
-7.01578975e-01 4.46729034e-01 -6.15166008e-01 1.72894001e+00
-2.04222426e-01 9.42096949e-01 4.47423667e-01 -8.12205195e-01
6.26777232e-01 1.45370737e-01 7.25194156e-01 -1.26671505e+00
3.87844406e-02 4.29624438e-01 9.21078697e-02 -6.56828701e-01
1.92729980e-01 -8.19334269e-01 6.22517884e-01 7.50826895e-01
9.12517831e-02 -2.57829338e-01 -2.36775190e-01 -1.92138553e-01
1.18333054e+00 2.96304375e-01 1.23603821e-01 -3.65900636e-01
-2.55666167e-01 4.43812422e-02 4.79643255e-01 7.31056929e-01
1.01419941e-01 7.09313631e-01 -7.43238721e-03 -6.48547590e-01
-1.74691093e+00 -1.01079190e+00 -5.43973804e-01 3.23075056e-01
-4.20452543e-02 9.43489466e-03 -1.88213944e-01 3.15015256e-01
-1.78797975e-01 2.27560803e-01 -8.18215981e-02 1.78544223e-01
-8.65969181e-01 -9.45484459e-01 2.70225912e-01 5.88567480e-02
6.58479393e-01 -7.64881015e-01 -9.25626218e-01 4.59938437e-01
5.24439430e-03 -1.45448530e+00 2.18083069e-01 2.63090342e-01
-8.06611598e-01 -9.10549402e-01 -6.72025740e-01 -3.11416894e-01
5.41581333e-01 3.42904270e-01 9.03596520e-01 -2.83857405e-01
-7.37924218e-01 1.10969335e-01 5.83240837e-02 -1.50625445e-02
-5.11623584e-02 -4.30791318e-01 3.59611869e-01 -2.93569326e-01
1.70290023e-01 -1.25161409e+00 -1.06623209e+00 6.16452768e-02
-5.75767994e-01 2.75965124e-01 9.01639104e-01 1.04901111e+00
1.03461385e+00 -2.22749814e-01 3.87696505e-01 -8.11752856e-01
1.64144322e-01 1.01182804e-01 -1.03589356e+00 -4.61368710e-01
-7.94022262e-01 2.33847544e-01 8.04270327e-01 -4.57207114e-01
-1.07910669e+00 -2.73348600e-01 -3.25720578e-01 -3.59310478e-01
-1.52767450e-01 4.47798342e-01 1.76107466e-01 -6.69194877e-01
9.20132399e-01 2.98231900e-01 3.59813482e-01 -4.92629170e-01
-1.48686901e-01 2.23317072e-01 9.57199812e-01 -4.21426564e-01
6.72849178e-01 9.46110427e-01 4.20961529e-01 -1.18399429e+00
-1.01507354e+00 -3.03435773e-01 -2.55732507e-01 -7.34929144e-02
9.25711453e-01 -8.25194180e-01 -1.10408568e+00 4.30471390e-01
-8.87647927e-01 -3.12742382e-01 -2.38201454e-01 1.07071447e+00
-5.67733765e-01 4.44184273e-01 -8.63853812e-01 -5.71807325e-01
-4.42405015e-01 -1.09756672e+00 9.10312891e-01 2.16771632e-01
-2.76956577e-02 -4.23571795e-01 -9.56267491e-02 4.63558137e-01
6.31449223e-01 3.20246518e-01 1.04607880e+00 1.96415618e-01
-1.35855186e+00 -2.15528533e-02 -4.99879807e-01 1.10938206e-01
5.97917773e-02 -3.46526474e-01 -1.05796838e+00 -4.88770336e-01
5.94248176e-01 -6.33248746e-01 8.67024720e-01 7.24128366e-01
1.13361478e+00 -2.30142593e-01 -3.19982290e-01 1.24478519e+00
1.78895998e+00 3.23259175e-01 6.56049967e-01 1.95772648e-01
6.94519460e-01 2.78902173e-01 9.38149169e-02 4.17479903e-01
-3.27000290e-01 6.27689600e-01 3.76395196e-01 -9.02382284e-02
-1.94717288e-01 8.01812783e-02 -2.47380957e-01 7.75068343e-01
-3.23626161e-01 -1.80931777e-01 -8.05052102e-01 2.39130020e-01
-1.26386392e+00 -1.16301298e+00 -2.16555879e-01 2.37864852e+00
9.46140707e-01 -6.39303625e-02 -5.47972739e-01 -1.06659159e-01
1.10579394e-01 5.54446280e-02 -9.81176317e-01 3.49917680e-01
-1.31544843e-01 5.93849659e-01 3.93688053e-01 4.67776686e-01
-5.41584551e-01 6.49865568e-01 6.56990910e+00 5.00569999e-01
-1.72729731e+00 5.89579307e-02 3.87657017e-01 -5.98839939e-01
-3.82097095e-01 -1.26364576e-02 -6.43763661e-01 4.21438932e-01
7.10416257e-01 1.99280947e-01 6.25967443e-01 3.72562677e-01
2.96187907e-01 -3.38662595e-01 -1.14066374e+00 1.35558641e+00
-4.66833383e-01 -1.97244012e+00 -1.92565531e-01 4.89250243e-01
5.51198900e-01 5.51813602e-01 2.72003293e-01 -5.31382859e-01
-2.69743931e-02 -1.13204134e+00 -1.54610323e-02 6.43166721e-01
1.41276038e+00 -6.67094171e-01 2.44523048e-01 4.45528537e-01
-2.59534031e-01 1.71744034e-01 -5.25417089e-01 -2.54970849e-01
3.72224808e-01 1.18650985e+00 -5.94866931e-01 2.40496770e-01
6.21436238e-01 4.78036612e-01 4.47478265e-01 9.43010092e-01
-1.11200372e-02 6.69258475e-01 -4.34620827e-01 1.75641090e-01
2.68293321e-01 -6.45288944e-01 7.20692217e-01 6.91301286e-01
3.93050939e-01 6.71988249e-01 -1.93165809e-01 1.24528646e+00
-2.56070554e-01 -5.78719437e-01 -5.24605632e-01 -3.09111536e-01
4.92193222e-01 1.28175747e+00 -4.47735518e-01 -6.84376135e-02
-2.30224699e-01 7.20949471e-01 3.16797584e-01 4.05524224e-01
-2.65765011e-01 -2.01109245e-01 7.26553202e-01 5.92412531e-01
1.48296282e-01 -6.54698670e-01 -4.84619409e-01 -1.25376153e+00
1.56444870e-02 -5.60078800e-01 -5.29314816e-01 -8.24772298e-01
-1.10709655e+00 2.82474697e-01 -4.11077917e-01 -9.58408475e-01
-1.20893337e-01 -7.39663720e-01 -3.21451187e-01 9.58302975e-01
-1.65141153e+00 -8.08144331e-01 -2.27079600e-01 2.64114048e-02
1.13197789e-02 1.22114718e-02 7.88048804e-01 4.22620922e-01
-2.75433868e-01 4.97475713e-02 6.51939094e-01 -1.41541928e-01
7.00556815e-01 -8.86774719e-01 2.85748631e-01 7.18767583e-01
-7.10303895e-03 6.57272756e-01 7.85803974e-01 -3.61635655e-01
-1.89442432e+00 -6.60639703e-01 4.42653894e-01 -1.57124579e-01
5.68236589e-01 -7.96229541e-02 -1.09246159e+00 4.95537788e-01
1.49846658e-01 1.45073310e-01 7.46053457e-01 -1.18017018e-01
-3.44348729e-01 8.02622363e-02 -1.34479010e+00 3.98843974e-01
9.61823165e-01 -6.68668091e-01 -1.66699693e-01 4.86480653e-01
3.76683056e-01 -5.41269958e-01 -9.13726509e-01 4.27034110e-01
6.37124956e-01 -1.26348984e+00 8.66907597e-01 -2.10052803e-01
9.47436273e-01 -9.80884656e-02 -5.51819801e-03 -9.72068429e-01
-6.74127698e-01 -9.66443956e-01 -3.18693444e-02 5.01050115e-01
1.81202386e-02 -9.21221673e-01 9.94135559e-01 4.84586030e-01
-4.65149462e-01 -9.67195153e-01 -1.06809127e+00 -7.32009947e-01
-7.58042261e-02 -1.03200629e-01 2.34973773e-01 9.69698489e-01
-1.84789658e-01 3.88873488e-01 -4.83952641e-01 2.10944876e-01
1.26584756e+00 4.10633028e-01 4.95549649e-01 -1.18649518e+00
-6.52567446e-01 -2.87290722e-01 -3.68840426e-01 -1.39437580e+00
-2.33167827e-01 -5.48536003e-01 -4.66681272e-02 -1.26305568e+00
2.72027254e-01 -5.94415307e-01 1.99012339e-01 1.95617810e-01
2.65939981e-01 5.05011797e-01 2.53006443e-02 7.72743642e-01
-2.67997503e-01 6.02974951e-01 1.58795154e+00 2.00493708e-01
-2.52485718e-03 -3.35526854e-01 -4.09468800e-01 5.43533325e-01
5.40365636e-01 -4.49198991e-01 -2.31538668e-01 -8.62486720e-01
3.10258240e-01 4.10939664e-01 4.55176115e-01 -1.39508307e+00
5.32924116e-01 -1.35093823e-01 6.30425334e-01 -3.79427671e-01
6.70292795e-01 -3.58400762e-01 3.58798832e-01 4.36369359e-01
-1.07831292e-01 -5.55342734e-01 -4.84774671e-02 3.85029763e-01
-6.04011230e-02 -7.82547891e-02 1.36752319e+00 -4.35472161e-01
-2.83550262e-01 4.57574129e-01 -3.73204052e-01 2.01984146e-03
4.87135768e-01 -2.31234297e-01 -5.96335530e-01 -1.89626023e-01
-4.20990348e-01 -2.20853895e-01 8.22179854e-01 -6.99253738e-01
8.97763848e-01 -9.22960699e-01 -4.63123173e-01 4.07328367e-01
-1.83238417e-01 2.95609117e-01 3.95470023e-01 1.02484643e+00
-8.64024282e-01 4.04053509e-01 -4.67990488e-01 -7.45933056e-01
-9.26422119e-01 1.52993590e-01 3.75505596e-01 -1.48873433e-01
-1.38890147e+00 9.46978390e-01 1.83118358e-01 -2.46475488e-01
-2.92814821e-01 9.08866450e-02 3.59564662e-01 -5.96911967e-01
8.79276156e-01 2.13721305e-01 -4.80190553e-02 -1.28575182e-02
8.06036666e-02 5.48647463e-01 -3.38813096e-01 -1.35843679e-01
1.84121764e+00 9.38587338e-02 -2.64308423e-01 2.21833333e-01
1.15950727e+00 -4.89430837e-02 -1.71818113e+00 -4.81099516e-01
-2.51007408e-01 -5.76064050e-01 5.31310916e-01 -7.69628167e-01
-9.03474748e-01 9.85690355e-01 3.75998735e-01 -2.99005389e-01
1.02688122e+00 1.53993994e-01 1.04015493e+00 7.68071473e-01
6.06427789e-01 -7.30529368e-01 9.98764038e-02 3.45990628e-01
3.97947311e-01 -1.18630421e+00 4.44924235e-01 -3.31888683e-02
1.07977249e-01 1.01179075e+00 1.73983321e-01 -1.77532762e-01
2.19388545e-01 5.51938713e-01 -3.93963344e-02 -6.28203511e-01
-6.28341794e-01 1.96626395e-01 -1.07495770e-01 6.49958789e-01
2.99630582e-01 -1.92363381e-01 -7.30251819e-02 -2.14890689e-01
-4.78034019e-02 1.67008683e-01 6.29396141e-01 8.27785373e-01
-7.29557157e-01 -8.92600298e-01 -1.13693066e-01 7.36823916e-01
-6.38144195e-01 -1.14474095e-01 1.04848176e-01 5.52463174e-01
-3.76723826e-01 3.49772632e-01 2.11384878e-01 9.06802062e-03
-1.44345865e-01 -2.47244298e-01 9.07616913e-01 -5.03942966e-01
1.41058192e-01 3.90930533e-01 -8.48234743e-02 -7.54418850e-01
-5.29055893e-01 -4.12344754e-01 -1.46410239e+00 -5.71800292e-01
-1.79891795e-01 -9.48218256e-02 7.93258309e-01 1.19445503e+00
6.12352431e-01 2.79876351e-01 3.41070324e-01 -1.01329350e+00
-4.57170755e-01 -6.78828359e-01 -6.87026083e-01 -7.90648349e-03
6.57811224e-01 -3.92426103e-01 -3.79332095e-01 -2.50615448e-01] | [12.832693099975586, -2.7858383655548096] |
b24848e7-a13b-456a-9376-39efa0138236 | toward-fast-and-accurate-neural-chinese-word | 1903.04190 | null | https://arxiv.org/abs/1903.04190v2 | https://arxiv.org/pdf/1903.04190v2.pdf | Toward Fast and Accurate Neural Chinese Word Segmentation with Multi-Criteria Learning | The ambiguous annotation criteria lead to divergence of Chinese Word Segmentation (CWS) datasets in various granularities. Multi-criteria Chinese word segmentation aims to capture various annotation criteria among datasets and leverage their common underlying knowledge. In this paper, we propose a domain adaptive segmenter to exploit diverse criteria of various datasets. Our model is based on Bidirectional Encoder Representations from Transformers (BERT), which is responsible for introducing open-domain knowledge. Private and shared projection layers are proposed to capture domain-specific knowledge and common knowledge, respectively. We also optimize computational efficiency via distillation, quantization, and compiler optimization. Experiments show that our segmenter outperforms the previous state of the art (SOTA) models on 10 CWS datasets with superior efficiency. | ['Wei Chu', 'Kunlong Chen', 'Taifeng Wang', 'Xingyi Cheng', 'Weipeng Huang'] | 2019-03-11 | null | null | null | null | ['compiler-optimization'] | ['computer-code'] | [ 2.57413507e-01 -6.97161034e-02 -6.43831849e-01 -5.30908287e-01
-9.90815818e-01 -9.63522255e-01 9.04852003e-02 -1.95203274e-01
-6.53718650e-01 6.15403652e-01 5.52755892e-01 -6.67111456e-01
1.27998710e-01 -6.12872720e-01 -5.43922663e-01 -2.48072982e-01
5.80304742e-01 5.12384892e-01 4.93372709e-01 -1.19055100e-01
2.27991685e-01 -2.16012821e-01 -7.31563509e-01 4.82496947e-01
1.43437278e+00 9.83238876e-01 6.89837158e-01 3.38939101e-01
-6.20690346e-01 5.36374569e-01 -5.32572925e-01 -6.70415878e-01
1.67164057e-01 -3.82943422e-01 -1.49760616e+00 -1.88090384e-01
6.30067512e-02 -2.49474421e-01 1.19510807e-01 1.18889236e+00
2.97803760e-01 -8.85470882e-02 5.32024622e-01 -7.08791137e-01
-1.13560426e+00 1.61601508e+00 -4.93601382e-01 1.70498580e-01
-3.15924026e-02 5.41631766e-02 1.57493031e+00 -6.97738171e-01
8.15293491e-01 9.36995327e-01 5.78233838e-01 7.74618268e-01
-1.04625165e+00 -8.07942629e-01 5.08790791e-01 3.55653763e-01
-1.44637740e+00 8.40959605e-03 6.93200052e-01 -2.59766012e-01
9.37741876e-01 2.00097084e-01 6.45830512e-01 1.13442338e+00
-5.29504120e-01 1.59649479e+00 1.01746464e+00 -2.86731035e-01
1.98668391e-01 -5.80739900e-02 7.41655469e-01 3.74047697e-01
3.91755432e-01 -4.81245011e-01 -5.57951033e-01 1.97389826e-01
5.43661058e-01 -3.44138086e-01 -3.53986621e-01 -4.38316204e-02
-1.21911871e+00 9.31432605e-01 1.28512889e-01 3.35742056e-01
-1.39722660e-01 -1.35526881e-01 5.65178812e-01 1.73760071e-01
1.85103267e-01 6.39808834e-01 -1.29721522e+00 -5.33128977e-01
-8.62562478e-01 8.03533383e-03 8.36629987e-01 1.77962935e+00
9.15083230e-01 -2.95823179e-02 -2.83773124e-01 8.75807166e-01
3.72143596e-01 2.82629699e-01 9.67240512e-01 -8.73975277e-01
8.39231789e-01 8.85674417e-01 -1.55895889e-01 -2.02431723e-01
-3.56580645e-01 -5.09740055e-01 -5.71699202e-01 -8.96240354e-01
2.29956433e-01 -4.63320583e-01 -1.25219190e+00 1.74730647e+00
1.19166851e-01 1.31126717e-01 2.58717775e-01 8.29395711e-01
8.33243072e-01 6.74833596e-01 3.19880068e-01 1.41548723e-01
1.47610307e+00 -9.93762553e-01 -7.64989793e-01 -4.63743985e-01
9.56989884e-01 -6.09191000e-01 1.45328975e+00 2.43787259e-01
-9.47884679e-01 -4.45645273e-01 -8.96053255e-01 -4.25854594e-01
-5.33486366e-01 3.55833560e-01 6.98407769e-01 9.77524102e-01
-6.22150779e-01 2.84165651e-01 -8.66869807e-01 -1.22847809e-02
6.28366232e-01 2.03309596e-01 2.49509364e-01 6.77590370e-02
-1.64349139e+00 5.68155050e-01 1.01984847e+00 -5.03057688e-02
-5.90817928e-01 -1.01590240e+00 -8.62037480e-01 1.01265833e-01
6.32064104e-01 -5.23809969e-01 1.34951615e+00 -8.15967739e-01
-1.84849358e+00 6.99167967e-01 2.70226877e-02 -5.04140496e-01
-1.89786125e-02 -6.45841897e-01 -4.72207695e-01 -1.12480290e-01
3.33421141e-01 7.22802997e-01 2.88159430e-01 -8.48624408e-01
-7.55420029e-01 -1.57178655e-01 -5.16182296e-02 4.05675083e-01
-4.98944283e-01 1.02389187e-01 -9.50406969e-01 -7.84960866e-01
1.03189461e-02 -6.94268107e-01 -3.35252404e-01 -7.00817406e-01
-7.69136667e-01 -3.18570703e-01 5.46028495e-01 -8.32257688e-01
1.66266537e+00 -2.06249785e+00 3.62873405e-01 2.01259367e-02
1.00722872e-02 4.07706887e-01 -1.08806238e-01 2.07668528e-01
4.00793016e-01 5.76505423e-01 -5.15279770e-01 -7.74801597e-02
2.49582171e-01 6.85816348e-01 -4.08685505e-01 -9.91021190e-03
4.49818134e-01 1.14190304e+00 -8.60214233e-01 -8.24663758e-01
-1.96437553e-01 -1.08323619e-01 -7.17463732e-01 2.02089801e-01
-5.68861425e-01 3.07828397e-01 -7.23562598e-01 8.74581158e-01
8.50088656e-01 -3.33341211e-01 4.92359072e-01 -1.01761185e-01
-3.60089354e-02 7.94550955e-01 -1.28323913e+00 2.41515565e+00
-3.15884739e-01 3.15176278e-01 -2.42062360e-01 -8.91096473e-01
7.94448316e-01 4.82652634e-02 1.04394116e-01 -7.40853250e-01
2.75557458e-01 4.43386108e-01 8.69276598e-02 -2.34501734e-01
1.01626194e+00 1.96717128e-01 -6.58767402e-01 2.80696243e-01
3.98776889e-01 -1.43924892e-01 3.12373400e-01 3.16016138e-01
7.91972995e-01 3.64850700e-01 2.55169302e-01 -5.10040820e-01
5.38034856e-01 2.88314253e-01 1.22470438e+00 4.50977206e-01
-1.65058032e-01 4.35092598e-01 6.31479740e-01 -1.53530583e-01
-6.50585413e-01 -1.07494211e+00 -9.38334465e-02 1.40568841e+00
3.99628103e-01 -6.09261453e-01 -8.80301893e-01 -9.29756701e-01
-2.71333128e-01 8.79297137e-01 -4.73189175e-01 7.05367699e-02
-8.72709930e-01 -7.30247021e-01 9.97926414e-01 1.06223941e+00
7.29394257e-01 -7.21658885e-01 -6.32505476e-01 2.70919889e-01
-4.35777038e-01 -1.37000930e+00 -7.14436233e-01 6.36191368e-01
-8.89792621e-01 -1.05306292e+00 -6.89787507e-01 -1.00647664e+00
2.85917848e-01 -1.01149036e-02 1.23266375e+00 -3.67328852e-01
2.57178843e-01 -2.26134751e-02 -6.90208077e-01 -3.00366104e-01
-1.34251148e-01 1.00521815e+00 -4.32008177e-01 -3.92271489e-01
7.59043097e-01 -2.14474760e-02 -4.62439477e-01 3.70179594e-01
-9.18157160e-01 3.06152821e-01 6.94670081e-01 9.78479803e-01
8.10937762e-01 -4.00539488e-01 4.25419748e-01 -1.56214941e+00
6.82537913e-01 -6.33229077e-01 -7.44342387e-01 5.64472079e-01
-7.23805904e-01 3.93733382e-01 6.80109322e-01 -3.42595458e-01
-1.47103667e+00 3.99947315e-02 -5.07525429e-02 -9.61155258e-03
-1.82223275e-01 7.60510206e-01 -6.61822855e-01 5.54334283e-01
2.47095093e-01 3.39978248e-01 -8.10674191e-01 -9.93186593e-01
8.80045235e-01 9.93207812e-01 5.93941927e-01 -1.05406141e+00
2.08601162e-01 1.00457385e-01 -8.11157882e-01 -3.68178129e-01
-9.26201642e-01 -7.14528918e-01 -9.56251740e-01 4.81692672e-01
9.96968389e-01 -1.17083383e+00 2.43233163e-02 6.01547718e-01
-1.27750397e+00 -4.57943201e-01 -4.32829589e-01 3.00637156e-01
-1.89352468e-01 2.69802749e-01 -8.13346684e-01 -1.80213258e-01
-4.31587070e-01 -1.21874094e+00 1.06558192e+00 5.81942141e-01
-1.45906672e-01 -9.18100655e-01 -2.07198076e-02 4.36837822e-01
2.26301715e-01 -2.27000326e-01 8.53997231e-01 -9.66379344e-01
-5.37051141e-01 4.72909302e-01 -4.72151846e-01 5.55516183e-01
9.53783914e-02 -2.74912655e-01 -7.42093921e-01 5.69083467e-02
-4.14462507e-01 -3.96919578e-01 1.03229642e+00 1.50206015e-01
1.46583009e+00 -1.66970521e-01 -3.18760872e-01 9.81852889e-01
1.46893370e+00 2.98835486e-01 4.13416088e-01 2.88580328e-01
9.66310561e-01 3.02522629e-01 6.86842620e-01 2.84223676e-01
8.12309086e-01 1.51839599e-01 2.15377472e-02 1.06869541e-01
-2.61715025e-01 -2.27938563e-01 2.95688242e-01 1.57198513e+00
2.01967016e-01 -2.45893165e-01 -1.14673650e+00 1.06490910e+00
-1.57798898e+00 -3.29030424e-01 -1.42212063e-01 1.51805437e+00
1.52937233e+00 5.82135729e-02 -2.13828817e-01 -2.40722626e-01
7.10440993e-01 1.44080475e-01 -7.77864158e-01 -7.47164309e-01
-3.81217510e-01 5.67203879e-01 9.82799470e-01 1.86050281e-01
-1.10018182e+00 1.58000135e+00 6.06474733e+00 1.16426313e+00
-8.14969480e-01 3.42699498e-01 5.84399641e-01 3.45585495e-01
-8.33876669e-01 2.19518110e-01 -1.26761925e+00 6.52047515e-01
8.60660851e-01 -7.05473870e-02 1.15109988e-01 7.57463932e-01
-6.86692536e-01 1.27681226e-01 -8.10407400e-01 6.17790759e-01
-2.39390358e-01 -1.44373965e+00 2.83827018e-02 -3.17936510e-01
1.04022622e+00 4.48093235e-01 4.07560021e-02 4.83043402e-01
1.00587285e+00 -5.94329119e-01 8.04734826e-01 6.77760839e-02
1.02519381e+00 -6.43592238e-01 5.42657435e-01 1.05026796e-01
-1.30091679e+00 6.97100312e-02 -3.51838231e-01 1.63539469e-01
1.06879964e-01 2.27103055e-01 -8.58640552e-01 7.69092143e-01
6.00258648e-01 8.30672503e-01 -5.33785045e-01 5.65687954e-01
-5.74924767e-01 1.25656199e+00 -2.81549275e-01 -2.31772348e-01
7.42801011e-01 -1.88941911e-01 1.10503487e-01 1.80114377e+00
1.27019593e-02 1.67665984e-02 1.09719038e-01 9.96908605e-01
-2.86605924e-01 8.28579292e-02 1.23211496e-01 -1.83295041e-01
7.81892180e-01 8.97616982e-01 -7.40513325e-01 -4.05428082e-01
-7.46452928e-01 1.13849187e+00 3.96724969e-01 3.92905921e-01
-8.10104907e-01 -6.37461901e-01 9.33855534e-01 -5.94425559e-01
7.46739566e-01 -2.75215924e-01 -6.98819876e-01 -1.29063666e+00
-1.81133941e-01 -7.33884573e-01 7.99748659e-01 -2.46781930e-01
-1.08704889e+00 3.89432192e-01 -1.00293225e-02 -1.05231059e+00
7.63678774e-02 -7.12238252e-01 -2.35853672e-01 8.50562513e-01
-2.09135461e+00 -1.47299588e+00 1.46908522e-01 6.74277365e-01
6.91408038e-01 -1.35666966e-01 7.66272008e-01 3.28839660e-01
-7.67150164e-01 7.36000896e-01 2.85708666e-01 5.95369816e-01
5.66235900e-01 -1.65030074e+00 8.05251122e-01 1.11716044e+00
2.99987435e-01 6.72747910e-01 6.05565347e-02 -7.68804014e-01
-1.40443408e+00 -1.19085395e+00 8.20439637e-01 -1.77901179e-01
7.01089859e-01 -3.79513204e-01 -9.04544294e-01 8.56725812e-01
4.69341099e-01 -9.31640342e-02 1.21978796e+00 4.31843400e-01
-6.87117994e-01 1.08033240e-01 -6.16926491e-01 4.94484812e-01
1.23060715e+00 -5.25763333e-01 -1.02740908e+00 -2.48635620e-01
1.31233001e+00 -7.46110082e-01 -9.36793089e-01 4.23772819e-02
3.74273658e-01 -4.30963546e-01 6.67381644e-01 -8.20402086e-01
3.71221155e-01 -2.11032823e-01 -3.93558770e-01 -1.39859962e+00
-3.37967783e-01 -7.91057706e-01 -3.54179926e-02 1.50524414e+00
7.25802600e-01 -3.32703620e-01 7.53195047e-01 5.10343969e-01
-4.71455783e-01 -6.04927838e-01 -9.84896958e-01 -7.97781944e-01
6.14496112e-01 -7.80861139e-01 1.27730870e+00 1.12274921e+00
9.06720012e-02 3.91733736e-01 -7.38467500e-02 1.59508258e-01
1.07425869e-01 5.93190193e-01 2.80083835e-01 -8.63676846e-01
-1.81737885e-01 -5.76211095e-01 -4.55621444e-03 -1.68593442e+00
1.69803098e-01 -1.02682436e+00 3.57681699e-02 -1.33580434e+00
1.16077848e-01 -7.83422053e-01 -7.25249648e-01 6.60179138e-01
-4.62640941e-01 -1.00095391e-01 1.81820527e-01 -1.91229582e-02
-1.07574475e+00 3.92292470e-01 1.28680372e+00 -1.69872791e-01
-2.57337213e-01 -3.95663440e-01 -1.22660351e+00 7.68721700e-01
9.63332593e-01 -3.52906466e-01 -3.20538729e-01 -1.37581205e+00
4.63429332e-01 -2.97862530e-01 -3.01800668e-01 -7.64760077e-01
4.53550398e-01 -3.93761724e-01 5.73987886e-02 -7.45453477e-01
-1.29902631e-01 -6.52877629e-01 -4.09489244e-01 3.12036961e-01
-4.48796004e-01 -2.04615533e-01 5.10121100e-02 4.76711899e-01
-2.47441694e-01 -3.10028940e-01 5.01446545e-01 -1.93217099e-01
-1.31915712e+00 1.99658036e-01 -3.02165866e-01 8.86510253e-01
7.28299558e-01 -1.04650214e-01 -2.28967801e-01 4.15986419e-01
-3.46348286e-01 5.62581360e-01 2.27012783e-01 4.66801167e-01
5.17516196e-01 -1.17299962e+00 -6.72663450e-01 2.69492239e-01
2.53822029e-01 5.83496809e-01 -1.61653832e-02 3.26048315e-01
-4.00362045e-01 5.14044285e-01 -1.04881696e-01 -4.27408069e-01
-7.68838584e-01 1.30777985e-01 -5.21897897e-02 -5.87971807e-01
-2.90152431e-01 1.31674087e+00 1.38077224e-02 -7.34115720e-01
6.39079213e-02 -9.87633109e-01 -2.77616829e-01 1.90935075e-01
2.32568875e-01 4.26007509e-01 -1.41490579e-01 -4.32402432e-01
-4.59321499e-01 5.07805645e-01 -4.04942930e-01 9.39837471e-02
1.09452856e+00 -3.55416358e-01 1.29696757e-01 2.87356943e-01
1.08220577e+00 -1.18797839e-01 -1.28971541e+00 -7.06209660e-01
5.47476172e-01 -2.24546492e-01 1.19329914e-01 -1.12157607e+00
-1.21875715e+00 9.73460674e-01 1.75539643e-01 -2.55101442e-01
1.35283196e+00 -3.98932099e-02 1.47583187e+00 2.41355389e-01
2.04576418e-01 -1.78529680e+00 -3.15815121e-01 1.02761590e+00
1.09537601e-01 -1.20854902e+00 -2.75304019e-01 -5.15223086e-01
-1.18457997e+00 9.96061385e-01 7.82866120e-01 1.16087385e-01
5.21206141e-01 3.61816317e-01 2.43903901e-02 -2.27327403e-02
-5.66486001e-01 -6.16239965e-01 1.95236444e-01 7.10426867e-01
3.32263052e-01 5.33273935e-01 -6.49741471e-01 1.60066497e+00
-1.42832994e-01 6.05907962e-02 3.30796748e-01 9.19473350e-01
-2.17468724e-01 -1.41848516e+00 2.09196135e-01 4.78119880e-01
-6.12714529e-01 -5.19374490e-01 -3.54632378e-01 5.29556990e-01
5.38420498e-01 6.91686869e-01 4.03284431e-02 -4.55166608e-01
4.25042920e-02 1.68504298e-01 6.13106601e-02 -7.52156615e-01
-8.11677754e-01 -1.70749202e-02 1.69893742e-01 -4.99094874e-01
-3.61361623e-01 -8.46864820e-01 -1.52320874e+00 1.92221954e-01
-5.82912505e-01 2.54535854e-01 3.57967228e-01 8.42513502e-01
4.35433418e-01 5.93240857e-01 3.63263249e-01 1.75881997e-01
-4.87065762e-01 -7.84858406e-01 -6.20587528e-01 2.58316755e-01
-1.32345691e-01 -2.50657320e-01 2.78172553e-01 2.19622314e-01] | [9.963030815124512, 10.11161994934082] |
863a84b3-f1fe-4670-a7ad-c8dde0c56bdb | audio-visual-speech-enhancement-with | 2306.06495 | null | https://arxiv.org/abs/2306.06495v1 | https://arxiv.org/pdf/2306.06495v1.pdf | Audio-Visual Speech Enhancement With Selective Off-Screen Speech Extraction | This paper describes an audio-visual speech enhancement (AV-SE) method that estimates from noisy input audio a mixture of the speech of the speaker appearing in an input video (on-screen target speech) and of a selected speaker not appearing in the video (off-screen target speech). Although conventional AV-SE methods have suppressed all off-screen sounds, it is necessary to listen to a specific pre-known speaker's speech (e.g., family member's voice and announcements in stations) in future applications of AV-SE (e.g., hearing aids), even when users' sight does not capture the speaker. To overcome this limitation, we extract a visual clue for the on-screen target speech from the input video and a voiceprint clue for the off-screen one from a pre-recorded speech of the speaker. Two clues from different domains are integrated as an audio-visual clue, and the proposed model directly estimates the target mixture. To improve the estimation accuracy, we introduce a temporal attention mechanism for the voiceprint clue and propose a training strategy called the muting strategy. Experimental results show that our method outperforms a baseline method that uses the state-of-the-art AV-SE and speaker extraction methods individually in terms of estimation accuracy and computational efficiency. | ['Shigeo Morishima', 'Keitaro Tanaka', 'Tomoya Yoshinaga'] | 2023-06-10 | null | null | null | null | ['speech-enhancement', 'speech-extraction'] | ['speech', 'speech'] | [ 0.3264381 -0.07979473 0.00756943 -0.16222504 -1.1460321 -0.31054142
0.2491533 0.09201301 -0.29537877 0.5333175 0.27621046 -0.22826807
0.24561931 -0.30495888 -0.4147279 -0.7607765 0.1883137 -0.03896622
0.5558414 0.1446196 0.1776544 0.36807662 -1.8397728 0.63629985
0.6386946 1.0169967 0.8870429 1.0866566 -0.22955568 0.60477173
-1.0681547 0.09115969 -0.11553784 -0.548727 -0.21224342 0.3745711
0.32146758 -0.47980866 -0.40584797 1.2807442 0.74444664 0.09434908
0.41110462 -1.2376524 -0.15573251 0.45097974 -0.63242143 0.5321406
0.76576555 0.04966417 0.63863415 -1.1844649 0.51500213 1.1777611
0.39802963 0.5057423 -0.9524376 -0.93934745 0.19278339 0.2337057
-1.6190498 -1.0240306 0.95350385 -0.32850105 0.68024653 0.5714149
0.56252486 1.091862 -0.11985666 0.96070695 0.7285715 -0.52340454
0.24493225 0.56282645 0.17708339 0.31608313 -0.39860743 0.08332489
-0.65399826 -0.38242748 0.6533426 -0.35929933 -0.8109398 0.09362632
-0.98849916 0.34403664 -0.25806212 0.47415698 -0.62651557 -0.299105
0.23828582 0.15066908 0.37778687 -0.28970796 -0.07865708 -0.41510156
-1.3148248 -0.09590866 0.58604926 1.0099081 0.28535816 0.29514238
-0.3092659 0.926366 0.38685477 0.61070794 0.33026615 -0.44443467
0.65467286 0.09110124 0.4601916 -0.8282738 -0.13119724 -0.368432
-0.64266914 0.31938633 0.312798 -0.21250096 -0.80822384 1.5685811
0.5160943 0.5673115 -0.0063217 0.9838092 1.1679803 1.1811354
-0.17014809 -0.9174384 1.4853873 -0.86135674 -1.3186157 -0.30256677
-0.08116832 -0.9468507 0.83644974 0.6916586 -1.0483891 -0.88135797
-0.9981084 0.54934233 0.08579443 0.7469731 -0.13595866 0.8202473
-1.0101988 -0.02049509 -0.67310417 -0.23079495 -0.06471074 0.36282283
-0.41655183 0.22308344 -1.0275836 0.3466358 0.01337638 0.07343278
-1.0370572 -0.28196746 -0.91028595 0.3059255 0.4724793 -0.16057047
1.4208363 -1.4974996 -1.5321608 0.48553377 -0.72264653 -0.06199807
0.3657785 -0.02254735 -1.1056125 0.519287 -0.08099669 0.26515988
1.7183067 -1.2964305 -0.9633818 -0.12670854 -0.36743784 0.20445175
-0.36265117 0.5107812 -0.9191504 -0.69131064 0.08721222 -0.38876474
0.43369254 -0.12993206 -0.5264131 -0.0507454 1.2114997 -1.1444232
1.6735485 -2.708964 -0.19983326 0.10383877 0.11427784 0.7022866
-0.1800749 0.33057758 -0.2456933 -0.32558787 0.17036563 -0.44764796
-0.43526867 -0.3462779 -0.16777223 0.47241575 0.0444012 0.04738797
-0.894752 -0.7008159 0.22904347 0.6623811 -0.19039665 0.52519864
0.3255742 0.38623223 -0.2222436 0.54734075 0.8133768 0.16903624
-0.04165995 -0.2031596 -0.2312262 0.50611156 -1.720197 1.1471893
-0.35573626 0.9065657 0.81307405 -0.6582065 0.92199355 0.91800994
0.06472043 -0.30051813 -0.03083921 0.17386532 0.07909162 -0.7975433
0.5092426 0.22310582 0.41558582 0.24404296 0.02361378 0.28675553
-0.03662344 0.25617743 0.9506334 -0.24568541 0.42442554 0.23138036
0.9045647 -0.58551264 0.53614557 0.7336382 -0.5194664 0.77979225
0.2941547 0.32807806 -0.74923974 -0.93510485 0.05632164 1.1604755
0.07565369 -0.6475762 -0.9720744 -0.5788923 -0.39894894 0.6644422
-0.28844616 0.03665022 -0.49326658 -0.11984951 0.36165392 0.33041465
0.20233163 -1.1046294 -0.22976908 0.4241269 -0.5956224 -0.9739656
-1.0562774 -0.02170717 -0.3137535 -0.6250538 -0.9165386 -0.83214545
0.51838684 0.529959 0.5428822 -0.19545276 0.08721881 0.5383532
-0.23119311 -0.3324518 -0.7865925 -0.51098686 0.08245286 0.705275
0.32700124 -0.4765045 -0.36756194 0.36039338 -0.4903738 0.04773692
0.26571655 0.72420144 0.42928913 0.3435621 0.6787288 -0.2812191
0.8065956 -0.34808972 -0.58209354 0.05738106 0.08571836 -0.41734472
0.49472237 -1.178651 -1.2217267 0.11915506 -0.3607349 -0.61047256
-0.53392214 0.27907664 -0.5046975 0.36671102 0.31127697 0.56151634
-0.17559887 -0.5601792 -0.21187046 1.5085026 0.83121705 0.09916037
0.43584082 0.23577175 -0.6486489 -1.4970055 -0.17517991 -0.8668218
-0.21539909 -0.52443093 0.6412352 -0.9723228 -0.6475448 0.5621289
-1.543081 0.18110454 0.16975918 0.7986063 -0.13815138 0.48896065
-0.45881566 -1.5614511 -0.10793389 -1.2732695 1.243407 0.29798663
-0.3146165 -0.4246407 -0.17412591 0.15860586 0.22453164 -0.47816285
0.2941967 -0.81817037 -0.4230742 -0.29476038 -0.03616854 0.32561153
0.46054643 0.0875435 -1.5460805 -0.13940515 0.2585065 0.3477323
0.62104774 0.73010355 0.8591412 -0.30003318 -0.58255935 0.34944162
0.8156564 0.9276709 0.6115371 -0.13678837 0.32268825 0.6324668
0.73520017 0.5350749 0.03862154 0.96415335 0.30733818 -0.19361456
-0.29053053 -0.23919734 0.8234809 0.8433313 0.02810013 -0.59234667
-0.4930879 0.53285265 -1.5433229 -1.1267036 -0.26774237 2.4736018
0.6741023 0.22937933 0.31659755 0.531079 1.3101048 0.2501217
-0.28864062 -0.21763952 -0.05356819 -0.04047823 -0.08633962 0.72184354
-1.2285945 0.50850403 5.404799 1.1356249 -1.3728292 0.22423746
0.26446912 -0.18919013 -0.06011471 -0.50442344 -0.9325629 0.69589204
1.0212193 0.03618781 0.27296913 0.68948495 0.70882624 -0.31069034
-1.1614554 1.4857705 0.2818303 -0.8173128 -0.4079718 -0.11304951
0.03952475 -0.47259986 -0.02601201 0.15592085 -0.5373505 -0.66794246
1.0489751 0.20204759 0.969217 -0.67370546 0.5219677 0.49674106
-1.6217662 -0.07939414 0.01499579 0.20774084 0.31854632 0.41658753
-0.9825746 0.351607 0.6981999 0.16264078 -0.03664896 1.1357151
-0.3097819 0.9419816 -0.30038577 0.09176716 -0.11837795 0.1481321
1.2632304 1.4360553 0.62305444 0.0722859 -0.03647711 0.6998467
0.21162508 0.23013127 -0.597457 -0.14739937 0.79155874 1.1178163
-0.560189 -0.443731 -0.4729295 1.1062367 -0.39441857 0.67306554
-0.76305497 -0.82530487 0.5073357 0.24248189 0.5472465 0.1365708
0.18360633 -0.7788478 0.21248044 -0.94964665 0.10694163 -1.2707174
-0.6694822 0.8817652 -0.24433306 -1.4813442 -0.5463644 -0.11400557
-0.7482881 1.0825446 -0.9893634 -0.7254654 -0.17177166 0.60442585
0.93256634 -0.24536738 0.6375349 0.5458051 -0.43843424 0.7086089
-0.04573368 0.10450913 0.8074309 -0.8702849 0.15468068 0.9667128
0.15770152 0.27952766 0.8506056 -0.76240176 -1.0503916 -0.67647064
1.1599123 0.0798718 0.32967633 -0.6837447 -1.1012443 0.53114504
0.22894236 -0.1802482 0.40119168 -0.43497604 0.1698435 -0.1872233
-0.996802 0.49327913 0.51278037 -0.7355063 -0.7480977 -0.15502836
0.6580977 -0.41931784 -0.09148536 -0.08424731 0.59300584 -0.97333485
0.9309991 -0.12690483 -0.12382013 -0.5411574 -0.00935031 -1.26189
-0.03725641 -0.9507624 -0.14440735 1.791646 0.24086753 -0.38544974
0.31987625 0.17150412 -0.0824789 -0.10565208 -1.0355109 -0.5944023
-1.0205947 -0.71960056 0.4296487 0.7775525 0.09070957 0.4783193
-0.79006827 0.6856054 0.33866814 -0.31476513 0.68668485 -0.97685796
-0.36080956 -0.14298081 -0.30002192 -1.187999 -0.06376362 -0.19750056
0.21193953 -1.3917353 0.00858217 0.27095637 -0.27002192 0.08295286
-0.16524182 -0.15452291 0.07392664 -0.11669399 -0.4223192 0.4383126
1.0343417 -0.25946945 -0.796186 0.49103117 -0.35012335 0.8262042
0.31791657 -0.5528759 -0.5068318 0.11380894 -0.23374695 0.86949486
0.24580863 -0.80454415 0.44706646 0.13891967 0.21331477 -0.964145
0.78735983 -0.85680205 0.08764029 0.11337411 -0.16346611 -0.2835521
0.34086257 0.60716903 -0.501454 -0.41937515 0.6651873 0.2367018
-0.43418378 -0.09559958 -1.0824227 -0.36079544 0.7673194 -0.5859314
-0.22400825 -0.9057615 -0.89839184 -0.17108664 -0.04248111 0.25274506
0.9780323 -1.2622112 -0.67648953 0.6224233 -0.14638127 -0.42155132
0.47737473 1.0370731 -0.01845277 0.22281308 0.16339317 -0.64397943
-2.0594027 0.5773651 0.28865656 0.25983804 -0.6212859 0.8991382
0.63326895 0.24701841 0.76793396 -0.39666286 -0.609361 0.1355123
1.2751313 0.4145907 -0.04664491 -0.879338 -0.50540173 0.21494073
0.05127325 -0.5188314 1.036377 -0.5030312 0.22095612 0.7399069
1.0086458 0.6760636 -1.0972131 -0.232279 -0.3672205 -0.57438314
0.32568935 -0.5911093 -0.8947064 1.066576 0.86383694 0.35334513
1.4581952 0.06177254 0.49223575 -0.03047867 0.13270727 -1.060359
0.08926294 0.0207178 1.017268 -1.0090464 -0.35237858 -0.69477606
-0.66428924 1.1303238 0.4133845 0.5770829 0.6291509 0.4591828
0.1241194 0.22819324 -0.79494536 -0.30761963 0.48366824 1.0116282
0.20943247 -0.1444537 0.13662411 1.0860636 0.03310403 -0.23521492
0.6801997 0.7043558 -0.70930374 -0.6691497 -1.0089487 0.05375566
-0.74480236 -0.14931066 -0.4093454 0.51482147 0.03981095 1.4443899
0.23882389 -0.4083032 0.3101545 0.22431886 0.23646127 -0.5012944
-0.45767868 1.0754082 0.22752427 -0.26931593 -0.3515555 -0.6923575
-0.98750585 0.19403933 -0.6015435 0.23979877 0.6851577 0.75793976
0.04712501 0.75642717 0.6086397 -1.0024265 -0.11992552 -1.2387146
-0.9473429 0.08443331 0.89676094 -0.49665472 -0.603488 0.31917277] | [14.535749435424805, 5.272113800048828] |
e2cec697-cca3-419b-aacc-9ce8cd741273 | sifter-a-task-specific-alignment-strategy-for | 2306.12280 | null | https://arxiv.org/abs/2306.12280v1 | https://arxiv.org/pdf/2306.12280v1.pdf | SIFTER: A Task-specific Alignment Strategy for Enhancing Sentence Embeddings | The paradigm of pre-training followed by fine-tuning on downstream tasks has become the mainstream method in natural language processing tasks. Although pre-trained models have the advantage of generalization, their performance may still vary significantly across different domain tasks. This is because the data distribution in different domains varies. For example, the different parts of the sentence 'He married Smt. Dipali Ghosh in 1947 and led a very happy married life' may have different impact for downstream tasks. For similarity calculations, words such as 'led' and 'life' are more important. On the other hand, for sentiment analysis, the word 'happy' is crucial. This indicates that different downstream tasks have different levels of sensitivity to sentence components. Our starting point is to scale information of the model and data according to the specifics of downstream tasks, enhancing domain information of relevant parts for these tasks and reducing irrelevant elements for different domain tasks, called SIFTER. In the experimental part, we use the SIFTER to improve SimCSE by constructing positive sample pairs based on enhancing the sentence stem and reducing the unimportant components in the sentence, and maximize the similarity between three sentences. Similarly, SIFTER can improve the gate mechanism of the LSTM model by short-circuiting the input gate of important words so that the LSTM model remembers the important parts of the sentence. Our experiments demonstrate that SIFTER outperforms the SimCSE and LSTM baselines. | ['Qiuhong zhai', 'XiaoYu Zhang', 'Chaoming Liu', 'Wenhao Zhu', 'Chao Yu'] | 2023-06-21 | null | null | null | null | ['sentence-embeddings', 'sentence-embeddings', 'sentiment-analysis'] | ['methodology', 'natural-language-processing', 'natural-language-processing'] | [ 1.17888689e-01 -1.21008821e-01 4.22720760e-02 -7.60642648e-01
-3.85988444e-01 -4.76596355e-01 4.62385088e-01 4.73138630e-01
-7.53566265e-01 5.63533783e-01 8.00822794e-01 -2.81965077e-01
6.55851364e-02 -9.00551140e-01 -4.57561851e-01 -6.49863422e-01
3.05482209e-01 1.97015226e-01 1.63850725e-01 -9.81229186e-01
5.35287440e-01 1.83339179e-01 -1.08370888e+00 7.72997618e-01
7.41526663e-01 7.74557054e-01 5.04626811e-01 1.78191304e-01
-4.85356897e-01 5.01882553e-01 -9.22969222e-01 -5.21227777e-01
2.88660433e-02 -5.05848289e-01 -8.99006486e-01 -4.31889951e-01
2.76843887e-02 -1.58349201e-01 -7.38977119e-02 1.11802542e+00
5.79085886e-01 3.79760265e-01 5.31774938e-01 -7.78967440e-01
-1.05350447e+00 1.17281127e+00 -5.56243598e-01 3.83977920e-01
1.17076389e-01 -3.38871740e-02 1.18539321e+00 -9.35663104e-01
4.47795093e-01 1.55625367e+00 6.44609869e-01 6.13160551e-01
-9.99612451e-01 -6.77134037e-01 5.58813512e-01 2.69299567e-01
-9.60097611e-01 -3.33125412e-01 6.45387650e-01 -7.17379525e-02
1.23165882e+00 -4.76915278e-02 4.02168959e-01 1.12116623e+00
6.77024603e-01 6.19318187e-01 9.80877936e-01 -3.26519996e-01
5.50448932e-02 1.31293491e-01 4.10877198e-01 1.48779094e-01
7.92358592e-02 -1.36170954e-01 -5.47956169e-01 1.40489444e-01
3.60857129e-01 2.25628480e-01 -1.84523091e-01 3.43322694e-01
-1.11557841e+00 9.63517070e-01 6.20724559e-01 8.08073580e-01
-3.74363273e-01 -3.07182580e-01 7.43283749e-01 5.46151876e-01
4.66516167e-01 5.93502522e-01 -8.89104128e-01 7.05238506e-02
-6.14856362e-01 -2.58027427e-02 5.98343849e-01 7.22451806e-01
6.40014708e-01 -6.92862570e-02 -5.27791262e-01 9.54722285e-01
1.70119151e-01 1.45769849e-01 9.24132884e-01 -5.86999118e-01
7.98363984e-01 5.11418283e-01 -3.72001588e-01 -9.13655341e-01
-3.20674062e-01 -6.00214422e-01 -9.79241490e-01 -2.60473311e-01
3.22294325e-01 -3.45406264e-01 -9.04018879e-01 2.03131485e+00
-2.21522078e-01 -4.37700957e-01 3.61808926e-01 7.44188309e-01
9.69437301e-01 8.48104000e-01 5.08561134e-01 -1.61061242e-01
1.40600252e+00 -8.88774693e-01 -6.96818113e-01 -8.80702615e-01
1.01761115e+00 -8.22364569e-01 1.44434488e+00 2.55925860e-02
-9.82471228e-01 -7.98861384e-01 -9.35499191e-01 -4.30436432e-01
-6.42985702e-01 -1.62043609e-02 3.60670954e-01 1.51323110e-01
-9.93674099e-01 8.51285338e-01 -2.69683689e-01 -4.86364067e-01
2.42923006e-01 2.92890251e-01 -1.96554199e-01 -1.13178648e-01
-1.90506566e+00 1.38199830e+00 6.05821192e-01 -8.35237466e-03
-3.80951256e-01 -6.76822305e-01 -9.56238866e-01 4.61294144e-01
-8.44488442e-02 -8.36650252e-01 1.30539906e+00 -1.21372724e+00
-1.23986661e+00 8.79847884e-01 -4.00461704e-01 -4.27256227e-01
-4.55590747e-02 -2.87635803e-01 -3.55797499e-01 -2.46589661e-01
2.75083423e-01 6.35537565e-01 7.45425284e-01 -7.69045174e-01
-5.99504054e-01 -5.93058646e-01 1.03952572e-01 5.18586040e-01
-5.75436771e-01 3.83850455e-01 -1.80101991e-01 -8.71527970e-01
6.90329447e-02 -5.35414815e-01 -3.13792586e-01 -6.33175552e-01
-1.11586116e-01 -3.47095013e-01 6.52928591e-01 -6.88701153e-01
1.38877285e+00 -2.34241819e+00 -6.34523705e-02 -6.32030666e-02
9.05649085e-03 2.15729773e-01 -5.20088613e-01 5.55157721e-01
-3.32796872e-01 2.22606137e-01 -1.73571274e-01 -2.21599892e-01
-1.59327179e-01 1.75790414e-01 -4.56511199e-01 1.74139589e-02
3.90780419e-01 9.96458709e-01 -9.72385228e-01 -2.91337252e-01
-1.05416559e-01 3.55908126e-01 -4.08780247e-01 -1.60356134e-01
1.21708907e-01 2.30582640e-01 -2.39376754e-01 1.53498277e-01
5.30996740e-01 -4.86714467e-02 1.18971221e-01 -1.36148900e-01
-1.03602502e-02 1.00766289e+00 -7.42760539e-01 1.61005485e+00
-5.64176083e-01 6.08106554e-01 -5.48537150e-02 -1.18732333e+00
1.10143864e+00 2.05882847e-01 -3.84705998e-02 -1.12693810e+00
3.82419884e-01 1.17316611e-01 5.86217463e-01 -3.29616994e-01
5.92453837e-01 -5.95515609e-01 -3.45055729e-01 4.99438822e-01
-5.07629141e-02 -2.49195933e-01 2.12498516e-01 4.68201101e-01
8.00690949e-01 -2.10312739e-01 3.33618343e-01 -4.04745847e-01
5.59809148e-01 -1.75617382e-01 5.96685708e-01 4.97137457e-01
-2.29878053e-01 4.14142489e-01 5.83292186e-01 -2.53869351e-02
-7.84329951e-01 -9.12086189e-01 1.07156359e-01 1.51164854e+00
1.59281003e-03 -4.06169891e-01 -6.87614679e-01 -5.12022138e-01
-2.59264320e-01 1.14171982e+00 -6.77168190e-01 -7.27217436e-01
-6.80153549e-01 -7.28094399e-01 2.71036893e-01 7.81999648e-01
6.03573024e-01 -1.30219340e+00 -1.42394423e-01 3.65090638e-01
-3.69879007e-01 -8.36675763e-01 -5.45996368e-01 5.06593168e-01
-9.46939349e-01 -4.77982312e-01 -5.09743392e-01 -1.05407691e+00
6.51857793e-01 4.85495061e-01 1.38842261e+00 -1.66021939e-02
3.52593213e-01 -4.23496187e-01 -4.74202305e-01 -6.37878001e-01
-3.28301638e-01 2.49575183e-01 -5.65355234e-02 -2.82807767e-01
7.95082331e-01 -4.63987261e-01 -3.75142038e-01 -3.89108807e-02
-8.04993272e-01 -1.35778815e-01 6.31457269e-01 8.06009769e-01
2.03292593e-01 1.47453338e-01 8.83124650e-01 -9.96311963e-01
1.21665907e+00 -4.61538404e-01 1.35417968e-01 -1.23286620e-02
-4.33721125e-01 2.46250018e-01 1.03652334e+00 -3.76283795e-01
-1.21802485e+00 -4.11741406e-01 -4.72370088e-01 1.22718275e-01
4.74400856e-02 7.02332377e-01 -3.73577893e-01 5.20927548e-01
7.10649729e-01 1.84918523e-01 -1.46721557e-01 -2.27263123e-01
1.51003405e-01 5.55347264e-01 2.15138227e-01 -4.47878391e-01
6.13865912e-01 2.48272985e-01 -4.45519328e-01 -5.32842934e-01
-1.16821754e+00 -3.32116365e-01 -5.47170162e-01 2.07390532e-01
6.94146276e-01 -8.37459803e-01 -3.60350162e-01 5.15691340e-01
-1.47047210e+00 -2.63132811e-01 -2.90198117e-01 3.95567209e-01
9.46734194e-03 2.27755502e-01 -9.12198186e-01 -3.05109382e-01
-5.31769395e-01 -1.06028414e+00 8.17631483e-01 3.52369130e-01
-6.79126143e-01 -1.16531420e+00 -3.28268439e-01 -4.78364341e-02
5.19725263e-01 -3.84637177e-01 1.34747267e+00 -1.00972688e+00
2.50891387e-01 4.12993841e-02 -3.05987597e-01 6.84815586e-01
2.91024864e-01 -2.04609603e-01 -9.64643478e-01 -3.23213972e-02
4.77205783e-01 -1.05779544e-01 1.12418044e+00 5.97319603e-01
8.79869342e-01 -2.65348852e-01 -2.78954893e-01 2.16125116e-01
1.15704727e+00 2.93558627e-01 7.02348590e-01 4.76174086e-01
4.25291002e-01 1.02136409e+00 7.56771743e-01 1.38182342e-01
3.73210192e-01 1.98177338e-01 -1.37755767e-01 -1.66808352e-01
1.69141160e-03 -2.23746568e-01 7.96492279e-01 9.32135820e-01
3.95543337e-01 -1.28495261e-01 -7.13071823e-01 6.26741707e-01
-1.66586030e+00 -1.01140714e+00 -2.80137897e-01 2.04306531e+00
9.70015407e-01 4.58203048e-01 -2.77561128e-01 1.23454079e-01
8.75596583e-01 3.48821133e-01 -3.37376803e-01 -9.99099195e-01
-2.34873623e-01 4.09741551e-01 1.91036344e-01 4.85790998e-01
-7.34161675e-01 1.31765413e+00 5.84570503e+00 8.37763727e-01
-1.16215336e+00 1.08684622e-01 6.74491525e-01 -7.14428872e-02
-4.10579026e-01 1.05264582e-01 -1.15075183e+00 5.58385134e-01
9.59512591e-01 -4.81332660e-01 6.67824745e-02 6.69871807e-01
6.46888137e-01 -1.79259136e-01 -1.12900245e+00 7.08830714e-01
1.09281145e-01 -1.02436137e+00 2.03406617e-01 -3.78984928e-01
6.03111565e-01 4.72727530e-02 -2.30920129e-02 8.21945846e-01
1.87397480e-01 -8.66066933e-01 5.74623227e-01 2.04763576e-01
3.13749492e-01 -8.56658697e-01 9.14916277e-01 4.74525183e-01
-1.02568197e+00 -9.75435451e-02 -5.33623576e-01 -5.73553503e-01
1.28106400e-01 7.52060771e-01 -5.49450517e-01 7.31402487e-02
7.64817417e-01 6.74584091e-01 -5.76777995e-01 4.62781727e-01
-5.95817804e-01 3.69068772e-01 1.42340856e-02 -2.91941017e-01
3.61728460e-01 -2.14855701e-01 2.36861601e-01 1.40051126e+00
2.08154395e-01 1.70574069e-01 -1.74662799e-01 7.94820487e-01
-2.41520345e-01 2.65058666e-01 -7.23023951e-01 -4.73634200e-03
6.09800756e-01 1.18347502e+00 -5.17246842e-01 -5.63154221e-01
-4.48435485e-01 9.56079483e-01 3.78917754e-01 4.02470887e-01
-6.93134844e-01 -6.44685090e-01 7.33373225e-01 4.53511775e-02
1.43489718e-01 -3.54448147e-02 -7.94061244e-01 -9.69510138e-01
-3.09289582e-02 -8.50903571e-01 3.85940194e-01 -6.88363254e-01
-1.49134696e+00 6.19619787e-01 -2.86484241e-01 -8.96787226e-01
-1.65647015e-01 -5.80510437e-01 -9.40579653e-01 1.21637404e+00
-1.48907852e+00 -8.74591172e-01 1.78474322e-01 6.01473987e-01
8.24592829e-01 5.19048609e-02 7.17735350e-01 8.42963085e-02
-4.29165781e-01 4.93225515e-01 2.77769566e-03 2.91375071e-01
1.01075351e+00 -9.97791886e-01 7.42821395e-01 9.36262131e-01
-3.06255013e-01 1.07294810e+00 6.02903485e-01 -6.66549444e-01
-7.37233102e-01 -1.13387120e+00 1.67133737e+00 -1.73609838e-01
7.47736156e-01 -3.29768628e-01 -1.12365699e+00 6.47858620e-01
4.19183344e-01 -5.96854508e-01 6.98575497e-01 4.04658109e-01
-3.06808233e-01 -1.38426155e-01 -8.04260969e-01 8.90873134e-01
9.53572333e-01 -5.72506547e-01 -1.20563340e+00 1.58643454e-01
9.13436890e-01 -3.06450017e-03 -6.17082179e-01 3.65682691e-01
8.56042728e-02 -9.61150408e-01 7.72813678e-01 -7.55684316e-01
7.49574661e-01 4.63201590e-02 3.42847593e-02 -1.84954369e+00
-6.49420440e-01 -8.97420570e-02 3.53704214e-01 1.49864054e+00
6.57911122e-01 -7.09769130e-01 3.65150392e-01 6.00773871e-01
-3.91301155e-01 -6.14934742e-01 -4.45792735e-01 -6.28994226e-01
4.67648596e-01 -4.73874658e-01 5.44149339e-01 9.51064408e-01
1.79124832e-01 1.09781682e+00 1.74242854e-01 -2.41617620e-01
-4.28691320e-02 7.79511780e-02 2.54708797e-01 -1.12322259e+00
-1.43046781e-01 -7.41862178e-01 2.84692168e-01 -1.07336211e+00
3.39463115e-01 -9.84181464e-01 1.20396286e-01 -1.71853340e+00
2.22820550e-01 -1.34036884e-01 -3.33707899e-01 5.84688365e-01
-5.55427670e-01 -1.22003071e-01 3.19785327e-01 -8.11437294e-02
-1.63004890e-01 4.77538437e-01 1.34018660e+00 -1.72061548e-01
-3.26416165e-01 -9.65028703e-02 -1.33785939e+00 8.36866498e-01
1.02446663e+00 -5.23129165e-01 -4.64427680e-01 -8.18256080e-01
3.46887410e-01 -3.58277112e-01 5.24373576e-02 -6.81038439e-01
8.80979151e-02 -2.79871020e-02 5.12838364e-01 -3.83920908e-01
2.03399926e-01 -4.70182329e-01 -5.63158035e-01 6.87398970e-01
-5.24416149e-01 2.51249880e-01 5.21182418e-01 1.02584317e-01
-4.34176296e-01 -3.95813495e-01 9.45942283e-01 -4.43923473e-01
-7.39152908e-01 -1.88779098e-03 -4.83001232e-01 3.44484985e-01
5.89064300e-01 -2.00743720e-01 -3.86020035e-01 -3.94468725e-01
-5.04103482e-01 1.42165139e-01 1.82967335e-01 8.16290498e-01
6.86083615e-01 -1.10259521e+00 -9.01716471e-01 2.90744126e-01
-7.00098425e-02 -8.12178031e-02 2.55497903e-01 6.32694006e-01
1.11940242e-01 3.72439921e-01 -2.64685005e-01 -8.69899169e-02
-1.21082854e+00 4.59047496e-01 1.01717286e-01 -4.41156685e-01
-2.17490360e-01 1.12864876e+00 6.37205899e-01 -4.25020278e-01
1.92319211e-02 -5.00197947e-01 -5.28290927e-01 5.29393435e-01
6.37698472e-01 -5.19714095e-02 3.22285593e-01 -5.77010453e-01
-4.29802179e-01 2.88862795e-01 -5.77963948e-01 -2.45814119e-02
1.31504381e+00 -2.51192272e-01 -3.34202558e-01 4.96306419e-01
1.30200195e+00 -1.18900754e-01 -6.35752738e-01 -3.16866577e-01
1.12383284e-01 -1.25271425e-01 6.97169006e-02 -6.39110863e-01
-8.63049626e-01 1.12728298e+00 1.30809575e-01 3.40180136e-02
1.23709631e+00 -6.73270002e-02 1.07997584e+00 3.56083840e-01
8.44799802e-02 -1.41947722e+00 -1.42018227e-02 1.31317997e+00
9.34022903e-01 -1.13635039e+00 -2.64718831e-01 -8.83902088e-02
-9.41286564e-01 1.02783346e+00 8.19403112e-01 -7.39926696e-02
4.35109735e-01 2.30499789e-01 1.16407029e-01 -7.17922449e-02
-8.16381872e-01 -1.56048730e-01 7.27935657e-02 3.71616691e-01
8.12149346e-01 -1.54332891e-01 -7.96413600e-01 8.75797510e-01
-4.82559562e-01 -3.10762286e-01 3.74289155e-01 7.46191978e-01
-5.89011967e-01 -1.15944254e+00 -3.80833715e-01 5.41763842e-01
-4.78216201e-01 -7.80759990e-01 -6.02558792e-01 5.60923159e-01
1.81550801e-01 9.82429385e-01 2.40615997e-02 -3.71076196e-01
3.26779962e-01 2.01953411e-01 1.00083545e-01 -1.06385338e+00
-1.05949259e+00 -7.05702007e-02 7.92612582e-02 -1.50170311e-01
-1.27125785e-01 -6.38333559e-01 -1.63347805e+00 -4.58365023e-01
-1.82201520e-01 1.81133687e-01 4.50723529e-01 1.33808744e+00
3.87168944e-01 6.45776093e-01 5.27838886e-01 -6.45255089e-01
-7.43767738e-01 -1.25967026e+00 -5.31886458e-01 6.01307571e-01
7.42092580e-02 -3.14029604e-01 -2.46045515e-01 -4.43110764e-02] | [11.084193229675293, 9.089661598205566] |
0899687e-6c17-4c4f-b83a-6ba2c84d6d51 | behavior-retrieval-few-shot-imitation | 2304.08742 | null | https://arxiv.org/abs/2304.08742v2 | https://arxiv.org/pdf/2304.08742v2.pdf | Behavior Retrieval: Few-Shot Imitation Learning by Querying Unlabeled Datasets | Enabling robots to learn novel visuomotor skills in a data-efficient manner remains an unsolved problem with myriad challenges. A popular paradigm for tackling this problem is through leveraging large unlabeled datasets that have many behaviors in them and then adapting a policy to a specific task using a small amount of task-specific human supervision (i.e. interventions or demonstrations). However, how best to leverage the narrow task-specific supervision and balance it with offline data remains an open question. Our key insight in this work is that task-specific data not only provides new data for an agent to train on but can also inform the type of prior data the agent should use for learning. Concretely, we propose a simple approach that uses a small amount of downstream expert data to selectively query relevant behaviors from an offline, unlabeled dataset (including many sub-optimal behaviors). The agent is then jointly trained on the expert and queried data. We observe that our method learns to query only the relevant transitions to the task, filtering out sub-optimal or task-irrelevant data. By doing so, it is able to learn more effectively from the mix of task-specific and offline data compared to naively mixing the data or only using the task-specific data. Furthermore, we find that our simple querying approach outperforms more complex goal-conditioned methods by 20% across simulated and real robotic manipulation tasks from images. See https://sites.google.com/view/behaviorretrieval for videos and code. | ['Chelsea Finn', 'Dorsa Sadigh', 'Suraj Nair', 'Maximilian Du'] | 2023-04-18 | null | null | null | null | ['few-shot-imitation-learning', 'open-question'] | ['methodology', 'natural-language-processing'] | [ 2.87533879e-01 2.25481749e-01 -3.79765898e-01 -2.88269579e-01
-9.47513938e-01 -9.56876993e-01 5.24232030e-01 -4.77220640e-02
-8.39767158e-01 9.50238883e-01 8.83295983e-02 -1.05777629e-01
-2.37489238e-01 -3.41993272e-01 -1.11671114e+00 -8.04696500e-01
-8.90595280e-03 9.15780365e-01 3.98259491e-01 -2.15393096e-01
1.41233891e-01 4.85333055e-01 -1.70362163e+00 -2.24389389e-01
9.27534342e-01 6.84833109e-01 7.92923152e-01 6.54976070e-01
4.71597314e-01 7.50824690e-01 -2.24646971e-01 3.57771158e-01
5.51326752e-01 -3.14120770e-01 -8.42640877e-01 1.56667665e-01
4.19188321e-01 -6.69235289e-01 -4.41752821e-01 1.00839663e+00
3.85603547e-01 6.57154083e-01 5.07329881e-01 -1.19562626e+00
-3.05811524e-01 5.35835743e-01 -9.79568511e-02 1.99019432e-01
2.76722074e-01 9.11626995e-01 7.81873226e-01 -4.61176246e-01
8.61582339e-01 1.09202111e+00 -4.88191135e-02 8.19299161e-01
-1.32685268e+00 -6.55849695e-01 5.52514791e-01 2.13937342e-01
-7.32794821e-01 -4.41883504e-01 5.52994430e-01 -7.36394167e-01
8.35818648e-01 -2.48249575e-01 7.24680662e-01 1.44348657e+00
-1.09045625e-01 8.66022646e-01 1.14335597e+00 -5.49700074e-02
4.45963234e-01 -7.77112618e-02 1.46646857e-01 8.10789406e-01
1.07160702e-01 5.88078916e-01 -4.86750871e-01 4.19492610e-02
8.12615871e-01 2.08555803e-01 -3.58024806e-01 -9.20235813e-01
-1.28315949e+00 5.56401253e-01 3.47028226e-01 -6.21047430e-02
-5.31340837e-01 2.77963936e-01 2.20128804e-01 6.67458355e-01
-1.08770020e-01 8.58562768e-01 -7.94525206e-01 -3.78432214e-01
-2.35628441e-01 5.21431923e-01 7.68015921e-01 1.17935836e+00
1.14230907e+00 -5.90418354e-02 -1.98033258e-01 6.10817730e-01
-1.33597637e-02 4.65951473e-01 4.15941358e-01 -1.53776681e+00
6.16709352e-01 4.67387676e-01 5.26877582e-01 -2.29771286e-01
-3.35501552e-01 1.61567479e-02 1.03028417e-01 5.82969427e-01
7.36481965e-01 -4.90607381e-01 -1.30186486e+00 2.22665858e+00
6.37099802e-01 -1.10559575e-01 2.60262247e-02 1.14497781e+00
1.55057594e-01 3.96675795e-01 1.53050661e-01 3.12420558e-02
1.02242637e+00 -1.21563458e+00 -4.18159276e-01 -6.12182140e-01
5.47341764e-01 -2.32187107e-01 1.25790405e+00 4.23542798e-01
-8.63939464e-01 -5.23096800e-01 -8.02845716e-01 -2.42889926e-01
-2.61888623e-01 1.14592053e-02 5.78800857e-01 -1.88761055e-01
-9.69466686e-01 6.65084839e-01 -1.20873308e+00 -6.60761535e-01
4.07984316e-01 6.00502849e-01 -5.96404910e-01 -2.64611721e-01
-8.41257513e-01 1.06166470e+00 5.70653379e-01 -2.71120310e-01
-1.99558353e+00 -4.72574979e-01 -9.78912354e-01 -7.07377419e-02
1.29855204e+00 -6.17514908e-01 1.74084473e+00 -8.77474308e-01
-1.79601872e+00 4.84889209e-01 7.76680466e-03 -3.45462322e-01
3.96139205e-01 -4.87575948e-01 3.49686444e-01 3.25938761e-01
2.64087468e-01 8.62046182e-01 8.26453090e-01 -1.27886379e+00
-9.22180951e-01 -6.12557232e-01 6.22258484e-01 5.07728815e-01
-1.04409285e-01 -3.70595694e-01 -5.80883324e-01 -2.87855119e-01
-1.01573095e-01 -1.42147124e+00 -3.33785087e-01 1.38236582e-01
-1.56805933e-01 -4.25595641e-01 6.78539693e-01 -4.49625552e-01
4.33531910e-01 -2.00986767e+00 7.04039633e-01 -6.07641302e-02
2.17279971e-01 1.23355441e-01 -3.32536280e-01 5.29821277e-01
2.18716189e-01 -4.30521250e-01 -2.88965911e-01 -1.72223553e-01
4.40980382e-02 4.37760890e-01 -2.22402111e-01 4.01661396e-01
1.16839387e-01 8.13918769e-01 -1.39734232e+00 -2.24330679e-01
3.48443836e-01 -1.44022668e-03 -8.01727116e-01 4.78384167e-01
-7.57165074e-01 1.14081872e+00 -9.25028682e-01 4.79189694e-01
-7.95028433e-02 -1.38950914e-01 2.83189356e-01 9.18565169e-02
-4.33940105e-02 3.00512344e-01 -9.09930587e-01 1.97795117e+00
-5.00462770e-01 4.99347299e-01 3.91254187e-01 -1.26247466e+00
4.90682900e-01 1.93440810e-01 6.02912962e-01 -3.87248278e-01
1.73835114e-01 1.24111228e-01 2.44753867e-01 -1.05818796e+00
1.16679609e-01 5.91337308e-02 -1.15797006e-01 5.11441827e-01
4.83663082e-01 -2.44043469e-01 3.16474587e-01 1.65473759e-01
1.48254061e+00 7.83744872e-01 2.23372087e-01 3.91756184e-02
2.25337986e-02 5.81294119e-01 5.53556085e-01 9.10852194e-01
-5.14082670e-01 8.03366378e-02 3.34341168e-01 -8.51174369e-02
-1.04766726e+00 -1.03658056e+00 3.88403773e-01 1.41189158e+00
2.08667651e-01 -1.68768410e-03 -6.32189691e-01 -8.29244256e-01
2.31120989e-01 6.29184067e-01 -7.00600803e-01 -3.75703156e-01
-6.98430181e-01 4.44558710e-02 5.42524308e-02 4.99157667e-01
2.77612388e-01 -1.43402612e+00 -9.09515560e-01 1.05014302e-01
1.14233624e-02 -1.10784948e+00 -4.76753294e-01 7.16203213e-01
-8.99767101e-01 -1.25894189e+00 -5.50757527e-01 -8.50762844e-01
9.35965538e-01 2.75082916e-01 7.97841251e-01 -3.17308575e-01
-2.60029593e-03 8.28632176e-01 -2.85608321e-01 -2.03340873e-01
-3.32400769e-01 7.00692236e-02 1.79385409e-01 -3.41319442e-01
2.30446756e-01 -6.40972972e-01 -5.25520802e-01 1.11669227e-01
-6.42403901e-01 2.12724376e-02 8.40720594e-01 9.60294366e-01
5.55563390e-01 -2.98790425e-01 4.93892193e-01 -8.91920030e-01
4.89593595e-01 -6.19801641e-01 -8.62151682e-01 6.18962571e-02
-4.87462848e-01 4.57256734e-01 7.56148994e-01 -8.48227024e-01
-1.02340722e+00 6.19153619e-01 3.96926343e-01 -7.85075426e-01
-4.93594199e-01 3.56142342e-01 -8.57957155e-02 1.95032075e-01
6.34633064e-01 1.31221324e-01 2.32161433e-01 -5.42592466e-01
3.54014367e-01 3.30498695e-01 6.37324095e-01 -1.11759675e+00
7.26715446e-01 4.00042564e-01 -2.81617790e-01 -4.20264661e-01
-8.39761078e-01 -5.15710294e-01 -5.57662487e-01 -2.86924958e-01
9.52904999e-01 -8.09946775e-01 -9.56961930e-01 2.72982597e-01
-7.75622427e-01 -1.24026084e+00 -4.93440509e-01 7.70136833e-01
-1.03428805e+00 -8.68967995e-02 -5.25604367e-01 -6.53461039e-01
3.36596906e-01 -1.55876338e+00 1.03401637e+00 3.03033262e-01
-9.67029668e-03 -6.36098921e-01 1.44756734e-01 4.03034717e-01
4.69586849e-02 3.34661417e-02 7.26795197e-01 -8.42766166e-01
-8.41209948e-01 -5.47850505e-02 9.74915549e-02 2.25379214e-01
3.23281020e-01 -4.62570399e-01 -7.86247432e-01 -5.50450027e-01
-5.09900134e-03 -9.45755124e-01 7.69772470e-01 1.63116172e-01
1.08511066e+00 -2.04683289e-01 -4.65571940e-01 3.72030079e-01
1.14882100e+00 3.10335606e-01 -2.74576545e-02 3.04288536e-01
6.09069407e-01 7.80382633e-01 1.00739050e+00 2.21025333e-01
5.68633735e-01 6.11871779e-01 6.09018564e-01 4.04113680e-01
9.61532444e-02 -5.00520885e-01 6.19900346e-01 2.27617145e-01
-3.76996323e-02 2.97258627e-02 -7.61402726e-01 6.74784243e-01
-2.20251036e+00 -8.00904036e-01 6.11982882e-01 2.19157052e+00
1.07377410e+00 7.88304880e-02 2.69062728e-01 -4.16533142e-01
5.17925560e-01 -2.33932346e-01 -1.41446483e+00 1.01734109e-01
4.64435667e-01 1.20923683e-01 6.21964693e-01 5.59433460e-01
-9.87847626e-01 1.16180098e+00 5.40715790e+00 3.70939970e-01
-1.07684898e+00 -5.98921953e-03 3.50604616e-02 -3.19774956e-01
5.44515364e-02 2.20036283e-01 -7.02967167e-01 3.36017609e-01
7.62762427e-01 -1.12447664e-01 1.06192231e+00 1.00424838e+00
4.67126399e-01 -4.85529929e-01 -1.64287686e+00 5.92832088e-01
-2.49901414e-01 -7.11758077e-01 -2.58867204e-01 1.52083039e-01
6.07155681e-01 4.96362478e-01 -8.69738385e-02 7.33357072e-01
1.10773098e+00 -6.83131814e-01 7.34123170e-01 4.42767560e-01
4.86885577e-01 -1.15361236e-01 1.28330037e-01 8.47036541e-01
-7.53094077e-01 -5.25249302e-01 -1.19443856e-01 -2.23457143e-01
-5.34189381e-02 -7.63622969e-02 -8.28764200e-01 1.80270955e-01
7.90159702e-01 7.45569706e-01 -3.19460005e-01 8.87856901e-01
-5.05013287e-01 5.51226377e-01 -5.25712729e-01 1.47678116e-02
3.69028986e-01 -3.53198230e-01 6.07295811e-01 4.75569218e-01
1.91112161e-01 2.67820865e-01 6.48176610e-01 7.85229921e-01
2.07375363e-02 -4.03333455e-01 -9.80178654e-01 -1.73957601e-01
5.81688404e-01 1.11685801e+00 -3.61874998e-01 -5.13738692e-01
-3.41415495e-01 8.28947365e-01 8.53073001e-01 7.22967386e-01
-6.05651557e-01 3.77815776e-02 6.96465492e-01 -2.91372567e-01
4.28849012e-01 -4.92066026e-01 1.95149615e-01 -1.13281012e+00
1.10744976e-01 -1.07444370e+00 3.91862422e-01 -8.16308141e-01
-1.01492834e+00 1.42970502e-01 2.96124279e-01 -1.30480576e+00
-4.73169923e-01 -6.83883607e-01 -2.57415831e-01 5.90820193e-01
-1.54866970e+00 -7.52595305e-01 -3.82904053e-01 7.73238540e-01
8.48974407e-01 1.89827047e-02 5.63722670e-01 -2.79427953e-02
-5.59290648e-01 1.17240198e-01 -1.66035313e-02 5.19030914e-03
8.52017343e-01 -1.39076722e+00 -1.44976199e-01 5.22478163e-01
-9.11757872e-02 6.66554451e-01 8.38938594e-01 -6.16383076e-01
-1.78064680e+00 -8.31478715e-01 1.37761030e-02 -6.63352609e-01
7.58279324e-01 -2.54906744e-01 -7.83933163e-01 1.20340359e+00
7.45953768e-02 1.32940650e-01 1.44586071e-01 -1.72537044e-01
-1.22359641e-01 -3.30625102e-02 -1.04836881e+00 8.86196136e-01
1.36795843e+00 -3.86259198e-01 -8.13792050e-01 4.31178719e-01
9.11141872e-01 -5.27913749e-01 -6.50275230e-01 2.65771151e-01
3.85094076e-01 -5.76366842e-01 6.22720361e-01 -1.01485550e+00
3.09192181e-01 -3.26141477e-01 -6.73052855e-03 -1.64862895e+00
-4.64795008e-02 -7.69578397e-01 -1.36132464e-01 7.00762451e-01
4.39525127e-01 -8.14007401e-01 7.45899200e-01 7.90279210e-01
-3.79870951e-01 -6.95762157e-01 -5.46899915e-01 -8.43570113e-01
1.53221503e-01 -2.39854082e-01 2.59513557e-01 7.15411782e-01
1.52635187e-01 4.06115174e-01 -1.76778182e-01 2.72158653e-01
5.14915109e-01 2.66305566e-01 1.15795743e+00 -9.76645470e-01
-3.93713653e-01 -1.15448326e-01 6.45090342e-02 -1.48807251e+00
4.71839756e-01 -8.66985857e-01 6.04229629e-01 -1.43393350e+00
1.88551679e-01 -6.11049652e-01 -2.46814311e-01 9.20977414e-01
-2.08117455e-01 -4.16365266e-01 2.18778372e-01 2.74364263e-01
-7.21162260e-01 6.27644062e-01 1.73052979e+00 -9.60186124e-02
-4.18486357e-01 -3.35523561e-02 -4.76779103e-01 6.79237485e-01
9.19979811e-01 -5.26827872e-01 -7.69568741e-01 -6.02688432e-01
-1.72101960e-01 3.06611478e-01 4.37869489e-01 -9.70408440e-01
4.43083167e-01 -6.08871400e-01 7.27785900e-02 -6.22855164e-02
5.02764523e-01 -1.06300271e+00 -2.91837901e-01 4.52702194e-01
-7.69139886e-01 -1.32722348e-01 1.09145418e-01 9.14472818e-01
1.44941166e-01 -2.76211858e-01 6.01671576e-01 -4.26494867e-01
-1.01279521e+00 4.31952834e-01 -4.74267185e-01 2.74156094e-01
1.31121147e+00 3.99651825e-02 -4.01356369e-01 -4.33129817e-01
-1.10526538e+00 7.91013360e-01 7.12186933e-01 5.18153548e-01
2.48967335e-01 -8.57376695e-01 -2.70462841e-01 3.59258130e-02
3.01892638e-01 3.65343571e-01 -3.36040556e-02 7.68688142e-01
6.07366487e-02 2.83044517e-01 -4.54428285e-01 -5.91067314e-01
-8.11702013e-01 7.27163970e-01 3.05427939e-01 -4.43324223e-02
-6.75550699e-01 5.98048687e-01 3.97014588e-01 -7.87496805e-01
4.80363548e-01 -4.98080283e-01 -9.68709067e-02 -2.79602855e-01
6.72242697e-03 2.54236758e-01 -2.27345288e-01 -2.65982181e-01
-1.09428741e-01 3.58315647e-01 -1.04148924e-01 -3.85595351e-01
1.37537324e+00 -3.52055281e-02 3.25781554e-01 5.51765800e-01
9.30790544e-01 -2.99109131e-01 -2.17331815e+00 -3.79487842e-01
6.27064779e-02 -3.25943649e-01 -1.89699382e-01 -9.33570027e-01
-9.49244976e-01 6.28812730e-01 3.10607761e-01 -1.77386478e-02
8.76989245e-01 2.89756954e-01 4.36901480e-01 1.04108155e+00
8.01453829e-01 -1.34853351e+00 4.13880527e-01 4.55195308e-01
8.65931213e-01 -1.50472808e+00 -1.70683801e-01 -4.25926223e-02
-7.31527507e-01 7.59181499e-01 1.13033760e+00 -4.25805479e-01
3.87510806e-01 1.14156313e-01 4.09247214e-03 -2.86504149e-01
-1.05726540e+00 -6.06126845e-01 -1.42438829e-01 7.32707679e-01
-2.51156092e-01 -6.35150746e-02 1.58691421e-01 3.75732958e-01
-8.57569501e-02 2.09921524e-01 3.88277382e-01 1.41427457e+00
-5.72840691e-01 -1.03297210e+00 -1.62923515e-01 6.50681496e-01
2.39526313e-02 3.41553241e-01 -3.05418253e-01 7.93843210e-01
-4.55007069e-02 8.04946125e-01 -1.86055198e-01 -2.12216601e-01
3.68489057e-01 1.13925874e-01 6.58828795e-01 -1.11588109e+00
-2.95510650e-01 -4.60751355e-02 5.93334287e-02 -9.73404646e-01
-3.34851027e-01 -8.46643865e-01 -1.48260462e+00 3.04926932e-01
-8.52511376e-02 1.23749822e-02 4.85228181e-01 1.01729488e+00
4.49565172e-01 5.85867465e-01 4.93222356e-01 -1.23151243e+00
-1.03937232e+00 -9.66607869e-01 -2.05146164e-01 5.04562259e-01
7.65907228e-01 -1.17518771e+00 -4.85435724e-01 2.32057363e-01] | [4.320559024810791, 1.2315562963485718] |
c675e770-70dd-4158-a808-81a324866828 | view-consistent-metal-segmentation-in-the | 2112.02101 | null | https://arxiv.org/abs/2112.02101v1 | https://arxiv.org/pdf/2112.02101v1.pdf | View-Consistent Metal Segmentation in the Projection Domain for Metal Artifact Reduction in CBCT -- An Investigation of Potential Improvement | The positive outcome of a trauma intervention depends on an intraoperative evaluation of inserted metallic implants. Due to occurring metal artifacts, the quality of this evaluation heavily depends on the performance of so-called Metal Artifact Reduction methods (MAR). The majority of these MAR methods require prior segmentation of the inserted metal objects. Therefore, typically a rather simple thresholding-based segmentation method in the reconstructed 3D volume is applied, despite some major disadvantages. With this publication, the potential of shifting the segmentation task to a learning-based, view-consistent 2D projection-based method on the downstream MAR's outcome is investigated. For segmenting the present metal, a rather simple learning-based 2D projection-wise segmentation network that is trained using real data acquired during cadaver studies, is examined. To overcome the disadvantages that come along with a 2D projection-wise segmentation, a Consistency Filter is proposed. The influence of the shifted segmentation domain is investigated by comparing the results of the standard fsMAR with a modified fsMAR version using the new segmentation masks. With a quantitative and qualitative evaluation on real cadaver data, the investigated approach showed an increased MAR performance and a high insensitivity against metal artifacts. For cases with metal outside the reconstruction's FoV or cases with vanishing metal, a significant reduction in artifacts could be shown. Thus, increases of up to roughly 3 dB w.r.t. the mean PSNR metric over all slices and up to 9 dB for single slices were achieved. The shown results reveal a beneficial influence of the shift to a 2D-based segmentation method on real data for downstream use with a MAR method, like the fsMAR. | ['Björn W. Kreher', 'Florian Kordon', 'Andreas Maier', 'Tristan M. Gottschalk'] | 2021-12-03 | null | null | null | null | ['metal-artifact-reduction'] | ['medical'] | [ 1.73618838e-01 2.67322689e-01 4.96720970e-01 -7.17710704e-02
-8.58364403e-01 -1.46931127e-01 2.35286802e-01 3.98500293e-01
-8.36051047e-01 6.00638509e-01 -7.98465125e-03 -1.56548470e-01
-4.30336267e-01 -6.29563093e-01 -7.16991842e-01 -9.46778834e-01
-9.56577882e-02 3.34022671e-01 5.27688622e-01 -3.24355587e-02
1.69849679e-01 7.90568888e-01 -1.31912518e+00 2.98641145e-01
5.58911860e-01 1.07354736e+00 4.78405356e-01 1.06799439e-01
-8.85106549e-02 1.96428508e-01 -4.85160321e-01 -3.46010119e-01
6.07654333e-01 -3.82596403e-01 -3.80806983e-01 9.63623300e-02
2.04007521e-01 -1.59869552e-01 2.37056181e-01 1.00737202e+00
7.29703605e-01 -2.24331543e-01 4.81782764e-01 -3.91423315e-01
1.92933828e-01 7.61343181e-01 -4.37021613e-01 -8.39212611e-02
2.86201388e-01 1.99667737e-01 1.24789454e-01 -8.76439571e-01
9.45981503e-01 5.77853322e-01 8.21140885e-01 2.00662836e-01
-1.53684723e+00 -5.42281568e-01 -3.07062715e-01 8.28809962e-02
-1.31265163e+00 5.21110445e-02 9.99476671e-01 -8.53694141e-01
4.87452954e-01 4.20171350e-01 9.90418136e-01 5.35527647e-01
6.27687991e-01 3.27596106e-02 1.57391024e+00 -5.50141692e-01
3.76621664e-01 3.83745253e-01 2.43373192e-03 3.77417088e-01
7.02711642e-01 1.26964957e-01 2.36610565e-02 1.05198324e-01
5.85843205e-01 -1.98594451e-01 -6.31214201e-01 -7.26941109e-01
-9.72753584e-01 3.75627458e-01 4.36501116e-01 8.89594913e-01
-6.79657817e-01 -1.99158993e-02 6.04896009e-01 4.15138267e-02
2.79596478e-01 6.68542325e-01 -1.20555788e-01 1.30327225e-01
-1.14169335e+00 -2.04214901e-01 5.30328751e-01 4.31576818e-01
5.11579692e-01 -4.97244857e-02 -1.03330776e-01 5.42772949e-01
2.86764950e-01 2.58160770e-01 6.23908162e-01 -5.09143114e-01
-4.40981202e-02 6.18804574e-01 -1.96828961e-01 -8.99954200e-01
-7.47334421e-01 -6.96627617e-01 -5.56392252e-01 8.31780493e-01
6.32910371e-01 -8.46035033e-02 -1.05177903e+00 1.32517898e+00
4.65326667e-01 -4.88956541e-01 -1.39985487e-01 1.18679607e+00
5.42968929e-01 2.13070840e-01 -3.71936150e-02 -6.75771654e-01
1.36758649e+00 -3.09546769e-01 -8.17946494e-01 4.91473414e-02
6.47554040e-01 -9.52126265e-01 1.05362427e+00 1.02452052e+00
-1.25540388e+00 -3.58803451e-01 -1.28698826e+00 5.19076228e-01
-4.24109995e-02 1.95118502e-01 2.82426681e-02 9.18273568e-01
-9.11454260e-01 8.13627958e-01 -9.56200957e-01 -2.14939252e-01
2.35164717e-01 4.89289075e-01 -5.49311697e-01 -1.58344023e-02
-8.11098754e-01 1.31669211e+00 2.11562008e-01 4.03353542e-01
-4.05240715e-01 -7.63514996e-01 -5.06760120e-01 -5.79923205e-02
2.48390332e-01 -3.64576042e-01 7.72373617e-01 -9.75946426e-01
-1.59740806e+00 9.33241427e-01 4.34411556e-01 -5.36638856e-01
1.13334858e+00 -1.61266416e-01 -1.25602290e-01 3.56957436e-01
-8.19517374e-02 1.11257518e-02 5.89119256e-01 -1.59481144e+00
-3.80151086e-02 -5.95671594e-01 -2.07033366e-01 -1.47630394e-01
-9.16264057e-02 -1.83527708e-01 -3.06815892e-01 -7.17992485e-01
5.28647363e-01 -1.02837992e+00 -4.18625712e-01 1.71682268e-01
-1.31399602e-01 6.13377094e-01 3.52600217e-01 -9.61280942e-01
1.14137876e+00 -2.16946268e+00 -8.64941925e-02 6.09187305e-01
4.55325618e-02 3.14647332e-02 2.93605477e-01 2.05141395e-01
-2.80172795e-01 -2.75794238e-01 -4.91014957e-01 -1.04641050e-01
-5.75897574e-01 -1.66578412e-01 4.94802356e-01 8.70369792e-01
-2.99277633e-01 1.88011721e-01 -5.05353272e-01 -4.25302744e-01
3.76941711e-01 6.87949121e-01 -4.80218768e-01 9.33148935e-02
2.84875989e-01 6.70110703e-01 -1.82028994e-01 1.67793587e-01
9.06555355e-01 2.24990472e-01 3.58278215e-01 -5.97744882e-01
-4.39450413e-01 -3.65625694e-02 -1.15658224e+00 1.86790848e+00
-8.50051343e-01 3.77644300e-01 3.94599855e-01 -7.42105067e-01
1.13133705e+00 6.20021045e-01 9.04630482e-01 -9.38421071e-01
3.96779150e-01 8.68169308e-01 4.02933419e-01 -5.42183995e-01
-4.61095572e-02 -6.50481701e-01 4.94032145e-01 -9.92841739e-03
-8.97733197e-02 -2.05803588e-01 3.03200129e-02 -2.14048892e-01
7.99382031e-01 2.18074828e-01 1.69497997e-01 -7.34200001e-01
6.64771914e-01 4.67566624e-02 2.53653139e-01 2.66266227e-01
-3.22472909e-03 9.88167822e-01 4.57008213e-01 -3.05936247e-01
-6.98056042e-01 -8.98257613e-01 -5.62364876e-01 -8.98504779e-02
1.80750370e-01 1.60823495e-03 -9.42074895e-01 -4.54947650e-01
-2.49313444e-01 6.96191013e-01 -5.51533699e-01 -4.06720579e-01
-7.62734294e-01 -6.75482869e-01 -1.23002082e-02 1.05147492e-02
2.25256592e-01 -8.53662133e-01 -1.33074617e+00 4.79307622e-01
-2.52656415e-02 -9.00732219e-01 1.46417394e-01 5.52100480e-01
-1.38280749e+00 -9.22910273e-01 -1.03713715e+00 -3.86735678e-01
6.66844904e-01 -2.34387770e-01 7.73695230e-01 7.09625036e-02
-4.57305908e-01 3.40576530e-01 -4.08016950e-01 -2.48663917e-01
-6.94028795e-01 -3.93074036e-01 -2.03502104e-01 7.39537552e-02
-3.42631042e-01 -5.82225025e-01 -8.17796290e-01 2.52275229e-01
-1.01546168e+00 -3.38591188e-02 8.52167308e-01 6.84396863e-01
6.48160160e-01 -3.25859368e-01 3.66622061e-01 -8.92439961e-01
2.41049334e-01 1.20790675e-01 -7.57720590e-01 -9.84722599e-02
-9.32211816e-01 1.31655201e-01 5.04455984e-01 -3.46472085e-01
-1.03342652e+00 3.17470938e-01 -3.67855579e-01 -4.72541988e-01
-2.01453835e-01 4.71385270e-01 -5.59290685e-02 -3.19237590e-01
9.22898173e-01 -9.15630832e-02 3.32081079e-01 -4.26307440e-01
-2.27914646e-01 2.63292551e-01 1.58580914e-01 -2.17510507e-01
5.29230773e-01 5.65331817e-01 2.40596399e-01 -6.07817054e-01
-5.71375825e-02 -4.16133702e-01 -6.19645298e-01 -7.22262025e-01
9.70255911e-01 -4.00799572e-01 -2.75998116e-01 3.07936132e-01
-1.08733511e+00 -2.46700332e-01 -5.58293641e-01 9.71815646e-01
-6.04170740e-01 5.18611670e-01 -3.77675265e-01 -7.11623967e-01
-5.04679859e-01 -1.74887323e+00 5.91879964e-01 -1.09741099e-01
-1.91022918e-01 -6.56575859e-01 2.69195493e-02 2.83215404e-01
4.07603800e-01 5.73199749e-01 8.85657072e-01 -6.07295692e-01
-3.96886379e-01 -5.11117578e-01 2.79014319e-01 5.41501939e-01
1.83855459e-01 -2.31618255e-01 -8.33786011e-01 -1.35167062e-01
5.80585182e-01 4.36932862e-01 4.86880571e-01 6.46245360e-01
8.22430074e-01 1.14978977e-01 -7.07895234e-02 4.24173415e-01
1.90391028e+00 4.38841105e-01 7.84756839e-01 5.14076889e-01
2.10556164e-01 6.81361675e-01 7.67144859e-01 3.68458003e-01
-5.31397045e-01 8.81970048e-01 6.87043607e-01 -1.68014362e-01
-4.63122666e-01 2.84456611e-01 1.76285684e-01 6.11326635e-01
-2.95356303e-01 2.40622282e-01 -8.15825284e-01 4.75822181e-01
-1.13243592e+00 -4.67259228e-01 -5.46484053e-01 2.78747606e+00
5.57144940e-01 4.25619334e-01 -1.63255215e-01 5.36421597e-01
6.65367424e-01 -1.96037427e-01 1.26240693e-03 -5.17692089e-01
1.05647871e-03 3.44204694e-01 5.49768567e-01 4.40427899e-01
-6.20222867e-01 6.83955029e-02 5.05765295e+00 7.35668063e-01
-1.66214693e+00 2.70901769e-01 2.58483350e-01 -1.33698642e-01
-1.93221018e-01 -8.34076107e-02 -1.71742260e-01 5.79169154e-01
7.11326838e-01 1.83418244e-01 -7.22688884e-02 7.00404882e-01
5.88614166e-01 -8.40463936e-01 -7.23362267e-01 9.59845781e-01
1.40256556e-02 -1.18067825e+00 -1.60249487e-01 1.63756251e-01
3.63228798e-01 -3.74077469e-01 -2.24320650e-01 -2.97230542e-01
-8.01152706e-01 -5.37315249e-01 9.19224203e-01 6.03775144e-01
6.85642779e-01 -6.17354095e-01 1.09393311e+00 -1.76092926e-02
-7.53693938e-01 7.52926096e-02 -3.43929268e-02 3.08372647e-01
4.19705778e-01 1.03236020e+00 -8.90269458e-01 6.92893982e-01
6.44644380e-01 8.17927718e-02 -4.71502930e-01 1.46790946e+00
-7.68756866e-02 4.72207218e-01 -3.95253748e-01 2.21805245e-01
-1.32496115e-02 -4.20838088e-01 8.01862299e-01 1.21165133e+00
5.71382940e-01 5.79193421e-03 -4.43056673e-01 7.48408556e-01
4.02167439e-01 5.99899292e-01 -3.61389220e-01 5.75671911e-01
-1.51271552e-01 1.46908200e+00 -1.18485343e+00 -1.30975291e-01
-3.53842258e-01 8.00080061e-01 -4.00205851e-01 4.31150496e-02
-7.25105166e-01 -1.09893177e-02 -1.89460535e-02 1.04636228e+00
1.43358305e-01 -4.03480604e-02 -8.12433839e-01 -4.97756690e-01
4.29840535e-01 -6.13989651e-01 2.37652749e-01 -8.01501334e-01
-6.87108397e-01 7.58088648e-01 6.14306107e-02 -1.55859077e+00
8.54053125e-02 -6.43390119e-01 -4.97131675e-01 8.75970304e-01
-9.66556966e-01 -6.67669296e-01 -2.82139421e-01 2.88382232e-01
2.41852179e-01 3.67189556e-01 3.70871246e-01 4.79793280e-01
-5.40305115e-02 2.68142700e-01 -5.53011708e-02 -2.34742060e-01
6.85400724e-01 -1.02622890e+00 -6.90857768e-01 8.67017150e-01
-2.05342025e-01 3.67892116e-01 9.96663451e-01 -7.81698763e-01
-1.09251285e+00 -3.85832697e-01 4.03148562e-01 7.24677891e-02
2.95407563e-01 3.12587284e-02 -9.54368770e-01 2.34901264e-01
1.29299685e-01 -7.43512660e-02 4.69799548e-01 -3.06744963e-01
1.92476094e-01 -3.11021835e-01 -1.44067240e+00 5.01256347e-01
5.83915532e-01 6.43308535e-02 -5.75509071e-01 -1.88867807e-01
1.83657482e-01 -3.89061570e-01 -1.18892646e+00 7.98869014e-01
7.42441893e-01 -1.40196323e+00 7.87849963e-01 1.11710444e-01
3.77645850e-01 -2.34746262e-01 2.29655534e-01 -1.14210892e+00
-3.61091085e-02 -1.50113344e-01 5.59577942e-01 7.79981613e-01
5.89117706e-01 -6.06629133e-01 7.06607580e-01 6.16210580e-01
-6.08180523e-01 -7.51247883e-01 -1.40209794e+00 -7.33861983e-01
3.21861394e-02 -4.20161694e-01 -2.54106581e-01 6.63681865e-01
2.19727792e-02 -3.09774280e-01 1.86447531e-01 1.85390875e-01
4.17183608e-01 -6.73221648e-02 4.50456321e-01 -1.27540874e+00
-3.11815411e-01 -4.18135136e-01 -8.05371225e-01 -1.76537573e-01
-6.32541537e-01 -8.41427326e-01 1.39948487e-01 -1.61565304e+00
-1.80563658e-01 -4.07196224e-01 -3.77301127e-01 -1.62621271e-02
1.45491287e-01 3.54314536e-01 3.21674436e-01 1.06871523e-01
3.63451481e-01 8.44083428e-02 1.49733067e+00 1.44419357e-01
-3.45308810e-01 1.72211707e-01 -3.25764306e-02 7.80066550e-01
4.69678581e-01 -6.88543379e-01 -1.20689921e-01 -1.72380432e-01
2.70878762e-01 2.28114486e-01 3.09476614e-01 -1.48916602e+00
1.28877655e-01 4.46333915e-01 2.56462663e-01 -5.53074419e-01
3.39579523e-01 -1.45291245e+00 7.42318332e-01 1.14819181e+00
-6.61794394e-02 -1.45008773e-01 3.41571629e-01 1.32683381e-01
-2.52726644e-01 -5.28916419e-01 1.25164592e+00 -2.31614307e-01
-2.92488635e-01 -2.91531444e-01 -4.70427811e-01 -4.56832886e-01
1.06069374e+00 -7.82661140e-01 5.68395436e-01 -1.11850046e-01
-1.37318897e+00 -6.23328865e-01 4.34123486e-01 -1.46182850e-01
5.54231584e-01 -8.54883015e-01 -5.42589247e-01 8.96770135e-02
-8.81421939e-02 -6.97561279e-02 6.95355713e-01 1.79505408e+00
-9.10842896e-01 2.08496843e-02 -4.39775020e-01 -6.93485022e-01
-1.25483835e+00 7.40245938e-01 7.82006800e-01 -4.34807003e-01
-8.05115640e-01 5.10304630e-01 1.50318116e-01 9.89397839e-02
-3.28720175e-02 -5.44167459e-01 -1.54515281e-01 2.44193301e-01
1.04576759e-02 4.73323375e-01 8.65832984e-01 -5.01165628e-01
-3.38497996e-01 7.70143330e-01 2.25216806e-01 -2.70608425e-01
1.17931783e+00 -9.52524766e-02 -1.36572467e-02 4.63355422e-01
9.23122168e-01 5.21024704e-01 -1.00500250e+00 2.74179846e-01
1.88276116e-02 -4.41111356e-01 2.90741235e-01 -1.01923645e+00
-1.46805835e+00 8.05518806e-01 1.18733680e+00 -9.08478647e-02
1.38578331e+00 -1.44660473e-01 3.38406950e-01 -2.60147303e-01
7.42410362e-01 -9.35698152e-01 -2.62860864e-01 -2.95436829e-01
1.15525270e+00 -7.27454126e-01 2.01822028e-01 -7.49869168e-01
-4.18648869e-01 1.36995018e+00 2.70511478e-01 -1.82725579e-01
7.61034608e-01 5.65731585e-01 3.86660874e-01 -1.40150487e-01
1.34001717e-01 -1.11682639e-01 1.98320597e-01 2.84630716e-01
5.20936549e-01 -6.57953769e-02 -1.37609994e+00 6.25807703e-01
-3.37250195e-02 1.97846845e-01 7.17331529e-01 8.81317854e-01
-2.49083087e-01 -9.19812024e-01 -6.81157291e-01 2.57368326e-01
-5.83982170e-01 7.83750340e-02 1.50231794e-02 1.13629460e+00
2.67410010e-01 6.17768466e-01 -1.37979567e-01 -1.22055225e-01
8.05921257e-01 5.63323684e-02 5.70381165e-01 -4.84037429e-01
-1.23819554e+00 4.18182939e-01 -6.64140880e-02 -7.14429379e-01
-3.75782073e-01 -7.52045393e-01 -1.32934749e+00 3.73960316e-01
-4.74910408e-01 9.72495452e-02 1.19454575e+00 7.57315755e-01
-5.49928471e-02 6.82397485e-01 4.20576900e-01 -7.64130473e-01
-2.79319286e-01 -8.49915862e-01 -6.77719533e-01 4.02482599e-01
2.91383117e-02 -6.93092644e-01 -4.42583054e-01 -8.84410441e-02] | [13.608577728271484, -2.700577735900879] |
1f509ae2-e329-48b3-a242-2e4b7bf63d68 | game-theoretic-mixed-experts-for | 2211.14669 | null | https://arxiv.org/abs/2211.14669v2 | https://arxiv.org/pdf/2211.14669v2.pdf | Game Theoretic Mixed Experts for Combinational Adversarial Machine Learning | Recent advances in adversarial machine learning have shown that defenses considered to be robust are actually susceptible to adversarial attacks which are specifically customized to target their weaknesses. These defenses include Barrage of Random Transforms (BaRT), Friendly Adversarial Training (FAT), Trash is Treasure (TiT) and ensemble models made up of Vision Transformers (ViTs), Big Transfer models and Spiking Neural Networks (SNNs). We first conduct a transferability analysis, to demonstrate the adversarial examples generated by customized attacks on one defense, are not often misclassified by another defense. This finding leads to two important questions. First, how can the low transferability between defenses be utilized in a game theoretic framework to improve the robustness? Second, how can an adversary within this framework develop effective multi-model attacks? In this paper, we provide a game-theoretic framework for ensemble adversarial attacks and defenses. Our framework is called Game theoretic Mixed Experts (GaME). It is designed to find the Mixed-Nash strategy for both a detector based and standard defender, when facing an attacker employing compositional adversarial attacks. We further propose three new attack algorithms, specifically designed to target defenses with randomized transformations, multi-model voting schemes, and adversarial detector architectures. These attacks serve to both strengthen defenses generated by the GaME framework and verify their robustness against unforeseen attacks. Overall, our framework and analyses advance the field of adversarial machine learning by yielding new insights into compositional attack and defense formulations. | ['Marten van Dijk', 'Caiwen Ding', 'Sohaib Ahmad', 'Kaleel Mahmood', 'Ethan Rathbun'] | 2022-11-26 | null | null | null | null | ['adversarial-defense'] | ['adversarial'] | [ 4.92451400e-01 2.32034802e-01 4.42182958e-01 1.81805402e-01
-6.43212616e-01 -1.56698632e+00 8.96595597e-01 -4.35470074e-01
-1.81270003e-01 4.96393055e-01 -3.76392692e-01 -7.34757841e-01
-4.09582406e-01 -1.27318883e+00 -6.82652652e-01 -8.69971573e-01
-3.30169499e-01 2.82964110e-01 3.13357174e-01 -8.15630734e-01
2.33038530e-01 6.84314013e-01 -9.26701367e-01 1.20758928e-01
7.61906862e-01 7.08578348e-01 -6.73472464e-01 8.66832256e-01
4.18206900e-01 9.64252830e-01 -9.45977926e-01 -1.11289847e+00
8.72088552e-01 -4.58415419e-01 -6.06045067e-01 -4.72585976e-01
2.74359044e-02 -4.09793049e-01 -5.40955782e-01 1.47007668e+00
5.86759508e-01 -2.03228340e-01 7.04282701e-01 -1.77734733e+00
-7.62759566e-01 8.36129785e-01 -3.10513437e-01 -1.80760995e-02
-2.67670918e-02 5.50272822e-01 7.49183834e-01 -1.35939732e-01
2.94305801e-01 1.39173520e+00 5.80837488e-01 1.31900942e+00
-1.08202064e+00 -1.13787580e+00 1.12774394e-01 -1.52850986e-01
-1.02038109e+00 -1.44569516e-01 7.23737001e-01 -3.01090539e-01
6.30019128e-01 7.09192336e-01 3.17383945e-01 1.75933719e+00
4.83255327e-01 3.72271568e-01 1.30922091e+00 -1.94760770e-01
5.58984518e-01 -2.03217170e-03 -1.53485477e-01 5.07803082e-01
4.32184160e-01 9.99241650e-01 1.98931918e-02 -5.90154707e-01
8.26674819e-01 -2.94445217e-01 -1.90707266e-01 -3.97828341e-01
-4.82633889e-01 1.24866796e+00 4.93021458e-01 1.45320848e-01
-1.38148293e-01 3.06186795e-01 5.44102788e-01 8.24991226e-01
6.76183924e-02 1.02120602e+00 -3.61552179e-01 3.75293046e-01
-3.15437526e-01 3.89411539e-01 1.02962995e+00 4.21689659e-01
1.55364364e-01 6.26323521e-01 4.05312143e-02 2.44188651e-01
1.95420355e-01 6.06767416e-01 6.02391697e-02 -1.03132415e+00
1.81811243e-01 1.33054525e-01 -1.06135696e-01 -8.89411390e-01
-1.83610126e-01 -5.42237818e-01 -8.20484400e-01 9.20398414e-01
4.99229074e-01 -5.06045461e-01 -7.93011546e-01 2.20477176e+00
1.20556712e-01 2.01489270e-01 4.10622030e-01 4.35433686e-01
2.76170224e-01 2.19654605e-01 1.27894834e-01 1.02224715e-01
1.09403968e+00 -4.62451518e-01 -9.95172486e-02 -2.47811407e-01
3.64263237e-01 -2.14048848e-01 6.31276727e-01 4.31228131e-01
-9.67000842e-01 -1.29676402e-01 -1.22313237e+00 8.25090528e-01
-5.93804657e-01 -8.77719045e-01 4.37399715e-01 1.70789015e+00
-1.04071009e+00 6.60694182e-01 -6.54830158e-01 -1.27180398e-01
5.57428002e-01 7.07954645e-01 -2.30636746e-01 3.84239048e-01
-1.70414484e+00 1.09613657e+00 2.22541988e-01 -3.28525454e-01
-1.77953148e+00 -5.53997815e-01 -6.32740676e-01 -2.79980898e-02
4.02049035e-01 -8.96493077e-01 9.45528090e-01 -1.37418091e+00
-1.51874506e+00 9.07333076e-01 8.29009116e-01 -1.01122570e+00
8.17199528e-01 1.87979594e-01 -3.43342662e-01 3.41277011e-02
-2.95747399e-01 8.53171125e-02 1.10308611e+00 -1.30306756e+00
-2.14938790e-01 -4.75795567e-01 9.87286747e-01 -2.46624425e-02
-3.93742889e-01 4.49135274e-01 5.19853175e-01 -9.47659552e-01
-3.78182143e-01 -1.06077981e+00 -5.55656672e-01 -1.05202921e-01
-6.07984185e-01 3.29742938e-01 8.89753938e-01 -1.75323859e-01
7.09742904e-01 -1.87667751e+00 2.74519920e-01 4.67817754e-01
4.75871652e-01 7.26595104e-01 -4.48786408e-01 4.92718279e-01
-2.60236293e-01 6.01391494e-01 -3.19289774e-01 2.62383014e-01
2.64840871e-01 3.44215453e-01 -8.93119991e-01 4.33464497e-01
2.43864968e-01 1.04184580e+00 -8.06234300e-01 1.30427510e-01
1.14116617e-01 1.34821802e-01 -7.13997722e-01 4.09403294e-01
4.48940769e-02 5.67185819e-01 -6.36387348e-01 6.55208528e-01
7.28748441e-01 3.43261331e-01 1.38750032e-01 9.74018648e-02
2.86703199e-01 -2.23660633e-01 -9.37768281e-01 7.02630460e-01
-1.28279909e-01 1.94903359e-01 4.59297836e-01 -1.24635339e+00
7.65149474e-01 1.65538594e-01 2.40984987e-02 -3.76563191e-01
4.87900674e-01 1.95161760e-01 3.89330894e-01 -5.64598255e-02
-1.22796550e-01 -3.40739340e-01 -5.69417596e-01 5.84878683e-01
1.66002810e-01 -3.24314803e-01 -3.18779230e-01 3.19416910e-01
1.69016385e+00 -2.17916504e-01 3.08483988e-01 -3.03248942e-01
6.07810616e-01 -2.27993190e-01 5.33365369e-01 1.29229283e+00
-5.80661774e-01 1.19190738e-01 8.71428549e-01 -2.48309940e-01
-9.29376543e-01 -1.53876877e+00 3.82740527e-01 1.11879814e+00
1.33894920e-01 -1.24694027e-01 -1.08079064e+00 -1.00897014e+00
-1.67548805e-01 5.91243625e-01 -8.22031081e-01 -8.13066304e-01
-5.04458249e-01 -8.76710057e-01 1.60096526e+00 4.60698158e-01
6.47571087e-01 -9.53276575e-01 -5.23123562e-01 -2.61322912e-02
3.03186297e-01 -7.98421919e-01 1.02396468e-02 5.54898918e-01
-4.14026588e-01 -1.49983048e+00 -3.39708954e-01 -3.67486715e-01
4.19126421e-01 -4.20425013e-02 9.95311081e-01 8.84275660e-02
-5.26457764e-02 6.36547804e-01 -4.65338618e-01 -8.35158288e-01
-9.76373136e-01 -3.43537062e-01 4.49670285e-01 -1.47507444e-01
-2.22833436e-02 -9.79292154e-01 -2.41158828e-01 4.49684888e-01
-1.20076180e+00 -5.24308920e-01 4.59270716e-01 9.83013272e-01
-6.70551360e-02 9.42001864e-02 5.51099837e-01 -1.00056779e+00
1.00307155e+00 -6.10369563e-01 -8.20776999e-01 4.11892474e-01
-2.84654200e-01 2.38597635e-02 1.22658968e+00 -7.14567244e-01
-6.79950416e-01 -3.89953673e-01 -2.87904888e-01 -5.69022536e-01
1.96875110e-02 -5.21948859e-02 -6.15711272e-01 -9.97546852e-01
1.22304225e+00 2.03820229e-01 -1.50882557e-01 1.29596666e-01
5.22305012e-01 2.01275513e-01 3.86863232e-01 -9.94181633e-01
1.66668522e+00 4.47330803e-01 2.15494394e-01 -2.88487107e-01
-3.99773210e-01 5.32890856e-01 -2.73793250e-01 -2.01390207e-01
7.39850104e-01 -3.82143706e-01 -9.82200861e-01 8.56199205e-01
-9.48763371e-01 -4.22486454e-01 -1.96326837e-01 -1.54810786e-01
-6.87173963e-01 4.31772560e-01 -6.91280067e-01 -1.10318351e+00
-4.52089101e-01 -1.14296603e+00 3.03459555e-01 2.38725901e-01
1.53221175e-01 -9.84309971e-01 1.50308490e-01 1.69638142e-01
6.75423563e-01 8.52582037e-01 9.05816674e-01 -1.42794073e+00
-4.04535085e-01 -2.62166589e-01 2.05186576e-01 5.79091251e-01
-1.22264124e-01 1.21931501e-01 -1.07650971e+00 -4.78588015e-01
3.94754469e-01 -6.12228096e-01 6.12993956e-01 7.68544823e-02
1.03129065e+00 -6.73042059e-01 -6.66251481e-02 8.69023860e-01
1.48360097e+00 5.92499554e-01 7.22501576e-01 4.77211952e-01
5.09746671e-01 7.21600294e-01 3.78739610e-02 1.84383959e-01
-1.40542999e-01 3.65348935e-01 1.17345750e+00 9.19220001e-02
5.10253429e-01 -1.04434893e-01 6.92664027e-01 1.80262849e-01
-2.18660995e-01 -4.59631234e-01 -8.68617654e-01 -4.26667510e-03
-1.50573123e+00 -1.36497974e+00 1.90299034e-01 2.07754564e+00
4.15214300e-01 5.32347322e-01 2.08419248e-01 2.29334682e-01
7.09936023e-01 1.35003239e-01 -7.93398499e-01 -8.98830414e-01
-4.38036740e-01 8.27037811e-01 8.33750129e-01 3.34193200e-01
-1.37822318e+00 1.04273283e+00 6.80524111e+00 1.06675339e+00
-8.39749396e-01 2.32628420e-01 5.80007315e-01 4.11774777e-02
-3.64151001e-01 2.05081448e-01 -2.49487981e-01 2.19173104e-01
1.08692658e+00 -5.61294138e-01 8.30755174e-01 8.64733517e-01
-1.78677395e-01 6.18961751e-01 -9.68386471e-01 4.48752552e-01
-1.39629215e-01 -1.23412883e+00 2.51612037e-01 2.80788541e-01
5.63092411e-01 -1.39218569e-01 6.46505177e-01 7.29398847e-01
1.41805542e+00 -1.30992198e+00 6.13754094e-01 -6.17643185e-02
4.20438915e-01 -9.82961595e-01 4.90219921e-01 4.19428200e-01
-9.63509679e-01 -5.11045992e-01 -3.34752709e-01 -1.06773749e-01
-3.04492891e-01 -1.91442534e-01 -1.77282751e-01 7.41305709e-01
3.30080479e-01 -5.30329123e-02 -4.38524038e-01 4.02292252e-01
-3.56720746e-01 5.67972004e-01 -7.78359920e-02 1.21432133e-01
5.64972639e-01 7.75885060e-02 9.45405662e-01 6.09110534e-01
-2.66253240e-02 2.95788378e-01 2.08708756e-02 1.01274502e+00
-5.60325049e-02 -4.21664953e-01 -9.63751018e-01 -1.36020372e-03
5.91594636e-01 1.21918976e+00 -5.80567122e-01 6.63364753e-02
1.17488075e-02 9.09624279e-01 7.62423426e-02 2.26314664e-01
-1.29334927e+00 -4.54302818e-01 1.18654716e+00 -2.79907674e-01
-1.01572774e-01 6.35979325e-02 -2.89489090e-01 -1.21146131e+00
-4.10855651e-01 -1.58901989e+00 5.38098335e-01 -3.60520750e-01
-1.69372642e+00 6.94761574e-01 -1.34737357e-01 -1.24249184e+00
-3.09478402e-01 -8.69461536e-01 -1.28545928e+00 7.64149427e-01
-8.55809212e-01 -1.50252533e+00 2.51197904e-01 1.06321847e+00
1.42614186e-01 -6.07324600e-01 1.03221714e+00 -5.12966029e-02
-5.59792697e-01 1.06885350e+00 1.01377673e-01 5.85549414e-01
2.79518217e-01 -1.22137344e+00 7.80186772e-01 1.34312832e+00
8.72494578e-02 7.23246872e-01 8.30766976e-01 -4.50684845e-01
-1.27683711e+00 -1.18567371e+00 -2.15942532e-01 -8.13687563e-01
1.02181256e+00 -4.96596783e-01 -6.16108775e-01 5.90077341e-01
-7.35160932e-02 -5.64989150e-02 6.99436247e-01 -2.50886321e-01
-9.05032277e-01 1.28563255e-01 -1.72843862e+00 9.42994773e-01
1.11609817e+00 -7.78080463e-01 -4.94868159e-01 1.51572078e-01
6.78525865e-01 -1.50360674e-01 -5.36919773e-01 4.57839757e-01
5.96927226e-01 -1.18951643e+00 1.35545135e+00 -1.14218521e+00
2.64493406e-01 1.13389837e-02 -3.72764140e-01 -1.41055894e+00
-1.78351060e-01 -9.01440978e-01 -3.20479982e-02 1.04639506e+00
3.18364985e-02 -1.10889208e+00 6.15090787e-01 6.75630748e-01
5.81036173e-02 -5.42412698e-01 -1.08317029e+00 -1.17384088e+00
9.58006620e-01 -4.53871727e-01 5.55565655e-01 1.01175773e+00
-2.24259168e-01 -2.08416551e-01 -5.53942740e-01 4.08645481e-01
1.00201869e+00 -5.35904467e-01 8.02698314e-01 -1.14634073e+00
-5.51234961e-01 -7.21219122e-01 -7.21393108e-01 -1.28104359e-01
3.29407692e-01 -7.80005753e-01 -4.45589691e-01 -6.95723295e-01
-1.01979263e-01 -2.97641814e-01 -4.72900182e-01 5.29657722e-01
-3.34106386e-02 3.59608948e-01 5.33751369e-01 1.05233513e-01
-1.85711637e-01 1.10332690e-01 7.72173166e-01 -3.36951971e-01
2.11374372e-01 1.98906898e-01 -1.22157884e+00 1.00311053e+00
9.51375902e-01 -6.62211180e-01 -4.93307680e-01 -5.70818409e-02
1.91923663e-01 -1.15403369e-01 7.88882554e-01 -1.05932665e+00
9.75917503e-02 -2.82924563e-01 1.67876124e-01 3.57690752e-01
2.62100309e-01 -6.47995651e-01 2.55571693e-01 1.07826078e+00
-3.74125987e-01 4.93736891e-03 1.90619811e-01 5.90927064e-01
2.72402793e-01 -1.30376801e-01 1.22667265e+00 -3.94392252e-01
-3.87637228e-01 3.82035285e-01 -5.75157225e-01 3.36648852e-01
1.43248868e+00 -1.19189695e-01 -6.57153130e-01 -5.09438038e-01
-8.40067744e-01 3.79941948e-02 5.51105618e-01 2.09696025e-01
4.10911024e-01 -1.07624364e+00 -8.45578551e-01 1.68578997e-01
-2.12290004e-01 -7.99185932e-01 2.37431347e-01 3.15424412e-01
-3.87016833e-01 -4.36472781e-02 -7.43174136e-01 1.15040325e-01
-1.27791977e+00 8.66964281e-01 8.88346851e-01 -6.57831311e-01
6.12547509e-02 1.18237364e+00 6.97504461e-01 -5.62541366e-01
-3.42370979e-02 3.68106365e-01 -1.10752031e-01 -2.20576763e-01
4.19215590e-01 3.27605575e-01 -2.48427778e-01 -4.96445358e-01
-5.06862164e-01 3.35380405e-01 5.49065880e-02 -1.92395881e-01
9.54068840e-01 4.00208205e-01 -1.18966296e-01 -3.21857363e-01
7.46358514e-01 8.59300196e-02 -9.21975911e-01 2.00059742e-01
-2.24001572e-01 -2.27445066e-01 -4.73577499e-01 -8.99500251e-01
-1.02250874e+00 9.05923486e-01 5.33718884e-01 8.07920396e-01
1.25634944e+00 -3.93314272e-01 4.24275577e-01 3.18976551e-01
6.85421228e-01 -5.74035764e-01 1.87237665e-01 6.61347508e-01
7.84373581e-01 -6.69822037e-01 -6.36958599e-01 -1.98349103e-01
-5.61427832e-01 9.61540401e-01 7.90727377e-01 -7.97366679e-01
5.50396502e-01 5.89621246e-01 4.77520935e-02 -1.06117278e-02
-6.33169830e-01 -1.89544693e-01 -9.66137424e-02 1.30068374e+00
-4.56171870e-01 1.21431001e-01 -1.45345986e-01 9.53671694e-01
-2.10474208e-01 -7.37081110e-01 6.73168719e-01 7.23549843e-01
-2.46400967e-01 -1.40335298e+00 -6.56679213e-01 1.64325684e-01
-8.31863582e-01 -1.08116239e-01 -1.01454246e+00 7.67794847e-01
2.43471190e-01 1.19068575e+00 -4.51735318e-01 -1.01867580e+00
2.34710842e-01 -2.27254212e-01 5.33192337e-01 -3.20970774e-01
-1.30986714e+00 -4.97809768e-01 -1.12546749e-01 -5.92212677e-01
-1.18428253e-01 -9.34013426e-02 -5.57658315e-01 -7.83832848e-01
-3.09796512e-01 2.16581002e-01 2.81908780e-01 8.71601164e-01
-2.98838201e-03 5.55596113e-01 1.03855610e+00 -6.92768693e-01
-1.31718493e+00 -3.68595451e-01 -5.01804233e-01 9.53213573e-02
1.12380562e-02 -6.23584926e-01 -8.28792512e-01 -4.76233989e-01] | [5.5822577476501465, 7.724758148193359] |
94f7e20c-6a68-4c47-9615-a6135b45eb62 | multiplier-bootstrap-based-exploration | 2302.01543 | null | https://arxiv.org/abs/2302.01543v1 | https://arxiv.org/pdf/2302.01543v1.pdf | Multiplier Bootstrap-based Exploration | Despite the great interest in the bandit problem, designing efficient algorithms for complex models remains challenging, as there is typically no analytical way to quantify uncertainty. In this paper, we propose Multiplier Bootstrap-based Exploration (MBE), a novel exploration strategy that is applicable to any reward model amenable to weighted loss minimization. We prove both instance-dependent and instance-independent rate-optimal regret bounds for MBE in sub-Gaussian multi-armed bandits. With extensive simulation and real data experiments, we show the generality and adaptivity of MBE. | ['Rui Song', 'Branislav Kveton', 'Haoyu Wei', 'Runzhe Wan'] | 2023-02-03 | null | null | null | null | ['multi-armed-bandits'] | ['miscellaneous'] | [-2.06561431e-01 -1.85312331e-02 -8.22395921e-01 -3.55204731e-01
-1.29905570e+00 -6.38062596e-01 2.34406039e-01 3.42608541e-02
-3.77603740e-01 1.59259152e+00 -1.38669536e-01 -8.52394640e-01
-9.93866086e-01 -7.04517484e-01 -7.84129024e-01 -6.46358073e-01
-4.96564656e-01 7.57066131e-01 -1.87421188e-01 2.04643250e-01
2.71495610e-01 6.65812671e-01 -8.46027374e-01 -9.07358006e-02
1.00781119e+00 1.49966371e+00 -4.06435765e-02 6.28011286e-01
2.10034428e-03 5.98504603e-01 -2.55982727e-01 -6.22392058e-01
4.95503604e-01 -4.03910726e-01 -8.31376970e-01 -6.63023535e-03
-2.06495836e-01 -4.72253561e-01 -1.57729879e-01 9.61065888e-01
4.65491980e-01 2.06880257e-01 4.98022974e-01 -1.27795899e+00
-2.98021257e-01 1.02945244e+00 -9.38215613e-01 4.10816431e-01
-1.07706795e-02 -3.53717178e-01 1.30885613e+00 -3.32369179e-01
1.98078364e-01 1.30669272e+00 4.93062824e-01 3.31726938e-01
-1.51065254e+00 -6.18061125e-01 3.06912273e-01 1.11339226e-01
-8.86026323e-01 -2.59200573e-01 6.93459511e-01 -7.76211843e-02
4.33602780e-01 5.19510150e-01 6.73135102e-01 7.58469939e-01
-2.68972576e-01 1.29738104e+00 1.52862573e+00 -5.03091753e-01
7.33125806e-01 1.96469784e-01 9.92261246e-02 2.89389670e-01
6.31824195e-01 4.57082361e-01 -5.14184296e-01 -5.68525314e-01
8.42250526e-01 -6.63151592e-02 -1.04340807e-01 -8.74955535e-01
-5.48685193e-01 1.16640985e+00 3.51912081e-01 -3.95587772e-01
-5.96667945e-01 5.77491701e-01 2.83841074e-01 4.06816393e-01
8.29163134e-01 2.44423792e-01 -3.56172293e-01 -4.67320979e-01
-1.28305507e+00 5.02668560e-01 9.71596539e-01 1.04998374e+00
3.68859559e-01 5.09978347e-02 -4.39822197e-01 7.97182083e-01
1.82942078e-01 5.57135642e-01 2.29679476e-02 -9.13549542e-01
7.98412919e-01 -1.38848871e-01 9.26704526e-01 -5.73106468e-01
-3.79969805e-01 -9.33990836e-01 -5.80841184e-01 1.15006067e-01
4.73635346e-01 -2.18070850e-01 -6.55376554e-01 1.76409066e+00
3.48744065e-01 -1.35894999e-01 -9.28076878e-02 9.04910982e-01
-1.05257139e-01 2.89756000e-01 -1.79924563e-01 -7.50225246e-01
6.92224085e-01 -9.78418291e-01 -5.01854420e-01 -2.82689035e-01
4.65212375e-01 -3.42072427e-01 8.28244627e-01 6.20616913e-01
-1.47563267e+00 4.15705115e-01 -7.83572197e-01 4.89153802e-01
1.91939622e-01 -3.86866421e-01 1.16561902e+00 1.16083062e+00
-3.97572100e-01 5.99192083e-01 -8.16630065e-01 1.06863938e-01
9.06878173e-01 1.96966663e-01 1.89381644e-01 -3.06135744e-01
-6.74405217e-01 6.93546414e-01 4.18338120e-01 2.31145956e-02
-8.69405091e-01 -7.46008217e-01 -3.62747848e-01 1.92123875e-01
7.31247962e-01 -5.11687398e-01 1.68700910e+00 -6.89588189e-01
-1.61162317e+00 2.71121234e-01 -2.42348269e-01 -1.15083611e+00
1.08025539e+00 -1.85173705e-01 9.70812440e-02 -7.80899171e-03
-1.49863120e-02 -6.76387399e-02 6.22171819e-01 -1.23918903e+00
-8.07810485e-01 -2.96935707e-01 2.31235534e-01 5.25905080e-02
-9.73867700e-02 5.80762550e-02 3.57781574e-02 -7.11406052e-01
1.20671071e-01 -7.81434059e-01 -5.86777806e-01 -2.40331143e-01
-4.65882838e-01 5.37978746e-02 -6.35374561e-02 -4.44741964e-01
1.42022586e+00 -1.68551517e+00 -1.40982851e-01 7.01980233e-01
-2.33745188e-01 -1.01801328e-01 4.00320813e-02 3.51174980e-01
3.28532696e-01 1.57668397e-01 -2.75133938e-01 -3.32433224e-01
3.97029668e-01 4.39906657e-01 -7.06285775e-01 5.72689712e-01
-2.43642539e-01 9.39531505e-01 -1.09430003e+00 -3.20217818e-01
-1.53667585e-03 -3.77793401e-01 -7.05951035e-01 1.79311201e-01
-4.12591994e-01 3.84611078e-02 -8.78044724e-01 9.49489594e-01
8.26100826e-01 -4.91449028e-01 3.41778547e-02 5.67699611e-01
4.68766652e-02 1.97928891e-01 -1.22330451e+00 1.19242728e+00
-6.77028060e-01 1.50478050e-01 4.19437945e-01 -1.43138123e+00
5.94166756e-01 -1.36117125e-02 2.82636613e-01 -5.90670466e-01
1.62372112e-01 5.18862605e-01 -3.02621305e-01 -7.51682669e-02
4.64632094e-01 -5.51493227e-01 -8.15507323e-02 8.14779937e-01
-3.38469118e-01 -1.28107667e-01 3.29234928e-01 3.22385551e-03
9.40617681e-01 -8.53707418e-02 4.43037182e-01 -2.74205297e-01
-1.66243285e-01 -1.39964372e-01 4.37990010e-01 1.37357903e+00
-2.78668493e-01 2.58339375e-01 5.32350838e-01 -3.50025326e-01
-8.36398721e-01 -1.15368390e+00 -4.36993271e-01 1.22503304e+00
1.39772266e-01 1.09817192e-01 -2.16646880e-01 -5.52986145e-01
6.83288932e-01 1.08105648e+00 -6.59921050e-01 2.73460120e-01
-3.44104096e-02 -1.09909666e+00 7.42296129e-02 3.48885536e-01
3.66148889e-01 -4.45604593e-01 -6.19525969e-01 6.94694936e-01
3.87490056e-02 -7.98938453e-01 -2.41385460e-01 3.75668138e-01
-1.07267261e+00 -8.29879642e-01 -8.40572059e-01 1.70822486e-01
2.62622416e-01 2.96195179e-01 1.04023135e+00 -4.08274889e-01
-1.58524856e-01 6.58872306e-01 -2.40843460e-01 -7.82051623e-01
-2.49798037e-02 3.58664207e-02 -1.35069400e-01 -6.85731322e-02
3.79256494e-02 -6.61612630e-01 -6.95624173e-01 3.22715461e-01
-5.32508016e-01 -1.90670773e-01 5.32895207e-01 9.35085595e-01
7.65372694e-01 -2.44825199e-01 9.59706903e-01 -7.67976344e-01
9.61969852e-01 -7.10060894e-01 -1.19353187e+00 5.62843084e-01
-8.90529215e-01 1.82459190e-01 1.90797329e-01 -3.67491782e-01
-9.49210584e-01 -2.34994516e-01 2.29694381e-01 -2.34754398e-01
4.11952436e-01 8.78808022e-01 3.17053169e-01 -2.26581171e-01
5.05093932e-01 1.74879760e-01 -2.11700037e-01 -5.54267704e-01
4.60293740e-01 5.62486291e-01 2.13923573e-01 -1.15107048e+00
4.93690431e-01 4.85005140e-01 2.50565767e-01 -3.74527991e-01
-1.43843269e+00 -1.83087781e-01 2.23296568e-01 2.01486517e-02
1.19088292e-01 -6.61933243e-01 -8.31324935e-01 -2.14624643e-01
-7.49026239e-01 -4.19251144e-01 -4.44240093e-01 7.87061632e-01
-1.40869606e+00 1.76947281e-01 -1.70384362e-01 -1.81530082e+00
-3.83258879e-01 -7.48070061e-01 3.65747005e-01 1.66697294e-01
4.26673964e-02 -8.37706506e-01 1.23491131e-01 3.01573724e-01
4.89213943e-01 1.14449434e-01 5.65612912e-01 -5.96680224e-01
-8.30486774e-01 -3.00175428e-01 -5.19110143e-01 2.23818496e-01
-2.26567179e-01 -4.99504715e-01 -5.24478495e-01 -4.78050053e-01
-1.28377974e-01 -6.72311127e-01 1.02989876e+00 9.45761025e-01
1.51841140e+00 -6.37558937e-01 -3.34519625e-01 7.08908200e-01
1.39165878e+00 3.15900682e-03 2.98585564e-01 8.23340118e-01
-1.46788403e-01 2.57743508e-01 9.59039807e-01 1.12260413e+00
-7.04316143e-03 4.70923364e-01 5.81111372e-01 3.51467758e-01
5.98917663e-01 -1.95417121e-01 -1.94582358e-01 -4.96727861e-02
-3.49666417e-01 -3.06080788e-01 -5.29073060e-01 8.69237840e-01
-2.26114869e+00 -1.13822055e+00 3.87057513e-01 2.57901812e+00
1.06473005e+00 2.60273010e-01 4.36350137e-01 3.80369462e-02
6.06449008e-01 -7.79036134e-02 -8.84759486e-01 -6.78620756e-01
-5.45262769e-02 3.12317163e-01 1.11654055e+00 3.47381204e-01
-9.50382531e-01 5.09212613e-01 7.63173246e+00 1.31373680e+00
-4.49084669e-01 2.34568521e-01 9.42370117e-01 -7.91591704e-01
-4.06398237e-01 -6.56089261e-02 -6.77475929e-01 3.86622518e-01
7.70291388e-01 -6.08497083e-01 9.44707274e-01 1.23532581e+00
2.46534616e-01 -2.82113105e-01 -1.01639521e+00 9.15723979e-01
-6.67233348e-01 -1.46969247e+00 -4.31232750e-01 2.37602517e-01
1.04088748e+00 5.75869679e-02 1.09101340e-01 2.34056443e-01
9.71325457e-01 -1.04718029e+00 8.21104705e-01 3.86610270e-01
6.20091498e-01 -1.26979434e+00 5.27502477e-01 4.23562527e-01
-4.92693484e-01 -5.04865468e-01 -5.29904306e-01 -6.17147274e-02
4.50500906e-01 9.05857623e-01 -6.05772972e-01 6.22414768e-01
5.72189033e-01 1.13562167e-01 2.92716891e-01 1.66102302e+00
-1.17928460e-01 9.67648089e-01 -7.65557945e-01 -3.10606211e-01
6.66706443e-01 -3.42200607e-01 5.12316346e-01 1.16259170e+00
6.41996324e-01 2.29384545e-02 1.40550867e-01 8.20121467e-01
-8.46410915e-02 6.11213259e-02 -1.46796227e-01 -1.79758117e-01
6.53494418e-01 8.94018352e-01 -5.21592736e-01 -1.01656795e-01
-2.06266627e-01 5.18666208e-01 5.40088475e-01 4.99287218e-01
-8.40600073e-01 -1.31273270e-01 6.94592357e-01 -6.48462623e-02
7.30416417e-01 -1.32977560e-01 -4.87922519e-01 -7.64611542e-01
1.38110086e-01 -5.34727395e-01 6.44207954e-01 -3.03719193e-01
-1.54756701e+00 -7.62917474e-02 3.10829639e-01 -8.05437863e-01
-5.04131377e-01 -4.33750093e-01 -4.03809309e-01 8.87508512e-01
-1.76394832e+00 -9.14046288e-01 4.39106703e-01 3.22387934e-01
1.14249654e-01 -1.95344448e-01 5.63546956e-01 -1.20444775e-01
-4.35744256e-01 7.96400964e-01 1.09999752e+00 -4.50822443e-01
1.66688755e-01 -1.34483349e+00 -1.80101946e-01 5.50940514e-01
2.10294370e-02 4.02298182e-01 1.05369973e+00 -3.94953519e-01
-1.43136132e+00 -8.23805511e-01 2.23744631e-01 9.04822350e-02
1.12982619e+00 -2.25315392e-02 -3.28984946e-01 7.77269065e-01
-3.34723860e-01 2.40415916e-01 7.80525208e-01 5.99241793e-01
-4.25692827e-01 -2.54512459e-01 -1.42167580e+00 3.70990902e-01
1.10529268e+00 -1.61286116e-01 -2.76065081e-01 4.83691871e-01
4.84016001e-01 -4.63560164e-01 -8.27840924e-01 3.83571893e-01
7.46987045e-01 -9.92980301e-01 1.01367867e+00 -9.94477391e-01
1.17264494e-01 4.17323381e-01 -4.77256715e-01 -1.38240302e+00
-1.93292037e-01 -1.05124450e+00 -7.45864034e-01 8.77190173e-01
5.20198464e-01 -8.68678212e-01 9.65613961e-01 8.45966280e-01
2.05079943e-01 -1.06163180e+00 -1.52716041e+00 -1.31443679e+00
3.80713165e-01 -7.96064079e-01 7.65552461e-01 5.48601151e-01
1.69471413e-01 -3.00599784e-01 -7.35690236e-01 -7.87098929e-02
1.01609802e+00 8.14058900e-01 6.10594749e-01 -9.60261345e-01
-7.80904472e-01 -7.21949160e-01 9.15562660e-02 -1.28602469e+00
-1.17412880e-02 -5.52468240e-01 -1.22447297e-01 -1.47333741e+00
4.45479155e-01 -9.13578153e-01 -7.51610279e-01 1.24946460e-01
1.11655720e-01 -1.28912106e-01 1.07970186e-01 -5.17556258e-02
-8.37505758e-01 7.31489480e-01 1.11010540e+00 -2.33752187e-02
-1.38875738e-01 8.05000305e-01 -8.60175133e-01 3.40623796e-01
9.37458456e-01 -6.23819888e-01 -5.15440881e-01 -1.10662945e-01
4.95611280e-01 3.65174353e-01 3.53875577e-01 -5.04610896e-01
-2.57564187e-01 -7.24142849e-01 1.75434425e-01 -8.38623941e-01
3.04507196e-01 -6.44903839e-01 -6.85594529e-02 4.60473269e-01
-5.75963140e-01 -5.11245728e-01 1.50446862e-01 1.08369327e+00
1.96753994e-01 -5.99982381e-01 7.09985077e-01 -7.16586784e-02
-1.63935665e-02 4.91926759e-01 -6.91838935e-02 3.38107169e-01
1.00560558e+00 1.19307004e-01 -1.90637067e-01 -7.62507975e-01
-6.68290317e-01 4.57812458e-01 -5.04198000e-02 -3.05097431e-01
3.08016241e-01 -1.42938435e+00 -7.57827163e-01 -3.74503434e-01
-2.69008707e-02 -1.15035236e-01 1.42312273e-01 7.78396010e-01
-3.20536457e-02 6.69313550e-01 2.35762462e-01 -2.18463093e-01
-8.20872843e-01 7.14043736e-01 1.79961190e-01 -6.75888062e-01
-1.24825083e-01 9.14783359e-01 -2.66347408e-01 -6.06138743e-02
4.90217507e-01 -2.04429641e-01 5.11504114e-01 -1.00435972e-01
6.18225038e-01 7.40384758e-01 -1.18398212e-01 2.49577329e-01
-9.23336297e-03 -1.16073556e-01 -9.28642526e-02 -4.51369792e-01
1.55889285e+00 -2.50818282e-01 3.65634561e-02 3.76941055e-01
6.97517574e-01 -2.26605922e-01 -1.42079639e+00 -4.77886260e-01
2.18267530e-01 -1.02014613e+00 3.86979520e-01 -9.08539772e-01
-6.69838846e-01 5.88273168e-01 3.55884552e-01 7.10761726e-01
8.76030505e-01 -1.45029919e-02 5.31180799e-01 6.26795888e-01
8.41018319e-01 -1.32700789e+00 -3.37323397e-01 2.16479704e-01
9.29051340e-01 -1.25583458e+00 2.73193806e-01 1.28497547e-02
-3.66764098e-01 7.66171038e-01 -2.60986425e-02 -5.16747832e-02
7.22411633e-01 6.73632994e-02 -6.78583086e-01 2.49042124e-01
-7.89534509e-01 -2.24672094e-01 6.72962442e-02 4.40825045e-01
7.12339673e-03 4.63792175e-01 -6.17438614e-01 1.02260482e+00
-2.58307844e-01 2.44189501e-01 3.01009655e-01 9.43432629e-01
-6.88621700e-01 -1.03754711e+00 -4.42956775e-01 8.29533041e-01
-6.98505521e-01 -3.47358771e-02 8.93105790e-02 7.15800166e-01
-8.18582892e-01 9.71378803e-01 -1.42212152e-01 2.26793930e-01
-6.75711408e-02 -1.86722487e-01 8.98564994e-01 -1.48821741e-01
5.06379195e-02 7.11804777e-02 3.99006307e-01 -5.25036097e-01
-3.58825058e-01 -6.53233111e-01 -8.17363441e-01 -6.21586978e-01
-5.94512880e-01 4.12736267e-01 8.63298416e-01 8.42887223e-01
2.05897495e-01 5.00055635e-03 9.30819094e-01 -6.57340586e-01
-1.55162227e+00 -8.24636579e-01 -1.02806211e+00 -3.35566364e-02
5.02178252e-01 -8.98325086e-01 -4.50094074e-01 -8.85908067e-01] | [4.513027667999268, 3.2638907432556152] |
7a18994c-5f5a-4a10-8ac3-f1a251c9800d | sentiment-analysis-for-arabic-in-social-media | 1911.05483 | null | https://arxiv.org/abs/1911.05483v1 | https://arxiv.org/pdf/1911.05483v1.pdf | Sentiment Analysis for Arabic in Social Media Network: A Systematic Mapping Study | With the expansion in tenders on the Internet and social media, Arabic Sentiment Analysis (ASA) has assumed a significant position in the field of text mining study and has since remained used to explore the sentiments of users about services, various products or topics conversed over the Internet. This mapping paper designs to comprehensively investigate the papers demographics, fertility, and directions of the ASA research domain. Furthermore, plans to analyze current ASA techniques and find movements in the research. This paper describes a systematic mapping study (SMS) of 51 primary selected studies (PSS) is handled with the approval of an evidence-based systematic method to ensure handling of all related papers. The analyzed results showed the increase of both the ASA research area and numbers of publications per year since 2015. Three main research facets were found, i.e. validation, solution, and evaluation research, with solution research becoming more treatment than another research type. Therefore numerous contribution facets were singled out. In totality, the general demographics of the ASA research field were highlighted and discussed | ['Mohamed Elhag M. Abo', 'Ram Gopal Raj', 'Atika Qazi', 'Abubakar Zakari'] | 2019-10-26 | null | null | null | null | ['arabic-sentiment-analysis'] | ['natural-language-processing'] | [-1.91548541e-01 1.80838704e-01 -7.61096060e-01 -1.03897057e-01
-1.46660671e-01 -6.97665811e-01 4.81950194e-01 3.28685164e-01
-3.17267179e-01 3.66973877e-01 3.31010580e-01 -3.06637496e-01
-3.69021803e-01 -5.00349820e-01 -2.85831213e-01 -4.71923172e-01
2.49873653e-01 2.01214343e-01 -1.72851011e-01 -7.55959034e-01
1.15197599e+00 2.54596502e-01 -1.55380750e+00 4.42886233e-01
6.68407679e-01 7.89073765e-01 3.57285351e-01 -7.21125752e-02
-5.64786971e-01 5.29884279e-01 -8.32091570e-01 -8.28438640e-01
-1.18109927e-01 -2.28587359e-01 -7.99253762e-01 1.92224860e-01
-2.14748323e-01 3.60568643e-01 4.72710341e-01 9.94148672e-01
5.26668668e-01 -3.11969399e-01 1.82642415e-01 -1.45306957e+00
-6.81333721e-01 7.63608336e-01 -9.55010891e-01 2.42158622e-01
6.56205535e-01 -7.06409097e-01 9.91995037e-01 -1.17403162e+00
1.24493122e+00 1.20946181e+00 4.60559547e-01 -2.22457334e-01
-3.98267567e-01 -8.48372638e-01 -3.81304957e-02 1.72509879e-01
-1.07413805e+00 -1.33576706e-01 7.89760053e-01 -4.22947019e-01
9.64812040e-01 2.70449728e-01 8.26126456e-01 7.06745207e-01
4.67271805e-01 4.19654876e-01 1.78323543e+00 -7.11871624e-01
7.50716254e-02 9.92916286e-01 2.86133647e-01 5.80773987e-02
7.36280799e-01 -7.35398889e-01 -9.66260552e-01 -5.74342385e-02
1.58152077e-02 -2.12211907e-01 1.79197997e-01 1.81276232e-01
-8.20171833e-01 8.89438033e-01 -8.51444453e-02 6.76132917e-01
-7.41823912e-01 -8.79826128e-01 7.66632080e-01 5.69963157e-01
7.49504685e-01 3.98925662e-01 -8.08409750e-01 -6.83131695e-01
-7.75181830e-01 1.94140762e-01 9.85551536e-01 9.35152650e-01
1.14587024e-01 -2.05821797e-01 3.10255706e-01 8.81542325e-01
5.65908194e-01 4.19427365e-01 4.77169603e-01 -6.19232059e-01
5.07254004e-01 1.13691139e+00 5.58281764e-02 -1.62796330e+00
-3.74912232e-01 -5.66920161e-01 -3.74066353e-01 -2.22471789e-01
-1.03712536e-01 -3.84693325e-01 -1.98182449e-01 1.00598204e+00
1.54569447e-01 -1.02000904e+00 -2.48842314e-02 6.06662393e-01
1.13310409e+00 7.67292321e-01 -3.66996927e-03 -6.16857052e-01
1.68929017e+00 -5.58159947e-01 -1.33533287e+00 -1.36715129e-01
5.04750967e-01 -1.35529292e+00 8.58980954e-01 9.20990109e-01
-1.41161966e+00 -2.44197175e-01 -1.06772935e+00 3.52892756e-01
-9.91219282e-01 1.61251053e-01 6.40562236e-01 1.06538999e+00
-9.81823802e-01 2.86100805e-03 -3.19627494e-01 -8.63106191e-01
2.99353629e-01 2.64338136e-01 -9.67662409e-02 1.14243038e-01
-1.05106223e+00 1.29584801e+00 1.13011263e-01 2.44775385e-01
2.94837326e-01 -4.49971348e-01 -3.61614615e-01 -5.47153354e-01
2.49315679e-01 -1.22032985e-01 6.82177365e-01 -1.02003264e+00
-1.27988064e+00 9.89014864e-01 -4.20044392e-01 -7.44751543e-02
-8.23279768e-02 -1.86808541e-01 -8.96557808e-01 1.49649769e-01
5.75264156e-01 1.71500817e-01 1.50005773e-01 -1.11520624e+00
-9.93466258e-01 -6.36629045e-01 1.69974975e-02 2.28915125e-01
-7.99484789e-01 1.10290253e+00 -3.54718924e-01 -7.29240835e-01
5.01883268e-01 -8.61059725e-01 -3.87681238e-02 -7.74729431e-01
-2.83984747e-02 -4.08600122e-01 7.41262615e-01 -8.06081116e-01
1.73191106e+00 -1.89164591e+00 -1.23821132e-01 3.95541966e-01
-1.60876736e-01 -2.10823640e-01 5.28761804e-01 1.23847747e+00
1.09623857e-01 5.54488301e-01 3.13149303e-01 2.15085134e-01
3.05753890e-02 4.76793535e-02 -1.13261983e-01 6.14121020e-01
-1.21881440e-01 5.46251059e-01 -6.67172074e-01 -4.60220128e-01
-2.45791271e-01 2.35268444e-01 1.55518904e-01 -5.38134098e-01
4.31502640e-01 -6.42375648e-02 -6.15672886e-01 1.11575103e+00
8.79707277e-01 -8.82210582e-02 3.78566921e-01 -3.36851537e-01
-9.42384720e-01 5.53112626e-01 -1.13816321e+00 1.27592623e+00
-5.42381823e-01 7.30946958e-01 5.02606273e-01 -8.01656485e-01
1.22900510e+00 3.15845460e-01 7.03842282e-01 -9.35432196e-01
2.93097764e-01 6.09855473e-01 9.70552266e-02 -5.24275482e-01
7.44553566e-01 1.64748788e-01 -1.75043553e-01 4.09708351e-01
-1.70511886e-01 4.88208830e-02 3.06087196e-01 3.01016301e-01
2.87114024e-01 1.16574906e-01 3.61064881e-01 -7.78861761e-01
4.51506793e-01 4.24057305e-01 3.83373350e-01 9.00075287e-02
8.44823103e-03 1.11390511e-02 6.90897524e-01 2.73733269e-02
-6.82280779e-01 -6.74512088e-02 -5.73234618e-01 9.57340121e-01
1.60147995e-01 -7.83444107e-01 -6.24862313e-01 -2.40653157e-01
-2.42838517e-01 4.55431938e-01 -6.43461525e-01 3.92280757e-01
-8.09517801e-02 -1.06297445e+00 -8.01338702e-02 1.97440740e-02
6.75100803e-01 -1.16797268e+00 -8.94878328e-01 2.25764364e-02
-8.17555338e-02 -9.59651113e-01 4.72386569e-01 1.80519640e-01
-9.41129267e-01 -8.63436341e-01 -7.21963346e-01 -9.38953936e-01
4.24558878e-01 4.10731077e-01 9.19544280e-01 -2.22338840e-01
1.06437184e-01 1.95945248e-01 -9.34974790e-01 -1.00071824e+00
-5.18638752e-02 3.86113226e-01 -1.89533774e-02 -4.21351522e-01
9.49792385e-01 -1.20528169e-01 -2.55109608e-01 5.10148764e-01
-5.86990654e-01 -4.96046871e-01 6.99490488e-01 3.16848248e-01
1.33122355e-01 5.95208228e-01 1.09327316e+00 -9.02959228e-01
1.20660746e+00 -9.89369571e-01 -3.03159691e-02 -9.37865376e-02
-1.35956728e+00 -8.53597224e-01 -3.24049920e-01 3.08946788e-01
-1.15226400e+00 -7.29064226e-01 1.04113258e-01 6.28394842e-01
7.47768432e-02 1.36804199e+00 -1.82696700e-01 -2.42132209e-02
5.07505298e-01 -4.58741486e-01 3.96030784e-01 -2.72651464e-01
-7.39610046e-02 1.35056174e+00 -2.65786976e-01 -2.97168374e-01
3.37971658e-01 3.67938757e-01 -3.66629213e-01 -1.02096152e+00
-5.37552238e-01 -7.17536032e-01 -3.82912517e-01 -6.08432531e-01
5.46520054e-01 -8.65572870e-01 -6.22291028e-01 2.52137631e-01
-7.43898511e-01 3.86321485e-01 1.18613638e-01 5.09138584e-01
2.66274750e-01 5.34509718e-02 -4.99010444e-01 -1.01368010e+00
-3.90479505e-01 -1.22187912e+00 3.33025604e-01 2.66451031e-01
-9.42895651e-01 -7.48950064e-01 -1.99362949e-01 6.83497071e-01
3.34079534e-01 2.68228978e-01 5.91357708e-01 -4.55386013e-01
3.86988312e-01 -2.77125925e-01 -1.16439372e-01 -4.06933427e-02
3.74154896e-01 3.05185497e-01 -5.31883299e-01 -1.81935385e-01
3.97403479e-01 2.21300703e-02 2.20333748e-02 5.03675878e-01
6.32029951e-01 -3.92714441e-01 -4.07021761e-01 -4.05892760e-01
1.35342944e+00 7.67226875e-01 7.52376258e-01 1.53264678e+00
-2.38590036e-02 1.38940001e+00 1.44150615e+00 6.01888180e-01
4.28457379e-01 5.10451913e-01 3.49772304e-01 1.32647142e-01
4.80880350e-01 1.97320923e-01 6.22152984e-01 1.23518443e+00
-3.78704250e-01 3.66213582e-02 -9.96153355e-01 7.09598541e-01
-1.38140428e+00 -6.76200449e-01 -8.35291386e-01 1.65530670e+00
4.89781499e-01 4.79805171e-01 5.00781000e-01 3.62803131e-01
5.49936771e-01 -3.95300016e-02 7.45506510e-02 -1.10838068e+00
-5.60429394e-01 2.90041298e-01 7.79212594e-01 -1.09360099e-01
-5.52311242e-01 6.04055822e-01 6.53217268e+00 5.15708387e-01
-8.75733674e-01 3.31714004e-02 6.38465703e-01 1.16474241e-01
-5.77530026e-01 1.27326429e-01 -7.39006519e-01 5.09424269e-01
9.36039686e-01 -1.75939500e-01 -2.00370356e-01 7.48148084e-01
5.85763216e-01 -4.44361478e-01 7.30037391e-02 6.57548845e-01
2.78627183e-02 -1.30620897e+00 -2.32158676e-01 3.49590242e-01
8.10900211e-01 -1.07669428e-01 4.64726090e-01 -2.36787871e-02
-1.09414235e-02 -6.58595562e-01 8.55093658e-01 7.31800869e-02
3.51076514e-01 -7.84502506e-01 1.43506968e+00 -5.64651005e-02
-5.68680346e-01 -2.56743610e-01 -1.89253449e-01 -5.54070652e-01
2.97174156e-01 5.36403060e-01 -4.32265103e-01 9.48244393e-01
1.58282530e+00 1.01385951e+00 -4.71482575e-01 6.10435426e-01
1.15940109e-01 1.03006005e+00 4.54315636e-03 -5.39731383e-01
3.82232726e-01 -7.77695656e-01 3.64915729e-01 1.20505488e+00
3.25091243e-01 -6.06396496e-02 -4.10534471e-01 3.62147033e-01
4.35114175e-01 7.75841177e-01 -4.19452220e-01 -5.07382929e-01
7.44289756e-01 1.41381979e+00 -1.30934894e+00 1.47673920e-01
-7.96715260e-01 4.27085102e-01 -6.77064896e-01 6.68419152e-02
-2.71622121e-01 -5.47071815e-01 1.36131555e-01 4.01020914e-01
2.54439235e-01 2.54033804e-02 -1.07088637e+00 -3.66297603e-01
1.74597472e-01 -1.12174559e+00 3.00058573e-01 -7.19754875e-01
-1.02012432e+00 3.62183928e-01 9.54342186e-02 -1.18617678e+00
3.57625425e-01 -4.37746108e-01 -1.06146030e-01 1.01686311e+00
-1.04827595e+00 -9.61959243e-01 6.16289340e-02 -5.64863048e-02
5.24868071e-01 -5.41794419e-01 6.12498462e-01 2.70898581e-01
-6.07565582e-01 9.29264799e-02 2.49179639e-02 -4.66512740e-01
6.84103429e-01 -9.10567999e-01 -1.08389348e-01 3.54315728e-01
-6.92188382e-01 1.07484186e+00 8.26598346e-01 -1.06751561e+00
-1.47363079e+00 3.99825349e-02 1.33167958e+00 -3.05860102e-01
1.05622268e+00 -2.72220261e-02 -2.25126520e-01 3.22058618e-01
9.35119212e-01 -1.13584518e+00 1.10215330e+00 3.24261755e-01
4.40422237e-01 -9.53669176e-02 -1.15902460e+00 5.86889923e-01
3.59126806e-01 -1.83997959e-01 -5.60102522e-01 8.00901279e-02
1.46427706e-01 -1.61548927e-01 -1.15745151e+00 5.87360449e-02
7.52193451e-01 -6.96532547e-01 6.03922069e-01 -2.31355980e-01
7.07548618e-01 6.44018054e-02 -1.23573922e-01 -8.89275670e-01
-1.34535134e-01 -6.21879637e-01 3.34801555e-01 1.67217577e+00
7.03999877e-01 -8.06879282e-01 8.57461452e-01 4.76179719e-01
-3.54791641e-01 -9.78677392e-01 -5.12222528e-01 -2.45255977e-02
-4.95757349e-03 -4.11406875e-01 4.07401234e-01 1.35467732e+00
7.36552835e-01 4.48988855e-01 1.88592494e-01 -4.03722525e-01
1.75362259e-01 5.32630607e-02 4.47605312e-01 -1.35062301e+00
5.64677417e-01 -6.89720333e-01 -2.44141951e-01 -5.44275582e-01
-3.37271959e-01 -6.07434928e-01 -7.07689643e-01 -2.00005221e+00
1.29596531e-01 -3.54580969e-01 -1.01970308e-01 9.20076668e-02
4.42599714e-01 1.81206241e-02 -3.31733935e-02 3.61256003e-01
-1.82805255e-01 -1.72810644e-01 1.16274190e+00 1.55979291e-01
-4.00888920e-01 1.18125960e-01 -1.62104034e+00 6.17808640e-01
1.01671827e+00 -3.25010031e-01 -5.11260569e-01 1.94055736e-01
1.58551633e+00 -1.42783955e-01 -2.36787751e-01 -1.81319937e-01
1.59630552e-02 -2.69974291e-01 9.28100795e-02 -1.15763474e+00
-1.29236847e-01 -1.06508374e+00 1.88273236e-01 2.68147051e-01
-1.58159375e-01 8.24876785e-01 3.79437715e-01 -2.10483208e-01
-4.39207584e-01 -5.69087863e-01 -1.12157173e-01 -1.06697746e-01
-6.63341224e-01 -4.06735301e-01 -7.90626943e-01 -2.24771768e-01
1.05275249e+00 -8.93580675e-01 -6.13754429e-03 -4.68360424e-01
-7.08198011e-01 1.00191109e-01 3.47632706e-01 6.88075721e-01
3.60638201e-01 -9.76680636e-01 -5.71298838e-01 -5.78664243e-01
-7.30018690e-02 -3.33578765e-01 2.78516859e-01 1.26859128e+00
-6.46331787e-01 6.37093365e-01 -5.76464355e-01 -5.91762215e-02
-1.01474953e+00 2.42273450e-01 -3.92503113e-01 6.73153326e-02
-2.64359206e-01 5.06151795e-01 -4.49367136e-01 -2.36366510e-01
2.48941462e-02 7.63964206e-02 -1.21305966e+00 1.02725482e+00
2.65690684e-01 1.05467451e+00 4.01718587e-01 -1.19354045e+00
-5.24182856e-01 5.01788259e-01 -1.53557137e-01 -3.01178813e-01
1.61986697e+00 -5.90182602e-01 -8.69815886e-01 8.08194101e-01
9.98611093e-01 3.07453573e-01 -7.71770626e-03 3.83086771e-01
4.48731989e-01 -2.87132055e-01 1.22759923e-01 -1.25575459e+00
-1.07258022e+00 2.83030629e-01 5.24407446e-01 7.49803603e-01
1.12665892e+00 -8.52593631e-02 3.62947464e-01 -2.55434602e-01
2.85936415e-01 -1.78303230e+00 -2.46337503e-01 1.23309806e-01
1.08235300e+00 -9.14405704e-01 3.37916374e-01 -7.07349062e-01
-1.06877208e+00 1.28178787e+00 3.13935280e-01 3.23097736e-01
1.15039277e+00 5.06053090e-01 2.14235038e-01 -8.46992254e-01
-2.18007028e-01 4.07601923e-01 1.73841920e-02 3.68814051e-01
1.02118421e+00 -1.74183846e-01 -1.50753140e+00 1.12727308e+00
-6.41699970e-01 1.53746203e-01 8.27980161e-01 1.21780777e+00
-1.47414044e-01 -1.22269070e+00 -5.06856918e-01 5.76453924e-01
-1.26808512e+00 1.09762058e-01 -7.70153880e-01 1.15800393e+00
-5.65643385e-02 1.44101548e+00 -4.77462038e-02 -4.55966055e-01
4.41676885e-01 -2.96953231e-01 -6.05987944e-02 -3.48759055e-01
-1.10233772e+00 1.70331314e-01 4.52506006e-01 -9.23688337e-02
-1.20041108e+00 -1.26517677e+00 -1.00394273e+00 -3.01594228e-01
-6.81356907e-01 6.88891232e-01 1.42553759e+00 7.64570832e-01
5.35084605e-01 4.75821584e-01 6.45550787e-01 -1.46372139e-01
3.08941212e-02 -1.18019462e+00 -8.49626958e-01 -2.64538646e-01
-3.43968719e-01 -7.00008929e-01 -2.38834217e-01 -2.23308951e-01] | [10.882221221923828, 6.876919746398926] |
70125e2e-5eb9-49ad-a7ec-657a8ad40a18 | simulating-realistic-mri-variations-to | 2111.00837 | null | https://arxiv.org/abs/2111.00837v1 | https://arxiv.org/pdf/2111.00837v1.pdf | Simulating Realistic MRI variations to Improve Deep Learning model and visual explanations using GradCAM | In the medical field, landmark detection in MRI plays an important role in reducing medical technician efforts in tasks like scan planning, image registration, etc. First, 88 landmarks spread across the brain anatomy in the three respective views -- sagittal, coronal, and axial are manually annotated, later guidelines from the expert clinical technicians are taken sub-anatomy-wise, for better localization of the existing landmarks, in order to identify and locate the important atlas landmarks even in oblique scans. To overcome limited data availability, we implement realistic data augmentation to generate synthetic 3D volumetric data. We use a modified HighRes3DNet model for solving brain MRI volumetric landmark detection problem. In order to visually explain our trained model on unseen data, and discern a stronger model from a weaker model, we implement Gradient-weighted Class Activation Mapping (Grad-CAM) which produces a coarse localization map highlighting the regions the model is focusing. Our experiments show that the proposed method shows favorable results, and the overall pipeline can be extended to a variable number of landmarks and other anatomies. | ['Srinivasa Rao Kundeti', 'Deepam Gautam', 'Sumit Sharma', 'Razeem Ahmad Ali Mattathodi', 'Shrey Singla', 'Muhammad Ilyas Patel'] | 2021-11-01 | null | null | null | null | ['brain-landmark-detection'] | ['computer-vision'] | [ 1.2377848e-01 5.7651818e-01 -3.4001213e-02 -6.1622161e-01
-7.4493873e-01 -4.9066198e-01 2.5982437e-01 3.7341487e-01
-5.2661830e-01 5.3527546e-01 2.4072768e-01 -3.3208624e-01
3.4236345e-02 -4.6924940e-01 -4.8063794e-01 -4.5393810e-01
-3.2730669e-01 7.7564830e-01 4.6853742e-01 -1.7473189e-02
3.2752949e-01 8.3977139e-01 -8.3839756e-01 6.7456864e-02
6.5367788e-01 6.9267279e-01 4.7432801e-01 2.1791998e-01
-2.0193745e-01 5.3556103e-01 -3.4778664e-01 -2.5339732e-01
3.7795058e-01 -3.6621407e-01 -1.0151412e+00 8.5454278e-02
3.5486352e-01 -2.0161210e-01 3.0095823e-02 1.1984137e+00
4.6060273e-01 -1.5311612e-01 6.2797040e-01 -7.5735658e-01
-2.4219827e-01 6.3746047e-01 -9.9181366e-01 5.9799165e-01
8.1982568e-02 6.2082361e-02 4.5118889e-01 -9.3505311e-01
7.8659064e-01 8.0410683e-01 5.8097708e-01 5.9289217e-01
-1.1202177e+00 -5.9751409e-01 1.3304994e-01 1.9453922e-02
-1.2733169e+00 -1.3730870e-01 7.2409743e-01 -6.4626759e-01
3.4690470e-01 3.2195595e-01 7.6768094e-01 7.2603029e-01
3.5490382e-01 4.9387470e-01 1.3429687e+00 -3.0134389e-01
2.1967319e-01 -7.4935113e-03 8.8803343e-02 9.6160018e-01
-3.8443461e-02 -2.4680552e-01 -2.5007868e-01 -6.5134443e-02
1.2403173e+00 9.6275210e-02 -5.2284127e-01 -6.5640414e-01
-1.4628972e+00 5.9525675e-01 9.6009874e-01 5.9227896e-01
-6.2239861e-01 -1.2587340e-01 2.1559519e-01 -1.5953638e-01
3.0166170e-01 4.9456134e-01 -1.7219622e-01 3.0838174e-01
-1.2268543e+00 -2.0294201e-01 6.2841237e-02 6.7401516e-01
5.6909341e-01 -3.9098654e-02 -1.5928411e-01 8.1739295e-01
2.0797975e-01 -2.3622854e-02 8.3739531e-01 -7.5207120e-01
3.4039229e-01 6.0491920e-01 -1.8534975e-01 -4.8920897e-01
-8.3804822e-01 -4.9391627e-01 -8.6117381e-01 2.9401374e-01
4.5717353e-01 1.2209951e-01 -1.4332877e+00 1.1889701e+00
4.5489272e-01 1.1186135e-01 -5.5159944e-01 1.1265721e+00
7.8537071e-01 2.3424353e-02 3.0914804e-01 -8.7192588e-02
1.5647053e+00 -8.2023340e-01 -4.7523919e-01 -5.0397372e-01
7.1636409e-01 -5.8535326e-01 1.2395599e+00 2.3469869e-02
-1.0997306e+00 -2.9680070e-01 -9.0661258e-01 1.7301818e-02
-1.8725450e-01 1.6812108e-01 6.2091362e-01 6.0767150e-01
-1.1826086e+00 4.3011570e-01 -1.2211041e+00 -5.4426789e-02
8.9377165e-01 4.6474963e-01 -6.6396618e-01 -6.4689033e-02
-7.4755675e-01 1.2275449e+00 3.2881528e-01 3.6576465e-01
-8.2148677e-01 -9.5623636e-01 -9.1560775e-01 -1.4662701e-01
1.4006835e-01 -4.9022347e-01 8.6719763e-01 -5.0082457e-01
-1.0792273e+00 1.3504467e+00 -1.5262501e-01 -2.7080381e-01
7.9835474e-01 3.1368890e-01 -4.2913325e-02 3.8591617e-01
2.5672242e-01 9.1648966e-01 5.8217490e-01 -1.2932442e+00
-2.4821183e-01 -7.5513196e-01 1.2869474e-02 1.8915960e-01
2.2678342e-01 8.1329882e-02 -3.1048095e-01 -7.7891737e-01
6.8640178e-01 -8.2144266e-01 -6.0179353e-01 1.2460181e-01
-3.9541647e-01 2.6910406e-01 5.7872528e-01 -1.0233617e+00
7.2058374e-01 -1.9982089e+00 -1.0553182e-01 5.5441016e-01
5.4064512e-01 -1.0875779e-01 1.4738889e-01 -3.6238739e-01
-4.1851354e-01 7.6271236e-02 -5.1446956e-01 -1.9031249e-01
-5.7282329e-01 5.2731126e-03 2.0154542e-01 5.5220199e-01
3.2958876e-02 9.8243868e-01 -8.5435885e-01 -7.0793819e-01
3.3120495e-01 3.9146137e-01 -3.5198382e-01 6.0517214e-02
3.6753851e-01 1.2593099e+00 -5.8502156e-01 6.3603806e-01
6.2876564e-01 -1.8863779e-01 1.4257634e-01 -3.7379733e-01
1.2694717e-02 5.4242145e-03 -8.7139708e-01 2.1157219e+00
-5.4792660e-01 2.8285015e-01 1.7342487e-01 -8.4784436e-01
8.1667751e-01 2.3903999e-01 7.1461171e-01 -6.7548561e-01
2.7876875e-01 2.3208800e-01 3.5592023e-01 -4.5310572e-01
-1.3847136e-01 -7.1269475e-02 3.8120967e-01 4.8391303e-01
1.7453900e-02 -2.7752778e-01 5.0395016e-02 1.6545734e-01
8.0937874e-01 7.3926508e-02 2.2847287e-01 -4.6507719e-01
4.9703452e-01 5.3305030e-02 2.7058029e-01 4.1545811e-01
-3.1679806e-01 1.0038542e+00 5.0154817e-01 -6.3347560e-01
-8.3729148e-01 -1.0408506e+00 -4.9612591e-01 6.7503744e-01
2.8954426e-02 -1.1898591e-01 -9.9790096e-01 -1.0808717e+00
-4.7315529e-01 5.7105750e-01 -8.9967006e-01 1.1897887e-02
-7.9357970e-01 -8.8255697e-01 1.9461480e-01 6.1151308e-01
3.1249762e-01 -1.1961679e+00 -1.0999095e+00 6.7308657e-02
-1.7782985e-01 -9.0690881e-01 -5.1358002e-01 2.0338975e-01
-1.0116887e+00 -1.2843442e+00 -1.1004757e+00 -9.9432361e-01
1.3805314e+00 -5.0107200e-02 8.9497906e-01 2.3272365e-01
-6.7665780e-01 1.0602559e-01 -4.0208887e-02 -2.0790774e-01
-2.8541216e-01 -1.3095918e-01 -2.4908590e-01 -2.7485409e-01
4.3212142e-02 -5.0962603e-01 -8.6758107e-01 3.4061274e-01
-7.3734909e-01 1.6189623e-01 5.5451065e-01 8.2518733e-01
7.5532728e-01 -4.1848990e-01 2.2260800e-01 -1.1215121e+00
5.7393241e-01 -2.2034389e-01 -4.6705699e-01 4.4488603e-01
-3.0309206e-01 9.7265448e-03 2.5812197e-01 -3.2581282e-01
-7.5779969e-01 3.1447372e-01 -4.2230046e-01 -3.0165216e-01
-3.5153356e-01 4.2420650e-01 2.0458774e-01 -5.0882643e-01
6.7790157e-01 6.0760748e-02 1.4574021e-02 -3.5520765e-01
1.7452019e-01 1.3727468e-01 5.6969714e-01 -3.9854887e-01
5.6389046e-01 5.1565623e-01 1.2832007e-01 -2.7795595e-01
-5.4452211e-01 -3.5926154e-01 -1.2460450e+00 -2.5539955e-01
9.3732989e-01 -3.9311868e-01 -3.4386370e-01 -2.2031632e-01
-1.1549652e+00 -2.9004291e-01 -3.5253984e-01 6.8972200e-01
-3.2925364e-01 2.1233685e-01 -4.2826802e-01 -3.1727457e-01
-4.7144124e-01 -1.9966875e+00 1.1457733e+00 2.5181919e-01
-3.2929400e-01 -1.0589595e+00 -2.3381884e-01 1.6597804e-01
4.8486921e-01 2.8194115e-01 1.2432619e+00 -6.8204254e-01
-3.9395526e-01 -2.9030120e-01 -2.1213646e-01 3.8185599e-03
1.8544362e-01 -3.8164291e-01 -8.5961527e-01 -7.4621215e-02
2.4500518e-04 -9.5095798e-02 5.0454235e-01 7.7659762e-01
1.5660527e+00 2.5738332e-01 -3.8747489e-01 7.3833060e-01
9.9306291e-01 2.1491769e-01 4.9393809e-01 3.2934493e-01
6.9708276e-01 7.2423261e-01 3.6429057e-01 -4.7383074e-02
3.2983762e-01 3.7962911e-01 6.2515086e-01 -6.5618354e-01
-2.9106531e-01 -5.3761918e-02 -2.1265255e-01 5.9144461e-01
-3.8834751e-01 5.1656783e-01 -1.1799612e+00 5.0983584e-01
-1.3201439e+00 -4.3372038e-01 -8.8561609e-02 2.3633218e+00
7.6684737e-01 1.0455081e-01 9.1773063e-02 -1.3364929e-01
6.4121968e-01 -7.7342750e-03 -4.7950739e-01 2.9290471e-02
3.4284496e-01 5.4614937e-01 6.4819050e-01 5.9899163e-01
-1.0362645e+00 7.9856360e-01 6.8198457e+00 5.8458138e-01
-1.5243026e+00 4.3969676e-01 9.1890341e-01 -6.3265641e-03
-1.6726613e-01 -1.3116138e-01 -3.1109494e-01 3.0042303e-01
3.3241037e-01 1.5584998e-01 1.6463581e-01 9.2415428e-01
2.5225002e-01 -2.2443923e-01 -1.0960467e+00 1.0401378e+00
-2.1916742e-02 -1.3486912e+00 -2.5294462e-01 4.3561682e-02
4.2965996e-01 8.6410575e-02 -7.4695073e-02 1.0256049e-01
-6.2873192e-02 -1.0964862e+00 4.8796886e-01 3.1991106e-01
9.8871034e-01 -4.8082319e-01 7.0700592e-01 3.2364818e-01
-9.4321746e-01 3.5660988e-01 -3.3522069e-01 6.5827143e-01
2.8007200e-01 2.9454282e-01 -1.2754381e+00 3.6994916e-01
6.6744483e-01 1.4531586e-01 -7.8056252e-01 1.3348963e+00
-3.9719898e-01 2.2339918e-01 -1.0417385e-01 5.4935479e-01
2.3045273e-01 -2.1215257e-01 2.9083490e-01 1.0923654e+00
2.0367606e-01 5.0610125e-02 2.2565660e-01 8.9458686e-01
-8.5226912e-03 3.7194526e-01 -3.6027592e-01 6.4763254e-01
1.0097083e-01 1.6551828e+00 -1.4440279e+00 -2.1344066e-01
-9.6449718e-02 9.3973291e-01 2.6591080e-01 3.3673972e-01
-7.1200532e-01 2.7549539e-02 -2.3505550e-02 6.1106473e-01
-2.9935849e-01 -1.6808568e-01 -4.9966645e-01 -9.5423061e-01
3.5914823e-02 -5.7862043e-01 3.1855524e-01 -8.0047023e-01
-8.0216736e-01 9.9183732e-01 1.4149477e-01 -9.5214474e-01
-2.7573615e-01 -5.6183577e-01 -8.5406226e-01 1.1031808e+00
-1.4663990e+00 -1.0639569e+00 -5.1608521e-01 5.4538786e-01
4.9015030e-01 6.4776717e-03 8.0472958e-01 3.1370223e-01
-2.5380260e-01 3.5888016e-01 -5.4141390e-01 3.3383760e-01
6.3269728e-01 -1.2582748e+00 3.0256283e-01 8.0847996e-01
1.7033243e-01 8.8654560e-01 3.6127722e-01 -7.6195794e-01
-7.6724541e-01 -8.4439945e-01 3.6348018e-01 -3.0638787e-01
5.3544724e-01 -2.6591638e-01 -9.8397380e-01 8.3693039e-01
-8.2331226e-04 4.6903536e-01 4.9370340e-01 -1.6907145e-01
6.2147487e-02 2.3382795e-01 -1.4400016e+00 5.0878936e-01
9.4856650e-01 -2.1199229e-01 -5.8991474e-01 5.8308291e-01
1.6436911e-01 -8.0273116e-01 -9.1594273e-01 5.1293510e-01
3.9832947e-01 -7.4583220e-01 9.9252218e-01 -4.9774599e-01
2.2045161e-01 -2.7677572e-01 5.2064687e-01 -1.6037991e+00
-1.9158471e-01 -1.5164429e-01 3.6066103e-01 5.9463733e-01
4.8992267e-01 -5.2365881e-01 9.7794974e-01 7.4172693e-01
-2.7319905e-01 -8.7476724e-01 -1.1550330e+00 -2.3778580e-01
-2.3698753e-02 -2.2314580e-01 5.3253281e-01 1.0783504e+00
-3.8330298e-02 -1.8887417e-02 1.0112265e-02 1.8431288e-01
5.3136468e-01 -1.6419341e-01 3.2754323e-01 -1.2204741e+00
7.0661098e-02 -7.1875739e-01 -4.9525058e-01 -8.8034600e-01
-5.8318317e-02 -1.2101914e+00 -1.9850901e-01 -1.7171588e+00
1.5140608e-01 -7.5149965e-01 -3.5783145e-01 5.8434230e-01
-2.1107829e-01 6.1615413e-01 1.8971916e-02 1.8340027e-01
-2.2265743e-02 1.8675229e-01 1.8352212e+00 -1.8025625e-01
1.5332494e-02 -1.9096769e-02 -4.7283864e-01 9.6409780e-01
6.6043669e-01 -5.6875038e-01 -4.2518100e-01 -5.5777484e-01
-3.0752769e-01 1.2227689e-01 3.8163891e-01 -8.1279898e-01
1.5731919e-01 1.0959483e-01 7.3799920e-01 -4.4924650e-01
1.9597811e-01 -9.0426260e-01 -1.6943738e-01 4.5341459e-01
-2.7500159e-01 4.5568806e-01 1.0192680e-01 -1.8373000e-02
-1.8025154e-01 -4.1715428e-01 9.6343386e-01 -6.2203979e-01
-4.2265633e-01 5.5800462e-01 -3.0532844e-02 -2.1802967e-02
1.0697296e+00 -2.6519644e-01 2.9049542e-02 -1.5945792e-01
-1.3063589e+00 1.6287677e-01 3.1419384e-01 2.0048416e-01
7.6451010e-01 -9.8077196e-01 -6.7271155e-01 3.2406783e-01
3.6819126e-03 2.5037161e-01 4.5679468e-01 1.5318155e+00
-1.0141599e+00 1.6800351e-01 -4.3117598e-01 -7.9453635e-01
-1.0001369e+00 2.2833055e-01 7.4287403e-01 -3.7112102e-01
-9.1143590e-01 8.8200909e-01 4.8431289e-01 -4.1813853e-01
2.3409854e-01 -5.0519383e-01 -4.0098631e-01 -2.3426573e-01
5.5795509e-01 -4.4259042e-02 5.4627985e-01 -8.0753195e-01
-5.6636369e-01 7.0727170e-01 -2.9067954e-01 6.9649093e-02
1.2075267e+00 2.0666257e-02 -7.1787007e-02 4.7893528e-02
8.4645057e-01 2.2723663e-01 -1.0040827e+00 -5.7337359e-02
8.0514602e-02 -4.3130398e-01 1.4615516e-01 -9.5201612e-01
-1.3697132e+00 1.1437324e+00 9.6633261e-01 -9.3543656e-02
8.3790904e-01 7.4086495e-02 3.8439953e-01 -3.7980759e-01
6.7525929e-01 -6.4408863e-01 -3.2784578e-01 6.4938299e-02
1.0808513e+00 -1.3829818e+00 6.8851677e-03 -5.6183594e-01
-8.6459678e-01 1.2171936e+00 7.4954164e-01 -1.0255154e-02
5.7652658e-01 3.0107901e-01 3.9363554e-01 -5.3422052e-01
1.5908460e-01 -3.2873545e-02 5.7347989e-01 7.1713686e-01
6.8215400e-01 2.3320537e-02 -3.0117613e-01 1.9250450e-01
-2.5395668e-01 -3.4855610e-01 3.5656324e-01 7.1890855e-01
-1.5258037e-01 -1.1344491e+00 -4.7905466e-01 5.1443624e-01
-5.9822071e-01 -1.9746965e-01 -9.8313600e-02 7.9572237e-01
1.1323956e-01 3.1015491e-01 8.8183999e-02 1.8325557e-01
3.7340033e-01 -3.9097030e-02 6.1617088e-01 -7.6181960e-01
-6.3074189e-01 3.3307698e-01 -3.1795615e-01 -5.1700598e-01
-9.2874363e-02 -3.5719234e-01 -1.6032118e+00 2.8555009e-01
-5.1533047e-02 1.3202764e-01 9.4130081e-01 1.0119721e+00
4.5677803e-02 8.2173014e-01 2.5212967e-01 -1.0837119e+00
-2.2983907e-03 -1.0580584e+00 -3.1404912e-01 3.6167201e-01
3.2550260e-02 -8.4410822e-01 3.6907051e-02 3.7246283e-02] | [14.489715576171875, -2.4153785705566406] |
f49bb10a-76e5-4d0b-a72d-ab6860c19852 | comprehensive-time-series-regression-models | 1412.5397 | null | http://arxiv.org/abs/1412.5397v3 | http://arxiv.org/pdf/1412.5397v3.pdf | Comprehensive Time-Series Regression Models Using GRETL -- U.S. GDP and Government Consumption Expenditures & Gross Investment from 1980 to 2013 | Using Gretl, I apply ARMA, Vector ARMA, VAR, state-space model with a Kalman
filter, transfer-function and intervention models, unit root tests,
cointegration test, volatility models (ARCH, GARCH, ARCH-M, GARCH-M,
Taylor-Schwert GARCH, GJR, TARCH, NARCH, APARCH, EGARCH) to analyze quarterly
time series of GDP and Government Consumption Expenditures & Gross Investment
(GCEGI) from 1980 to 2013. The article is organized as: (I) Definition; (II)
Regression Models; (III) Discussion. Additionally, I discovered a unique
interaction between GDP and GCEGI in both the short-run and the long-run and
provided policy makers with some suggestions. For example in the short run, GDP
responded positively and very significantly (0.00248) to GCEGI, while GCEGI
reacted positively but not too significantly (0.08051) to GDP. In the long run,
current GDP responded negatively and permanently (0.09229) to a shock in past
GCEGI, while current GCEGI reacted negatively yet temporarily (0.29821) to a
shock in past GDP. Therefore, policy makers should not adjust current GCEGI
based merely on the condition of current and past GDP. Although increasing
GCEGI does help GDP in the short-term, significantly abrupt increase in GCEGI
might not be good to the long-term health of GDP. Instead, a balanced,
sustainable, and economically viable solution is recommended, so that the
short-term benefits to the current economy from increasing GCEGI often largely
secured by the long-term loan outweigh or at least equal to the negative effect
to the future economy from the long-term debt incurred by the loan. Finally, I
found that non-normally distributed volatility models generally perform better
than normally distributed ones. More specifically, TARCH-GED performs the best
in the group of non-normally distributed, while GARCH-M does the best in the
group of normally distributed. | [] | 2019-08-17 | null | null | null | null | ['time-series-regression'] | ['time-series'] | [-1.09070337e+00 -7.22477632e-03 1.25443965e-01 1.54190615e-01
-2.25826293e-01 -6.53778851e-01 6.87912107e-01 -9.32625756e-02
-1.07298516e-01 8.35642278e-01 6.20076299e-01 -1.24681532e+00
-3.13840061e-01 -9.40454006e-01 -2.16050103e-01 -7.81463683e-01
-8.60058963e-02 2.73429662e-01 -2.40396857e-01 -2.97992259e-01
2.91114718e-01 4.15527195e-01 -1.04746783e+00 -8.23587954e-01
5.34422100e-01 7.40003824e-01 -8.89768079e-02 4.01388764e-01
-1.89929649e-01 7.87895977e-01 -5.37847638e-01 -1.41032770e-01
2.91601688e-01 -3.56319517e-01 -2.25121349e-01 -3.23685110e-01
-6.72401071e-01 -2.28408128e-01 -3.45027030e-01 9.22051013e-01
1.97392374e-01 2.39648178e-01 8.85993183e-01 -8.88405621e-01
-8.30922902e-01 7.30205953e-01 -5.97506523e-01 4.12750542e-01
1.65917128e-01 2.74486512e-01 3.92832279e-01 -7.05330849e-01
1.72350332e-01 1.33204782e+00 4.89938319e-01 -3.88480648e-02
-9.90181386e-01 -7.56017923e-01 3.39829594e-01 -5.75566590e-01
-1.03410816e+00 -1.62470657e-02 3.97219718e-01 -6.74320996e-01
1.29146385e+00 2.60397285e-01 7.33395696e-01 7.72922218e-01
6.79626882e-01 -2.58912951e-01 1.13926506e+00 -2.86611438e-01
3.72029603e-01 -2.52051875e-02 3.53370428e-01 -1.77208543e-01
6.40870154e-01 4.24583644e-01 4.38044310e-01 -2.89785177e-01
9.42666233e-01 2.05895126e-01 -7.58644193e-03 6.12965941e-01
-9.00900900e-01 9.06489849e-01 1.88089907e-02 6.64421439e-01
-9.89526510e-01 9.92837846e-02 3.30535948e-01 7.67813385e-01
4.99048889e-01 -7.96245933e-02 -6.93632305e-01 -2.77829379e-01
-6.77316129e-01 1.69852883e-01 8.61766577e-01 4.75092947e-01
5.10643125e-01 7.71360457e-01 1.44883633e-01 1.21193595e-01
4.05040741e-01 1.23403025e+00 8.44164968e-01 -6.61026299e-01
3.39911103e-01 3.04574937e-01 2.75214672e-01 -1.06797218e+00
-5.32156825e-01 -4.88931865e-01 -1.07888877e+00 3.81774247e-01
5.39916039e-01 -4.95837688e-01 -5.81102550e-01 1.79401898e+00
-2.07619548e-01 -9.86141153e-03 3.32437843e-01 2.56307214e-01
2.59581387e-01 9.11752224e-01 1.31616041e-01 -9.58859503e-01
1.13293087e+00 -4.11865950e-01 -8.68983150e-01 6.60882369e-02
7.13486254e-01 -7.48520553e-01 7.44978845e-01 2.40969658e-01
-1.13311374e+00 -3.32326800e-01 -2.70362258e-01 6.77341819e-01
-2.21821502e-01 -3.36490899e-01 5.59536219e-01 7.09784925e-01
-1.09692299e+00 4.23764795e-01 -9.86298680e-01 -8.88274312e-02
-4.22461331e-01 1.81002900e-01 -1.40091151e-01 5.48438191e-01
-9.33417678e-01 7.60997593e-01 2.93772638e-01 -1.81989018e-02
-4.15702373e-01 -3.41355383e-01 -6.82286084e-01 4.53080982e-01
1.05072774e-01 -4.93949831e-01 8.30295801e-01 -8.31314683e-01
-1.39747584e+00 3.02984565e-01 1.41847119e-01 -6.64358556e-01
1.55620366e-01 3.31491590e-01 -7.37041831e-01 -3.36791128e-01
2.37730183e-02 -2.99230397e-01 2.44195282e-01 -5.12728393e-01
-4.90841269e-01 -1.01546228e+00 -3.89128506e-01 -2.62673408e-01
-2.25110352e-03 4.53550518e-01 5.27707696e-01 -8.57011914e-01
1.86092541e-01 -1.08099592e+00 -1.83715001e-01 -1.27712679e+00
-8.87569636e-02 -4.42228109e-01 6.63811803e-01 -1.01615071e+00
1.36809349e+00 -2.13769197e+00 -5.00656962e-01 3.94975305e-01
-4.69531387e-01 5.17816581e-02 3.90456975e-01 7.04661012e-01
-6.32194161e-01 2.71550089e-01 -7.15999156e-02 1.61784515e-01
7.17205331e-02 1.54188663e-01 -5.81589460e-01 5.80468893e-01
-1.72333807e-01 7.38534093e-01 -5.16145945e-01 3.88461411e-01
1.39531717e-01 2.15766095e-02 -8.29653963e-02 -3.88514400e-02
2.71885484e-01 4.25469756e-01 -2.55276918e-01 6.40175819e-01
5.93532801e-01 -1.33074289e-02 -8.60121325e-02 5.65902889e-01
-9.13178980e-01 5.61432242e-02 -1.10808194e+00 5.51557720e-01
-4.07543451e-01 5.20119965e-01 -1.66554987e-01 -1.39186251e+00
1.41379666e+00 8.01715374e-01 2.72826225e-01 -7.31435299e-01
2.46339858e-01 7.01282859e-01 2.74859816e-01 -4.89240289e-01
2.29932249e-01 -3.08893919e-01 -1.13778681e-01 6.66319370e-01
-4.16726470e-01 4.20949236e-02 1.55663505e-01 8.97453539e-03
7.88402736e-01 -2.52417475e-01 5.46953380e-01 -5.68159699e-01
3.83504778e-01 -4.02804047e-01 6.14598274e-01 2.34647959e-01
7.59050995e-02 -8.25095549e-03 9.27498400e-01 -2.82459766e-01
-9.04523313e-01 -6.41194046e-01 3.17290202e-02 6.50234222e-01
-5.13102114e-01 5.32041907e-01 -4.99347329e-01 -6.00094944e-02
2.65398920e-01 1.23462009e+00 -4.42366511e-01 2.16306880e-01
-4.99688238e-01 -9.00734842e-01 1.46657541e-01 3.91612858e-01
5.04399776e-01 -1.18383312e+00 -6.30891740e-01 4.22549546e-01
3.91420037e-01 -3.55555594e-01 -1.03857584e-01 1.86007135e-02
-1.05054438e+00 -6.45563424e-01 -9.81178164e-01 -7.14821890e-02
3.03690791e-01 8.19317102e-02 9.99354780e-01 -1.05651855e-01
4.41389799e-01 3.49060714e-01 -1.73330873e-01 -7.33906329e-01
-2.39378884e-01 -5.59604287e-01 1.06897965e-01 -3.55019569e-01
2.58526564e-01 -7.51317739e-01 -5.91293931e-01 5.20675816e-02
-6.13047779e-01 -7.42693484e-01 -1.14520229e-02 6.21621847e-01
1.40566185e-01 2.86687225e-01 1.18004882e+00 -3.12116981e-01
5.09455025e-01 -8.66089463e-01 -9.03047085e-01 -1.10720312e-02
-1.19297588e+00 -3.83033067e-01 3.48131955e-01 -3.12730849e-01
-1.31877816e+00 -7.23781526e-01 -7.34530240e-02 -1.12693265e-01
1.90241989e-02 1.12957895e+00 4.07081656e-02 4.68627781e-01
2.90322214e-01 -1.53617747e-02 9.05849189e-02 -6.17088735e-01
3.57915796e-02 6.88387752e-01 6.80231214e-01 -2.55107999e-01
5.03115237e-01 4.11628336e-01 -7.85763469e-03 -4.91793275e-01
1.33135661e-01 -3.01000267e-01 -1.17376767e-01 8.12927186e-02
7.02693760e-01 -1.32193542e+00 -5.62718153e-01 7.75623620e-01
-6.29871964e-01 -3.12500149e-01 -2.26649299e-01 1.06301951e+00
-3.43105614e-01 1.62861958e-01 -6.10781550e-01 -1.47767460e+00
-5.70957601e-01 -9.64310110e-01 3.45776565e-02 6.13925219e-01
-1.69932812e-01 -1.39011157e+00 1.84076205e-01 -2.12488338e-01
6.41190767e-01 6.00296080e-01 9.71423388e-01 -9.07909036e-01
-8.58521834e-02 -2.98679173e-01 -6.27356470e-02 3.89115691e-01
2.63627648e-01 5.98740816e-01 -4.33059871e-01 -4.30939764e-01
3.97935152e-01 2.36981794e-01 4.85193968e-01 1.02230859e+00
3.16007614e-01 -4.63356346e-01 -6.62607094e-03 5.56803584e-01
1.40587151e+00 1.01647854e+00 8.45697820e-01 8.91230941e-01
4.50360514e-02 5.50207257e-01 5.53430915e-01 7.17240274e-01
5.16398072e-01 1.60466954e-01 3.77750903e-01 1.68946348e-02
4.13128853e-01 -3.51371057e-02 6.32215858e-01 6.82411611e-01
-4.52594042e-01 -8.22413862e-02 -8.98566365e-01 6.77171171e-01
-1.49291348e+00 -1.13777065e+00 -7.28619814e-01 2.40365076e+00
2.77050853e-01 2.74163157e-01 5.09814024e-01 2.07827929e-02
4.04389858e-01 -1.33090764e-01 -2.63424456e-01 -8.53018641e-01
-4.20422882e-01 1.35140181e-01 8.17761183e-01 3.42829019e-01
-3.78835410e-01 5.83905160e-01 5.40248680e+00 2.00826123e-01
-1.33120561e+00 -2.42062986e-01 9.77857411e-01 4.21979308e-01
-5.73375225e-01 6.31965518e-01 -5.49918652e-01 8.53833318e-01
1.55063093e+00 -1.04948521e+00 2.42687333e-02 8.51739764e-01
7.29920924e-01 -4.49615240e-01 -1.54818922e-01 5.96531451e-01
-6.76653743e-01 -8.46889496e-01 -5.80506325e-01 6.72456324e-01
8.07826221e-01 -6.59232065e-02 2.88432270e-01 6.03553355e-01
4.63726461e-01 -7.00884700e-01 6.49949372e-01 7.02090919e-01
4.70658600e-01 -1.07733095e+00 1.01787567e+00 4.98277575e-01
-8.63459229e-01 -5.57083011e-01 -4.96878922e-01 -6.86274648e-01
3.88500124e-01 7.38318682e-01 -6.02698922e-02 7.53030181e-01
4.92450058e-01 1.90248549e-01 -1.69527307e-01 4.77951705e-01
4.03447822e-03 9.71371531e-01 -2.67478466e-01 4.45076078e-01
5.34127057e-01 -9.91114736e-01 4.43319410e-01 7.86395967e-01
1.02725184e+00 6.89005136e-01 -3.16900611e-01 4.14043814e-01
5.87098181e-01 1.46571532e-01 -8.58798504e-01 -1.67606920e-01
5.26403666e-01 7.36900330e-01 -6.44687831e-01 -6.72385097e-01
-7.38514841e-01 9.81862172e-02 -4.57937062e-01 7.02542186e-01
-4.19021100e-01 -1.36451021e-01 4.54402357e-01 1.81895837e-01
8.38914439e-02 -1.39183670e-01 -5.20978630e-01 -9.14735019e-01
-2.51039565e-01 -8.91473532e-01 5.48654974e-01 -4.99682456e-01
-8.32960904e-01 3.24868232e-01 1.78863198e-01 -7.20380306e-01
-1.03315008e+00 -1.01339400e-01 -1.18445122e+00 1.41348004e+00
-1.05196059e+00 -6.93600953e-01 2.49292701e-01 4.74802673e-01
1.56968862e-01 -4.92040694e-01 7.35491097e-01 -9.91020277e-02
-7.12105811e-01 1.67952687e-01 5.80420852e-01 -1.32244229e-01
3.20161104e-01 -1.04691827e+00 5.50171018e-01 9.05517876e-01
-2.53196836e-01 8.59531641e-01 7.64611721e-01 -1.02536905e+00
-9.30807889e-01 -6.51827216e-01 1.13245618e+00 6.15466647e-02
9.60669577e-01 4.85405296e-01 -1.28141510e+00 1.00499153e+00
3.47723037e-01 -3.29091996e-01 2.49662176e-01 -1.12221271e-01
-1.97854955e-02 1.43841878e-01 -1.10228908e+00 2.65515625e-01
1.42776236e-01 -1.62088573e-01 -5.16654074e-01 1.17311999e-01
6.46935701e-01 1.18341058e-01 -1.19492996e+00 1.27404332e-01
5.88419437e-01 -1.21920073e+00 6.79926395e-01 -5.01868606e-01
-2.10177694e-02 1.36896744e-01 -2.00281397e-01 -1.35296726e+00
-4.86068130e-01 -7.87949026e-01 2.48898119e-01 1.52223825e+00
2.77060419e-01 -1.46397436e+00 1.62909269e-01 9.39538479e-01
-1.32235751e-01 -3.47653180e-01 -1.06257963e+00 -1.18281269e+00
6.13903165e-01 -1.15730807e-01 1.11466873e+00 1.03498650e+00
3.46735083e-02 -2.20609725e-01 -2.92120934e-01 1.14916891e-01
4.02985871e-01 1.34614669e-02 8.12902927e-01 -1.12060857e+00
-1.56244278e-01 -6.64148927e-01 -4.95373271e-02 -3.99078310e-01
1.37146503e-01 -3.25873941e-01 -7.66454577e-01 -1.30946279e+00
2.17325501e-02 -2.49858737e-01 -1.04406141e-01 5.89137115e-02
-1.46477818e-01 -5.88006496e-01 1.47276074e-01 2.99454510e-01
7.64489293e-01 2.58564442e-01 7.05947816e-01 4.05892551e-01
-5.69071591e-01 4.41207558e-01 -6.16510153e-01 7.60734618e-01
8.91605556e-01 -2.18488976e-01 -4.85182405e-01 -2.79287118e-02
4.37252641e-01 1.08684242e+00 2.43175641e-01 -4.87351924e-01
-1.51534881e-02 -5.20994961e-01 4.46940912e-03 -8.35126162e-01
-2.23171294e-01 -6.34478986e-01 1.02349222e+00 6.84970617e-01
4.17155027e-01 8.66349578e-01 8.15095752e-02 4.39491510e-01
-2.75794089e-01 -2.67553210e-01 4.80844766e-01 -8.67597386e-02
1.09121511e-02 1.97312474e-01 -7.24302828e-01 -2.10358843e-01
9.35778856e-01 -2.55689919e-01 -1.83250904e-01 -7.99659133e-01
-8.11646461e-01 1.88327581e-01 6.84031785e-01 1.55078754e-01
1.53910993e-02 -1.05738044e+00 -7.20686316e-01 1.25203237e-01
-5.57799280e-01 -3.73767078e-01 3.28958988e-01 1.09783661e+00
-1.71980307e-01 7.85690963e-01 -1.04802363e-01 1.63385466e-01
-8.80290627e-01 7.06144154e-01 1.34845346e-01 -3.35554123e-01
-7.22696543e-01 5.57737887e-01 4.18508083e-01 2.43821412e-01
-1.15061559e-01 -4.29982245e-01 -3.01987022e-01 5.71085751e-01
6.46936655e-01 7.58310556e-01 -6.35158420e-02 -7.80149579e-01
-1.51127994e-01 4.49100047e-01 3.57477397e-01 -4.18652207e-01
1.43112230e+00 -5.88939607e-01 -3.10902387e-01 8.37558389e-01
8.96333337e-01 -1.31878972e-01 -1.00279546e+00 1.29890904e-01
3.40787590e-01 -2.31012523e-01 9.09397975e-02 -4.41063881e-01
-1.13039422e+00 5.37059486e-01 3.13564479e-01 6.65040135e-01
1.06625891e+00 -2.27704197e-01 5.57842433e-01 -1.53334200e-01
2.99444556e-01 -9.31519985e-01 -3.21855545e-01 5.98818898e-01
9.02639091e-01 -8.57192934e-01 -1.40156969e-01 3.33962649e-01
-6.07315660e-01 1.08318806e+00 -2.06561550e-01 -2.80059963e-01
1.11622405e+00 1.29787564e-01 -2.54957080e-02 -2.65662730e-01
-9.14628267e-01 -9.52772424e-03 -3.55942965e-01 4.09859121e-01
5.92189610e-01 4.93329793e-01 -8.11176598e-01 6.39943063e-01
-5.66316545e-01 -3.73730779e-01 8.21159840e-01 4.46617693e-01
-5.27250051e-01 -7.21277893e-01 -8.85604203e-01 3.47590387e-01
-9.56700146e-01 -3.41908894e-02 2.22381160e-01 1.26838613e+00
-3.54428679e-01 9.80102837e-01 5.60955226e-01 4.49656606e-01
3.39716405e-01 2.17003331e-01 -2.66839892e-01 -1.73340380e-01
-6.10798419e-01 7.11400509e-01 -1.44340107e-02 -5.95632866e-02
-4.83670272e-02 -1.11907864e+00 -1.32517946e+00 -7.09783256e-01
-4.34074283e-01 1.63751960e-01 7.86674261e-01 8.27801049e-01
-1.01933122e-01 6.94120899e-02 1.10509706e+00 -3.41875345e-01
-1.06236959e+00 -1.39760447e+00 -1.16776061e+00 3.24163726e-03
1.19604819e-01 -4.73913461e-01 -9.98854697e-01 -1.55414239e-01] | [5.514752388000488, 4.0551533699035645] |
be3280b7-a933-4bd2-a82f-917212627ecc | deep-features-for-cbir-with-scarce-data-using | 2205.08935 | null | https://arxiv.org/abs/2205.08935v1 | https://arxiv.org/pdf/2205.08935v1.pdf | Deep Features for CBIR with Scarce Data using Hebbian Learning | Features extracted from Deep Neural Networks (DNNs) have proven to be very effective in the context of Content Based Image Retrieval (CBIR). In recent work, biologically inspired \textit{Hebbian} learning algorithms have shown promises for DNN training. In this contribution, we study the performance of such algorithms in the development of feature extractors for CBIR tasks. Specifically, we consider a semi-supervised learning strategy in two steps: first, an unsupervised pre-training stage is performed using Hebbian learning on the image dataset; second, the network is fine-tuned using supervised Stochastic Gradient Descent (SGD) training. For the unsupervised pre-training stage, we explore the nonlinear Hebbian Principal Component Analysis (HPCA) learning rule. For the supervised fine-tuning stage, we assume sample efficiency scenarios, in which the amount of labeled samples is just a small fraction of the whole dataset. Our experimental analysis, conducted on the CIFAR10 and CIFAR100 datasets shows that, when few labeled samples are available, our Hebbian approach provides relevant improvements compared to various alternative methods. | ['Giuseppe Amato', 'Claudio Gennaro', 'Fabrizio Falchi', 'Claudio Gallicchio', 'Davide Bacciu', 'Gabriele Lagani'] | 2022-05-18 | null | null | null | null | ['content-based-image-retrieval', 'unsupervised-pre-training'] | ['computer-vision', 'methodology'] | [ 1.94387183e-01 -1.38177797e-01 -5.26470691e-02 -4.03849304e-01
-4.36310291e-01 -3.10171604e-01 9.13725674e-01 8.91913921e-02
-9.63429570e-01 6.95758641e-01 5.07559590e-02 -3.60593125e-02
-5.18087566e-01 -7.06520200e-01 -7.56831765e-01 -1.10607088e+00
2.72902906e-01 7.49743283e-01 1.60071731e-01 -1.92312643e-01
3.18582118e-01 9.74565923e-01 -1.53582489e+00 3.10766071e-01
4.29608226e-01 1.35173535e+00 3.28596085e-01 5.49805522e-01
-1.82986200e-01 1.23809934e+00 -4.92843330e-01 -2.09745586e-01
2.97924072e-01 -4.18914646e-01 -8.88996363e-01 -2.29290217e-01
1.55089870e-01 -3.52525562e-01 -3.10689002e-01 8.92671645e-01
5.08341908e-01 3.91042650e-01 7.66678572e-01 -9.53585386e-01
-3.88386577e-01 7.21194506e-01 -2.58598864e-01 5.05451083e-01
-5.47715068e-01 4.66010496e-02 1.02694023e+00 -9.70781088e-01
4.87253994e-01 1.26068008e+00 2.22028941e-01 9.13610160e-01
-1.36818480e+00 -6.85944438e-01 -4.08664644e-02 3.45795721e-01
-1.26996148e+00 -5.87000906e-01 1.04928100e+00 -5.99856496e-01
5.95789492e-01 6.42324537e-02 7.71030903e-01 8.45711768e-01
-1.00753464e-01 1.02301884e+00 1.10634482e+00 -7.42628336e-01
6.82052612e-01 3.55934113e-01 5.93034446e-01 5.17200589e-01
1.47642046e-01 3.00941169e-01 -6.29658818e-01 -7.48702139e-03
6.98584497e-01 1.10271927e-02 -7.96847194e-02 -4.01991755e-01
-9.35960352e-01 1.01878977e+00 7.91120708e-01 6.40669942e-01
-6.06792629e-01 4.52976316e-01 1.72285989e-01 1.13398820e-01
4.64587063e-01 3.46936315e-01 -2.57301062e-01 3.76052052e-01
-1.13995588e+00 8.94695744e-02 5.61282456e-01 3.83086473e-01
9.56840158e-01 2.92307494e-04 -2.52233326e-01 9.50656414e-01
2.11644992e-01 3.90139163e-01 9.22341287e-01 -7.53837407e-01
1.29100094e-02 4.59535658e-01 -1.15388092e-02 -8.35712850e-01
-2.85405308e-01 -4.59039807e-01 -9.39945042e-01 1.19202785e-01
3.96423280e-01 -2.34421529e-02 -9.48272705e-01 1.45910859e+00
1.87346652e-01 -9.44560096e-02 1.31696180e-01 9.78423178e-01
7.68388867e-01 6.05654836e-01 1.12453394e-01 -8.11760724e-02
1.04913282e+00 -9.41471219e-01 -4.40005809e-01 -1.02918409e-01
3.10956776e-01 -2.69153744e-01 9.74823833e-01 5.50075769e-01
-7.68549919e-01 -5.48661292e-01 -1.03382361e+00 9.34743434e-02
-4.18606848e-01 3.47025752e-01 5.53176880e-01 5.87679684e-01
-1.12706137e+00 7.26489425e-01 -7.27918506e-01 -2.10385859e-01
6.86196268e-01 6.16900444e-01 -6.89067468e-02 -2.11898372e-01
-9.51406658e-01 5.66907704e-01 6.67674184e-01 2.49499843e-01
-1.25868905e+00 -4.79131222e-01 -1.71779886e-01 3.63596410e-01
7.75262415e-02 -5.07821023e-01 9.16798234e-01 -1.43196476e+00
-1.75371492e+00 1.00622249e+00 5.64401522e-02 -8.48481417e-01
2.72895604e-01 -2.33706951e-01 1.97846040e-01 3.83659959e-01
-5.16410828e-01 1.03298521e+00 9.52456057e-01 -1.40406656e+00
-4.30789918e-01 -4.60779667e-01 1.62096262e-01 2.43776198e-02
-5.75135708e-01 4.84241731e-02 -1.44464061e-01 -5.14935255e-01
9.02493950e-03 -9.67380464e-01 -1.65402576e-01 -7.90627897e-02
-1.93549201e-01 -3.18234295e-01 3.41644347e-01 -3.77116591e-01
7.88365841e-01 -2.38344097e+00 7.11819157e-02 4.79759246e-01
1.13245890e-01 4.24720198e-01 -1.09701850e-01 1.32961661e-01
1.76218282e-02 3.26985074e-03 5.97244799e-02 -1.33083416e-02
-2.32732251e-01 1.17310904e-01 -4.70697999e-01 2.88045555e-01
9.83961672e-02 9.38568950e-01 -7.99421906e-01 -4.58874404e-01
1.62295103e-01 3.58877391e-01 -4.84616548e-01 1.03132084e-01
-2.68835664e-01 3.63841176e-01 -4.94328380e-01 1.53600007e-01
2.11417675e-01 -3.24395567e-01 -3.59859914e-02 -2.01576948e-01
1.39018208e-01 -4.32096459e-02 -7.07546830e-01 1.47951293e+00
-2.95839667e-01 8.70554447e-01 -2.19831243e-01 -1.39099920e+00
1.06101918e+00 2.02572376e-01 5.11371613e-01 -7.55323112e-01
3.25074226e-01 1.48657843e-01 3.01094532e-01 -1.69940606e-01
-1.52054474e-01 -1.25737801e-01 5.23513556e-01 4.86534029e-01
4.93237227e-01 5.18422313e-02 2.58651674e-01 -2.41200645e-02
9.32703197e-01 -1.17838807e-01 1.55795887e-01 -5.26565135e-01
7.12472737e-01 2.20169231e-01 2.46159270e-01 8.06924999e-01
-8.62278044e-02 3.45293283e-01 2.42262602e-01 -7.73408711e-01
-9.21881795e-01 -7.96247542e-01 -3.82715538e-02 1.31998515e+00
2.76190788e-02 1.47571102e-01 -1.02829123e+00 -4.16409105e-01
-3.51068676e-01 7.27302611e-01 -7.18077004e-01 -4.63602513e-01
-4.08308327e-01 -1.13684082e+00 4.75781471e-01 3.05321097e-01
7.35538423e-01 -1.54655313e+00 -8.57451379e-01 1.72599718e-01
2.43019968e-01 -9.50813115e-01 1.72495507e-02 3.59124571e-01
-1.08471119e+00 -7.46454000e-01 -8.53278100e-01 -7.38609731e-01
5.95291197e-01 1.43203333e-01 8.13279033e-01 -1.07209934e-02
-3.92122090e-01 3.79203111e-01 -2.94664085e-01 -5.63459396e-01
-1.52463585e-01 -8.89702886e-02 -1.85111649e-02 2.84720570e-01
6.42233789e-01 -5.00839710e-01 -8.97368908e-01 9.93898138e-02
-9.79007661e-01 5.40377281e-04 7.71432936e-01 9.54595506e-01
7.11589634e-01 5.35225496e-02 5.02724349e-01 -8.09787869e-01
5.39691329e-01 -8.27350393e-02 -7.16937542e-01 3.05733919e-01
-7.12509036e-01 3.51560175e-01 5.69281638e-01 -6.95734441e-01
-1.03154981e+00 1.52175933e-01 -1.39190495e-01 -5.09460151e-01
-2.65775979e-01 5.44579864e-01 2.72995587e-02 -3.08608741e-01
8.95179689e-01 5.47242701e-01 -1.36866376e-01 -4.62616622e-01
1.93402320e-01 5.23741364e-01 2.39006534e-01 -6.26268208e-01
4.46215004e-01 6.10059559e-01 -6.63294718e-02 -8.76272142e-01
-9.70222116e-01 -3.90229493e-01 -7.25338936e-01 -3.27685654e-01
8.12826753e-01 -7.39221752e-01 -7.29411304e-01 3.84168774e-01
-8.82751048e-01 -7.17518747e-01 -4.96241063e-01 7.59757280e-01
-5.42642891e-01 -2.12560624e-01 -6.73314452e-01 -8.40331137e-01
-7.10892022e-01 -9.14275527e-01 5.85096836e-01 3.75921071e-01
1.64446712e-01 -8.24061394e-01 1.46244824e-01 4.68872726e-01
3.00849587e-01 -2.45005816e-01 1.18325102e+00 -1.26468480e+00
-3.47087622e-01 -1.64667279e-01 -2.21292660e-01 5.73982418e-01
-2.12232083e-01 -1.92644030e-01 -1.34971809e+00 -1.40568301e-01
2.14942500e-01 -5.20271778e-01 1.10583842e+00 4.91534352e-01
1.35212934e+00 -3.54602337e-01 -1.54974237e-01 3.92145246e-01
1.41186178e+00 2.57576436e-01 4.58037049e-01 4.24198598e-01
5.00419974e-01 7.25654423e-01 2.96223611e-01 2.17657104e-01
6.58224523e-02 3.21614236e-01 3.00750911e-01 1.02124810e-01
-2.27387786e-01 -1.10632159e-01 9.56355035e-02 6.46048903e-01
-9.63794291e-02 -1.85699686e-01 -1.02868104e+00 4.61472303e-01
-1.76692677e+00 -8.76450062e-01 4.45511967e-01 2.12802887e+00
7.88613915e-01 1.20259307e-01 4.07938324e-02 3.27904493e-01
6.31835938e-01 -2.38227263e-01 -6.98418379e-01 4.21227142e-02
-1.60956666e-01 4.09819573e-01 3.99686247e-01 2.98417300e-01
-1.03243470e+00 1.11972880e+00 5.07740450e+00 1.14450455e+00
-1.55292559e+00 2.72552192e-01 9.24388945e-01 -2.16526777e-01
-6.73074797e-02 -2.42455065e-01 -7.68286526e-01 2.27626681e-01
1.13271463e+00 1.81413531e-01 6.64110601e-01 8.03066611e-01
2.30577514e-01 -1.03236780e-01 -1.04529321e+00 1.28479421e+00
-3.44678834e-02 -1.41380632e+00 3.26300412e-01 6.51858822e-02
6.81856751e-01 4.38922524e-01 -1.60786659e-02 2.55763531e-01
4.38412547e-01 -8.50955546e-01 7.13799953e-01 5.94549000e-01
2.33568147e-01 -7.68314064e-01 7.21459031e-01 7.88339138e-01
-4.37743366e-01 -3.40802610e-01 -7.31848359e-01 3.44005406e-01
-4.01456624e-01 8.48220408e-01 -5.67502499e-01 5.12008592e-02
7.98527896e-01 4.27535325e-01 -6.03895187e-01 1.15639877e+00
-1.04741156e-01 9.55165744e-01 -3.47372741e-01 -3.94633681e-01
3.04671735e-01 -2.41512090e-01 2.29800716e-01 1.07117224e+00
-2.03159302e-02 1.93212122e-01 -2.02936560e-01 8.88908505e-01
-2.17446208e-01 3.07437599e-01 -1.94570452e-01 -1.65423542e-01
1.40632078e-01 1.34713066e+00 -1.08761942e+00 -3.44915479e-01
9.21518952e-02 7.03099906e-01 4.58723634e-01 6.09239399e-01
-4.07434374e-01 -1.80349067e-01 3.33011523e-02 -1.16549678e-01
3.82961035e-01 9.72570572e-03 -1.86376959e-01 -1.06521392e+00
-4.41576600e-01 -6.78734303e-01 2.23660007e-01 -7.86281645e-01
-9.94971514e-01 6.99568629e-01 -1.20989837e-01 -6.35085225e-01
-2.92889833e-01 -8.64633858e-01 -3.61403316e-01 5.84316015e-01
-1.44109213e+00 -1.12088192e+00 -1.72825322e-01 8.14703226e-01
5.72112620e-01 -3.86317909e-01 6.70187056e-01 2.47351944e-01
-5.85930467e-01 3.55416775e-01 5.51358104e-01 3.57581168e-01
2.20974714e-01 -9.58149433e-01 -2.69707829e-01 6.00089371e-01
7.58952379e-01 5.62861919e-01 4.93996948e-01 -1.36021540e-01
-1.21747601e+00 -8.89960766e-01 5.28810322e-01 6.29783571e-02
4.89408463e-01 -4.47272807e-01 -6.74433351e-01 4.05725896e-01
-5.68937249e-02 2.26526096e-01 6.76723540e-01 -1.34436399e-01
-1.49283454e-01 -7.10127056e-01 -1.03747535e+00 3.36867809e-01
5.38872242e-01 -3.88454109e-01 -4.84024912e-01 3.59227926e-01
3.11868995e-01 3.03127587e-01 -6.30932570e-01 3.26879323e-01
6.12498164e-01 -7.65771627e-01 1.00772619e+00 -6.37712419e-01
2.74445146e-01 -7.70037062e-03 -1.82701856e-01 -1.16551757e+00
-2.88791090e-01 -2.46620685e-01 -2.26422399e-01 1.11369586e+00
2.33928174e-01 -3.07895213e-01 8.68744195e-01 5.81368506e-01
3.50581467e-01 -7.33804166e-01 -7.91049123e-01 -5.95237911e-01
1.19117521e-01 -8.54088515e-02 1.32070839e-01 4.61705029e-01
-3.62381458e-01 5.85186839e-01 -5.49581088e-02 -3.56544435e-01
6.20907128e-01 1.91626355e-01 5.26628256e-01 -1.49229300e+00
-3.88331383e-01 -6.31388724e-01 -2.31126875e-01 -9.12233353e-01
1.86813429e-01 -7.38956690e-01 2.81100929e-01 -1.10855126e+00
4.92393762e-01 -4.27558720e-01 -7.56306589e-01 3.43908131e-01
1.17712334e-01 1.47254735e-01 2.57897526e-01 5.56105375e-01
-4.27459449e-01 4.13597286e-01 9.17972922e-01 -2.94593066e-01
-1.49099454e-01 1.83582217e-01 -3.93827349e-01 7.04609096e-01
7.37752438e-01 -6.44144893e-01 -5.26647806e-01 -4.17959064e-01
1.29024655e-01 -2.25924805e-01 5.47647655e-01 -1.15549338e+00
3.95361900e-01 -3.98937017e-02 6.37918413e-01 -4.54884291e-01
3.26271325e-01 -7.88182616e-01 -2.03266248e-01 7.01737046e-01
-7.92468071e-01 -4.91674304e-01 8.57166126e-02 5.82924366e-01
-2.66637266e-01 -6.83568478e-01 9.43704784e-01 -2.29075477e-01
-6.44127011e-01 2.67094880e-01 -4.88976538e-01 -5.38134538e-02
6.68237805e-01 -7.60030514e-03 -5.71740903e-02 -2.92161882e-01
-6.62461162e-01 -2.14121073e-01 -1.03071079e-01 1.04374299e-02
6.57770872e-01 -9.61751282e-01 -5.05881608e-01 -1.75551310e-01
-9.44790468e-02 -8.94616842e-02 4.17957269e-03 7.07403421e-01
-5.45732021e-01 6.09703064e-01 -2.57774860e-01 -5.76879025e-01
-9.21414495e-01 9.67215955e-01 2.88335353e-01 -3.44412595e-01
-3.10763747e-01 1.05157757e+00 5.33678055e-01 -1.59850270e-01
2.32657149e-01 -2.36928016e-02 -5.81766069e-01 1.14255778e-01
4.53543752e-01 2.69972086e-01 1.99850127e-01 -4.90378678e-01
-1.36828244e-01 2.31825650e-01 -2.43871644e-01 -3.83040845e-01
1.60426760e+00 -2.61299405e-02 -1.56818122e-01 4.85424280e-01
1.17102540e+00 -4.38874215e-01 -1.09354615e+00 -4.16181594e-01
4.10812199e-01 1.06155530e-01 4.44810838e-01 -7.27681339e-01
-1.26041400e+00 1.25109673e+00 9.28217769e-01 -9.05857608e-02
1.17142034e+00 -1.08998135e-01 4.11365092e-01 1.02091420e+00
3.47280562e-01 -1.00103021e+00 1.41231388e-01 3.51492614e-01
7.73142457e-01 -1.13621593e+00 -2.11479530e-01 1.96409002e-01
-4.37218755e-01 1.26283360e+00 4.60463047e-01 -5.94283342e-01
8.87306929e-01 -1.73446953e-01 -9.38020870e-02 -2.40111619e-01
-7.31711388e-01 -2.19549462e-01 5.05521595e-01 1.32155254e-01
1.93719879e-01 -2.71002561e-01 -2.16730341e-01 5.91894925e-01
-1.39026746e-01 2.74245352e-01 1.27457246e-01 6.86372042e-01
-6.11479104e-01 -6.37608767e-01 -1.60975412e-01 5.91390848e-01
-5.75659692e-01 -3.75259250e-01 -4.65932786e-01 5.86903214e-01
-1.59479424e-01 7.72552431e-01 -4.03800718e-02 -1.25012606e-01
-1.39734849e-01 7.92786032e-02 6.24194682e-01 -4.21382636e-01
-6.34428740e-01 2.92319387e-01 -3.54734898e-01 -3.93819749e-01
-6.06997311e-01 -3.45531464e-01 -1.10894156e+00 5.98595068e-02
-3.32227498e-01 2.92775869e-01 8.51171732e-01 1.29300034e+00
1.28890157e-01 3.16568613e-01 4.87433702e-01 -1.03292561e+00
-5.34439564e-01 -9.74717021e-01 -5.34943223e-01 3.93126488e-01
2.30510846e-01 -5.24963439e-01 -3.21899354e-01 4.22736436e-01] | [9.30739688873291, 2.877955675125122] |
4857e4b6-a449-4226-a47a-5ef0bd1b2deb | convolutional-neural-networks-demystified-a | 2108.11663 | null | https://arxiv.org/abs/2108.11663v3 | https://arxiv.org/pdf/2108.11663v3.pdf | Convolutional Neural Networks Demystified: A Matched Filtering Perspective Based Tutorial | Deep Neural Networks (DNN) and especially Convolutional Neural Networks (CNN) are a de-facto standard for the analysis of large volumes of signals and images. Yet, their development and underlying principles have been largely performed in an ad-hoc and black box fashion. To help demystify CNNs, we revisit their operation from first principles and a matched filtering perspective. We establish that the convolution operation within CNNs, their very backbone, represents a matched filter which examines the input signal/image for the presence of pre-defined features. This perspective is shown to be physically meaningful, and serves as a basis for a step-by-step tutorial on the operation of CNNs, including pooling, zero padding, various ways of dimensionality reduction. Starting from first principles, both the feed-forward pass and the learning stage (via back-propagation) are illuminated in detail, both through a worked-out numerical example and the corresponding visualizations. It is our hope that this tutorial will help shed new light and physical intuition into the understanding and further development of deep neural networks. | ['Danilo Mandic', 'Ljubisa Stankovic'] | 2021-08-26 | null | null | null | null | ['physical-intuition'] | ['reasoning'] | [ 2.99315065e-01 -7.03818426e-02 4.26932305e-01 -2.99644321e-01
1.27623960e-01 -2.46429473e-01 4.69887376e-01 -1.60959139e-01
-3.17268729e-01 3.94949853e-01 -1.36204258e-01 -4.18303251e-01
-4.16951120e-01 -7.93671966e-01 -5.19770265e-01 -8.66331637e-01
-6.64212108e-01 -4.58881855e-01 1.36232138e-01 -3.67848217e-01
1.19947411e-01 9.73309577e-01 -1.33782554e+00 2.94456899e-01
9.91520137e-02 1.46916616e+00 -1.26742005e-01 5.37898839e-01
1.04654774e-01 6.41694069e-01 -3.18502665e-01 -2.64895171e-01
2.89486617e-01 -2.53027052e-01 -6.87584400e-01 -2.01225713e-01
1.93388507e-01 -1.92822635e-01 -9.12891269e-01 1.03305411e+00
4.98312831e-01 1.70015991e-01 4.93721098e-01 -1.02074873e+00
-6.23751819e-01 3.44534785e-01 -2.15775728e-01 4.13227111e-01
-3.04341346e-01 2.09435210e-01 7.93302894e-01 -1.08905351e+00
3.39925319e-01 1.01961875e+00 1.08466327e+00 4.92783904e-01
-1.13909674e+00 -4.21990901e-01 3.53735611e-02 1.19685093e-02
-9.95990574e-01 -3.77259284e-01 9.70214665e-01 -4.41772878e-01
1.05858946e+00 1.57526195e-01 8.55236411e-01 7.21267581e-01
4.07588661e-01 5.78809917e-01 6.94813132e-01 -6.37734830e-01
2.88870156e-01 -8.50415602e-02 4.14457142e-01 5.60459614e-01
2.45039716e-01 4.41014588e-01 -5.64651430e-01 1.45851066e-02
1.19886541e+00 9.79396477e-02 -2.53895670e-01 -5.14451444e-01
-9.48228061e-01 7.48777747e-01 8.24280381e-01 3.91784102e-01
-2.92591661e-01 3.34025711e-01 3.08643281e-01 5.11306942e-01
1.80453420e-01 4.73487854e-01 -3.63831878e-01 2.34060317e-01
-9.47661757e-01 3.75207186e-01 6.31013632e-01 4.68516886e-01
6.72088504e-01 3.19756478e-01 2.64109343e-01 6.86012685e-01
2.66555518e-01 2.33091280e-01 1.74279317e-01 -8.07874203e-01
7.40964040e-02 3.06395233e-01 -1.27298057e-01 -1.16857922e+00
-7.13800609e-01 -9.36454058e-01 -1.23654616e+00 6.98157549e-01
4.52634156e-01 -6.15549028e-01 -7.67971814e-01 1.54027295e+00
-1.28178507e-01 4.97124940e-02 -3.01390607e-02 1.01472545e+00
9.21015322e-01 5.75547099e-01 -3.32022905e-01 7.46492073e-02
1.28176379e+00 -6.22525632e-01 -6.66812539e-01 -2.02641889e-01
2.96302855e-01 -5.73186100e-01 6.57842457e-01 3.58994573e-01
-1.43197131e+00 -6.80131674e-01 -1.34701872e+00 -1.65666595e-01
-5.16029894e-01 2.80978065e-02 7.55570471e-01 4.11848128e-01
-9.85614300e-01 1.23881197e+00 -1.07969320e+00 -3.51361424e-01
7.00872540e-01 4.83746260e-01 -1.78644076e-01 4.01987463e-01
-1.27992976e+00 8.09616208e-01 -1.52250007e-01 5.75627446e-01
-5.81063807e-01 -8.85127664e-01 -6.53612852e-01 1.92657024e-01
-1.43062249e-01 -6.48230553e-01 1.16466725e+00 -9.37361419e-01
-1.46756923e+00 7.24131525e-01 -6.80488870e-02 -7.79261112e-01
2.02153549e-01 -5.20536266e-02 -5.17841816e-01 1.96310386e-01
-4.51739758e-01 3.68833363e-01 8.04950237e-01 -1.07849300e+00
-3.77347350e-01 -2.83578455e-01 3.90070491e-02 -2.37044156e-01
-3.26014876e-01 1.95749149e-01 -9.09989625e-02 -8.24346066e-01
6.10838234e-01 -5.73096454e-01 -4.67184097e-01 4.18075949e-01
-4.22057837e-01 1.87521681e-01 8.10386479e-01 -3.35666448e-01
1.19931996e+00 -2.31701612e+00 -1.06231205e-01 5.32441080e-01
7.12885916e-01 3.99840206e-01 1.90320104e-01 6.97624981e-01
-5.68225682e-01 -1.02562420e-01 -2.77891904e-01 -3.02645385e-01
-2.03306720e-01 -2.08255172e-01 -3.05324972e-01 6.40850961e-01
4.98587668e-01 1.06521082e+00 -5.57640612e-01 6.70125559e-02
3.62781525e-01 8.67805064e-01 -3.38068217e-01 -1.32607281e-01
3.60082984e-01 3.38921607e-01 -2.01405704e-01 4.58777398e-01
6.76963747e-01 -3.89934957e-01 -2.60268822e-02 -6.51312947e-01
-5.08526921e-01 3.02405924e-01 -1.09221208e+00 1.41893828e+00
-2.56419539e-01 1.31764948e+00 4.68534082e-01 -1.51726758e+00
8.27043235e-01 2.26697728e-01 4.83170420e-01 -7.81068683e-01
2.95401871e-01 1.45989165e-01 3.49368572e-01 -5.96890688e-01
1.83752209e-01 -1.80494860e-01 1.40958235e-01 3.20028603e-01
3.28292817e-01 1.94432005e-01 4.13493887e-02 1.54036907e-02
1.10360968e+00 -1.98401794e-01 1.16156265e-01 -6.38994813e-01
3.44659984e-01 1.40002994e-02 1.82137445e-01 7.53241956e-01
-1.50880352e-01 7.59102583e-01 5.99680901e-01 -9.27093685e-01
-1.00907254e+00 -1.04046631e+00 -4.59310561e-01 4.81572598e-01
2.76361173e-03 -2.03333035e-01 -8.36762249e-01 1.46288142e-01
1.72336139e-02 -1.33237854e-01 -7.99667299e-01 -2.62812644e-01
-7.08849430e-01 -9.53495622e-01 4.40711886e-01 6.53555453e-01
6.55094683e-01 -1.15922236e+00 -9.44344103e-01 3.15356791e-01
4.11647826e-01 -9.59913373e-01 3.37621212e-01 6.63073182e-01
-1.06662989e+00 -1.02511477e+00 -8.34244132e-01 -9.90622938e-01
6.98345304e-01 1.37245640e-01 9.28828061e-01 2.71788627e-01
-4.84910369e-01 6.63944259e-02 5.57209216e-02 -3.92620683e-01
-7.22184405e-02 -1.53249502e-01 1.77544340e-01 4.13954183e-02
3.78711432e-01 -1.15154457e+00 -7.19253123e-01 1.37177706e-01
-6.91539943e-01 7.83722475e-02 6.30885065e-01 6.49892509e-01
1.54580653e-01 1.81870446e-01 2.94058442e-01 -5.66121995e-01
7.57285655e-01 -8.39139670e-02 -6.35540187e-01 -7.08945394e-02
-1.20651469e-01 -1.40596613e-01 6.38061106e-01 -1.75627902e-01
-7.01083541e-01 -1.87065542e-01 -3.06263387e-01 -2.49117389e-01
-1.80718824e-01 4.46497232e-01 3.43523771e-02 -6.05680287e-01
7.03172445e-01 -5.47599345e-02 1.75780490e-01 -7.54434109e-01
6.39256015e-02 3.60327274e-01 7.36509979e-01 -2.85748869e-01
9.12461519e-01 8.52460623e-01 3.44780862e-01 -1.13527942e+00
-4.69657212e-01 -1.89162493e-01 -1.00507104e+00 -3.99016291e-01
6.55197144e-01 -4.60352391e-01 -9.77775156e-01 7.64254332e-01
-1.18352878e+00 -5.01721203e-01 -4.46325779e-01 4.20374751e-01
-3.11139047e-01 -1.21710151e-01 -8.06547821e-01 -6.69203103e-01
-8.74581560e-02 -1.02824175e+00 4.04131740e-01 3.70376170e-01
-2.25118220e-01 -1.25613880e+00 -1.47570491e-01 -3.79352182e-01
7.87888825e-01 2.86282241e-01 1.05405164e+00 -4.15161312e-01
-4.03309554e-01 -2.41362378e-01 -3.93932641e-01 5.28304577e-01
-4.00709510e-02 2.06579566e-01 -1.38353527e+00 -1.44928053e-01
2.46686712e-01 4.40365449e-02 9.72796440e-01 8.42976213e-01
1.27059937e+00 3.52808787e-03 -3.09499711e-01 8.71178627e-01
1.61957574e+00 1.81038871e-01 6.87019467e-01 2.58853018e-01
4.12606090e-01 7.26949155e-01 -4.66465265e-01 2.50870854e-01
-3.17325741e-01 2.97331035e-01 5.66557586e-01 -7.38447487e-01
-1.20962471e-01 2.22683758e-01 -1.12942003e-01 6.98503792e-01
-2.54703343e-01 1.47662356e-01 -7.96127737e-01 2.90399939e-01
-1.54548955e+00 -9.27791238e-01 -2.79472142e-01 2.08128929e+00
3.27173233e-01 4.29333508e-01 -6.00940660e-02 6.43242359e-01
6.31210327e-01 9.20326412e-02 -5.10022759e-01 -3.58114213e-01
-2.78288931e-01 4.89517599e-01 4.49774265e-01 3.42683882e-01
-1.16314030e+00 3.96170557e-01 7.41814089e+00 4.50909674e-01
-1.52226925e+00 -2.62110233e-01 4.46623564e-01 -8.35152119e-02
1.28834009e-01 -2.26972967e-01 -5.59157014e-01 1.64237380e-01
6.58879340e-01 1.72334582e-01 3.97481263e-01 5.25155544e-01
5.11859298e-01 -5.93044655e-03 -1.10228288e+00 9.73048806e-01
-4.70827490e-01 -1.88986874e+00 -2.07900614e-01 7.73626938e-02
5.04396975e-01 2.61278927e-01 1.33938551e-01 -8.96329880e-02
-4.87769991e-02 -1.06290627e+00 7.18911469e-01 6.72645509e-01
4.71405864e-01 -4.42262262e-01 6.22306645e-01 5.22181354e-02
-1.10268497e+00 -1.27027586e-01 -3.02789986e-01 -6.52228951e-01
1.43136919e-01 9.80580688e-01 1.15708262e-01 2.44085222e-01
8.94687593e-01 7.78712332e-01 -1.95549548e-01 1.13770568e+00
2.25405186e-01 4.33093995e-01 -3.11982244e-01 -1.52905449e-01
4.83433157e-01 -1.98782096e-03 4.79389876e-01 1.34559309e+00
-3.86259928e-02 3.04478675e-01 -4.23555285e-01 1.26751828e+00
-7.03191310e-02 -3.61770123e-01 -3.88470471e-01 -2.92204004e-02
1.80223659e-01 1.43777049e+00 -9.98799503e-01 -2.05414817e-01
-7.13637471e-01 5.03607571e-01 2.08824072e-02 7.11271942e-01
-5.23284972e-01 -9.79352355e-01 1.05131114e+00 2.97301888e-01
2.19901800e-01 -4.54775780e-01 -7.47425020e-01 -8.83503973e-01
1.95222035e-01 -3.98107618e-01 -2.18194053e-01 -7.77478158e-01
-1.17531896e+00 4.80210692e-01 -1.57636955e-01 -1.08202982e+00
7.67875239e-02 -1.19998693e+00 -1.01900244e+00 1.00912225e+00
-1.50073230e+00 -5.32168388e-01 -3.59141946e-01 3.47018570e-01
9.48164538e-02 -5.21642007e-02 7.38735735e-01 4.41396415e-01
-7.32062459e-01 2.08914742e-01 2.31400922e-01 6.66625679e-01
-1.15392439e-01 -8.19293916e-01 7.67899692e-01 8.06248963e-01
-2.54811272e-02 9.37994540e-01 8.47135723e-01 -7.51455501e-02
-1.34983385e+00 -6.61535859e-01 5.30832469e-01 1.86412096e-01
7.99494982e-01 -4.99986559e-01 -1.03743303e+00 5.03061831e-01
2.79419303e-01 2.08849832e-01 4.40066457e-01 2.89646927e-02
8.88950527e-02 -1.94260910e-01 -9.46151018e-01 5.48092067e-01
9.73573804e-01 -3.68775100e-01 -5.14939845e-01 9.96524617e-02
3.22428852e-01 -3.00803572e-01 -6.84752584e-01 4.10607308e-01
8.19853187e-01 -1.39971900e+00 1.19353199e+00 -6.25602543e-01
4.62611318e-01 -7.20985830e-02 5.47672100e-02 -1.05497038e+00
-4.24012005e-01 -9.06119049e-01 3.10511366e-02 6.46813393e-01
4.03765321e-01 -7.74759054e-01 8.03176761e-01 4.91313219e-01
-2.86635637e-01 -1.20771933e+00 -8.15843821e-01 -7.79195845e-01
2.36656711e-01 -8.57866824e-01 3.60190719e-01 6.44929647e-01
2.02345878e-01 -1.09968118e-01 1.60362452e-01 2.20854953e-01
5.85333407e-01 -2.65084535e-01 3.15734893e-01 -1.40913808e+00
6.14463538e-02 -8.60853910e-01 -6.54965878e-01 -1.20667279e+00
-3.72448444e-01 -4.41202968e-01 -2.57724583e-01 -1.37809193e+00
-2.94737667e-01 -3.35835278e-01 -3.84298235e-01 2.17083514e-01
5.55234194e-01 6.31380558e-01 6.41759112e-02 2.86951035e-01
4.71390188e-02 1.44467071e-01 9.90835369e-01 1.91141844e-01
-1.57098487e-01 2.45928302e-01 -5.01426339e-01 1.01553965e+00
8.38029981e-01 -1.30242512e-01 -4.96570580e-02 -5.09068429e-01
3.51060838e-01 -3.08819860e-01 9.97283399e-01 -1.45178032e+00
4.71586943e-01 1.41931430e-01 6.12206101e-01 -4.57717031e-01
5.72672009e-01 -8.33298862e-01 -1.86503917e-01 5.09992182e-01
-4.61807519e-01 -1.72478855e-02 3.46444368e-01 4.93433177e-02
-3.09049755e-01 -1.83706477e-01 9.86679196e-01 -2.18265772e-01
-6.30335093e-01 2.31753796e-01 -4.85541999e-01 -1.10226408e-01
5.86821675e-01 -4.58166152e-01 -6.47132937e-03 -4.24695939e-01
-1.06176174e+00 -2.56339870e-02 5.86322770e-02 -5.14007211e-02
5.81188083e-01 -1.24314404e+00 -1.99271530e-01 8.47924232e-01
-4.08486038e-01 -2.98883263e-02 3.70685220e-01 8.84134948e-01
-6.27298534e-01 6.35266900e-01 -3.32695514e-01 -5.37007213e-01
-9.81497347e-01 2.04431400e-01 1.05488098e+00 1.65873036e-01
-8.98184836e-01 1.12068999e+00 1.44249380e-01 -1.33297557e-03
5.30200124e-01 -8.79338741e-01 -1.78721175e-01 -1.36288837e-01
7.64916599e-01 3.83520514e-01 4.02594805e-01 -3.71485680e-01
-3.59089971e-01 8.82323325e-01 3.39670837e-01 -2.26090997e-01
1.59680164e+00 3.52851339e-02 -2.03217044e-01 5.10315061e-01
1.40105832e+00 -3.27830255e-01 -1.53011751e+00 -3.04243505e-01
-2.04997212e-01 5.67726744e-03 3.40467930e-01 -5.39080858e-01
-1.37302721e+00 1.50590599e+00 4.16651040e-01 7.78544486e-01
1.16213322e+00 -1.08952314e-01 7.43731081e-01 4.44758445e-01
-9.17557478e-02 -9.53066468e-01 -1.32234961e-01 4.71419275e-01
8.53115976e-01 -7.94387400e-01 7.93952961e-03 -3.01009148e-01
1.60239220e-01 1.68941522e+00 2.94964105e-01 -5.36842167e-01
1.46943665e+00 7.84555078e-01 2.04178467e-02 -4.73169625e-01
-6.39791310e-01 -2.41557453e-02 3.43894750e-01 5.55993676e-01
5.83135247e-01 -3.00523460e-01 3.50781411e-01 6.43778443e-01
-1.70046762e-01 6.65910542e-02 2.40686402e-01 9.58835602e-01
-5.94040334e-01 -6.95616484e-01 -2.68275827e-01 3.77642721e-01
-5.22871614e-01 -2.94408888e-01 -6.18961416e-02 9.61777568e-01
1.48311302e-01 5.33000231e-01 4.08966154e-01 -2.81713039e-01
4.78023916e-01 -6.28185719e-02 4.65799034e-01 -3.54266971e-01
-5.73841274e-01 -4.04777825e-02 -2.56939530e-01 -5.19162834e-01
-4.53297108e-01 -5.07829845e-01 -1.36613488e+00 -5.46019912e-01
-2.12202743e-01 -7.69980252e-02 8.85507047e-01 1.14712274e+00
1.77844405e-01 9.87072468e-01 3.69849265e-01 -1.27558458e+00
-3.35236818e-01 -7.38282800e-01 -6.46569550e-01 -3.54633257e-02
7.29624093e-01 -5.56303918e-01 -5.96567333e-01 -4.63522822e-02] | [8.993414878845215, 2.3539631366729736] |
797320c7-4c03-4927-ba96-2d01584eb81e | multimodal-joint-attribute-prediction-and | 2009.07162 | null | https://arxiv.org/abs/2009.07162v1 | https://arxiv.org/pdf/2009.07162v1.pdf | Multimodal Joint Attribute Prediction and Value Extraction for E-commerce Product | Product attribute values are essential in many e-commerce scenarios, such as customer service robots, product recommendations, and product retrieval. While in the real world, the attribute values of a product are usually incomplete and vary over time, which greatly hinders the practical applications. In this paper, we propose a multimodal method to jointly predict product attributes and extract values from textual product descriptions with the help of the product images. We argue that product attributes and values are highly correlated, e.g., it will be easier to extract the values on condition that the product attributes are given. Thus, we jointly model the attribute prediction and value extraction tasks from multiple aspects towards the interactions between attributes and values. Moreover, product images have distinct effects on our tasks for different product attributes and values. Thus, we selectively draw useful visual information from product images to enhance our model. We annotate a multimodal product attribute value dataset that contains 87,194 instances, and the experimental results on this dataset demonstrate that explicitly modeling the relationship between attributes and values facilitates our method to establish the correspondence between them, and selectively utilizing visual product information is necessary for the task. Our code and dataset will be released to the public. | ['Bo-Wen Zhou', 'Tiangang Zhu', 'Yue Wang', 'Xiaodong He', 'Youzheng Wu', 'Haoran Li'] | 2020-09-15 | null | https://aclanthology.org/2020.emnlp-main.166 | https://aclanthology.org/2020.emnlp-main.166.pdf | emnlp-2020-11 | ['attribute-value-extraction'] | ['natural-language-processing'] | [ 2.44620964e-01 -1.61567897e-01 -7.87863553e-01 -9.24734294e-01
-3.04662526e-01 -8.43897760e-01 3.63374472e-01 4.04405862e-01
-2.09703550e-01 4.68416274e-01 2.39632964e-01 6.82001486e-02
-2.31462494e-01 -8.23112607e-01 -3.84855241e-01 -7.70604968e-01
1.90361053e-01 4.90723610e-01 -3.74128520e-01 -2.91049778e-01
1.23453662e-01 2.80856580e-01 -1.69355917e+00 6.04216099e-01
5.92852414e-01 1.37004137e+00 1.47622854e-01 1.31502852e-01
-4.07852083e-01 5.71506619e-01 -4.12310958e-01 -7.94199944e-01
1.45251796e-01 -4.66356054e-02 -6.92879915e-01 6.64881468e-01
9.74408388e-02 -4.29610610e-01 8.79535004e-02 1.25213802e+00
8.01169872e-02 3.54525670e-02 8.80378008e-01 -1.78356647e+00
-1.05951464e+00 7.46804416e-01 -8.22634280e-01 -5.30714095e-01
3.41052055e-01 8.61211196e-02 1.56109917e+00 -9.12988901e-01
6.20373249e-01 1.23931456e+00 2.29888663e-01 5.58150895e-02
-1.37679148e+00 -5.80653727e-01 4.22445297e-01 2.67279506e-01
-1.44034350e+00 -2.50697672e-01 1.21274006e+00 -4.06344235e-01
4.95455623e-01 1.05179690e-01 7.51919150e-01 1.06567550e+00
-4.33250159e-01 1.22143769e+00 7.24883795e-01 -1.37380838e-01
-2.14485060e-02 5.55288494e-01 6.82954043e-02 1.18029326e-01
2.46077389e-01 -1.30732939e-01 -4.84120578e-01 -5.08939754e-03
7.10398078e-01 3.35930586e-01 -3.99512723e-02 -8.41493726e-01
-1.56760240e+00 8.48228216e-01 2.30238587e-01 -1.42628208e-01
-4.78983343e-01 -2.08061859e-01 3.52295995e-01 3.23187023e-01
2.26404712e-01 2.71164656e-01 -6.34932816e-01 -1.04625784e-01
-2.43805096e-01 2.88778245e-01 6.90877557e-01 1.60005939e+00
9.71696913e-01 -5.01244664e-01 1.23679318e-01 1.13056898e+00
4.87111509e-01 6.41019285e-01 1.50566071e-01 -8.82762849e-01
5.71186423e-01 8.18378627e-01 5.34755290e-01 -1.23502362e+00
-2.92602599e-01 -1.67090639e-01 -5.68932950e-01 -1.71397537e-01
4.26348388e-01 9.92704183e-02 -7.24993944e-01 1.64020061e+00
4.39171344e-01 -4.53478575e-01 1.77891284e-01 1.13771343e+00
1.06875670e+00 5.97322226e-01 3.72095585e-01 -1.17140688e-01
1.65158939e+00 -7.74300277e-01 -1.02965212e+00 -2.81587452e-01
9.51944709e-01 -8.71715486e-01 1.28168452e+00 3.51122051e-01
-7.42471993e-01 -5.90654075e-01 -1.12178421e+00 -1.87625960e-01
-6.42502785e-01 4.33445394e-01 9.68307495e-01 2.50299960e-01
-1.17552526e-01 2.22982019e-01 -3.60312313e-01 -1.64920285e-01
3.82428080e-01 4.04144794e-01 -5.30677140e-01 -2.37374738e-01
-1.22102654e+00 6.78335130e-01 6.34843647e-01 1.61123425e-01
-3.94622236e-01 -3.30179453e-01 -1.26763916e+00 1.27089217e-01
6.10464334e-01 -3.78415167e-01 1.26779127e+00 -1.13784349e+00
-8.29369545e-01 6.34798646e-01 -1.13998719e-01 2.00269282e-01
1.02224417e-01 -1.33693010e-01 -6.08541489e-01 -2.29115468e-02
1.94855243e-01 8.82061183e-01 7.15167701e-01 -1.50284004e+00
-9.73218977e-01 -5.59831917e-01 4.09745902e-01 2.70538718e-01
-5.05272865e-01 -2.78664529e-01 -6.11594379e-01 -4.90646958e-01
3.54820251e-01 -9.16957915e-01 2.28571221e-02 4.50158194e-02
-4.22072858e-01 -3.53099555e-01 6.34119451e-01 -2.54240006e-01
8.37544978e-01 -2.51049709e+00 2.60907896e-02 3.88720512e-01
1.27512768e-01 -4.35038000e-01 -3.86650681e-01 3.73181581e-01
3.43265990e-03 -3.38561013e-02 -2.27728903e-01 -1.14384994e-01
2.83384144e-01 4.77046937e-01 -1.10483781e-01 2.11050630e-01
2.55687684e-01 1.08711076e+00 -8.89888346e-01 -7.91712046e-01
2.15959921e-02 4.91330832e-01 -1.64833531e-01 1.29197121e-01
-2.42982522e-01 2.07498536e-01 -8.75836015e-01 1.02150619e+00
8.30972850e-01 -3.37548941e-01 3.04753780e-01 -6.84607863e-01
4.11889464e-01 3.85749377e-02 -1.25045300e+00 1.53464544e+00
-3.72967631e-01 3.15357596e-01 -9.90239978e-02 -7.87944317e-01
1.15210271e+00 8.89110342e-02 6.76250279e-01 -7.51049995e-01
2.87585467e-01 -3.48157175e-02 6.80251271e-02 -6.40156269e-01
8.25784743e-01 -1.30192444e-01 -2.29332045e-01 4.21854019e-01
-1.70716077e-01 -1.63590237e-01 4.92321998e-01 1.47611931e-01
1.94030955e-01 3.34113032e-01 3.55940580e-01 1.93785056e-01
4.65741217e-01 2.16807127e-01 5.55795729e-01 3.14227104e-01
-1.64270535e-01 4.97293830e-01 7.98936248e-01 -5.25114655e-01
-9.97971773e-01 -9.04416621e-01 -2.56607205e-01 1.26752281e+00
7.40125418e-01 -5.92423975e-01 -2.94163734e-01 -9.43351686e-01
1.75401539e-01 6.93107426e-01 -5.64465523e-01 -1.37200460e-01
-2.26799294e-01 -6.31504595e-01 -2.05122352e-01 7.43378878e-01
3.55396241e-01 -1.14632380e+00 -1.51641443e-01 1.55675083e-01
-3.85885954e-01 -1.26315701e+00 -5.16199529e-01 -8.59620422e-02
-6.58796668e-01 -1.06781435e+00 -4.14167851e-01 -8.79316986e-01
8.46637726e-01 4.56421793e-01 1.26445663e+00 4.28278223e-02
2.53422052e-01 3.84359568e-01 -5.92287719e-01 -4.76024240e-01
-2.29114249e-01 1.55578300e-01 -1.95113435e-01 4.12578434e-01
1.03733301e+00 -3.69266272e-01 -5.97673476e-01 5.24522305e-01
-7.75042951e-01 2.40577161e-01 9.45004582e-01 8.92994761e-01
8.41446340e-01 1.61923379e-01 4.95942146e-01 -1.15419698e+00
5.00231326e-01 -5.07215858e-01 -3.69642824e-01 3.16305757e-01
-7.10356653e-01 -1.46209048e-02 3.23476791e-01 -6.53706193e-01
-1.09267592e+00 5.25626600e-01 1.55435726e-01 -1.47885367e-01
-2.89813936e-01 8.29921663e-01 -6.99344516e-01 4.66930747e-01
1.26258940e-01 1.82338227e-02 8.18841532e-03 -5.01852751e-01
5.28186738e-01 7.57035494e-01 2.86651701e-01 -4.94911402e-01
4.59325492e-01 3.09704781e-01 5.87823391e-02 -4.85478818e-01
-9.59127665e-01 -5.57281256e-01 -7.56671548e-01 -2.85541713e-02
5.76377451e-01 -9.15795922e-01 -9.13144231e-01 9.21966359e-02
-1.07520604e+00 3.26238215e-01 -1.79308012e-01 6.24991834e-01
-5.08985996e-01 2.61663854e-01 -4.21874434e-01 -6.04457319e-01
1.27468854e-01 -1.30048501e+00 1.16228533e+00 2.11290240e-01
-2.91631609e-01 -8.47374558e-01 -5.76423287e-01 4.53881741e-01
8.30778014e-03 -4.24354133e-04 1.21524096e+00 -7.23436952e-01
-4.49660420e-01 -2.32441679e-01 -4.92829978e-01 1.62479445e-01
3.97821605e-01 -1.24364980e-01 -8.16734910e-01 -3.50248888e-02
-1.57539845e-01 -3.41684073e-01 6.90489531e-01 1.17809407e-01
1.14560282e+00 -2.41970241e-01 -3.44255716e-01 2.96023011e-01
1.10302126e+00 3.24670643e-01 5.65880060e-01 2.64820427e-01
9.29471731e-01 1.39969981e+00 1.35631084e+00 6.49190366e-01
6.06055021e-01 7.75817156e-01 7.21816480e-01 -2.61437744e-01
3.08325022e-01 -3.29013914e-01 1.15925409e-02 4.02401835e-01
1.28366962e-01 -7.24015087e-02 -4.86900955e-01 5.42853057e-01
-1.98495317e+00 -5.45150816e-01 -2.74809718e-01 2.07755280e+00
8.91805351e-01 -4.42511998e-02 1.93854839e-01 -1.04939409e-01
7.85663068e-01 8.63152519e-02 -7.06264734e-01 -6.48097172e-02
-1.19463637e-01 -6.87937498e-01 3.14464480e-01 4.70346659e-02
-1.17425942e+00 7.40372777e-01 5.78887367e+00 7.54628539e-01
-5.92772663e-01 -1.77812755e-01 7.41237998e-01 -2.03897491e-01
-6.98763669e-01 -1.77229136e-01 -7.01658130e-01 4.82385278e-01
2.42921710e-01 -7.62785524e-02 1.96665853e-01 1.04605365e+00
-5.04295304e-02 8.73750299e-02 -1.43779767e+00 1.33051658e+00
-7.29667172e-02 -6.66378736e-01 3.32777709e-01 3.18514973e-01
3.76497716e-01 -7.50359058e-01 3.19730937e-01 2.21530288e-01
2.34429613e-01 -9.19145226e-01 8.14942956e-01 1.55927867e-01
6.05178416e-01 -9.88968730e-01 9.25433278e-01 8.44930038e-02
-1.35194266e+00 -1.54151320e-01 -5.02681375e-01 1.76660553e-01
2.81238705e-01 4.36454922e-01 -6.52536571e-01 3.82794589e-01
5.96599102e-01 1.16987121e+00 -6.09875500e-01 5.39268851e-01
-2.97874451e-01 7.91477188e-02 4.59181070e-02 -1.20318010e-01
3.75336111e-02 -5.44601738e-01 6.11582166e-03 8.58876109e-01
4.28770274e-01 1.55489549e-01 1.64187774e-01 1.06976509e+00
-1.34435833e-01 4.14546221e-01 -6.11887991e-01 -4.88603532e-01
4.45353240e-01 1.52277327e+00 -6.09019399e-01 -2.99480617e-01
-9.62318301e-01 6.34435534e-01 1.21240623e-01 5.71752191e-01
-9.12641227e-01 -4.81119245e-01 9.94522512e-01 -2.76708845e-02
4.70907569e-01 -1.51578426e-01 -4.84987468e-01 -1.03631282e+00
1.68284371e-01 -6.95787251e-01 4.08756316e-01 -9.44301784e-01
-1.54093051e+00 4.31470931e-01 6.68062568e-02 -1.80301785e+00
-3.56850863e-01 -7.45366335e-01 2.42634639e-01 6.24614775e-01
-1.49515069e+00 -1.52480936e+00 -3.19505066e-01 5.62059641e-01
4.03608531e-01 7.06169903e-02 5.33685625e-01 1.84415624e-01
-2.16056883e-01 4.53772575e-01 -1.71755001e-01 3.74859124e-01
8.43449652e-01 -1.01571214e+00 1.16167635e-01 1.97535992e-01
4.00596350e-01 7.86101699e-01 6.41401052e-01 -5.30576468e-01
-1.54324377e+00 -6.36516094e-01 8.44141304e-01 -3.42515558e-01
5.82830131e-01 -4.06306893e-01 -9.34988976e-01 6.51413381e-01
1.41840726e-01 -9.60945711e-02 9.95621502e-01 3.96969527e-01
-5.90464950e-01 -2.49755025e-01 -1.05687678e+00 5.90981305e-01
9.41207945e-01 -4.96396929e-01 -3.68240416e-01 2.04226255e-01
6.97261393e-01 -4.40935120e-02 -1.11233664e+00 3.39765251e-01
6.93162739e-01 -7.82353222e-01 1.09301019e+00 -6.05113506e-01
5.61867416e-01 -4.63630319e-01 -4.05064493e-01 -1.27683973e+00
-2.90618181e-01 1.61053017e-01 -8.59590173e-02 1.56400621e+00
8.95134985e-01 -4.83346909e-01 7.40776300e-01 1.27673995e+00
2.57852644e-01 -6.41179800e-01 -4.01124567e-01 -6.22548103e-01
-3.96708280e-01 -4.22517776e-01 1.36531711e+00 9.86218691e-01
6.34524822e-02 5.73347449e-01 -4.41001743e-01 1.43487960e-01
3.58349979e-01 6.82164133e-01 6.88649535e-01 -1.24949205e+00
-2.09837779e-01 -2.10022852e-01 -3.91601235e-01 -1.24488151e+00
1.13974102e-01 -5.82386911e-01 -1.60996377e-01 -1.48375285e+00
5.48440099e-01 -6.38558507e-01 -5.18705726e-01 7.52133548e-01
-3.19114253e-02 2.07774788e-01 2.08460808e-01 5.02412558e-01
-5.76868653e-01 5.97341776e-01 1.58347106e+00 -5.82898796e-01
-2.21905604e-01 -9.84823331e-02 -9.65839326e-01 5.69312096e-01
7.66356170e-01 -3.83275062e-01 -7.70734966e-01 -2.99815744e-01
5.88865757e-01 -5.52735589e-02 9.48811918e-02 -9.33921710e-02
-1.28849059e-01 -5.49621820e-01 7.06314683e-01 -6.09808147e-01
6.39980316e-01 -1.34170973e+00 3.66403125e-02 -1.31241828e-01
-4.60356444e-01 6.86860308e-02 -1.31071612e-01 4.89805073e-01
-5.87581813e-01 -4.16258961e-01 2.88723677e-01 -1.31889448e-01
-1.15561891e+00 3.72883648e-01 1.11407496e-01 -4.37619001e-01
9.15416121e-01 -3.36343110e-01 -2.92899042e-01 -7.20168054e-01
-8.36361885e-01 4.32262063e-01 5.80612957e-01 7.73446679e-01
6.89320087e-01 -1.75436544e+00 -4.83905017e-01 3.65838259e-01
8.26429307e-01 -1.56281680e-01 2.16400474e-01 7.16930091e-01
1.12780564e-01 1.99011359e-02 -2.13781029e-01 -7.02564776e-01
-1.31513751e+00 1.06902587e+00 -2.02499554e-01 2.44055524e-01
-1.28169626e-01 3.01434636e-01 5.84383547e-01 -3.04181308e-01
1.30030543e-01 -7.18325144e-03 -7.12845802e-01 5.36103368e-01
4.80708539e-01 1.91491024e-04 -2.71154791e-01 -9.99217510e-01
-4.76989895e-02 6.13786280e-01 -4.84604686e-01 -1.10003278e-01
1.21368253e+00 -6.28724039e-01 -1.56432725e-02 4.40238357e-01
1.35997629e+00 -3.03561926e-01 -1.41053212e+00 -5.36864698e-01
-2.12997217e-02 -7.80392587e-01 -1.03013434e-01 -7.68081784e-01
-1.35396159e+00 7.85139501e-01 2.98198372e-01 3.38192940e-01
1.30175674e+00 4.04163510e-01 7.80086219e-01 3.57601732e-01
4.21524346e-01 -1.11458349e+00 -1.01203201e-02 4.76208068e-02
7.28606939e-01 -1.67920411e+00 7.45438933e-02 -9.58705246e-01
-1.31333888e+00 1.00505614e+00 6.65733814e-01 6.34836495e-01
4.76727217e-01 -2.68212166e-02 2.22041696e-01 -2.77470618e-01
-5.30811965e-01 -3.79555345e-01 3.64679515e-01 7.48698235e-01
6.07794940e-01 2.87243038e-01 -9.48148593e-02 8.86200070e-01
-1.67618260e-01 -3.23494494e-01 3.44581515e-01 8.93351018e-01
-1.90514922e-01 -1.32776320e+00 -1.22249387e-01 3.57762992e-01
-1.77095145e-01 -7.49755576e-02 -5.10655403e-01 8.74628484e-01
1.54367417e-01 1.12599099e+00 1.63197517e-04 -3.26399505e-01
4.43323433e-01 -2.01004192e-01 3.15162838e-01 -3.16545695e-01
-1.02805980e-01 2.04736337e-01 2.83908963e-01 -3.57185811e-01
-5.32166600e-01 -6.90882444e-01 -1.26701725e+00 -2.22154662e-01
-2.17391357e-01 2.27142908e-02 7.74538696e-01 6.89094484e-01
4.12365586e-01 6.79441690e-02 7.01707661e-01 -5.51061809e-01
-4.07193929e-01 -7.81698704e-01 -1.03917587e+00 9.67442214e-01
1.56935573e-01 -1.04132831e+00 -1.72091946e-01 2.93468714e-01] | [10.65604019165039, 1.9439187049865723] |
efeaeb76-ca9b-4c86-bdb7-965ebaa5ec0c | neural-airport-ground-handling | 2303.02442 | null | https://arxiv.org/abs/2303.02442v1 | https://arxiv.org/pdf/2303.02442v1.pdf | Neural Airport Ground Handling | Airport ground handling (AGH) offers necessary operations to flights during their turnarounds and is of great importance to the efficiency of airport management and the economics of aviation. Such a problem involves the interplay among the operations that leads to NP-hard problems with complex constraints. Hence, existing methods for AGH are usually designed with massive domain knowledge but still fail to yield high-quality solutions efficiently. In this paper, we aim to enhance the solution quality and computation efficiency for solving AGH. Particularly, we first model AGH as a multiple-fleet vehicle routing problem (VRP) with miscellaneous constraints including precedence, time windows, and capacity. Then we propose a construction framework that decomposes AGH into sub-problems (i.e., VRPs) in fleets and present a neural method to construct the routing solutions to these sub-problems. In specific, we resort to deep learning and parameterize the construction heuristic policy with an attention-based neural network trained with reinforcement learning, which is shared across all sub-problems. Extensive experiments demonstrate that our method significantly outperforms classic meta-heuristics, construction heuristics and the specialized methods for AGH. Besides, we empirically verify that our neural method generalizes well to instances with large numbers of flights or varying parameters, and can be readily adapted to solve real-time AGH with stochastic flight arrivals. Our code is publicly available at: https://github.com/RoyalSkye/AGH. | ['Jie Zhang', 'Zhiguang Cao', 'Xianli Zhang', 'Yunwen Xia', 'Jianan Zhou', 'Yaoxin Wu'] | 2023-03-04 | null | null | null | null | ['combinatorial-optimization'] | ['methodology'] | [-3.26876640e-01 -2.55400501e-02 -1.07565127e-01 -1.27889752e-01
-6.53244913e-01 -7.51714706e-01 -2.43464291e-01 1.26832142e-01
-2.31347680e-01 9.19847608e-01 -2.76201010e-01 -8.27249706e-01
-7.35580206e-01 -1.19316304e+00 -9.12899613e-01 -6.04486823e-01
-4.35376018e-01 7.63848662e-01 -1.64479315e-01 -6.77125275e-01
7.13290274e-02 6.19926870e-01 -1.11726415e+00 -8.01349431e-02
7.57252932e-01 1.16700721e+00 2.11162925e-01 4.82284397e-01
1.48657724e-01 2.05108225e-01 -6.74748003e-01 -3.67356151e-01
8.26502681e-01 1.82340711e-01 -9.22636330e-01 3.01776707e-01
-1.92566201e-01 -1.94930121e-01 -3.25680137e-01 6.82498157e-01
3.24429929e-01 6.05352104e-01 9.94783118e-02 -1.65044141e+00
-5.71710587e-01 5.89223385e-01 -5.41003764e-01 1.55438676e-01
-7.64259370e-03 3.83579433e-01 1.24666870e+00 -3.78371626e-01
3.31022382e-01 1.01582944e+00 4.73421335e-01 2.94456780e-01
-9.38479602e-01 -3.96586746e-01 7.50227273e-01 7.11769089e-02
-1.33130729e+00 1.61196455e-01 5.37662387e-01 -8.40642378e-02
1.12960088e+00 2.71903276e-01 9.76179004e-01 6.29143119e-01
3.50470752e-01 8.57769191e-01 6.28980219e-01 8.10889229e-02
2.88673639e-01 -4.66686755e-01 -1.57212451e-01 7.68514812e-01
2.87295580e-01 1.90552205e-01 2.75626868e-01 1.97053179e-01
7.03216016e-01 2.18186572e-01 -2.25643262e-01 -2.10625693e-01
-9.94171739e-01 1.11945522e+00 6.39118969e-01 -9.99568626e-02
-6.14934385e-01 3.73124987e-01 4.38353330e-01 4.20851439e-01
1.66225150e-01 1.04058301e+00 -7.75191665e-01 7.25203454e-02
-7.19260693e-01 6.75197482e-01 8.63205135e-01 1.42967987e+00
5.69038570e-01 1.77260265e-01 -1.76043555e-01 6.06134117e-01
-1.79018974e-01 3.60869288e-01 -3.82702023e-01 -8.76080036e-01
8.82898927e-01 3.34481239e-01 4.76677746e-01 -1.10452080e+00
-8.07176769e-01 -6.55578613e-01 -6.97531641e-01 -1.60210311e-01
6.46586120e-02 -6.03176832e-01 -8.44242513e-01 1.49572086e+00
2.52087206e-01 5.94549552e-02 -1.12795830e-01 1.17309368e+00
3.18210244e-01 1.12386405e+00 -3.04880440e-01 -2.28616521e-01
1.01518548e+00 -1.43891907e+00 -6.13706052e-01 -3.30525696e-01
5.76414943e-01 -5.29093206e-01 7.39894748e-01 4.99636859e-01
-1.60416555e+00 -3.32088888e-01 -9.44516659e-01 1.36110261e-01
-5.76309502e-01 -1.57665496e-03 9.65095222e-01 2.52443254e-01
-1.01355505e+00 4.32514489e-01 -5.10873735e-01 1.74317196e-01
2.71026820e-01 7.78342724e-01 -6.21909834e-02 -4.35524702e-01
-1.43010771e+00 6.46301866e-01 5.73280931e-01 8.33109438e-01
-9.84188259e-01 -5.33160567e-01 -9.20002878e-01 4.69741017e-01
1.31183398e+00 -8.41675580e-01 1.41632378e+00 -5.59348583e-01
-1.35620165e+00 1.49784133e-01 3.03049237e-01 -4.15796578e-01
3.27120662e-01 -2.09052227e-02 -3.61925691e-01 -1.79521203e-01
-4.97226603e-02 4.93828475e-01 6.62060797e-01 -1.15478778e+00
-1.13087559e+00 1.28320679e-01 9.86115515e-01 9.22057331e-02
1.24464609e-01 -3.42190355e-01 -4.85836416e-01 -4.14686769e-01
-1.70916229e-01 -1.10570502e+00 -7.55729377e-01 -7.24422574e-01
-4.35492963e-01 -1.56638548e-01 9.56147835e-02 -5.63412428e-01
1.47063231e+00 -1.92598927e+00 5.33858001e-01 5.37049532e-01
1.91398188e-01 1.65445402e-01 -5.38300037e-01 8.31578612e-01
3.13505158e-02 1.79912344e-01 -1.85879722e-01 -2.51831982e-04
4.58707899e-01 4.76623625e-01 -3.48280609e-01 1.43695951e-01
4.11725342e-01 1.01640737e+00 -1.02678835e+00 7.69291148e-02
6.45144191e-03 -1.66678429e-01 -8.34653854e-01 2.40810290e-01
-4.53241050e-01 1.07922852e-01 -3.98778021e-01 1.07283890e+00
8.12346697e-01 -3.75462919e-02 3.37641329e-01 1.72563583e-01
-2.91801691e-01 3.80380787e-02 -1.14654636e+00 1.46901739e+00
-8.31456721e-01 2.51673788e-01 4.63553578e-01 -1.22343647e+00
7.06686616e-01 -6.79364577e-02 5.02197385e-01 -6.70398533e-01
3.66598725e-01 2.56150156e-01 7.21249580e-02 -3.95666063e-01
9.33861911e-01 -5.30607440e-02 -4.94602710e-01 4.77802828e-02
-3.10491443e-01 -1.60468414e-01 5.49710810e-01 -1.19648859e-01
1.05869234e+00 -3.63240540e-01 7.67425373e-02 2.88388785e-02
2.62896329e-01 1.49396911e-01 9.52323318e-01 6.05778813e-01
1.39500320e-01 8.96661654e-02 9.40773010e-01 -7.88905919e-01
-7.71724820e-01 -8.60135138e-01 3.31422031e-01 9.50828135e-01
3.34121943e-01 -2.57282436e-01 -5.77349186e-01 -5.75529397e-01
2.34449834e-01 6.26984417e-01 -5.25002241e-01 -1.80217437e-02
-5.62340915e-01 -7.28449702e-01 -4.98810410e-03 6.40426993e-01
2.06187725e-01 -9.33235586e-01 -5.41901827e-01 5.67420602e-01
-1.28972977e-01 -1.23098803e+00 -5.29083669e-01 3.15642953e-01
-4.65472788e-01 -1.03112042e+00 -3.99398625e-01 -8.08836460e-01
7.95368850e-01 5.57383001e-01 1.09318316e+00 3.95939618e-01
-3.62687469e-01 2.08316654e-01 -3.81871492e-01 -3.66973817e-01
2.76283473e-01 4.61240977e-01 -1.27125069e-01 -1.49444237e-01
-1.42467722e-01 -1.02457508e-01 -4.53574538e-01 4.78311330e-01
-9.27120924e-01 -8.77083987e-02 6.51207447e-01 1.05602849e+00
8.18640411e-01 7.71995842e-01 5.84194124e-01 -8.97119701e-01
8.65942717e-01 -6.47445083e-01 -1.17230344e+00 2.69426554e-01
-3.66461664e-01 -2.12657273e-01 9.49779332e-01 -2.19864901e-02
-5.48647344e-01 -2.81506360e-01 -8.97769257e-02 -5.84361851e-01
-8.37767683e-03 1.00904047e+00 -1.98099107e-01 -2.25513242e-02
-3.15192968e-01 -1.44596323e-01 -3.79299283e-01 8.94818157e-02
5.36714420e-02 1.04834378e-01 1.15222394e-01 -9.82875228e-01
8.71670663e-01 2.30653640e-02 3.36850226e-01 -2.89231837e-01
-8.67713928e-01 -2.98532695e-01 -1.77787915e-01 -1.09985903e-01
5.41599929e-01 -5.90595722e-01 -1.11737752e+00 5.74951731e-02
-9.54015434e-01 -6.45872951e-01 -8.51942748e-02 2.41426423e-01
-8.20204437e-01 -9.47308913e-03 -7.08215952e-01 -8.17710161e-01
1.28972828e-01 -1.49860156e+00 9.88638818e-01 1.01430446e-01
4.44424272e-01 -9.22125340e-01 -1.96382955e-01 5.12040019e-01
3.18193138e-01 3.71848673e-01 1.03790164e+00 -2.46748671e-01
-1.03687525e+00 -2.64894158e-01 -2.15440497e-01 1.15530789e-01
-7.45045245e-02 4.91352789e-02 -2.21967191e-01 -6.64876997e-01
-5.76653063e-01 -1.41795233e-01 5.76114774e-01 6.88176751e-01
1.65138602e+00 -5.81180632e-01 -2.35306367e-01 8.31428528e-01
1.59172809e+00 5.18494844e-01 5.54035485e-01 5.28717637e-01
4.22204077e-01 8.81663382e-01 1.29951370e+00 8.13056827e-01
4.83126193e-01 4.54007298e-01 9.31840777e-01 -2.62650102e-01
7.63594806e-01 4.37604710e-02 2.73081753e-02 8.43837917e-01
-4.16589260e-01 -9.01561260e-01 -9.40874100e-01 5.90054333e-01
-2.13076830e+00 -7.91581631e-01 1.13258652e-01 2.10293055e+00
1.55510619e-01 3.07117462e-01 1.65233448e-01 -1.81846060e-02
6.49465799e-01 1.08451039e-01 -4.69978034e-01 -9.33256567e-01
2.03997999e-01 1.50234178e-01 9.07746553e-01 5.27366161e-01
-1.17369974e+00 1.07945120e+00 5.57624245e+00 5.97354710e-01
-8.21730793e-01 -2.35716164e-01 5.34860432e-01 -6.02242470e-01
-1.97934434e-01 -6.72129989e-02 -8.23759198e-01 4.88230854e-01
9.70007122e-01 -1.06387064e-01 1.20862305e+00 5.82398057e-01
2.12013423e-01 3.30269903e-01 -9.97039199e-01 7.11564124e-01
-4.38136458e-01 -1.50968289e+00 -1.98573261e-01 3.05675566e-01
9.67370093e-01 -2.77690709e-01 3.12612653e-01 7.60864317e-01
4.20592397e-01 -1.02248156e+00 6.10635638e-01 2.16885343e-01
4.02163297e-01 -1.47568274e+00 1.20023131e+00 3.02905310e-02
-1.39871943e+00 -8.55444729e-01 -5.18452287e-01 -9.13803503e-02
5.10944128e-01 2.72307754e-01 -7.02101350e-01 1.08360553e+00
5.14119387e-01 4.51931894e-01 6.89185113e-02 1.15945590e+00
-7.24378526e-02 5.64248674e-02 -8.39151964e-02 -3.11177922e-03
1.08653760e+00 -4.42564666e-01 2.99680412e-01 8.34527016e-01
3.89484286e-01 2.67446518e-01 7.84662426e-01 6.60761774e-01
-3.39607894e-02 -1.42731592e-01 -5.62433541e-01 -3.66291136e-01
4.75073248e-01 1.34779918e+00 -8.76673162e-01 1.32276893e-01
-4.65758473e-01 4.59314406e-01 3.32320571e-01 6.83611810e-01
-1.42364848e+00 -6.50754809e-01 8.08280647e-01 7.00863972e-02
5.43515980e-01 -4.97510105e-01 1.79659322e-01 -7.34957933e-01
8.75697657e-02 -8.28006864e-01 4.46211666e-01 -6.20953083e-01
-1.25823987e+00 6.45752788e-01 5.71493572e-03 -1.14734650e+00
-2.56395992e-02 -1.02649844e+00 -5.18445671e-01 5.81843376e-01
-2.06570840e+00 -9.03102756e-01 -2.50949472e-01 3.91096979e-01
6.84523344e-01 -8.48308131e-02 2.85014242e-01 4.92973238e-01
-1.07780600e+00 4.34638739e-01 -5.64192608e-02 -9.67345014e-02
2.04110742e-01 -1.18990767e+00 4.61587191e-01 6.69393420e-01
-5.29853284e-01 5.75544238e-01 4.19538110e-01 -3.63040507e-01
-1.92457867e+00 -1.30747831e+00 4.45354670e-01 1.65868048e-02
8.48147631e-01 -2.77227759e-01 -4.22880083e-01 8.44470561e-01
1.26889408e-01 -1.27010390e-01 5.82340717e-01 3.96148920e-01
3.17210823e-01 -2.56482989e-01 -8.40437233e-01 6.62436724e-01
1.13365591e+00 1.18221559e-01 -2.48039529e-01 7.49266922e-01
1.11322343e+00 -8.47877085e-01 -8.76974583e-01 2.67862380e-01
6.55843690e-02 -6.06618106e-01 8.83656323e-01 -9.88148928e-01
3.75197083e-01 -1.95759416e-01 -1.95539534e-01 -1.56229579e+00
-6.98892772e-01 -7.82683194e-01 8.62351879e-02 7.49928236e-01
5.69869459e-01 -8.43171954e-01 4.32669550e-01 6.98189795e-01
-7.48790801e-01 -1.33065128e+00 -8.25599670e-01 -1.01975417e+00
-1.01207338e-01 -2.34129503e-01 1.27407193e+00 1.00007701e+00
-3.54594439e-01 -7.28868321e-02 -4.01354253e-01 7.12728918e-01
2.09073499e-01 6.96532786e-01 5.38789213e-01 -1.14505613e+00
-3.50766182e-01 -5.49046278e-01 3.39110009e-02 -8.59007895e-01
4.09698397e-01 -7.03949451e-01 2.26696014e-01 -1.76710653e+00
-5.77407360e-01 -8.11807156e-01 -5.58867931e-01 7.14018166e-01
3.56527269e-01 -4.96331960e-01 3.02992761e-01 -3.96724880e-01
-8.41399431e-01 5.51262319e-01 1.49693263e+00 -4.04104143e-01
-3.39231342e-01 2.15773687e-01 -7.65752196e-01 2.93236643e-01
9.08428371e-01 -2.26948231e-01 -5.62985718e-01 -7.87219644e-01
7.66129255e-01 6.57202959e-01 1.61121994e-01 -7.89766133e-01
8.52944851e-02 -7.78904438e-01 -1.11451536e-01 -5.03736198e-01
2.54158109e-01 -1.11383533e+00 3.74788940e-02 3.93069625e-01
-1.60818204e-01 8.58249784e-01 4.55570012e-01 5.09991467e-01
-2.97441721e-01 -2.66573280e-01 1.92065716e-01 -1.94024086e-01
-7.99234390e-01 9.07561958e-01 -4.67561007e-01 4.71231677e-02
1.26542759e+00 1.19944431e-01 -3.24340612e-01 -1.33140072e-01
-6.69759929e-01 1.18457794e+00 2.14809641e-01 4.00046557e-01
7.62640119e-01 -1.05893552e+00 -4.33939368e-01 1.66167289e-01
-1.42214939e-01 4.23922926e-01 6.36522114e-01 8.82564723e-01
-8.29284251e-01 7.76761234e-01 -4.74772304e-01 3.04736346e-02
-5.28469801e-01 1.13386357e+00 5.95812090e-02 -5.39416134e-01
-3.27266872e-01 7.81297266e-01 -1.25517347e-03 -5.13135016e-01
2.33538076e-01 -8.70518506e-01 2.19273850e-01 1.02381565e-01
1.82651713e-01 3.88248146e-01 1.05165832e-01 6.60321936e-02
-2.18782425e-01 3.78711253e-01 -1.32077292e-01 3.27743292e-01
1.49906933e+00 4.82174344e-02 -9.24251601e-02 -1.50538564e-01
6.89008594e-01 -2.16970086e-01 -1.21620083e+00 3.05878848e-01
-3.47856581e-01 -5.42526603e-01 1.56545356e-01 -6.95893526e-01
-1.66540945e+00 8.05879414e-01 -2.21724436e-01 4.55798298e-01
1.37361002e+00 -5.68921566e-01 1.23195589e+00 7.95647144e-01
7.20536470e-01 -1.30503619e+00 -2.86359042e-01 6.88515425e-01
7.70585418e-01 -9.61437404e-01 -1.72786787e-01 -3.79061133e-01
-6.46999776e-01 1.13864112e+00 7.91701138e-01 -2.45273948e-01
2.77917624e-01 2.41633117e-01 -2.56866693e-01 -2.17027515e-01
-1.11131954e+00 -3.29324484e-01 -1.19989542e-02 2.29336888e-01
-1.43991828e-01 3.90006870e-01 -2.96008110e-01 8.39960396e-01
-1.44082323e-01 -6.15073256e-02 6.84880078e-01 1.11348498e+00
-2.03544542e-01 -1.00100744e+00 -1.53564170e-01 2.75213927e-01
-4.90722172e-02 5.63811362e-02 -5.25133079e-03 1.08945107e+00
1.48301721e-01 8.41319561e-01 1.12399705e-01 -4.82457548e-01
5.02913296e-01 -4.20373470e-01 3.69613349e-01 -5.51662445e-01
-7.79963791e-01 -4.46154084e-03 2.91345209e-01 -7.39613056e-01
1.42239239e-02 -3.14029694e-01 -1.27542078e+00 -6.46874785e-01
-2.13623852e-01 4.42110568e-01 4.17537391e-01 7.24273622e-01
4.11578536e-01 1.09486735e+00 8.50075781e-01 -7.80950725e-01
-4.66385990e-01 -8.12899247e-02 -8.38168800e-01 -1.89633250e-01
3.68425012e-01 -1.02910244e+00 -4.75174822e-02 -6.89410031e-01] | [5.098991394042969, 2.780911445617676] |
39fd5392-7e99-4141-867e-579de580bdd5 | efficient-video-object-segmentation-via | 1802.01218 | null | http://arxiv.org/abs/1802.01218v1 | http://arxiv.org/pdf/1802.01218v1.pdf | Efficient Video Object Segmentation via Network Modulation | Video object segmentation targets at segmenting a specific object throughout
a video sequence, given only an annotated first frame. Recent deep learning
based approaches find it effective by fine-tuning a general-purpose
segmentation model on the annotated frame using hundreds of iterations of
gradient descent. Despite the high accuracy these methods achieve, the
fine-tuning process is inefficient and fail to meet the requirements of real
world applications. We propose a novel approach that uses a single forward pass
to adapt the segmentation model to the appearance of a specific object.
Specifically, a second meta neural network named modulator is learned to
manipulate the intermediate layers of the segmentation network given limited
visual and spatial information of the target object. The experiments show that
our approach is 70times faster than fine-tuning approaches while achieving
similar accuracy. | ['Aggelos K. Katsaggelos', 'Yanran Wang', 'Xuehan Xiong', 'Linjie Yang', 'Jianchao Yang'] | 2018-02-04 | efficient-video-object-segmentation-via-1 | http://openaccess.thecvf.com/content_cvpr_2018/html/Yang_Efficient_Video_Object_CVPR_2018_paper.html | http://openaccess.thecvf.com/content_cvpr_2018/papers/Yang_Efficient_Video_Object_CVPR_2018_paper.pdf | cvpr-2018-6 | ['video-instance-segmentation'] | ['computer-vision'] | [ 1.16283081e-01 5.10999300e-02 -4.12401944e-01 -5.21901786e-01
-5.10210156e-01 -5.54524958e-01 3.46345663e-01 -1.15405925e-01
-5.63553274e-01 5.56292355e-01 -3.21597785e-01 -7.24173784e-02
4.08026695e-01 -5.34669816e-01 -1.07620573e+00 -5.44632077e-01
8.23332146e-02 4.93447036e-01 9.00618434e-01 3.68076339e-02
4.00958508e-01 5.62462270e-01 -1.30755377e+00 4.63953614e-01
6.85947180e-01 9.48765934e-01 4.72032458e-01 8.86051118e-01
-2.02373967e-01 6.89674795e-01 -8.03103864e-01 -2.40846500e-02
3.79015893e-01 -6.02971256e-01 -1.13693941e+00 5.60372114e-01
5.97260892e-01 -4.29557860e-01 -2.89793134e-01 1.08226502e+00
9.82595235e-02 3.49653244e-01 2.49948472e-01 -9.48721886e-01
-2.99489260e-01 4.61534947e-01 -7.12677717e-01 5.57401717e-01
1.35067431e-03 4.28868741e-01 7.35327661e-01 -4.62176114e-01
5.83715439e-01 1.17897868e+00 4.66025472e-01 6.62754297e-01
-1.06901121e+00 -3.20562303e-01 6.16051137e-01 1.02537774e-01
-1.13760161e+00 -3.71318191e-01 6.15932882e-01 -3.65928382e-01
8.66163790e-01 -4.14040089e-02 8.46008420e-01 5.42052805e-01
2.42052183e-01 7.06391215e-01 7.08747149e-01 -2.08375841e-01
3.94779623e-01 -1.49238199e-01 8.41374137e-03 8.71740341e-01
1.15304198e-02 -2.32575331e-02 -3.40192080e-01 1.27720505e-01
1.13780189e+00 -1.39250651e-01 -6.31718934e-02 -5.01219094e-01
-1.21024585e+00 7.20887780e-01 6.76583290e-01 3.30427170e-01
-4.62362289e-01 6.69428110e-01 4.19679910e-01 1.54218286e-01
2.51598388e-01 5.05026698e-01 -7.87691057e-01 -1.87058151e-01
-1.33828509e+00 7.84586966e-02 6.89769030e-01 8.94051850e-01
9.60426867e-01 2.31072977e-01 -2.60529339e-01 4.65776205e-01
4.00445819e-01 -1.05766349e-01 5.39301395e-01 -1.38038623e+00
1.99044704e-01 5.67243338e-01 7.45719373e-02 -6.26618326e-01
-3.70641351e-01 -3.41516912e-01 -2.50369668e-01 5.12035370e-01
5.65676510e-01 -4.28929180e-01 -1.45036173e+00 1.47729278e+00
6.26016021e-01 2.05340505e-01 -3.92099261e-01 1.11150551e+00
5.16967177e-01 8.35702896e-01 3.04742604e-01 -9.38098654e-02
1.00843155e+00 -1.55170274e+00 -3.09328496e-01 -6.40727341e-01
3.71201336e-01 -5.38754344e-01 9.96657431e-01 2.37181589e-01
-1.37018251e+00 -6.65706098e-01 -1.06354296e+00 4.60251309e-02
-8.92899781e-02 2.72236727e-02 6.64449453e-01 5.68140626e-01
-1.09388506e+00 6.21872604e-01 -1.09127140e+00 -3.16197783e-01
8.17304969e-01 5.37524998e-01 -4.65325154e-02 1.89145416e-01
-5.34217179e-01 5.77113867e-01 8.15396369e-01 -1.60777736e-02
-1.27544665e+00 -8.19124460e-01 -6.76546574e-01 3.05615459e-02
5.49594700e-01 -8.20167899e-01 1.44762731e+00 -1.70378029e+00
-1.80511832e+00 8.68080378e-01 -1.72787994e-01 -4.77957636e-01
5.44784725e-01 -3.70521605e-01 2.62608141e-01 4.76236790e-01
-8.52504224e-02 1.13544309e+00 1.17919326e+00 -1.27671444e+00
-9.61132824e-01 -1.36475101e-01 5.20314693e-01 2.85594970e-01
4.06449735e-02 9.68371853e-02 -1.07374823e+00 -5.90394616e-01
4.85354336e-03 -8.52643788e-01 -7.41353691e-01 1.25340953e-01
-3.32902372e-01 5.34943305e-02 1.07100129e+00 -3.76426071e-01
1.08161926e+00 -2.08420777e+00 2.09944367e-01 9.42669660e-02
1.70227662e-01 5.76882541e-01 -2.17214480e-01 -2.29985770e-02
6.90911487e-02 2.55129393e-02 -2.23542586e-01 -1.68719649e-01
-4.60426033e-01 2.20408082e-01 7.10258335e-02 5.34761012e-01
1.20394923e-01 1.15121973e+00 -8.24130714e-01 -5.62045336e-01
3.17613333e-01 2.10291639e-01 -8.10589314e-01 3.57836813e-01
-8.20270598e-01 6.85111165e-01 -5.22943199e-01 4.23107177e-01
4.09699589e-01 -4.13682789e-01 2.58545607e-01 -2.19149530e-01
2.81120520e-02 3.17676598e-03 -1.05304682e+00 1.84747803e+00
-1.69113487e-01 6.67392433e-01 2.05625743e-01 -1.22227192e+00
5.94856620e-01 3.97169217e-03 5.88681698e-01 -3.94844651e-01
2.91989177e-01 -3.03246994e-02 1.23147264e-01 -6.29405499e-01
2.13347152e-01 6.97530285e-02 1.68094322e-01 3.59682202e-01
1.58779249e-01 -1.70186460e-01 4.09884393e-01 -4.52221148e-02
9.65323150e-01 4.29294795e-01 1.97610945e-01 -2.65417516e-01
3.93286765e-01 4.11253810e-01 5.78468204e-01 7.64379203e-01
-2.59965152e-01 7.10416555e-01 3.68309498e-01 -7.85396218e-01
-1.21988678e+00 -6.66215479e-01 3.28111768e-01 1.48780322e+00
4.11567628e-01 -2.45733544e-01 -1.36719298e+00 -8.64780724e-01
-2.40870550e-01 4.20862496e-01 -5.98891556e-01 -7.93460310e-02
-9.82232094e-01 -6.12682998e-01 3.41816366e-01 7.08556354e-01
7.66097903e-01 -1.21697879e+00 -1.30898988e+00 2.21064270e-01
4.41546291e-02 -1.20862973e+00 -7.20801771e-01 1.28056481e-01
-1.17841721e+00 -1.17596436e+00 -6.74403429e-01 -1.02239990e+00
9.37573850e-01 1.32457063e-01 1.15488732e+00 5.41878223e-01
-2.53894329e-01 1.59997672e-01 -1.74272373e-01 8.09772685e-02
-4.04313594e-01 3.65871221e-01 -4.37401325e-01 1.58570204e-02
1.80743128e-01 -2.99741894e-01 -6.60951376e-01 1.90688208e-01
-9.70262170e-01 7.04619586e-02 5.03252268e-01 6.34808719e-01
6.90832317e-01 -1.12298533e-01 3.32846165e-01 -9.26586926e-01
3.02828640e-01 -1.93523198e-01 -8.39025259e-01 1.60818487e-01
-2.63483465e-01 -1.53793739e-02 5.43644965e-01 -6.48391604e-01
-8.66330624e-01 5.67996621e-01 -9.67403874e-02 -5.99229574e-01
-3.48952353e-01 2.69361943e-01 -5.81036098e-02 -3.16291749e-01
5.63602149e-01 -8.14878494e-02 -1.48442909e-01 -3.17125082e-01
3.11609685e-01 8.99813622e-02 5.60614824e-01 -4.55356628e-01
6.78540945e-01 5.69418788e-01 -2.53116429e-01 -6.17577851e-01
-9.96137142e-01 -3.56823832e-01 -9.21237886e-01 -3.04389060e-01
1.11978102e+00 -6.05105400e-01 -6.37419641e-01 5.35324395e-01
-1.15521836e+00 -9.87685978e-01 -4.55455661e-01 2.29144454e-01
-8.38896930e-01 1.11259311e-01 -5.78044355e-01 -2.91675359e-01
-1.26693249e-01 -1.24936271e+00 1.00876558e+00 4.81900811e-01
-1.18246570e-01 -1.00882375e+00 -1.12093344e-01 2.67638803e-01
3.78495157e-01 1.81443706e-01 7.54344285e-01 -4.54436570e-01
-1.04331791e+00 -9.11318325e-03 -2.40231872e-01 1.58083051e-01
2.30182678e-01 2.58158386e-01 -6.72989964e-01 -2.74531215e-01
-3.41747217e-02 -2.45915323e-01 8.06536376e-01 8.21706295e-01
1.38680065e+00 -2.02636182e-01 -5.03815949e-01 9.40470219e-01
1.45963395e+00 3.87165576e-01 5.35712361e-01 5.89292347e-01
8.93422067e-01 2.54021376e-01 7.24156916e-01 2.85738170e-01
2.25699633e-01 6.00474775e-01 6.12004995e-01 -2.73659527e-01
-2.30956838e-01 9.76100340e-02 1.83619007e-01 4.42464687e-02
-5.18983305e-02 -2.83223391e-02 -8.31376970e-01 5.99991024e-01
-1.97019708e+00 -8.83170962e-01 2.31151015e-01 1.91919732e+00
9.05305326e-01 3.41690361e-01 2.61229396e-01 -3.63704652e-01
8.31141710e-01 2.24639058e-01 -8.59570324e-01 -4.68552321e-01
2.61747390e-01 2.86133066e-02 6.05246067e-01 5.48976243e-01
-1.30718541e+00 1.47001278e+00 7.72057295e+00 5.43624997e-01
-1.40982771e+00 1.35222599e-01 7.52859771e-01 -1.47052497e-01
7.12980852e-02 7.19153360e-02 -7.72083223e-01 3.47740471e-01
8.59058022e-01 2.40012687e-02 5.68002343e-01 6.88233137e-01
3.65652829e-01 -3.07854146e-01 -1.22019279e+00 8.31736982e-01
-3.02756149e-02 -1.70106590e+00 1.14648268e-01 -1.59646496e-01
1.00877881e+00 1.41201809e-01 -1.10196166e-01 1.38957854e-02
2.70500988e-01 -9.76483166e-01 8.62739444e-01 4.23597604e-01
4.56669062e-01 -6.63798809e-01 2.55063564e-01 5.00432491e-01
-9.33078885e-01 -2.03406081e-01 -2.27148309e-01 1.12688001e-02
2.91234612e-01 2.11832449e-01 -6.86600387e-01 -1.85936347e-01
8.46313059e-01 3.06162745e-01 -6.30350709e-01 1.40110123e+00
-2.40169570e-01 7.12516010e-01 -2.40171716e-01 2.09327877e-01
7.39346623e-01 -6.00074716e-02 5.92352271e-01 1.23812985e+00
-1.24148212e-01 1.82406098e-01 5.43906450e-01 7.45932519e-01
-1.73521996e-01 -1.92575991e-01 -2.43993402e-01 -6.69203550e-02
1.43343717e-01 1.24006903e+00 -1.15375173e+00 -5.75901568e-01
-2.85220176e-01 1.18600011e+00 3.22952956e-01 5.83673120e-01
-1.02919447e+00 -1.88815564e-01 5.24673939e-01 6.93762228e-02
8.86341095e-01 -3.95921558e-01 -4.43780363e-01 -8.14584196e-01
-6.19870536e-02 -9.86715376e-01 2.20504090e-01 -6.99994266e-01
-6.02437556e-01 5.54990232e-01 -9.27121714e-02 -8.58104289e-01
-3.98161083e-01 -4.57218647e-01 -6.98087811e-01 5.09704292e-01
-1.35878253e+00 -1.08238399e+00 -3.09554338e-01 5.24023533e-01
1.08347249e+00 -4.13730741e-02 4.22481030e-01 1.02830693e-01
-6.21385157e-01 3.84499431e-01 -1.07346267e-01 3.14643949e-01
3.32702100e-01 -1.22227490e+00 5.71129978e-01 8.70453954e-01
1.74351409e-01 4.03991312e-01 7.37636030e-01 -6.63507164e-01
-1.07801020e+00 -1.17993987e+00 2.55572498e-01 -1.42923534e-01
4.05885786e-01 -1.18892029e-01 -9.04129326e-01 8.92380059e-01
4.61075425e-01 3.20787549e-01 2.15928644e-01 -2.40458950e-01
-4.82263118e-02 -1.76520199e-02 -1.17083442e+00 5.92130184e-01
8.95141184e-01 -1.56566620e-01 -5.15618384e-01 2.76157320e-01
7.54418254e-01 -9.00729358e-01 -5.40760636e-01 2.68147588e-01
3.32810938e-01 -8.57623219e-01 8.73010933e-01 -9.55554783e-01
3.78182352e-01 -4.91085708e-01 1.28837392e-01 -1.16277039e+00
-3.07121038e-01 -6.27521336e-01 -2.68747240e-01 8.25599790e-01
3.99795920e-01 -1.54784963e-01 1.15455878e+00 7.00722992e-01
-1.79936543e-01 -8.40756536e-01 -6.48219526e-01 -5.11189103e-01
-7.43454043e-03 -1.52402490e-01 4.19970274e-01 6.03943765e-01
-4.18392807e-01 2.38122314e-01 -1.86926033e-02 2.86267191e-01
5.03099084e-01 3.25060338e-01 7.94481277e-01 -8.51837575e-01
-4.38711882e-01 -4.72303122e-01 -5.20459950e-01 -1.36194730e+00
2.04304710e-01 -3.82641941e-01 3.26154292e-01 -1.53516257e+00
1.46055892e-01 -2.62249559e-01 -1.29739136e-01 5.12790859e-01
-2.90610045e-01 5.21999061e-01 3.79919887e-01 1.41410440e-01
-9.41484630e-01 1.49253994e-01 1.40231538e+00 -2.95333266e-01
-4.64336067e-01 1.76068172e-01 -5.46086192e-01 1.08616984e+00
7.88821697e-01 -4.44140822e-01 -4.08165365e-01 -8.32332790e-01
-1.75299540e-01 -4.96922471e-02 3.79494220e-01 -1.00620520e+00
3.89242500e-01 -4.85352546e-01 5.96480429e-01 -4.32652295e-01
3.09965342e-01 -7.71557987e-01 -2.57565137e-02 5.54674089e-01
-3.92116755e-01 -3.30094099e-02 3.60759586e-01 3.66090596e-01
-1.88486636e-01 -4.88232821e-01 1.10158479e+00 -4.04043585e-01
-1.23577464e+00 4.09253329e-01 -5.47945380e-01 2.03994900e-01
1.32657409e+00 -5.25769055e-01 7.66212121e-02 -2.76657075e-01
-8.27750325e-01 2.24642202e-01 6.60418630e-01 3.48185301e-01
5.10067999e-01 -9.03902292e-01 -4.51835722e-01 8.76405928e-03
-3.70834678e-01 2.21591502e-01 -3.76127623e-02 5.28627872e-01
-8.03054929e-01 2.14279518e-01 -4.36030328e-01 -9.35513794e-01
-1.13068235e+00 6.55710161e-01 8.02336574e-01 -5.25114797e-02
-7.17622697e-01 9.86511409e-01 4.44862813e-01 -8.73515308e-02
2.87666410e-01 -3.35962981e-01 -1.02915183e-01 -2.73639023e-01
4.45792645e-01 1.51524425e-01 -6.35667443e-02 -5.61296165e-01
-2.43168056e-01 6.77410424e-01 -1.99229896e-01 -1.03954710e-01
1.17207897e+00 -2.75378615e-01 -3.86757702e-02 4.33016866e-01
1.08556199e+00 -2.30948612e-01 -2.07908106e+00 -6.48515001e-02
8.88016522e-02 -5.89716673e-01 -3.28258201e-02 -6.23508155e-01
-1.52870214e+00 6.43979549e-01 3.79590243e-01 1.13684289e-01
1.21643209e+00 9.71648544e-02 7.71827579e-01 4.70116854e-01
2.81998307e-01 -1.21331453e+00 2.72755027e-01 5.47274649e-01
4.63047504e-01 -1.18436611e+00 -5.67477196e-02 -3.45605016e-01
-5.88950336e-01 1.13790977e+00 9.70410168e-01 -4.24989164e-01
4.60269660e-01 2.84825891e-01 2.94203848e-01 -1.95944265e-01
-6.01036310e-01 -8.39929283e-02 2.27120325e-01 5.30021906e-01
2.82156974e-01 -3.84370089e-01 -2.72050053e-02 -3.28110121e-02
1.23874888e-01 1.17263414e-01 3.89520019e-01 9.65470612e-01
-8.00360501e-01 -9.09098387e-01 -2.35751688e-01 3.57570827e-01
-7.06606686e-01 1.92260712e-01 -2.07272038e-01 6.82286978e-01
1.12628669e-01 7.22043514e-01 1.89876735e-01 7.79673979e-02
1.62976965e-01 -1.65492252e-01 5.82122087e-01 -8.98056149e-01
-7.81400681e-01 2.22080499e-01 -2.52835214e-01 -9.70066786e-01
-7.10481107e-01 -5.52910388e-01 -1.61231887e+00 -3.59011665e-02
-1.43162370e-01 6.03160076e-02 4.93370295e-01 1.21889257e+00
2.17159450e-01 6.86784863e-01 5.30354321e-01 -1.24297333e+00
-2.20637858e-01 -4.48030144e-01 -2.03101471e-01 2.30264276e-01
5.11270165e-01 -3.74747336e-01 1.82659402e-01 6.33504570e-01] | [9.230489730834961, 0.06667330116033554] |
124233ee-2a00-442f-8fc2-d39f144c06ee | tanimoto-random-features-for-scalable | 2306.14809 | null | https://arxiv.org/abs/2306.14809v1 | https://arxiv.org/pdf/2306.14809v1.pdf | Tanimoto Random Features for Scalable Molecular Machine Learning | The Tanimoto coefficient is commonly used to measure the similarity between molecules represented as discrete fingerprints, either as a distance metric or a positive definite kernel. While many kernel methods can be accelerated using random feature approximations, at present there is a lack of such approximations for the Tanimoto kernel. In this paper we propose two kinds of novel random features to allow this kernel to scale to large datasets, and in the process discover a novel extension of the kernel to real vectors. We theoretically characterize these random features, and provide error bounds on the spectral norm of the Gram matrix. Experimentally, we show that the random features proposed in this work are effective at approximating the Tanimoto coefficient in real-world datasets and that the kernels explored in this work are useful for molecular property prediction and optimization tasks. | ['José Miguel Hernández-Lobato', 'Sukriti Singh', 'Sergio Bacallado', 'Austin Tripp'] | 2023-06-26 | null | null | null | null | ['property-prediction', 'molecular-property-prediction'] | ['medical', 'miscellaneous'] | [ 1.62191495e-01 -3.27780694e-01 -3.96688253e-01 -4.48465049e-01
-3.59373778e-01 -7.47310340e-01 4.46667820e-01 5.18660545e-01
-3.61016244e-01 8.90615106e-01 -2.40777478e-01 -3.69387597e-01
-6.25379860e-01 -8.34636152e-01 -4.05544728e-01 -8.20637286e-01
-6.31331146e-01 1.94941744e-01 5.02826691e-01 3.42088081e-02
5.39468825e-01 9.49114680e-01 -1.33286631e+00 2.60407962e-02
1.11540461e+00 9.21231627e-01 -7.97429308e-02 6.69889092e-01
1.78068787e-01 4.29978043e-01 -3.31180125e-01 -2.91259378e-01
2.96600670e-01 -2.63177127e-01 -7.35438168e-01 -3.26109558e-01
3.33875388e-01 9.66458619e-02 -9.04303640e-02 1.08363473e+00
3.15778613e-01 2.57874399e-01 1.31317067e+00 -1.10584819e+00
-6.87693596e-01 3.69445741e-01 -2.33735934e-01 -1.51854783e-01
6.67050123e-01 -2.51525134e-01 1.00133228e+00 -9.37397480e-01
4.86803383e-01 9.36905682e-01 1.05957758e+00 2.27946392e-03
-1.45227635e+00 -5.36050022e-01 -5.29628098e-01 3.92520338e-01
-1.60858428e+00 -1.34992495e-01 5.46306610e-01 -4.93346304e-01
8.13204825e-01 4.71622348e-01 4.31693792e-01 6.27225637e-01
4.30996299e-01 4.44486529e-01 1.23236430e+00 -5.03592312e-01
3.35935444e-01 3.31294805e-01 3.74972910e-01 9.85147655e-01
5.48381627e-01 1.98055491e-01 -2.83892602e-01 -6.85810268e-01
6.31750405e-01 2.49410138e-01 -4.57605153e-01 -7.70274937e-01
-1.31825304e+00 1.19436753e+00 3.19265664e-01 2.93901235e-01
-2.83756107e-01 -1.18635945e-01 3.98544490e-01 4.69312042e-01
4.06603992e-01 8.22757602e-01 -3.21050763e-01 -1.98995903e-01
-6.97484672e-01 2.81730980e-01 1.15503716e+00 7.38433659e-01
9.87397432e-01 -3.15358281e-01 9.46793035e-02 6.88860953e-01
1.19912192e-01 5.49314559e-01 2.93691874e-01 -5.46653152e-01
-8.63806829e-02 5.31635284e-01 1.54611260e-01 -1.11188471e+00
-4.73141223e-01 1.08383909e-01 -7.96664655e-01 1.04984865e-01
8.07707191e-01 2.84361988e-01 -5.13336778e-01 1.31027055e+00
4.47132379e-01 2.27608502e-01 8.00439268e-02 7.80166447e-01
3.43476892e-01 6.93729043e-01 -1.79800138e-01 -9.69958007e-02
9.74669218e-01 -7.20646083e-01 -4.60690290e-01 6.47467434e-01
9.60723102e-01 -1.10469222e+00 8.94444168e-01 4.78222013e-01
-5.41756511e-01 -2.55166352e-01 -1.09103346e+00 2.32979774e-01
-6.43585622e-01 1.40833244e-01 1.17704117e+00 1.03398740e+00
-7.76240528e-01 1.08844113e+00 -5.84500015e-01 -2.97270507e-01
-1.78774455e-04 5.30875444e-01 -7.11909652e-01 -1.02316231e-01
-1.13111889e+00 8.29597831e-01 4.68350112e-01 1.11175388e-01
-1.03632227e-01 -5.78872085e-01 -8.10697198e-01 1.25793517e-02
5.38465902e-02 -2.29521587e-01 7.08087921e-01 -5.34996331e-01
-1.70275354e+00 4.47493434e-01 5.60564660e-02 -3.03613842e-01
2.20445663e-01 1.82647943e-01 -3.72623175e-01 1.22489989e-01
-5.74264824e-02 4.68175672e-02 5.97571671e-01 -5.40269613e-01
-2.05891117e-01 -1.99818164e-01 -2.58637786e-01 -4.38257396e-01
-5.28078675e-01 -8.85479897e-02 2.87809342e-01 -5.93131304e-01
3.94261144e-02 -1.04985130e+00 -4.24593836e-01 3.67085300e-02
-3.34128320e-01 -4.21942055e-01 5.33496916e-01 -1.92961305e-01
1.07637620e+00 -1.99851859e+00 -4.32762206e-02 9.11371708e-01
1.37884438e-01 3.48814070e-01 -2.34683916e-01 6.48368418e-01
-2.15927124e-01 -3.46906111e-02 -2.76438087e-01 4.64075178e-01
3.09624821e-02 -3.07828393e-02 -4.39523160e-01 8.28430831e-01
1.89653307e-01 6.36187971e-01 -1.12380242e+00 -1.67801246e-01
-1.11889597e-02 6.14598036e-01 -4.17704344e-01 1.96978062e-01
3.12354982e-01 -2.40064919e-01 -4.94589359e-01 5.70110261e-01
8.18588912e-01 -1.92651987e-01 1.99666098e-02 -3.34514260e-01
-6.86350316e-02 8.68042856e-02 -1.19924390e+00 1.20231748e+00
-6.66349158e-02 5.91611624e-01 -5.76874554e-01 -1.06313133e+00
1.09635448e+00 8.79155993e-02 6.03419602e-01 -5.96105382e-02
-1.03102274e-01 4.11225438e-01 5.20000840e-03 -9.17439535e-02
3.51968765e-01 -2.14502320e-01 3.36488754e-01 3.76810461e-01
8.91919527e-03 -2.38310993e-02 2.66226292e-01 -9.18409079e-02
1.08667910e+00 -1.94185689e-01 6.73355401e-01 -6.40828967e-01
6.40271664e-01 -8.47770572e-02 1.06863238e-01 5.60360551e-01
9.55498964e-02 3.19736123e-01 5.64405382e-01 -5.88806450e-01
-9.88883078e-01 -1.19955492e+00 -5.67123711e-01 7.73017228e-01
-4.64045480e-02 -6.30375385e-01 -5.06760299e-01 -6.75395906e-01
4.72891361e-01 2.34906167e-01 -6.89651906e-01 -4.04224128e-01
3.11346967e-02 -8.35789800e-01 6.48229301e-01 3.73708427e-01
3.86191122e-02 -5.37563741e-01 -1.25998065e-01 2.00160652e-01
5.68352640e-01 -7.57439554e-01 -7.81172276e-01 5.15146077e-01
-9.28091824e-01 -1.29754388e+00 -5.49877346e-01 -5.14941931e-01
5.36997616e-01 3.75035197e-01 5.90488374e-01 -3.64399672e-01
-4.79274094e-01 1.88769221e-01 -4.46588129e-01 -2.53405571e-01
-4.41807836e-01 -4.42882665e-02 4.10427362e-01 2.04886511e-01
4.37022567e-01 -4.97521967e-01 -5.83189726e-01 5.90418696e-01
-1.01702559e+00 -4.27777618e-01 6.23329699e-01 1.12526798e+00
5.04935503e-01 -8.18197876e-02 4.68083680e-01 -7.98946381e-01
9.45506692e-01 -3.50694269e-01 -6.65978491e-01 3.09879750e-01
-1.01210976e+00 6.44981027e-01 8.51758599e-01 -1.03898454e+00
-4.26585287e-01 2.37896845e-01 1.09804288e-01 -2.87510335e-01
1.74076140e-01 6.32128298e-01 -1.71258952e-03 -9.32047486e-01
8.30692530e-01 2.18077391e-01 3.34868208e-02 -4.19957697e-01
3.98116440e-01 7.22844899e-01 1.44505985e-02 -7.71649480e-01
7.10100472e-01 2.43968651e-01 6.59603715e-01 -1.01631141e+00
-5.43091714e-01 -8.87391090e-01 -6.53393745e-01 4.12602335e-01
4.35217500e-01 -3.08056891e-01 -1.05296874e+00 1.26848370e-01
-9.17089403e-01 7.44165778e-02 -3.35444421e-01 8.36508989e-01
-6.63835645e-01 7.29036689e-01 -5.27982116e-01 -6.75088942e-01
-2.16572270e-01 -9.71732557e-01 7.93894649e-01 1.45458534e-01
-1.76801756e-01 -1.17172801e+00 5.46417832e-01 -3.03137451e-01
3.76794964e-01 1.65483832e-01 1.14617300e+00 -9.78279769e-01
-1.85441881e-01 -7.31554925e-01 -3.67162347e-01 4.23266649e-01
3.76574099e-01 2.78811663e-01 -7.65936553e-01 -3.13564867e-01
-2.49071106e-01 -1.33216202e-01 9.04187143e-01 2.93333024e-01
1.02918339e+00 -2.04525694e-01 -3.96882445e-01 9.40262318e-01
1.12237120e+00 3.24619710e-01 7.32956529e-01 1.06950939e-01
5.12356818e-01 2.19052121e-01 6.69759929e-01 3.55758995e-01
-1.52636319e-01 7.22352147e-01 -4.43104468e-02 7.03056669e-03
5.68155825e-01 -1.20883323e-01 3.57482582e-01 8.00746739e-01
-5.65540791e-01 2.83806980e-01 -7.01196015e-01 8.08618218e-02
-1.72852218e+00 -1.03769338e+00 -2.21226081e-01 2.75819588e+00
8.68306518e-01 -1.63475260e-01 1.46812007e-01 1.09229870e-01
5.28662980e-01 -1.88012362e-01 -3.04168075e-01 -7.56334186e-01
-1.55170500e-01 5.66554129e-01 7.98917472e-01 4.52133656e-01
-9.99858379e-01 7.14519680e-01 7.39610338e+00 1.12370646e+00
-1.21130049e+00 -3.49712163e-01 1.21207148e-01 4.15226251e-01
-7.77614713e-02 2.12776035e-01 -6.69792712e-01 3.19877923e-01
1.05898094e+00 -3.05388451e-01 2.14151859e-01 1.07656598e+00
-2.25955546e-01 -4.35201637e-02 -1.22568917e+00 1.13034463e+00
-2.18679294e-01 -1.31309426e+00 1.58326000e-01 3.87646616e-01
5.41001260e-01 -4.10638511e-01 8.87919515e-02 -3.11359353e-02
-3.96230072e-02 -1.23932409e+00 -6.30767345e-02 5.65185428e-01
7.89464772e-01 -1.08492076e+00 6.80703282e-01 -4.71435413e-02
-1.21836329e+00 2.84073800e-01 -9.09853399e-01 4.88825254e-02
-3.66727978e-01 9.45368648e-01 -1.26028204e+00 3.97388011e-01
-1.10016428e-01 7.35336185e-01 -6.57321155e-01 1.42946732e+00
2.26868749e-01 5.25534034e-01 -3.53034824e-01 -6.36591434e-01
2.17410237e-01 -6.98051333e-01 4.52951699e-01 1.27645636e+00
3.89702797e-01 -2.94519812e-01 1.17755964e-01 6.20100796e-01
2.74484366e-01 6.20274127e-01 -9.25346017e-01 -4.48288053e-01
1.73882484e-01 1.10712016e+00 -7.52372205e-01 -1.39118299e-01
-3.74619246e-01 1.24895322e+00 2.18745261e-01 4.03774321e-01
-6.68810487e-01 -1.18662632e+00 1.03201902e+00 1.27820581e-01
1.98556721e-01 -4.32005197e-01 3.00789028e-01 -1.25965679e+00
-9.76737496e-03 -5.53460538e-01 3.76357362e-02 -2.80275136e-01
-1.52146161e+00 4.99817193e-01 -1.05373859e-01 -1.31866181e+00
-2.62780041e-01 -1.26762855e+00 -3.13703060e-01 7.78325438e-01
-1.25206757e+00 -6.94657147e-01 2.07630396e-01 5.39321244e-01
-2.15334043e-01 -1.49193183e-01 1.32340062e+00 1.09117970e-01
-1.65513173e-01 6.72592044e-01 8.20703864e-01 -7.96680450e-02
1.11906517e+00 -1.41544974e+00 2.39690989e-01 1.42090380e-01
3.08748782e-01 1.16324282e+00 7.22769797e-01 -4.55975771e-01
-1.56535923e+00 -7.38674462e-01 4.93754894e-01 -1.45540029e-01
1.01172423e+00 -2.40518868e-01 -9.47054744e-01 1.88670784e-01
-5.19218683e-01 3.70203644e-01 1.22615182e+00 2.30878890e-01
-7.37759948e-01 -7.07641849e-03 -1.26481414e+00 3.58585894e-01
6.87578917e-01 -6.59109890e-01 -6.47194833e-02 5.94619274e-01
3.63674760e-01 1.36761650e-01 -1.46670806e+00 3.86945873e-01
1.06918275e+00 -7.89571524e-01 1.19085813e+00 -9.60678875e-01
-1.92969590e-01 -3.47127199e-01 -2.04922423e-01 -1.37410474e+00
-5.14388680e-01 -7.37100780e-01 -1.64034754e-01 5.55146873e-01
4.07060921e-01 -1.01831555e+00 7.42975593e-01 3.44677925e-01
3.55065912e-01 -1.11377966e+00 -9.28041697e-01 -1.24361074e+00
-5.45209646e-02 -1.67460665e-01 4.24696892e-01 1.06032062e+00
5.61066151e-01 2.06015036e-01 -3.36170554e-01 -5.30303717e-02
5.88566363e-01 2.46092781e-01 6.55158579e-01 -1.49010515e+00
-3.91089767e-01 -4.35272932e-01 -1.33247924e+00 -9.44237888e-01
1.47241890e-01 -9.70697701e-01 -2.98206210e-01 -8.10286820e-01
2.98989207e-01 -5.56748331e-01 -5.31318009e-01 1.14769228e-01
-1.81476817e-01 1.29028812e-01 -3.27746361e-01 7.89672788e-03
-3.36185008e-01 4.35818911e-01 1.08475137e+00 -2.04417229e-01
-4.52659220e-01 2.26596996e-01 -3.32622170e-01 5.86402833e-01
6.99417710e-01 -5.15061080e-01 -3.37933540e-01 5.71865976e-01
2.89445609e-01 -3.49522978e-01 6.55822754e-02 -9.07213807e-01
1.41548306e-01 -4.21588838e-01 5.48780084e-01 -1.46185786e-01
2.41443336e-01 -7.21239150e-01 5.85497469e-02 4.40508217e-01
-1.16692394e-01 -8.19121078e-02 -8.81169811e-02 6.96414351e-01
-1.69676587e-01 -3.87127340e-01 7.24167347e-01 3.68241102e-01
-2.86984742e-01 4.07742918e-01 -3.33457619e-01 -3.36900860e-01
1.02804041e+00 -5.25734067e-01 -4.07631211e-02 -9.24590603e-02
-4.84111965e-01 -3.82377684e-01 6.70160115e-01 -3.46026570e-02
8.74725640e-01 -1.48135591e+00 -3.72304499e-01 2.72129238e-01
3.26163083e-01 -5.67178249e-01 -3.15688133e-01 1.09716356e+00
-8.15813482e-01 6.83150351e-01 -3.46645921e-01 -5.14782488e-01
-1.38954413e+00 8.76745522e-01 2.47836336e-01 -3.53110582e-01
-2.90612906e-01 5.85102677e-01 5.33176214e-02 -3.21206391e-01
2.09574252e-02 -5.68513751e-01 1.23191468e-01 3.91287394e-02
7.59937644e-01 4.26928133e-01 3.21387239e-02 -4.39071357e-01
-4.40963686e-01 8.73912454e-01 -1.98371664e-01 2.01151133e-01
1.28142953e+00 5.09748042e-01 -3.81411552e-01 3.66959214e-01
1.67069030e+00 3.19824547e-01 -7.23194301e-01 -8.72226283e-02
2.91622102e-01 -6.71228945e-01 -3.13055962e-01 -2.43838683e-01
-4.86158967e-01 6.47520125e-01 7.68484294e-01 2.86919475e-01
1.00696039e+00 -1.87977031e-01 6.06708765e-01 1.10071504e+00
4.24420744e-01 -7.21355379e-01 -2.05370471e-01 4.84727323e-01
5.21362901e-01 -1.09388173e+00 2.48685464e-01 -7.90314138e-01
-3.13191444e-01 1.58194256e+00 -7.41994455e-02 -2.41385192e-01
8.86353970e-01 2.34792620e-01 -4.46357787e-01 7.46033490e-02
-3.62406552e-01 -2.62642443e-01 5.63551664e-01 7.72875071e-01
7.62631834e-01 2.55582005e-01 -6.58767164e-01 2.26353034e-01
-1.69564411e-02 9.82375070e-03 4.72118944e-01 9.02176976e-01
-6.65661395e-01 -1.51127589e+00 -2.75090754e-01 5.26015282e-01
-2.38574982e-01 -1.84947774e-01 -5.37331998e-01 4.39466357e-01
-2.73118317e-01 5.13504446e-01 -4.58873570e-01 -6.29878998e-01
2.15926707e-01 1.48928449e-01 8.34729910e-01 -3.95992398e-01
-2.14735866e-01 -4.10126597e-01 -1.17439389e-01 -5.89175582e-01
-1.28128290e-01 -3.70805055e-01 -1.07674623e+00 -3.28846246e-01
-6.51306152e-01 7.04852819e-01 7.61109710e-01 5.94950199e-01
2.41840020e-01 -3.36691558e-01 7.69674718e-01 -6.55521393e-01
-8.52200091e-01 -8.38889182e-01 -1.17405367e+00 3.20493370e-01
2.17949569e-01 -8.43094528e-01 -4.04442370e-01 -6.06094599e-02] | [7.57961893081665, 4.063313961029053] |
8e5b4ed5-6f59-4878-bb29-9373db7d3f4b | developing-all-skyrmion-spiking-neural | 1705.02995 | null | http://arxiv.org/abs/1705.02995v1 | http://arxiv.org/pdf/1705.02995v1.pdf | Developing All-Skyrmion Spiking Neural Network | In this work, we have proposed a revolutionary neuromorphic computing
methodology to implement All-Skyrmion Spiking Neural Network (AS-SNN). Such
proposed methodology is based on our finding that skyrmion is a topological
stable spin texture and its spatiotemporal motion along the magnetic nano-track
intuitively interprets the pulse signal transmission between two interconnected
neurons. In such design, spike train in SNN could be encoded as particle-like
skyrmion train and further processed by the proposed skyrmion-synapse and
skyrmion-neuron within the same magnetic nano-track to generate output skyrmion
as post-spike. Then, both pre-neuron spikes and post-neuron spikes are encoded
as particle-like skyrmions without conversion between charge and spin signals,
which fundamentally differentiates our proposed design from other hybrid
Spin-CMOS designs. The system level simulation shows 87.1% inference accuracy
for handwritten digit recognition task, while the energy dissipation is ~1
fJ/per spike which is 3 orders smaller in comparison with CMOS based IBM
TrueNorth system. | ['Deliang Fan', 'Zhezhi He'] | 2017-05-08 | null | null | null | null | ['handwritten-digit-recognition'] | ['computer-vision'] | [ 4.04978305e-01 -2.04296127e-01 3.19287121e-01 -5.55797890e-02
4.06175256e-01 -4.73525017e-01 6.42997205e-01 -1.37214944e-01
-5.02758443e-01 1.20160890e+00 -2.28476882e-01 -2.09470943e-01
2.01175790e-02 -7.62551904e-01 -1.21921790e+00 -1.12937498e+00
3.59505624e-01 3.30096424e-01 7.53467619e-01 -3.65363747e-01
1.06830657e+00 1.30869791e-01 -1.62370980e+00 2.78677732e-01
5.68520665e-01 1.25005484e+00 3.27397585e-01 7.28340268e-01
2.80836131e-02 8.60274017e-01 -7.50651240e-01 -2.83841819e-01
3.75798158e-02 -5.19184530e-01 -4.13824469e-02 -9.87626135e-01
2.99241751e-01 4.64554250e-01 -5.64110518e-01 1.32047045e+00
5.32530487e-01 -3.33665699e-01 6.78055704e-01 -1.17674935e+00
-1.12034869e+00 1.04840899e+00 -4.78752762e-01 6.46941006e-01
-1.98656768e-01 1.72690138e-01 -2.83242226e-01 -5.50111353e-01
6.86461926e-01 8.17643046e-01 8.25378418e-01 5.81943810e-01
-8.25076640e-01 -8.87616336e-01 -9.16460395e-01 2.13903129e-01
-1.23369229e+00 -4.05735791e-01 7.01754689e-01 1.34077758e-01
1.54760993e+00 2.65087903e-01 1.18655097e+00 9.35080945e-01
1.62163436e+00 1.17187947e-01 1.74086058e+00 -1.53446555e-01
9.16135728e-01 -4.86992568e-01 5.68965793e-01 5.08689106e-01
1.11299777e+00 -1.66391194e-01 -9.26713884e-01 1.08239204e-01
7.72233069e-01 -6.08730949e-02 1.98784042e-02 3.11787963e-01
-9.07529950e-01 -3.43841106e-01 5.21402419e-01 4.45560127e-01
-3.67832690e-01 1.13683975e+00 9.84253958e-02 -7.62977228e-02
-6.82877064e-01 7.90318027e-02 2.00130105e-01 -3.24512839e-01
-9.61091280e-01 9.18477029e-02 6.97024465e-01 5.51329374e-01
1.02694318e-01 6.44308805e-01 1.27809327e-02 2.81334281e-01
2.29402155e-01 1.48722243e+00 8.84592414e-01 -6.62808657e-01
-6.39567301e-02 6.67654753e-01 -1.79615900e-01 -7.66968906e-01
-2.35480666e-01 -4.76766795e-01 -1.29162228e+00 2.86197096e-01
-1.44087046e-01 1.04705758e-01 -1.52973032e+00 1.51115870e+00
-4.16745573e-01 5.35234809e-01 5.33072352e-01 9.11921442e-01
7.66223073e-01 6.70799553e-01 -3.16241503e-01 -8.99648145e-02
1.59217405e+00 -2.36604467e-01 -8.70521843e-01 -5.08458316e-02
-1.23861961e-01 -1.21967368e-01 4.52929378e-01 9.51567665e-02
-1.36980069e+00 -3.00238222e-01 -1.60867631e+00 1.34323493e-01
-5.43820500e-01 -2.54516006e-01 3.61097127e-01 5.59315801e-01
-1.14780366e+00 6.28986061e-01 -1.06253409e+00 -3.89422953e-01
3.43713522e-01 7.72193789e-01 2.11600274e-01 6.33098900e-01
-8.91446769e-01 8.88007998e-01 8.83167759e-02 1.09205469e-01
-6.61703348e-01 -2.56350189e-01 1.49938509e-01 2.79292352e-02
-6.89198732e-01 -7.54986703e-01 1.09273887e+00 -5.78011930e-01
-1.67076731e+00 8.26017320e-01 -3.31502080e-01 -9.72806990e-01
-3.94199729e-01 8.49141836e-01 -7.14701235e-01 7.20424280e-02
-1.45202637e-01 5.91091990e-01 3.64860594e-01 -8.51063609e-01
-6.79988936e-02 -7.13499784e-01 -6.71693385e-01 -2.14150459e-01
-2.60164201e-01 -3.52412343e-01 2.32322693e-01 -2.56301552e-01
5.76528013e-01 -1.15739536e+00 1.89510450e-01 -5.00164449e-01
-3.41248900e-01 -1.16731994e-01 1.07038343e+00 2.81612426e-01
6.09899163e-01 -1.88165236e+00 -2.40213573e-01 7.37906098e-02
-1.19618654e-01 3.47437114e-01 4.07755822e-01 2.99778700e-01
3.48159552e-01 -1.03142627e-01 -3.96164060e-01 3.44104201e-01
-3.68471980e-01 7.07609206e-02 -7.97942102e-01 3.65246058e-01
1.08289488e-01 1.33984125e+00 -6.36834979e-01 -6.57752454e-02
-3.24975967e-01 6.08960867e-01 9.42618549e-02 -7.17905521e-01
7.41118342e-02 3.49740624e-01 -2.47157738e-01 9.80236888e-01
1.27019203e+00 -2.81978667e-01 1.49471074e-01 -1.96831286e-01
-6.22740984e-01 2.29526326e-01 -6.32684648e-01 1.32385504e+00
3.16043526e-01 8.09784234e-01 -6.50080517e-02 -8.81156445e-01
1.26894486e+00 -1.36601731e-01 -2.32856363e-01 -1.69210064e+00
5.01404464e-01 9.32186246e-01 3.11386734e-01 -3.14715803e-01
4.39996570e-01 -1.55290678e-01 -7.22403750e-02 3.94078583e-01
1.36236683e-01 -1.01605132e-01 -3.42696197e-02 -1.52551726e-01
1.44306052e+00 -2.55932033e-01 -4.60115194e-01 -1.08112955e+00
7.09576011e-02 4.86040302e-02 4.04411167e-01 1.00262523e+00
-4.66410331e-02 2.36550599e-01 2.81316698e-01 -3.06151599e-01
-8.89613092e-01 -1.55531240e+00 -4.75565791e-01 2.83615794e-02
8.71692955e-01 4.03985143e-01 -8.69066715e-01 5.89333594e-01
-2.83255856e-02 5.31541407e-01 -4.15882885e-01 -2.55865067e-01
-6.18127823e-01 -1.25318122e+00 1.45301116e+00 5.38314283e-01
1.00229084e+00 -1.46213758e+00 -1.64548206e+00 4.42196876e-01
4.32482570e-01 -7.75064170e-01 4.02999312e-01 1.04982460e+00
-1.15371370e+00 -6.24003530e-01 -5.22310793e-01 -8.01181078e-01
8.03863943e-01 -2.24846944e-01 8.61948550e-01 -2.09711835e-01
-7.27367938e-01 -4.06718791e-01 1.67867303e-01 -5.05789280e-01
3.21758211e-01 -5.41320205e-01 6.74871325e-01 -4.48470742e-01
6.10847354e-01 -1.21548235e+00 -1.07791853e+00 -3.16568226e-01
-7.41933942e-01 6.02537673e-03 6.14243686e-01 1.91452727e-01
8.88896704e-01 8.53082687e-02 6.33230567e-01 -1.56265318e-01
4.82213438e-01 -3.87182206e-01 -4.48500633e-01 1.29921496e-01
-3.96994054e-01 3.73559713e-01 9.26050246e-01 -3.09476614e-01
-6.56705081e-01 -2.55780756e-01 2.64718592e-01 1.22486062e-01
3.62370163e-01 8.84975344e-02 4.48059887e-01 -3.74335974e-01
6.18157446e-01 8.53789270e-01 -3.28321010e-01 6.03524484e-02
-2.72843868e-01 7.98061728e-01 1.16332161e+00 -2.85788178e-01
1.76881149e-01 9.57316518e-01 6.84442461e-01 -7.06457615e-01
3.55152935e-01 6.34288490e-01 2.01350108e-01 -1.42334938e-01
6.94652498e-01 -6.47505820e-01 -1.30276716e+00 1.12546372e+00
-1.35333776e+00 -7.28157908e-02 3.30501869e-02 2.81865299e-01
-3.88151854e-01 -2.93142855e-01 -1.00376141e+00 -1.17463696e+00
-9.30867374e-01 -7.01631904e-01 6.38236642e-01 1.05546713e+00
2.51208544e-01 -4.02829409e-01 2.36527979e-01 -3.04410517e-01
9.53412771e-01 7.25435838e-02 8.58346820e-01 -7.18255863e-02
-8.26219440e-01 4.11028825e-02 -1.24456137e-01 -4.46991026e-01
-3.83697152e-01 -4.95847501e-02 -9.05747592e-01 5.66959158e-02
4.31483448e-01 -3.30139935e-01 1.07819211e+00 7.47408628e-01
6.49783552e-01 5.48168905e-02 -6.18932962e-01 4.03859317e-01
1.73175263e+00 6.89107120e-01 9.07102108e-01 9.65668857e-02
2.66153485e-01 -2.63339192e-01 -2.71298289e-01 6.39269948e-02
4.30988759e-01 2.07425222e-01 3.46706539e-01 5.29513597e-01
-1.00428335e-01 -1.35822713e-01 3.89889687e-01 1.33244693e+00
-1.61894336e-01 -3.98505539e-01 -1.08230841e+00 3.06553811e-01
-1.84391844e+00 -1.02198422e+00 -8.04076612e-01 2.02067518e+00
7.81839490e-01 2.40505874e-01 -6.09177470e-01 -6.83863368e-03
9.87819374e-01 -4.28696960e-01 -1.06278789e+00 -6.75368726e-01
-8.40929449e-01 7.75439084e-01 1.08198643e+00 -1.90824628e-01
6.08233772e-02 7.14005351e-01 5.67280197e+00 2.52348930e-01
-1.91771805e+00 8.89943615e-02 3.04738998e-01 -2.75043547e-01
-4.90032911e-01 1.01682186e-01 -7.40521193e-01 1.28982365e+00
1.35660326e+00 -2.00544462e-01 5.34544289e-01 1.57415420e-01
-7.03150332e-02 -5.89152753e-01 -5.55396497e-01 1.33479941e+00
-2.31865374e-03 -1.90099669e+00 1.41554177e-01 5.24008535e-02
7.50060976e-01 2.30150566e-01 3.16788435e-01 -1.42264932e-01
-1.21766366e-01 -1.06013644e+00 1.04966271e+00 1.13561189e+00
7.49785960e-01 -5.07493258e-01 7.43268609e-01 1.74838528e-01
-9.49492991e-01 -2.82739282e-01 -7.76890099e-01 -3.84976774e-01
2.78976560e-01 9.89002585e-01 -1.05796337e-01 1.05693005e-01
1.03232825e+00 1.17366791e-01 -4.59200978e-01 9.21586692e-01
6.39096558e-01 4.60615784e-01 -7.30482817e-01 -7.22009420e-01
2.80025601e-03 -2.43945941e-01 4.96613175e-01 9.69563663e-01
9.79640961e-01 2.79510528e-01 -9.83953178e-01 1.28254104e+00
-4.18336034e-01 -6.63654864e-01 -8.01737666e-01 -4.18718845e-01
9.58820641e-01 1.28471375e+00 -1.67006540e+00 -7.23977804e-01
6.45349205e-01 1.05715477e+00 -3.17682654e-01 -2.04283260e-02
-8.62275302e-01 -9.97622967e-01 5.90418577e-02 3.04979682e-01
3.60549390e-01 -2.71224201e-01 -1.07704449e+00 -6.59460485e-01
1.07184075e-01 -8.68199542e-02 -7.60442853e-01 -1.08853626e+00
-5.45735896e-01 6.83887064e-01 -8.30572605e-01 -7.04364359e-01
2.85578519e-01 -6.46499753e-01 -1.07082772e+00 4.53340501e-01
-9.49782193e-01 -5.21054029e-01 -3.28207284e-01 2.80881554e-01
-3.50110799e-01 -5.59315505e-03 5.53648710e-01 -8.92087724e-03
-4.59967971e-01 3.31513226e-01 4.02794302e-01 -2.64814049e-01
1.73359275e-01 -8.48466218e-01 6.95204496e-01 6.68369830e-01
-9.74617004e-02 1.04266441e+00 6.43263519e-01 -9.99877691e-01
-2.32100415e+00 -9.33888018e-01 5.94729364e-01 -2.59713411e-01
3.76587749e-01 -6.82438731e-01 -7.88583398e-01 -1.22313276e-01
6.51943326e-01 1.05107270e-01 2.95500159e-01 -9.94005442e-01
-1.37904108e-01 -2.54907191e-01 -1.47511148e+00 6.69771910e-01
1.38058162e+00 -4.84234244e-01 -5.66422522e-01 -1.04416408e-01
1.52737617e-01 -1.91675559e-01 -3.67068499e-01 6.60338581e-01
8.17532003e-01 -1.11971629e+00 2.27282107e-01 3.73407900e-02
2.80233473e-01 -9.16586220e-01 -4.02779251e-01 -7.55809665e-01
8.58533457e-02 -4.89245415e-01 -2.56666005e-01 9.36353207e-01
1.36930555e-01 -9.59240079e-01 7.39013135e-01 4.21346340e-04
-3.43101561e-01 -6.90478742e-01 -1.41764402e+00 -9.70387340e-01
-1.77708998e-01 2.41285771e-01 4.05622542e-01 4.58691716e-01
1.94162771e-01 3.98703329e-02 2.51840532e-01 -3.32184769e-02
9.87223744e-01 -9.57021788e-02 -2.71567553e-01 -1.05304790e+00
-2.36079935e-02 -3.04360002e-01 -9.54367042e-01 -5.44136822e-01
-1.10037498e-01 -1.12619829e+00 4.61392730e-01 -1.48348570e+00
4.46304202e-01 -3.17430347e-01 -8.70227575e-01 3.18651199e-01
6.60074890e-01 1.10329115e+00 -1.96742162e-01 3.25296760e-01
-5.77460051e-01 2.62273997e-01 6.80608094e-01 1.30270496e-01
7.62794614e-02 -8.22830558e-01 -3.36767077e-01 3.31862837e-01
6.67750597e-01 -7.59078503e-01 -3.62989396e-01 -5.48515737e-01
6.91889465e-01 1.02087177e-01 7.63002932e-01 -1.97313917e+00
9.87348080e-01 3.31972480e-01 6.41666174e-01 -6.90465510e-01
3.15235913e-01 -4.77665104e-02 7.96567500e-01 1.07001030e+00
9.09020826e-02 4.63419586e-01 2.02544421e-01 5.01143515e-01
2.03636870e-01 -1.83900490e-01 8.64005625e-01 5.73232435e-02
-2.80573368e-01 -2.99355000e-01 -8.56528521e-01 9.62566212e-03
9.49077189e-01 -9.60510135e-01 -1.25741446e+00 3.18724036e-01
-2.10929438e-02 -2.66004503e-01 7.50592947e-01 1.19233675e-01
8.39961946e-01 -1.24306500e+00 9.15165097e-02 4.68039751e-01
-5.10471225e-01 -4.70921338e-01 2.83635676e-01 1.14904141e+00
-6.01639450e-01 6.80039525e-01 -1.09505522e+00 -8.29769135e-01
-7.85962403e-01 -7.16744512e-02 1.45041257e-01 5.51740944e-01
-5.15758753e-01 9.14671659e-01 -7.01308489e-01 -2.18884628e-02
-2.67800927e-01 -3.95124584e-01 9.22453851e-02 -5.61810672e-01
3.16749156e-01 2.75937766e-01 1.04413569e-01 -3.83440852e-01
-9.49856639e-01 1.02331340e+00 2.21435353e-01 -6.50395274e-01
1.35583770e+00 2.42333233e-01 -8.69248211e-01 9.52707708e-01
7.85501957e-01 -3.47806722e-01 -8.04719210e-01 5.29465258e-01
9.30824131e-02 3.38175029e-01 -2.27671891e-01 -1.01927316e+00
-7.65654743e-01 7.99663305e-01 1.01601195e+00 -1.54224420e-02
1.02279186e+00 -3.15114260e-01 1.18938053e+00 8.96697700e-01
7.85463810e-01 -1.26970506e+00 1.51393324e-01 7.84204125e-01
4.03421193e-01 -1.22358151e-01 -6.09157979e-01 4.89001125e-01
-2.64623404e-01 1.21325123e+00 7.27507710e-01 -8.92522037e-01
6.03840590e-01 1.06355143e+00 -2.42471337e-01 -5.77626288e-01
-9.80004370e-01 4.14405555e-01 -6.37733698e-01 3.57944489e-01
2.22619548e-01 2.03971684e-01 -3.86711001e-01 7.21590698e-01
-4.21819508e-01 7.79241264e-01 1.01486444e+00 1.39320695e+00
-9.77062345e-01 -4.45418745e-01 -1.85658664e-01 7.85162032e-01
-3.62540036e-01 -1.68932065e-01 -4.18562740e-01 -1.51090056e-01
3.84141356e-01 5.34269214e-01 7.49110758e-01 -7.29758501e-01
7.91433081e-03 9.89303663e-02 1.03615046e+00 -1.27507960e-02
-4.35961992e-01 -5.71649671e-01 -6.32336318e-01 -3.08188677e-01
-8.66339132e-02 8.85327160e-02 -2.45451307e+00 -3.82475644e-01
-3.23882401e-02 2.71220058e-01 1.42825091e+00 6.98181093e-01
8.68630886e-01 6.66348755e-01 1.13688007e-01 -5.44966519e-01
-3.96580100e-01 -7.45603859e-01 -6.55120552e-01 -1.95016041e-01
1.00495115e-01 -5.80215096e-01 -1.72532722e-01 -1.77493200e-01] | [8.217039108276367, 2.4637224674224854] |
f43813df-f2d8-4c0b-ad69-fcd07d4590af | performance-optimized-deep-neural-networks | 2306.03779 | null | https://arxiv.org/abs/2306.03779v1 | https://arxiv.org/pdf/2306.03779v1.pdf | Performance-optimized deep neural networks are evolving into worse models of inferotemporal visual cortex | One of the most impactful findings in computational neuroscience over the past decade is that the object recognition accuracy of deep neural networks (DNNs) correlates with their ability to predict neural responses to natural images in the inferotemporal (IT) cortex. This discovery supported the long-held theory that object recognition is a core objective of the visual cortex, and suggested that more accurate DNNs would serve as better models of IT neuron responses to images. Since then, deep learning has undergone a revolution of scale: billion parameter-scale DNNs trained on billions of images are rivaling or outperforming humans at visual tasks including object recognition. Have today's DNNs become more accurate at predicting IT neuron responses to images as they have grown more accurate at object recognition? Surprisingly, across three independent experiments, we find this is not the case. DNNs have become progressively worse models of IT as their accuracy has increased on ImageNet. To understand why DNNs experience this trade-off and evaluate if they are still an appropriate paradigm for modeling the visual system, we turn to recordings of IT that capture spatially resolved maps of neuronal activity elicited by natural images. These neuronal activity maps reveal that DNNs trained on ImageNet learn to rely on different visual features than those encoded by IT and that this problem worsens as their accuracy increases. We successfully resolved this issue with the neural harmonizer, a plug-and-play training routine for DNNs that aligns their learned representations with humans. Our results suggest that harmonized DNNs break the trade-off between ImageNet accuracy and neural prediction accuracy that assails current DNNs and offer a path to more accurate models of biological vision. | ['Thomas Serre', 'Margaret Livingstone', 'Saloni Sharma', 'Michael Arcaro', 'Thomas Fel', 'Ivan F. Rodriguez', 'Drew Linsley'] | 2023-06-06 | null | null | null | null | ['object-recognition'] | ['computer-vision'] | [ 1.94641471e-01 -1.50374938e-02 1.45811617e-01 -4.05715346e-01
-5.58831841e-02 -5.83895445e-01 6.89818442e-01 -2.81909645e-01
-9.04555798e-01 3.70940536e-01 1.57615408e-01 -2.09501714e-01
7.25603700e-02 -4.60910887e-01 -9.17913735e-01 -6.95801675e-01
1.38989478e-01 3.40197980e-01 4.96735930e-01 -3.45739350e-02
2.93635160e-01 6.17983699e-01 -1.66712773e+00 6.63778245e-01
2.56575346e-01 1.13801920e+00 4.41336483e-01 6.48569107e-01
2.99378544e-01 8.12129438e-01 -5.13465047e-01 -1.30515069e-01
4.14788097e-01 -3.58711392e-01 -6.72777236e-01 -3.26777220e-01
8.07395637e-01 -2.87504196e-01 -4.94008541e-01 9.94177639e-01
4.04416025e-01 -3.57181370e-01 6.13052666e-01 -1.22683847e+00
-7.65491068e-01 3.19882631e-01 -1.40991420e-01 7.27025628e-01
-2.16737524e-01 7.18937576e-01 9.29085553e-01 -7.65147448e-01
6.63656294e-01 1.13764977e+00 7.07865775e-01 5.27714372e-01
-1.64601505e+00 -6.77096307e-01 -1.20563336e-01 1.98290378e-01
-1.16403449e+00 -5.71035385e-01 1.88375756e-01 -6.02662861e-01
1.37420118e+00 5.26515692e-02 1.32361495e+00 1.11248815e+00
6.50649726e-01 5.31195819e-01 1.14219725e+00 -2.05473062e-02
2.51037091e-01 3.48369405e-03 -2.44018555e-01 2.90505975e-01
4.09536958e-01 3.77814740e-01 -3.78721058e-01 9.30094495e-02
1.26494908e+00 1.21256769e-01 -4.24596250e-01 4.35919687e-03
-1.38505518e+00 7.61177599e-01 7.97511518e-01 5.65851808e-01
-4.58378166e-01 3.90939981e-01 1.62249580e-01 4.10968006e-01
-6.29428774e-02 9.99162197e-01 -4.11620259e-01 3.65036316e-02
-8.14828455e-01 3.04317087e-01 4.77363795e-01 2.83684760e-01
6.96939051e-01 3.16874385e-01 1.81691334e-01 7.39616632e-01
1.84486568e-01 2.23991483e-01 9.08167660e-01 -1.30098879e+00
-2.29629815e-01 8.36827695e-01 -2.29060248e-01 -9.53277111e-01
-4.92007941e-01 -6.41854465e-01 -7.25536644e-01 7.12488532e-01
6.68466806e-01 1.45236894e-01 -9.33618546e-01 1.85672665e+00
-3.63200724e-01 -2.96993591e-02 -6.30259067e-02 1.07972336e+00
4.36357409e-01 5.12572527e-01 2.44747326e-01 5.49449995e-02
1.15646565e+00 -2.53828168e-01 -1.13287650e-01 -7.18817592e-01
5.02625227e-01 -4.22323972e-01 9.07219827e-01 3.34545106e-01
-9.76776958e-01 -7.35481203e-01 -1.09997809e+00 -1.02547351e-02
-3.85725439e-01 -3.91392440e-01 6.80741608e-01 4.19712722e-01
-1.31413615e+00 6.33557677e-01 -7.37546265e-01 -7.12334633e-01
9.70462978e-01 6.49321437e-01 -5.36338985e-01 2.61569679e-01
-7.75809288e-01 1.30516088e+00 4.16967601e-01 -1.14133023e-02
-1.04720724e+00 -7.48835087e-01 -2.69857526e-01 1.45515695e-01
-3.10114294e-01 -7.42914081e-01 1.26990712e+00 -1.75899935e+00
-8.33500028e-01 1.34845936e+00 -1.09222181e-01 -7.83628285e-01
-7.79554471e-02 1.99362233e-01 -3.40464354e-01 1.44944012e-01
-9.45990682e-02 1.18490589e+00 5.04121304e-01 -1.22690284e+00
-6.03272736e-01 -6.15947187e-01 -1.38284326e-01 -1.53448991e-02
-1.10835798e-01 -5.33659346e-02 1.84204087e-01 -3.46793056e-01
2.13087946e-01 -8.73107374e-01 -2.08679889e-03 3.25469613e-01
1.65259108e-01 -1.53387263e-01 6.03752673e-01 -2.60894209e-01
6.21414483e-01 -2.06500125e+00 -2.13032916e-01 -1.80967897e-02
5.36434412e-01 5.61381161e-01 -5.15528917e-01 2.11711615e-01
-3.18907738e-01 3.01802814e-01 4.32938859e-02 4.19590831e-01
-2.36343503e-01 2.45705262e-01 -5.50076902e-01 4.76901978e-01
3.77194166e-01 1.29759789e+00 -5.36328375e-01 -1.09165773e-01
-4.57753465e-02 3.92813474e-01 -6.28883660e-01 3.24871466e-02
-2.39461273e-01 4.36744362e-01 5.90393580e-02 2.51645505e-01
4.12554920e-01 -4.21255231e-01 2.23091975e-01 -3.67017508e-01
-4.07743365e-01 2.77266353e-01 -4.87639755e-01 1.32574034e+00
-1.45998448e-01 1.15992558e+00 -2.95907676e-01 -1.28886747e+00
7.61956990e-01 3.77653629e-01 3.20708007e-01 -1.31404972e+00
3.89734536e-01 3.92800212e-01 1.02820647e+00 -4.08290684e-01
-2.79788464e-01 -2.78298765e-01 5.11826038e-01 3.45314115e-01
4.75875407e-01 -1.42720237e-01 -1.21886320e-01 -9.01207551e-02
1.24309969e+00 -1.68616638e-01 2.09093437e-01 -3.95270377e-01
-2.60224137e-02 1.85501173e-01 4.64676499e-01 8.99944246e-01
-1.99996769e-01 7.84379721e-01 5.56801558e-01 -9.49879766e-01
-1.40779853e+00 -1.12452650e+00 -3.73382241e-01 8.99798870e-01
-1.50792494e-01 2.14612231e-01 -6.70527995e-01 -8.00369016e-04
-2.27325149e-02 5.34886241e-01 -7.93997943e-01 -4.39531147e-01
-3.39679837e-01 -7.80880868e-01 7.23318100e-01 4.71203417e-01
6.27276778e-01 -1.34140575e+00 -1.25012684e+00 3.10814351e-01
3.68796229e-01 -1.07406378e+00 1.38485029e-01 4.60875928e-01
-8.93337965e-01 -1.09302378e+00 -7.44767845e-01 -1.06875944e+00
6.63244188e-01 1.65789247e-01 1.08071482e+00 1.32923096e-01
-7.07551301e-01 3.94626528e-01 8.18711147e-02 -6.97024822e-01
-2.64423192e-01 -9.50771645e-02 1.04967758e-01 -4.78991047e-02
6.56656027e-01 -1.02468777e+00 -9.68220890e-01 5.24280071e-01
-1.08169925e+00 5.41832484e-03 8.48662317e-01 7.39305556e-01
4.27679390e-01 -2.29257748e-01 6.82844698e-01 -6.57634377e-01
4.48707461e-01 -3.87331337e-01 -6.92943037e-01 3.71802077e-02
-5.05985379e-01 1.62322193e-01 6.23870552e-01 -5.84796667e-01
-6.45257473e-01 -2.12409105e-02 -1.31476477e-01 -2.22933501e-01
-2.76849419e-01 3.46193612e-01 2.77013719e-01 -4.00725037e-01
1.00675046e+00 4.15043533e-01 1.29221445e-02 -6.41264990e-02
-2.14834020e-01 2.93344170e-01 5.85191846e-01 -2.92118248e-02
5.90257943e-01 4.68141377e-01 -1.80112019e-01 -8.33040595e-01
-6.92124128e-01 -7.02943280e-02 -2.90008783e-01 -2.70333111e-01
1.03510678e+00 -8.12103748e-01 -1.33588588e+00 5.97686768e-01
-1.35938716e+00 -5.88365912e-01 -2.51408130e-01 5.76140225e-01
-5.73542893e-01 -4.26451713e-01 -4.40577447e-01 -5.38841069e-01
2.68136375e-02 -9.84200478e-01 5.25915384e-01 4.89295244e-01
-3.91215503e-01 -7.89765060e-01 3.49090338e-01 -1.64161455e-02
7.65879691e-01 -9.10561308e-02 1.23076332e+00 -8.44934165e-01
-5.50120890e-01 -1.01276122e-01 -6.39332116e-01 4.17000920e-01
-9.33094919e-02 -8.39192048e-02 -1.37375760e+00 1.61626227e-02
1.87372684e-01 -6.34631455e-01 1.02525520e+00 5.51083565e-01
1.23714340e+00 -4.75944310e-01 -3.04578513e-01 7.64838755e-01
1.65850198e+00 4.84250188e-01 9.93276298e-01 3.37981075e-01
2.33769611e-01 5.32243133e-01 -3.54356945e-01 -1.79882318e-01
8.66735801e-02 6.30462170e-01 5.14597297e-01 -1.37725607e-01
-1.23419039e-01 -2.14709297e-01 1.33031830e-01 4.03258741e-01
-2.87992001e-01 -1.21115930e-01 -1.19160175e+00 5.86097121e-01
-1.60402250e+00 -1.33186364e+00 1.93908960e-01 2.14722824e+00
8.19865465e-01 4.74126846e-01 1.07603021e-01 -1.82643950e-01
2.17138514e-01 1.17231188e-02 -1.00695741e+00 -4.67772424e-01
-4.26446080e-01 2.65111148e-01 5.35951912e-01 1.31097719e-01
-6.30635440e-01 8.62946451e-01 7.25506067e+00 2.37179026e-01
-1.54452741e+00 -1.21838488e-02 9.07469988e-01 -7.03580454e-02
-8.28156546e-02 -2.07164418e-02 -6.63711011e-01 3.37568998e-01
1.21464813e+00 -1.15974084e-01 7.34071374e-01 7.10103989e-01
1.02908790e-01 -2.46890917e-01 -1.38178980e+00 1.08139944e+00
-1.26440763e-01 -1.81156754e+00 4.12217081e-02 2.92525291e-01
5.72528183e-01 6.70756459e-01 3.35032105e-01 8.13122615e-02
4.06203538e-01 -1.63025534e+00 6.27976954e-01 6.04728699e-01
3.98520350e-01 -1.14571437e-01 6.06733263e-01 4.64266270e-01
-5.37198126e-01 -2.69130319e-01 -6.86751187e-01 -2.68021584e-01
-3.10713232e-01 9.87749845e-02 -9.52328265e-01 -7.71059334e-01
8.60694468e-01 5.37581921e-01 -7.54701138e-01 1.34635389e+00
3.27613384e-01 5.16769946e-01 -1.45284474e-01 5.89571428e-04
1.70639321e-01 2.32950374e-01 3.24454129e-01 9.17923748e-01
1.95926353e-01 2.32714042e-01 -4.29810256e-01 1.22494233e+00
-1.07249312e-01 -3.49167734e-01 -8.68485987e-01 -3.65340501e-01
3.87457311e-01 1.14002013e+00 -8.03597271e-01 -2.23955572e-01
-4.42104161e-01 7.27734983e-01 4.62967783e-01 6.23393893e-01
-4.14436907e-01 1.02068774e-01 7.61533558e-01 3.87791842e-01
2.47124285e-01 -1.78433403e-01 -4.45570856e-01 -8.84058356e-01
-2.62311101e-01 -9.71240044e-01 -1.87059447e-01 -1.12352800e+00
-1.18222487e+00 8.50615799e-01 -4.92788553e-01 -9.02524948e-01
-1.31353140e-01 -1.17044151e+00 -3.70776683e-01 7.87424743e-01
-1.08159912e+00 -6.72969282e-01 -1.95766032e-01 4.91387308e-01
3.11207741e-01 -1.17880568e-01 9.56424177e-01 -6.28225785e-03
-1.46534249e-01 2.89069414e-01 -3.78104970e-02 3.58413070e-01
4.71860111e-01 -9.20778930e-01 5.19616902e-01 4.66913074e-01
7.48978913e-01 6.38014376e-01 6.34058714e-01 -1.98122114e-01
-1.16616726e+00 -6.67121708e-01 5.19346774e-01 -7.07899868e-01
4.19339210e-01 -4.84665960e-01 -1.06574357e+00 8.67831111e-01
1.66260153e-01 2.26754308e-01 5.65011263e-01 -2.62242854e-01
-7.03685582e-01 -4.27096963e-01 -9.02143359e-01 6.17774367e-01
9.69675899e-01 -5.91014147e-01 -5.76631308e-01 1.35089174e-01
3.02021116e-01 2.58457750e-01 -6.06776536e-01 2.40336016e-01
9.16256130e-01 -1.20151472e+00 8.52082849e-01 -8.46561253e-01
3.46103638e-01 -6.81544840e-02 -3.54503602e-01 -1.40605283e+00
-5.79156101e-01 -1.76608041e-02 4.44501996e-01 5.97299814e-01
5.58761835e-01 -9.55065191e-01 7.06928611e-01 5.78848064e-01
-8.13389122e-02 -6.68998718e-01 -1.01828277e+00 -6.75228953e-01
1.50686800e-01 -1.30060866e-01 -1.71981137e-02 6.29216194e-01
-9.29667354e-02 3.37296158e-01 1.74891397e-01 -8.18403885e-02
3.66914123e-01 -3.32284898e-01 4.54197884e-01 -1.50001240e+00
-1.95017278e-01 -8.00906420e-01 -9.41114128e-01 -7.86820531e-01
-1.12110553e-02 -9.38288748e-01 -8.54319427e-03 -1.42117596e+00
3.31320792e-01 -1.83279857e-01 -3.42677474e-01 7.36067772e-01
3.06565821e-01 6.88466549e-01 2.78139293e-01 5.66079736e-01
-2.65360415e-01 -4.61943746e-02 1.16291273e+00 -4.99813706e-02
1.02776937e-01 -2.44779721e-01 -8.54662895e-01 8.33047509e-01
7.10083961e-01 -5.86761236e-01 -2.64349192e-01 -8.03354263e-01
4.40051258e-01 -3.35610271e-01 9.73314047e-01 -1.58691347e+00
4.05044407e-01 -3.80567834e-02 1.02637482e+00 1.31420836e-01
3.51726383e-01 -1.00652051e+00 2.87152648e-01 7.02635467e-01
-5.87683022e-01 9.29680392e-02 4.76896554e-01 2.28541195e-01
-1.76094308e-01 2.87621692e-02 1.18643391e+00 -4.58739638e-01
-9.29596663e-01 3.15006405e-01 -7.74790049e-01 1.40801981e-01
8.24901640e-01 -6.74517453e-01 -6.77295268e-01 -2.36659169e-01
-5.53661346e-01 -2.57626861e-01 5.42658269e-01 3.79429132e-01
6.79779530e-01 -1.06913948e+00 -5.61255634e-01 3.95982474e-01
-2.91818939e-03 -4.61095870e-01 -8.64561461e-03 8.57789159e-01
-4.76056695e-01 7.72581458e-01 -8.29680502e-01 -7.02218294e-01
-7.39515185e-01 2.91426629e-01 8.84036601e-01 2.71284312e-01
-3.45230758e-01 1.04134953e+00 7.57659018e-01 -9.65641961e-02
-6.45898432e-02 -2.59110332e-01 -7.94407204e-02 -7.58804753e-02
4.91701007e-01 -2.42608145e-01 -2.47869287e-02 -5.33147871e-01
-3.67561221e-01 4.62971032e-01 -2.46219993e-01 -2.14278623e-01
1.67268300e+00 2.91850477e-01 -3.75817567e-02 7.70950377e-01
1.04621005e+00 -6.47686481e-01 -1.55402243e+00 -3.28002614e-03
-1.60002094e-02 -2.09846616e-01 6.94763213e-02 -1.13202775e+00
-1.18707311e+00 1.07751346e+00 8.42878401e-01 2.53035009e-01
1.04464006e+00 1.22272409e-01 4.47347075e-01 3.64281923e-01
3.36605966e-01 -9.38353240e-01 3.73734146e-01 4.52455521e-01
1.02481198e+00 -1.27027202e+00 -3.30297589e-01 3.53686064e-01
-5.52705109e-01 1.06054509e+00 9.20208037e-01 -5.90549231e-01
5.36743104e-01 4.91579138e-02 8.08580816e-02 -3.63211274e-01
-1.09289753e+00 -6.87083509e-03 1.00108817e-01 7.41025150e-01
4.04430687e-01 -1.66981131e-01 1.11254789e-01 4.32424694e-01
-2.08193377e-01 1.51061296e-01 2.84734905e-01 4.77395892e-01
-8.36197555e-01 -6.81012809e-01 -1.30924255e-01 7.09205329e-01
-3.08843434e-01 -2.48177186e-01 -3.66777629e-01 8.29654992e-01
3.42068762e-01 4.34876770e-01 4.39667463e-01 -4.06342387e-01
1.75140560e-01 1.79122791e-01 7.12060750e-01 -7.23893881e-01
-6.57341242e-01 -3.24373990e-01 -4.28151131e-01 -6.23793781e-01
-3.01959693e-01 -4.07034248e-01 -1.21194446e+00 -1.49015769e-01
2.81220060e-02 -2.98649341e-01 6.88565135e-01 8.45149517e-01
4.28677648e-01 2.69507796e-01 2.63235837e-01 -9.73737478e-01
-4.59243089e-01 -7.96865523e-01 -4.94155616e-01 2.30439052e-01
4.52545166e-01 -3.10089231e-01 -3.88521105e-01 1.55512035e-01] | [9.633954048156738, 2.4523041248321533] |
976912f6-04b6-4f6d-8b13-b5c67eadfabb | one-shot-to-weakly-supervised-relation | null | null | https://openreview.net/forum?id=W0mr06PxTHp | https://openreview.net/pdf?id=W0mr06PxTHp | One-shot to Weakly-Supervised Relation Classification using Language Models | Relation classification aims at detecting a particular relation type between two entities in text, whose methods mostly requires annotated data. Data annotation is either a manual process for supervised learning, or automated, using knowledge bases for distant learning. Unfortunately, both annotation methodologies are costly and time-consuming since they depend on intensive human labour for annotation or for knowledge base creation. With recent evidence that language models capture some sort of relational facts as knowledge bases, one-shot relation classification using language models has been proposed via matching a given instance against examples. The only requirement is that each relation type is associated with an exemplar. However, the matching approach often yields incorrect predictions. In this work, we propose NoelA, an auto-encoder using a noisy channel, to improve the accuracy by learning from the matching predictions. NoelA outperforms BERT matching and a bootstrapping baseline on TACRED and reWiki80. | ['Sophia Ananiadou', 'Phong Le', 'Thy Thy Tran'] | 2021-06-22 | null | null | null | akbc-2021-10 | ['relation-classification'] | ['natural-language-processing'] | [ 1.82756558e-01 6.62981749e-01 -6.31843686e-01 -4.46781576e-01
-8.11058521e-01 -5.56485593e-01 7.83223391e-01 8.51000726e-01
-4.27621812e-01 1.12529719e+00 -7.32220560e-02 -2.89444387e-01
-1.83442682e-01 -1.14237392e+00 -7.75802672e-01 -3.29865485e-01
-2.79386615e-04 9.88503397e-01 4.03348446e-01 -2.64208287e-01
-1.55185357e-01 1.71919197e-01 -1.30060613e+00 4.68905538e-01
5.54410875e-01 1.22210681e+00 -1.30095556e-01 5.20373821e-01
-5.63092113e-01 1.40587258e+00 -4.06059831e-01 -1.33623600e+00
-1.65529177e-02 -2.93028682e-01 -1.50974071e+00 -2.25231156e-01
1.61348730e-01 1.74387157e-01 -2.54047275e-01 8.42578053e-01
2.83218443e-01 2.56041717e-03 6.93654835e-01 -1.28095758e+00
-6.12352669e-01 1.08537126e+00 -2.42110401e-01 1.19343154e-01
6.01523936e-01 -4.83920902e-01 1.52389276e+00 -8.64145994e-01
9.01134193e-01 7.05322087e-01 9.66685712e-01 3.55220765e-01
-1.37933159e+00 -5.52926064e-01 -2.46405795e-01 5.71816325e-01
-1.75389242e+00 -5.84794760e-01 4.89809334e-01 -4.69370842e-01
1.40497494e+00 1.58751190e-01 4.78288531e-01 9.87236857e-01
-3.74118924e-01 6.79240704e-01 6.51300073e-01 -8.30905080e-01
2.10682362e-01 4.97836798e-01 3.37483793e-01 5.50491095e-01
3.54939282e-01 4.00442183e-02 -7.31988847e-01 -2.64543325e-01
2.96355665e-01 -3.32574755e-01 -2.12301016e-01 -3.25366944e-01
-7.63972461e-01 6.10412359e-01 4.43186104e-01 5.80911994e-01
-3.97003800e-01 -1.25772163e-01 6.75530791e-01 4.46847469e-01
6.33483648e-01 7.11192369e-01 -9.02238548e-01 -1.19512402e-01
-7.67767608e-01 3.49620618e-02 1.36791074e+00 1.29068875e+00
1.09008265e+00 -8.22267354e-01 -4.13305424e-02 8.08059454e-01
4.12445441e-02 -1.11470439e-01 5.34330904e-01 -5.10330260e-01
5.93795776e-01 9.27497804e-01 7.96497092e-02 -7.97532916e-01
-4.37386781e-01 -2.52877831e-01 -8.03286433e-01 -4.30729955e-01
5.22095323e-01 -1.41394520e-02 -4.50663686e-01 1.28110480e+00
3.99709105e-01 3.49158674e-01 5.20806730e-01 4.42435801e-01
1.11104012e+00 2.38005847e-01 3.12011421e-01 -3.78509611e-01
1.35476422e+00 -9.40019310e-01 -8.87181818e-01 -2.05940142e-01
1.16214848e+00 -5.59651315e-01 7.19258070e-01 1.47150263e-01
-7.18473196e-01 -2.55673379e-01 -8.56573522e-01 -1.98655084e-01
-7.03019023e-01 9.32459608e-02 8.94769430e-01 5.68442106e-01
-4.46170986e-01 7.78690636e-01 -6.42857552e-01 -5.27740717e-01
4.13168669e-01 1.89280510e-01 -6.81229949e-01 1.46051288e-01
-1.54771650e+00 1.14591002e+00 9.89263296e-01 -8.25785175e-02
-1.30894542e-01 -4.81401891e-01 -1.04243755e+00 1.28723800e-01
7.48914897e-01 -3.99993211e-01 1.28615499e+00 -8.76610160e-01
-1.24005353e+00 1.12214291e+00 -9.87556279e-02 -1.08549178e+00
2.60495007e-01 -3.29406559e-01 -6.31474257e-01 -9.66864228e-02
6.42115995e-02 1.90206602e-01 2.64671147e-01 -1.12475967e+00
-6.64180160e-01 -1.65563434e-01 1.14654981e-01 -3.19416821e-02
-2.66405374e-01 1.45163730e-01 -2.82176852e-01 -4.23045248e-01
-1.23452311e-02 -6.72111869e-01 1.28817689e-02 -5.07314861e-01
-6.06287718e-01 -7.67039120e-01 3.47102135e-01 -5.22897899e-01
1.41620541e+00 -1.98792088e+00 -2.62863725e-01 3.06893110e-01
2.70221323e-01 3.18990499e-01 3.15827936e-01 6.11339450e-01
-2.13128835e-01 1.42952770e-01 3.92153999e-03 2.67267488e-02
-1.09224275e-01 4.20773983e-01 -4.22534674e-01 7.79117569e-02
3.43363196e-01 1.07318223e+00 -1.03178608e+00 -7.77774394e-01
-1.90368846e-01 2.59826511e-01 -2.26569831e-01 2.83960730e-01
-2.90013134e-01 8.90686810e-02 -8.67878646e-02 5.33182561e-01
1.23570606e-01 -5.93371153e-01 6.36404932e-01 -4.00189310e-01
3.83494377e-01 7.68052936e-01 -1.11142874e+00 1.49327946e+00
-6.04371488e-01 4.90697682e-01 -5.75848877e-01 -1.31533468e+00
1.10506725e+00 8.84534359e-01 3.44471544e-01 -2.42893204e-01
1.25149176e-01 5.59791028e-01 4.44525294e-03 -6.24010921e-01
4.03657019e-01 -3.28568190e-01 2.55597215e-02 1.62016302e-01
5.53834856e-01 -8.38661939e-02 4.27580774e-01 1.61339417e-01
1.29502499e+00 2.88168192e-01 9.26272750e-01 2.02489913e-01
4.66232568e-01 1.93993375e-01 7.07894027e-01 7.02583075e-01
1.81613475e-01 1.31281212e-01 4.09881413e-01 -5.61195374e-01
-8.20552886e-01 -5.37914872e-01 -2.91094065e-01 1.03582239e+00
6.58437982e-02 -8.53144109e-01 -2.38506973e-01 -1.07520795e+00
-1.41902445e-02 8.65886390e-01 -5.41727483e-01 -2.42268741e-01
-3.93007576e-01 -4.91335213e-01 8.10838938e-01 5.97989619e-01
2.65793622e-01 -1.05137324e+00 -3.01378816e-01 4.45740014e-01
-4.04186308e-01 -1.39399362e+00 1.54343247e-01 7.97279716e-01
-4.69399363e-01 -1.36795413e+00 -1.42650411e-01 -7.58199692e-01
4.28909034e-01 -3.37534040e-01 1.58699858e+00 1.81723475e-01
-6.73683211e-02 -1.62954837e-01 -6.35648668e-01 -3.07269186e-01
-3.98952365e-01 4.73942935e-01 -7.29601905e-02 -9.46909413e-02
9.72520947e-01 -4.25496519e-01 -3.14556696e-02 2.23416895e-01
-4.18262661e-01 2.71064267e-02 5.90505242e-01 1.11734736e+00
5.39938092e-01 2.48133317e-01 6.72103286e-01 -1.35798717e+00
3.42418134e-01 -4.96109605e-01 -2.87169039e-01 5.94454050e-01
-7.56390095e-01 2.52207041e-01 4.69813585e-01 -3.17498624e-01
-1.06263983e+00 3.89967680e-01 -1.34555483e-02 -2.00423256e-01
-3.21839541e-01 9.74675179e-01 -2.09705546e-01 1.75590336e-01
1.03523028e+00 -1.43564358e-01 -5.36172986e-01 -4.28949654e-01
4.38505918e-01 7.53957033e-01 5.70585728e-01 -6.20201230e-01
7.04314828e-01 -8.44999496e-03 -2.11787056e-02 -4.08478945e-01
-1.65992093e+00 -6.25970066e-01 -1.10925817e+00 1.79316223e-01
7.11597145e-01 -8.02884758e-01 -5.76358020e-01 -1.09292582e-01
-1.23586214e+00 -1.86621979e-01 -5.32022357e-01 4.93628860e-01
-4.90533829e-01 1.32807717e-02 -6.55582428e-01 -7.92109847e-01
-2.28655592e-01 -5.11102915e-01 8.02268445e-01 -3.23137492e-02
-6.35676146e-01 -9.84212875e-01 1.51278898e-01 5.01109719e-01
1.47770680e-02 -1.55243367e-01 7.78882444e-01 -1.45914972e+00
-3.67425412e-01 -7.64028132e-01 -1.82762787e-01 -1.22112744e-01
2.52215147e-01 -1.49858117e-01 -1.05618095e+00 1.62293285e-01
-5.53042710e-01 -8.06699872e-01 6.28938615e-01 -3.14934582e-01
9.91680443e-01 -5.58176279e-01 -7.35175788e-01 2.14530066e-01
1.17593825e+00 9.93987769e-02 3.70365500e-01 3.58238399e-01
5.43744504e-01 7.50009477e-01 7.91429341e-01 1.32740930e-01
4.37751055e-01 8.18526089e-01 -1.46213889e-01 1.17457122e-01
9.32297856e-02 -3.49985272e-01 -2.74936140e-01 5.32437444e-01
-3.82496148e-01 -2.09224358e-01 -1.25607049e+00 5.58588147e-01
-1.98047817e+00 -1.17463160e+00 -2.21139401e-01 2.19776034e+00
1.65064359e+00 3.67121339e-01 -1.20259546e-01 4.02723253e-01
5.20508945e-01 -4.76043373e-01 -1.71102017e-01 2.09808782e-01
-2.19933450e-01 3.15021694e-01 5.04575193e-01 4.85309452e-01
-1.23640990e+00 1.12081707e+00 5.47565842e+00 8.51417899e-01
-6.73137367e-01 2.16978997e-01 4.20050859e-01 3.38490397e-01
1.97917655e-01 3.03420305e-01 -9.86076415e-01 1.05871022e-01
1.03153372e+00 -2.83454418e-01 9.00165588e-02 7.86158323e-01
-5.94920397e-01 -1.45725548e-01 -1.47250009e+00 8.71178508e-01
-8.31149891e-02 -1.48982728e+00 -3.01260620e-01 -2.12031081e-01
4.04321432e-01 -1.45178527e-01 -7.64168680e-01 5.69662511e-01
7.44233787e-01 -1.04490709e+00 4.25003141e-01 5.83884239e-01
7.15442777e-01 -6.72954082e-01 1.23960745e+00 4.58547115e-01
-1.32770598e+00 1.67724088e-01 -2.14403927e-01 -1.44327253e-01
-1.16969757e-02 6.76823556e-01 -1.20409334e+00 7.42106438e-01
7.09859908e-01 7.71091044e-01 -5.78231871e-01 8.77719820e-01
-5.80636740e-01 7.28458524e-01 -2.08020687e-01 -4.03146893e-02
-2.65892595e-01 2.08933905e-01 1.09781876e-01 1.15724051e+00
-1.76047776e-02 2.05072388e-01 4.18215692e-01 5.65308630e-01
-2.43886679e-01 2.30126604e-01 -8.02913487e-01 -1.42923921e-01
7.55696058e-01 1.31057692e+00 -6.43802583e-01 -6.63814783e-01
-6.50296450e-01 8.92336190e-01 8.19964588e-01 8.88550580e-02
-5.24400949e-01 -6.87123060e-01 1.83042735e-01 -6.89937826e-03
2.15631798e-01 1.52779430e-01 -1.46018922e-01 -1.18435597e+00
-7.39743412e-02 -4.70850438e-01 6.68198526e-01 -5.64244032e-01
-1.53113198e+00 7.27043629e-01 -2.08913922e-01 -1.18012452e+00
-4.28523868e-01 -4.75615263e-01 -4.38710563e-02 5.63663721e-01
-1.38124299e+00 -1.18844926e+00 -2.83224434e-01 3.79456788e-01
1.56827658e-01 -8.70595425e-02 1.25289047e+00 5.74121058e-01
-4.85918641e-01 7.02458739e-01 -3.37062567e-01 6.76810145e-01
9.44468260e-01 -1.42992091e+00 1.48672119e-01 4.03919488e-01
7.08773077e-01 5.18441141e-01 7.56048620e-01 -7.16753900e-01
-8.73207629e-01 -1.04086041e+00 1.64091980e+00 -5.76376557e-01
1.05328536e+00 -2.94256568e-01 -1.36760175e+00 9.79619026e-01
9.34811607e-02 4.45027411e-01 1.12521625e+00 7.46547580e-01
-6.91294372e-01 -1.30956201e-02 -9.20555174e-01 1.49184823e-01
1.18613887e+00 -7.83066213e-01 -8.61769021e-01 5.63339591e-01
5.80414772e-01 -3.40650648e-01 -1.25904655e+00 3.21498424e-01
2.65519261e-01 -5.02263069e-01 7.55779147e-01 -7.99512982e-01
3.96554500e-01 -3.20580930e-01 -1.64172918e-01 -1.22841108e+00
-8.90835673e-02 -4.86369729e-01 -5.22740245e-01 1.64430606e+00
1.04437363e+00 -2.93895960e-01 6.60073161e-01 8.76185656e-01
3.30468446e-01 -5.39658725e-01 -8.22380602e-01 -9.87094104e-01
-2.18287632e-01 -6.25597656e-01 7.23936737e-01 1.40243196e+00
6.00508094e-01 1.08326948e+00 -2.65046835e-01 1.45999104e-01
1.94211349e-01 1.54128060e-01 8.10090244e-01 -1.71038449e+00
-3.59906405e-01 -1.71106279e-01 -7.03024566e-01 -5.49070656e-01
4.74636734e-01 -1.16266251e+00 5.03937118e-02 -1.35823131e+00
5.17141446e-02 -7.24713385e-01 -1.42952397e-01 9.71973181e-01
-1.64984629e-01 2.43675724e-01 -3.77645224e-01 3.13458353e-01
-7.20633805e-01 3.01749676e-01 5.33287406e-01 -1.83902204e-01
-8.89377818e-02 8.31650123e-02 -4.19161141e-01 8.44554722e-01
6.31708205e-01 -5.11902928e-01 -2.82941073e-01 1.12224042e-01
9.73700762e-01 1.94393322e-01 3.03599209e-01 -7.37840772e-01
5.73939681e-01 7.49100139e-03 1.05670229e-01 -4.45150018e-01
1.98431492e-01 -1.18418455e+00 1.93495005e-01 2.27147281e-01
-5.43106437e-01 -5.19572854e-01 -2.85400182e-01 6.96649969e-01
-5.34040391e-01 -5.46944797e-01 4.76839036e-01 1.80513784e-02
-7.28048205e-01 1.44078195e-01 5.07596955e-02 2.67006725e-01
9.82890606e-01 1.49445176e-01 -2.08661646e-01 -1.28497109e-01
-1.07856202e+00 -2.36149691e-02 1.18590787e-01 1.87625274e-01
1.96643680e-01 -1.27691245e+00 -5.35560071e-01 -1.12562969e-01
6.27692819e-01 2.45708212e-01 -4.39954370e-01 8.74672294e-01
-1.23231836e-01 3.00381541e-01 3.63521129e-01 -2.40064859e-01
-1.34153116e+00 6.88785851e-01 3.96414936e-01 -6.09853506e-01
-6.17871940e-01 1.01358116e+00 -5.21601081e-01 -5.32970607e-01
3.36354524e-01 -7.03273416e-02 -4.95096475e-01 3.69804591e-01
4.62985128e-01 1.45174218e-02 4.64644790e-01 -5.46330750e-01
-4.22272712e-01 1.89607382e-01 -1.51251346e-01 2.05184385e-01
1.26971543e+00 -7.08098710e-02 -2.46740207e-01 8.53411734e-01
8.16044807e-01 -5.91935851e-02 -6.52949154e-01 -8.99050593e-01
1.07198238e+00 -3.38176101e-01 -9.11867172e-02 -9.05429840e-01
-7.35345602e-01 4.90197301e-01 2.06412319e-02 6.07905805e-01
7.39192843e-01 5.39651215e-01 5.83218038e-01 9.81470883e-01
6.61701381e-01 -1.02529466e+00 -3.39214206e-01 7.50556052e-01
6.72388613e-01 -1.64282954e+00 1.05658486e-01 -8.25169086e-01
-7.00153351e-01 1.06655836e+00 6.55339956e-01 2.48197332e-01
8.32291663e-01 4.52391416e-01 -6.83196709e-02 -4.54807550e-01
-1.12446821e+00 -4.91716295e-01 4.73673910e-01 4.98462111e-01
1.06118011e+00 3.27143371e-02 -3.09530884e-01 8.53654563e-01
-4.64482486e-01 1.71441495e-01 1.05906993e-01 9.12164271e-01
-1.16008937e-01 -1.20020008e+00 1.58213019e-01 6.38154149e-01
-5.25699735e-01 -3.46771032e-01 -5.82348287e-01 6.42192662e-01
2.46610180e-01 1.03752017e+00 1.11417668e-02 -3.57793272e-01
1.90781251e-01 5.36067069e-01 3.51109177e-01 -1.02609885e+00
-6.38477683e-01 -3.84329557e-01 7.42441356e-01 -2.11752355e-01
-7.42083013e-01 -5.32110214e-01 -1.26776636e+00 -2.74142958e-02
-9.89796221e-01 4.59142894e-01 1.04654662e-01 1.29806828e+00
2.27562651e-01 3.03527504e-01 2.43230999e-01 -1.62508890e-01
-1.40403137e-01 -1.15638828e+00 -5.12808204e-01 5.17672598e-01
-7.79307261e-02 -8.59058380e-01 -5.46915159e-02 2.84047306e-01] | [9.414453506469727, 8.496979713439941] |
0bfe6a6c-8b31-4b65-8c46-807d599a7acf | general-board-game-concepts | 2107.01078 | null | https://arxiv.org/abs/2107.01078v1 | https://arxiv.org/pdf/2107.01078v1.pdf | General Board Game Concepts | Many games often share common ideas or aspects between them, such as their rules, controls, or playing area. However, in the context of General Game Playing (GGP) for board games, this area remains under-explored. We propose to formalise the notion of "game concept", inspired by terms generally used by game players and designers. Through the Ludii General Game System, we describe concepts for several levels of abstraction, such as the game itself, the moves played, or the states reached. This new GGP feature associated with the ludeme representation of games opens many new lines of research. The creation of a hyper-agent selector, the transfer of AI learning between games, or explaining AI techniques using game terms, can all be facilitated by the use of game concepts. Other applications which can benefit from game concepts are also discussed, such as the generation of plausible reconstructed rules for incomplete ancient games, or the implementation of a board game recommender system. | ['Cameron Browne', 'Dennis J. N. J. Soemers', 'Matthew Stephenson', 'Éric Piette'] | 2021-07-02 | null | null | null | null | ['board-games'] | ['playing-games'] | [-2.83563703e-01 2.83215612e-01 2.68328816e-01 1.03328384e-01
8.73354301e-02 -6.53546393e-01 8.67897809e-01 1.16850525e-01
-1.95570841e-01 4.26424146e-01 1.90375343e-01 -3.92165899e-01
-6.41549826e-01 -1.28590024e+00 -9.49632674e-02 -5.25428236e-01
-1.81703001e-01 6.66197002e-01 6.54650629e-01 -1.17394423e+00
5.39821923e-01 1.75504133e-01 -2.06202459e+00 1.88521370e-01
5.04488051e-01 5.26481330e-01 4.54581022e-01 6.06908500e-01
-4.55976874e-01 8.37719560e-01 -6.80312335e-01 -4.98127848e-01
2.98333198e-01 -8.03669989e-01 -7.35753417e-01 -1.74771234e-01
-4.63506967e-01 1.48742139e-01 1.17465090e-02 8.62052739e-01
2.11415738e-01 5.89607298e-01 5.83280265e-01 -1.03986466e+00
-1.52026445e-01 9.86393273e-01 1.65802076e-01 -2.16071755e-01
7.86314189e-01 4.61944304e-02 1.05906820e+00 -1.82077214e-01
7.65970170e-01 9.53066528e-01 5.15492320e-01 7.28183150e-01
-8.42077613e-01 -4.59135413e-01 7.39515051e-02 4.39487666e-01
-1.31549013e+00 9.14968923e-02 6.18835390e-01 -5.59983015e-01
7.78482080e-01 5.74649155e-01 1.35755765e+00 8.03356230e-01
2.32220545e-01 6.38890684e-01 7.31930614e-01 -7.46542811e-01
6.21675849e-01 1.43572748e-01 -1.94998801e-01 4.91532922e-01
6.24377094e-02 4.72230554e-01 -5.16048014e-01 -6.36813268e-02
9.79399681e-01 -6.28527999e-01 9.62738618e-02 -6.52533889e-01
-6.54068828e-01 1.25660050e+00 5.07503673e-02 5.06417572e-01
-2.07863122e-01 8.20010453e-02 4.91432548e-01 3.03811103e-01
2.19879389e-01 1.15496719e+00 -4.53244805e-01 -6.54347658e-01
-7.05605626e-01 7.37092555e-01 1.12417507e+00 6.96493983e-01
6.23997450e-01 1.38814330e-01 2.85710841e-01 6.56604946e-01
4.64558661e-01 -2.03028500e-01 6.92385197e-01 -9.56065714e-01
-3.56186211e-01 7.15666115e-01 -7.58199394e-02 -7.92598009e-01
-7.08095014e-01 -1.45897999e-01 -1.61766171e-01 7.65300810e-01
2.33230740e-01 -1.52454272e-01 -4.88357335e-01 1.45914602e+00
2.70477831e-01 1.22640349e-01 4.75058593e-02 6.19293392e-01
1.02678430e+00 5.23212969e-01 1.26507625e-01 -1.00518297e-02
1.27340972e+00 -5.44899881e-01 -2.18892857e-01 -3.41438800e-01
6.58161879e-01 -3.01923275e-01 9.05319631e-01 6.66682005e-01
-1.20478213e+00 -5.34493089e-01 -8.70147109e-01 3.29977393e-01
-7.54380286e-01 -3.04544955e-01 1.09837890e+00 9.75645900e-01
-1.11144781e+00 7.01887310e-01 -6.51600242e-01 -6.58273578e-01
-3.47744763e-01 5.58005095e-01 1.96135454e-02 4.63311106e-01
-1.45603037e+00 1.21167898e+00 9.75154161e-01 -4.69816983e-01
-4.65596408e-01 -7.06197381e-01 -8.25599372e-01 1.46673635e-01
5.75896621e-01 -8.35094631e-01 1.15201449e+00 -9.14876342e-01
-2.20646167e+00 7.85047948e-01 8.76737654e-01 -3.46847624e-01
2.03785509e-01 6.39542341e-02 -2.24534497e-01 -3.38856846e-01
7.37802982e-02 4.77297872e-01 2.83913940e-01 -9.50031281e-01
-8.98667812e-01 3.22241671e-02 8.78772974e-01 5.43249011e-01
2.56571770e-01 1.25732690e-01 -5.21409094e-01 -5.64539433e-01
-1.19992666e-01 -8.75482619e-01 -5.77642739e-01 -3.10145915e-01
9.29049700e-02 -1.78781182e-01 -8.52887109e-02 -2.21302688e-01
1.36510825e+00 -1.95582676e+00 4.94391620e-01 5.26152730e-01
4.78578098e-02 1.87677339e-01 7.56211877e-02 7.91003704e-01
-1.57166809e-01 3.33744325e-02 2.19760343e-01 4.92692441e-01
3.09926301e-01 2.41984472e-01 2.06285238e-01 -2.55734980e-01
-5.03570020e-01 5.40827155e-01 -9.99492168e-01 9.07611325e-02
6.84086382e-01 1.03881449e-01 -8.74030769e-01 -1.38041258e-01
-3.22148234e-01 1.93416953e-01 -6.34294033e-01 -1.06148742e-01
1.49929235e-02 2.66030014e-01 4.65757340e-01 3.55514407e-01
-4.98991042e-01 5.09750545e-01 -1.56050563e+00 1.78622127e+00
-4.96732205e-01 3.85748565e-01 -2.30306938e-01 -7.70088375e-01
1.04704237e+00 2.24765405e-01 4.14185345e-01 -5.81147075e-01
3.37881207e-01 3.52560021e-02 4.77280140e-01 -3.53638709e-01
7.59231567e-01 -2.82439888e-01 -3.96987557e-01 5.54436684e-01
1.75187603e-01 -6.86387956e-01 5.90213597e-01 1.99277233e-02
1.02047575e+00 6.97809696e-01 8.86969268e-01 -4.58380371e-01
4.98444736e-01 3.51522088e-01 1.10032938e-01 8.44476998e-01
3.20283532e-01 3.84918302e-01 3.88916641e-01 -5.10118663e-01
-7.02389956e-01 -7.19060302e-01 1.74404085e-01 1.40464008e+00
2.54782259e-01 -1.21619809e+00 -7.16970921e-01 8.80210996e-02
-4.75742489e-01 8.31229687e-01 -7.42748797e-01 -2.72030115e-01
-2.98962504e-01 -5.81128180e-01 3.48063648e-01 -5.74066676e-03
2.59614021e-01 -1.43430161e+00 -1.03513992e+00 7.05602825e-01
4.49855477e-02 -5.70123971e-01 2.42813855e-01 4.47709769e-01
-5.99862754e-01 -9.81459022e-01 -5.72411865e-02 -4.77890819e-01
-2.18383238e-01 -5.10404294e-04 1.41966975e+00 4.66106892e-01
-1.86056823e-01 4.06284571e-01 -1.00557220e+00 -6.28036618e-01
-6.88755274e-01 1.83632001e-01 -1.10997476e-01 -7.22438395e-01
1.27066404e-01 -7.06913829e-01 -2.75848746e-01 9.98282358e-02
-8.14074695e-01 4.74912167e-01 4.51923162e-02 4.78242517e-01
1.56786814e-01 3.44911069e-01 1.18759081e-01 -9.41908956e-01
9.90195155e-01 -3.79908651e-01 -5.15546083e-01 1.32134899e-01
-1.81318238e-01 2.02925116e-01 5.06605089e-01 -3.59747827e-01
-9.47937012e-01 -3.35488975e-01 -5.87183237e-01 3.45014334e-01
-2.98506856e-01 8.00417006e-01 -4.10817057e-01 -1.11293502e-01
1.00784266e+00 1.97743312e-01 -3.01745385e-02 -1.68556675e-01
7.26463795e-01 1.77287593e-01 1.62445664e-01 -8.35683763e-01
5.89878857e-01 1.11885909e-02 -1.48765266e-01 -8.34618092e-01
-1.34859934e-01 -2.75720835e-01 -3.40368658e-01 -3.99279833e-01
7.73127198e-01 -4.04851526e-01 -8.54032159e-01 2.86134064e-01
-7.58206904e-01 -5.37717462e-01 -7.63405800e-01 4.44852173e-01
-9.25929785e-01 7.95970485e-02 -2.30641469e-01 -7.01056659e-01
8.86959955e-02 -1.17152798e+00 5.30190587e-01 5.13007164e-01
-7.40465641e-01 -1.18232298e+00 4.74318087e-01 1.04035228e-01
2.56502509e-01 3.30763347e-02 1.07067609e+00 -7.23336697e-01
-1.84156373e-01 -6.68534711e-02 5.97315013e-01 -2.39913091e-01
-5.94559778e-03 2.12302506e-01 -3.89732331e-01 2.25096181e-01
-1.08819745e-01 1.50385797e-01 1.60803244e-01 5.38271368e-01
5.83180189e-01 2.18464956e-01 -2.51124084e-01 5.28395772e-01
1.44598198e+00 8.31159115e-01 1.05594039e+00 8.84509981e-01
1.25420213e-01 8.37847114e-01 7.20780194e-01 7.63261616e-01
1.79018095e-01 1.12381279e+00 4.45936173e-01 3.06541473e-01
1.33596405e-01 -5.65630123e-02 3.86052370e-01 7.10618138e-01
-7.30972528e-01 -2.38165364e-01 -9.03276086e-01 7.82095641e-02
-1.83201897e+00 -1.29446602e+00 -5.68610504e-02 2.01152563e+00
3.18907291e-01 3.67442876e-01 5.68809927e-01 1.78966567e-01
4.77892339e-01 2.40494665e-02 -6.86743855e-02 -9.29304957e-01
1.57940015e-01 8.29754889e-01 6.67080134e-02 6.47938728e-01
-6.86147749e-01 1.41217089e+00 6.82649946e+00 1.17752290e+00
-8.14209282e-01 -1.28607377e-01 1.83436945e-02 7.84428716e-02
-2.26818129e-01 2.46180356e-01 -4.06703591e-01 2.24619225e-01
7.66853511e-01 -5.12814581e-01 7.37235546e-01 7.92320669e-01
3.30309659e-01 -1.70956716e-01 -6.39666915e-01 7.70147681e-01
-4.14140746e-02 -1.53540480e+00 2.12367266e-01 7.90593699e-02
3.12950164e-01 -4.01670665e-01 -2.96849549e-01 6.70652866e-01
9.56896961e-01 -9.64518487e-01 9.42883074e-01 4.16071177e-01
3.30998629e-01 -9.17835712e-01 4.70972180e-01 2.08485007e-01
-1.36081052e+00 5.86913265e-02 -4.61589426e-01 -7.37777948e-01
-8.39229673e-02 -3.39761466e-01 -4.18402225e-01 1.10276949e+00
5.76343417e-01 4.30658817e-01 -4.95512605e-01 1.35076034e+00
-3.83551061e-01 3.34750265e-01 -2.72641480e-01 -5.31307697e-01
2.86338508e-01 -6.66306019e-01 4.96926755e-01 7.38069892e-01
5.60338676e-01 4.71893013e-01 3.46602872e-02 8.96207094e-01
8.92231405e-01 4.09901083e-01 -5.09881139e-01 2.09189087e-01
3.19772124e-01 1.08785045e+00 -1.17358160e+00 7.99803138e-02
-3.49130660e-01 7.24052966e-01 -2.06894502e-01 -5.26297279e-02
-8.42828393e-01 -3.73146981e-01 1.08652115e+00 2.31636286e-01
1.76659912e-01 9.70097780e-02 -1.52474910e-01 -8.70310545e-01
-5.76554716e-01 -1.16949093e+00 1.99689239e-01 -1.00935268e+00
-8.55557203e-01 6.84548616e-01 3.86135906e-01 -1.22386253e+00
-7.12474763e-01 -7.79270887e-01 -1.04852045e+00 5.14498651e-01
-3.50502908e-01 -9.64524806e-01 -1.22388385e-01 5.17768085e-01
4.68279004e-01 -6.67296529e-01 1.04665434e+00 -2.18745530e-01
-4.35560197e-02 -1.23656942e-02 5.93992658e-02 -3.24189037e-01
2.14936025e-02 -1.22022855e+00 4.07006055e-01 3.85115474e-01
4.41642761e-01 4.14948612e-01 1.16818798e+00 -3.53925198e-01
-7.34448433e-01 -3.30330372e-01 2.91301906e-01 -3.03320915e-01
8.16221058e-01 -1.97335452e-01 -4.58022654e-01 2.20447108e-01
5.98884597e-02 -1.00461340e+00 9.23268199e-01 5.96116662e-01
1.64312735e-01 2.54324079e-01 -7.14063764e-01 1.08862007e+00
1.11174047e+00 -1.19722620e-01 -7.20805407e-01 4.59119827e-02
2.60970265e-01 -4.61949110e-01 -6.66266382e-01 -1.27939239e-01
6.28702402e-01 -1.27256274e+00 1.18038547e+00 -5.23566306e-01
1.84357464e-01 -2.41093189e-01 5.75956404e-02 -1.60074162e+00
-7.36492991e-01 -6.81503594e-01 4.47445959e-01 9.10469830e-01
3.55224222e-01 -1.99218825e-01 9.50885057e-01 4.14856493e-01
-2.37775847e-01 -3.97336215e-01 -6.59702539e-01 -5.83779275e-01
3.51444334e-01 -9.74780500e-01 9.62366164e-01 7.23256767e-01
7.56608605e-01 3.56847644e-01 -3.18542540e-01 -2.50166684e-01
-5.76314866e-04 -1.94356009e-01 7.78378069e-01 -1.75566709e+00
-8.23418438e-01 -9.66813982e-01 -9.52674866e-01 -7.32097864e-01
-3.93982641e-02 -7.88053691e-01 -2.11632019e-03 -1.70218265e+00
-2.16767609e-01 -3.52075368e-01 1.91211402e-01 6.44938350e-02
-5.05048642e-03 5.89540042e-02 5.46472073e-01 -9.20708999e-02
-4.99976903e-01 3.00751626e-01 1.02201581e+00 2.17957348e-01
-5.65763116e-01 1.01684123e-01 -9.00171578e-01 9.77811754e-01
8.60637069e-01 -4.42807108e-01 -7.25297272e-01 5.63062653e-02
9.39246833e-01 2.02194959e-01 -1.09773166e-01 -1.39307511e+00
3.03720415e-01 -4.50821072e-01 -3.15055966e-01 9.50048789e-02
4.34810221e-01 -6.61219776e-01 8.76303375e-01 6.91411257e-01
-7.88314939e-02 8.68954696e-03 2.30631337e-01 1.48481667e-01
-3.22085656e-02 -8.29463899e-01 3.91466260e-01 -5.93195617e-01
-1.14866436e+00 -2.66904205e-01 -1.21384025e+00 -1.88922174e-02
1.19084370e+00 -9.02359426e-01 1.71584889e-01 -7.19848335e-01
-1.24758744e+00 -3.15750539e-02 3.83882254e-01 4.01025921e-01
3.44786823e-01 -1.03386438e+00 -4.85873967e-01 -3.33216861e-02
1.46411553e-01 -4.47725534e-01 1.35238126e-01 1.79094635e-02
-9.80565131e-01 1.26119182e-01 -8.39451253e-01 1.15657240e-01
-1.27753496e+00 3.25520873e-01 5.81511855e-01 -5.06669879e-01
-5.50975621e-01 6.70067787e-01 3.75547618e-01 -5.41482449e-01
-2.57861584e-01 -1.57109380e-01 -7.51897991e-01 1.48090810e-01
4.51701880e-01 1.98916674e-01 2.22440455e-02 -3.31067234e-01
-1.83591723e-01 4.51565325e-01 2.39167660e-01 -3.05459291e-01
1.41770589e+00 -7.95621017e-04 -8.78676623e-02 5.15966237e-01
-1.39906481e-01 -1.70047313e-01 -8.18986058e-01 3.30549598e-01
9.68506858e-02 -1.77980497e-01 -1.82493448e-01 -8.13055456e-01
-8.55311811e-01 4.21932966e-01 4.08563972e-01 5.67324400e-01
9.24533606e-01 8.40945542e-02 -1.71193630e-02 2.65684038e-01
9.63063300e-01 -1.03855753e+00 -1.13346860e-01 9.56723750e-01
6.33679986e-01 -4.97765779e-01 -1.58398613e-01 -3.50424707e-01
-7.47867823e-01 1.33492517e+00 6.11303568e-01 -2.85059988e-01
6.31517708e-01 3.14224988e-01 2.93443743e-02 -4.81814623e-01
-7.71868110e-01 -7.87502825e-01 6.28344417e-02 1.22427809e+00
5.74896455e-01 1.91036180e-01 -8.74176502e-01 1.02893162e+00
-8.99853945e-01 -3.46866436e-02 9.79092836e-01 8.14615190e-01
-6.32770479e-01 -1.57254231e+00 -2.87430048e-01 3.04252088e-01
-1.82733029e-01 -1.85938776e-01 -6.33143842e-01 1.28194737e+00
6.16960406e-01 8.80655408e-01 3.38713862e-02 -7.29935944e-01
3.42413306e-01 -7.31663108e-02 7.51571834e-01 -1.21315598e+00
-1.18987548e+00 -2.67621815e-01 2.35407338e-01 -2.74280220e-01
-3.99222314e-01 -5.16481817e-01 -1.13258982e+00 -7.86427259e-01
-3.90805155e-01 6.79405451e-01 4.69491988e-01 1.06491303e+00
-1.67474508e-01 6.75181389e-01 -1.59598023e-01 -9.77254987e-01
2.35264584e-01 -7.24339604e-01 -1.00038135e+00 1.51282236e-01
-8.50994825e-01 -8.95255864e-01 1.14788346e-01 -2.42616102e-01] | [3.4425103664398193, 1.4869986772537231] |
ecabae3d-8ec6-4ce1-89d4-f3341c10708a | reside-improving-distantly-supervised-neural | 1812.04361 | null | http://arxiv.org/abs/1812.04361v2 | http://arxiv.org/pdf/1812.04361v2.pdf | RESIDE: Improving Distantly-Supervised Neural Relation Extraction using Side Information | Distantly-supervised Relation Extraction (RE) methods train an extractor by
automatically aligning relation instances in a Knowledge Base (KB) with
unstructured text. In addition to relation instances, KBs often contain other
relevant side information, such as aliases of relations (e.g., founded and
co-founded are aliases for the relation founderOfCompany). RE models usually
ignore such readily available side information. In this paper, we propose
RESIDE, a distantly-supervised neural relation extraction method which utilizes
additional side information from KBs for improved relation extraction. It uses
entity type and relation alias information for imposing soft constraints while
predicting relations. RESIDE employs Graph Convolution Networks (GCN) to encode
syntactic information from text and improves performance even when limited side
information is available. Through extensive experiments on benchmark datasets,
we demonstrate RESIDE's effectiveness. We have made RESIDE's source code
available to encourage reproducible research. | ['Chiranjib Bhattacharyya', 'Rishabh Joshi', 'Shikhar Vashishth', 'Sai Suman Prayaga', 'Partha Talukdar'] | 2018-12-11 | reside-improving-distantly-supervised-neural-1 | https://aclanthology.org/D18-1157 | https://aclanthology.org/D18-1157.pdf | emnlp-2018-10 | ['relationship-extraction-distant-supervised'] | ['natural-language-processing'] | [ 8.39523152e-02 7.78185964e-01 -7.61628747e-01 -5.81352770e-01
-4.71746445e-01 -5.68446875e-01 3.52966875e-01 6.44300461e-01
-3.12861323e-01 1.15310264e+00 3.35309893e-01 -7.13354468e-01
-2.92316496e-01 -1.30278480e+00 -9.15733218e-01 -1.41498014e-01
-2.53698081e-01 6.36307180e-01 8.05179104e-02 -3.51751715e-01
-2.72444069e-01 5.00263691e-01 -9.12813604e-01 6.69549465e-01
6.97737575e-01 9.66795743e-01 -2.53267646e-01 3.16869110e-01
-3.37149411e-01 1.32792246e+00 -5.11357367e-01 -8.75919580e-01
1.51859567e-01 -2.93953985e-01 -1.47256565e+00 -4.43374306e-01
1.20752990e-01 -1.34613603e-01 -6.00494862e-01 8.79852474e-01
9.43919569e-02 -1.06225878e-01 6.33171678e-01 -1.24710739e+00
-7.56028056e-01 1.34320676e+00 -4.24820364e-01 4.40925568e-01
3.50444525e-01 -4.65058625e-01 1.52676058e+00 -7.10736632e-01
8.07883680e-01 8.04969370e-01 8.35976660e-01 1.94552064e-01
-1.07055223e+00 -5.58438420e-01 2.29037907e-02 4.78996277e-01
-1.54279304e+00 -5.54661334e-01 8.10365736e-01 -1.09747157e-01
1.86639762e+00 4.80932415e-01 4.02105510e-01 7.64991045e-01
3.63781676e-02 6.14712179e-01 6.49651825e-01 -5.65201283e-01
-1.64023489e-01 7.55370334e-02 5.95764220e-01 7.11429656e-01
7.05894053e-01 -6.38839602e-02 -4.65827018e-01 -2.70307839e-01
4.44939911e-01 -3.59508872e-01 -2.69853890e-01 -2.14954354e-02
-1.01181161e+00 5.44501066e-01 6.06714785e-01 2.65081525e-01
-1.95009753e-01 -1.50348708e-01 3.06207925e-01 3.68840873e-01
6.21026993e-01 6.59321845e-01 -1.03041899e+00 7.33657554e-02
-2.61278152e-01 2.76324973e-02 1.39220321e+00 1.48063278e+00
9.83627081e-01 -5.05059719e-01 3.69308554e-02 9.32548583e-01
5.22426218e-02 1.27752602e-01 2.99367249e-01 -4.10492033e-01
1.03212261e+00 1.05994844e+00 -1.29145503e-01 -1.25508106e+00
-6.01580918e-01 -3.77790540e-01 -9.19596672e-01 -5.90726018e-01
1.82576254e-01 -4.72427845e-01 -7.63092875e-01 1.40975142e+00
3.81202221e-01 1.03729539e-01 5.22474825e-01 6.31152391e-01
1.70185065e+00 4.50510770e-01 1.76708475e-02 -2.94153601e-01
1.39982462e+00 -9.26319480e-01 -9.46440339e-01 -2.85969198e-01
1.24666524e+00 -4.09595132e-01 4.77635711e-01 -1.67700738e-01
-7.10014880e-01 -3.76186334e-02 -1.00798297e+00 -3.41500908e-01
-8.43903899e-01 9.80918854e-03 1.10587287e+00 1.70221761e-01
-6.42159283e-01 7.17094302e-01 -7.85977960e-01 -2.30629042e-01
5.33030450e-01 6.62439227e-01 -8.46211970e-01 1.30896598e-01
-1.65414107e+00 1.08914995e+00 9.40382838e-01 2.46398702e-01
4.84625995e-02 -5.89178801e-01 -1.33790231e+00 2.05577552e-01
9.61592436e-01 -6.71866536e-01 1.04500377e+00 -5.12510240e-01
-1.05426764e+00 7.68272221e-01 -3.47054064e-01 -6.90088570e-01
-1.76290244e-01 -3.27463061e-01 -7.52084315e-01 -1.81641757e-01
-6.08416125e-02 3.87253463e-02 7.21644238e-02 -9.56741810e-01
-5.13270557e-01 -3.53308141e-01 3.98030162e-01 2.13385180e-01
-4.12216336e-01 2.84120351e-01 -3.39736700e-01 -4.13073599e-01
1.54768586e-01 -6.18091941e-01 1.67277362e-02 -7.20830500e-01
-1.25189793e+00 -5.88932753e-01 6.06234372e-01 -8.04294169e-01
1.56294787e+00 -1.61399043e+00 -4.29974422e-02 4.49114919e-01
6.50739253e-01 3.33760828e-01 5.05668111e-02 6.48765028e-01
-4.23382640e-01 3.03333700e-01 -1.28691524e-01 1.94586247e-01
-8.42744038e-02 5.03893197e-01 -9.68638211e-02 -5.68295121e-02
5.89235365e-01 1.45622635e+00 -8.20622265e-01 -7.75501728e-01
-3.19806814e-01 1.50328323e-01 -2.32759774e-01 2.45352119e-01
-3.21644604e-01 -1.47619292e-01 -4.45976794e-01 6.21343434e-01
4.15195525e-01 -6.86431408e-01 4.54549670e-01 -6.85010791e-01
3.67284954e-01 9.76994097e-01 -9.76035237e-01 1.14577663e+00
-3.90924484e-01 4.90750760e-01 -4.06760365e-01 -1.07275581e+00
9.78382647e-01 3.75004292e-01 3.86926472e-01 -4.16802257e-01
8.30356926e-02 4.27740850e-02 2.07612619e-01 -6.66442037e-01
4.29041058e-01 1.11988753e-01 8.80778804e-02 3.72644007e-01
1.81225926e-01 1.58663929e-01 4.88877386e-01 5.18548846e-01
1.50505555e+00 1.66243181e-01 9.81339216e-01 1.05733439e-01
3.92165124e-01 2.82215416e-01 7.39902973e-01 4.72970903e-01
3.14775586e-01 1.31618604e-01 7.48396158e-01 -5.78852594e-01
-6.68186903e-01 -7.06024468e-01 -1.42859653e-01 6.86677098e-01
-2.91864909e-02 -1.08247995e+00 -2.04481572e-01 -1.25776196e+00
9.95321423e-02 5.03708065e-01 -5.61115861e-01 -9.88407731e-02
-7.12274313e-01 -8.55847716e-01 6.56539977e-01 8.20126176e-01
3.40122432e-01 -9.90523160e-01 1.67531416e-01 2.63573170e-01
-3.00712466e-01 -1.48374009e+00 -7.46337324e-02 7.33650982e-01
-5.54807246e-01 -1.33762157e+00 7.38833174e-02 -7.80693650e-01
7.32698739e-01 -1.85992807e-01 1.56389964e+00 9.80364010e-02
-2.57576704e-02 -2.67976195e-01 -5.08667409e-01 -3.58927339e-01
-2.54703164e-01 6.15505159e-01 -1.91157907e-01 -5.57403028e-01
1.04092157e+00 -6.81124389e-01 4.57588583e-02 7.44786710e-02
-4.97292101e-01 2.15734243e-01 5.46151340e-01 8.87367845e-01
5.48948407e-01 3.44517469e-01 4.93374705e-01 -1.57532644e+00
6.25949383e-01 -5.99715114e-01 -4.42293555e-01 4.92841363e-01
-7.66667843e-01 3.10951233e-01 4.78032023e-01 -1.86091915e-01
-1.04670131e+00 -3.47097218e-02 2.39828005e-02 1.04364015e-01
-2.75010169e-01 1.29066253e+00 -4.93361741e-01 1.83947548e-01
6.68926537e-01 -2.33946547e-01 -4.66016620e-01 -4.03879404e-01
3.21031272e-01 7.13063002e-01 5.62039196e-01 -8.18970382e-01
7.94412255e-01 4.96951528e-02 1.00969456e-01 -3.40141028e-01
-1.38168108e+00 -3.74571979e-01 -9.18071985e-01 5.56339264e-01
5.01155078e-01 -8.47001970e-01 -6.79134369e-01 -2.60690060e-02
-1.31770444e+00 -3.64699990e-01 -2.24261314e-01 4.53184307e-01
-5.11688851e-02 1.09579653e-01 -1.03663051e+00 -5.74992895e-01
-3.83571804e-01 -4.78566647e-01 6.35501921e-01 1.92697957e-01
-5.68833590e-01 -1.16826057e+00 -9.23333317e-02 3.31557065e-01
-7.56005868e-02 2.19344884e-01 1.01307952e+00 -1.30624127e+00
-4.34994251e-01 -2.09695682e-01 -5.06672800e-01 2.14966871e-02
6.90895855e-01 -8.56578127e-02 -7.19288766e-01 2.33951211e-01
-6.19208872e-01 -3.32195103e-01 7.56605744e-01 -1.46562308e-01
1.05625606e+00 -8.79985988e-01 -8.16096783e-01 7.64533937e-01
1.17075872e+00 1.77448094e-01 4.47398424e-01 4.78750557e-01
1.18703723e+00 7.36525178e-01 5.53307652e-01 1.54633656e-01
6.48272455e-01 4.90163684e-01 3.28137390e-02 -1.78181767e-01
-1.35535210e-01 -3.37760329e-01 -2.39122644e-01 8.35264981e-01
-3.30651104e-01 -3.28979045e-01 -1.17124069e+00 4.60524023e-01
-1.78036082e+00 -8.48946393e-01 -3.90764832e-01 1.60465527e+00
1.72330439e+00 2.40196109e-01 -1.86388046e-01 2.55715758e-01
5.28447688e-01 -1.04149349e-01 -3.95867079e-01 -2.34591976e-01
-3.27305436e-01 4.10837740e-01 7.18811274e-01 6.43191934e-01
-1.32931554e+00 1.19993937e+00 5.01101446e+00 6.82767987e-01
-6.27657056e-01 -1.95522413e-01 4.86512333e-01 1.76828802e-01
-4.36720282e-01 1.51506752e-01 -1.13626599e+00 -2.65867310e-03
8.73813510e-01 -2.16555312e-01 2.63124526e-01 6.72554433e-01
-3.86897951e-01 5.56275621e-02 -1.52184975e+00 6.78839326e-01
-4.08115201e-02 -1.60822022e+00 -1.56754166e-01 7.77583420e-02
5.48872590e-01 9.82810929e-02 -4.68425691e-01 3.10582429e-01
8.04883540e-01 -1.22166717e+00 -4.36476944e-03 2.49798611e-01
7.62489855e-01 -9.39969122e-01 1.25139308e+00 1.95195884e-01
-1.31338668e+00 2.65031099e-01 -1.95272118e-01 -2.87749201e-01
-4.96715568e-02 1.01702988e+00 -1.27301085e+00 1.12500691e+00
6.32467389e-01 1.02771187e+00 -5.97514629e-01 4.50315922e-01
-5.96421838e-01 5.49948275e-01 -4.30488169e-01 -9.40818787e-02
-8.93052518e-02 -6.36788532e-02 2.85338789e-01 1.33634973e+00
-2.39440560e-01 5.86739123e-01 2.23364346e-02 7.34516740e-01
-3.08082551e-01 2.13542804e-01 -8.80486846e-01 -3.56444985e-01
5.56014180e-01 1.30138671e+00 -5.46694934e-01 -5.01954436e-01
-6.10020816e-01 6.45667493e-01 9.50342417e-01 6.02904856e-01
-4.10046607e-01 -8.25992167e-01 3.90189022e-01 -1.61702186e-01
2.69861162e-01 -1.44541254e-02 -3.81014377e-01 -1.21111047e+00
2.33758479e-01 -6.29407465e-01 8.02166581e-01 -6.75647914e-01
-1.48238087e+00 8.58058691e-01 1.93626806e-01 -7.75882125e-01
-3.64525884e-01 -5.67530155e-01 -1.86703563e-01 9.90184307e-01
-1.47929490e+00 -1.36068094e+00 -5.51388264e-02 5.06841063e-01
-5.14768064e-02 -1.83225319e-01 1.03208160e+00 2.47987658e-01
-8.87998879e-01 9.03878570e-01 -4.67040867e-01 9.69034970e-01
6.82341635e-01 -1.41589296e+00 5.64011157e-01 6.19660556e-01
5.76372445e-01 9.44196582e-01 3.84632081e-01 -1.08286202e+00
-1.26344550e+00 -1.24289691e+00 1.50506425e+00 -5.95028818e-01
1.09712160e+00 -2.71094531e-01 -1.01720786e+00 1.32733488e+00
-3.02214595e-03 4.26375121e-01 1.04578912e+00 8.83600950e-01
-7.45918393e-01 -1.51126578e-01 -7.96590149e-01 4.82190341e-01
1.41239941e+00 -7.37690926e-01 -8.78120542e-01 4.79047894e-01
1.11550879e+00 -7.36761928e-01 -1.30962074e+00 8.59155536e-01
2.44130254e-01 -3.40054154e-01 9.48371053e-01 -1.16834497e+00
7.16410995e-01 -2.14907184e-01 1.17767025e-02 -1.25739336e+00
-2.01407000e-01 -4.41121936e-01 -1.08059609e+00 1.48427033e+00
1.12297821e+00 -7.24114478e-01 1.02773213e+00 8.49221766e-01
1.90602262e-02 -1.11272538e+00 -5.92657804e-01 -8.51269603e-01
-1.58404633e-01 -3.23156685e-01 9.60714340e-01 1.43741071e+00
6.82577372e-01 9.08998907e-01 -8.11275542e-02 4.00359541e-01
2.17415825e-01 4.88472998e-01 7.07428634e-01 -1.30717051e+00
-5.34113646e-01 -6.32301718e-02 -3.88683528e-01 -7.43189275e-01
6.96301401e-01 -1.29340351e+00 -1.91002578e-01 -1.72502923e+00
2.49732733e-01 -7.15444744e-01 -2.59835005e-01 1.23180842e+00
-2.84800828e-01 -2.98995432e-03 -3.39206666e-01 9.09834877e-02
-5.03906786e-01 2.34000310e-01 1.17919075e+00 -3.72720182e-01
-2.53691286e-01 -1.35413051e-01 -9.23482120e-01 6.84787154e-01
8.65437806e-01 -6.41265929e-01 -4.86300439e-01 -2.64800370e-01
6.26640022e-01 -4.20458429e-02 1.37575582e-01 -4.44176644e-01
3.85033190e-01 -3.64264786e-01 3.73333573e-01 -5.82308590e-01
1.26261085e-01 -8.60202670e-01 1.82976685e-02 -2.13573389e-02
-4.68722403e-01 -2.54557818e-01 1.42739207e-01 3.90178621e-01
-4.38701004e-01 -5.42664863e-02 -4.47233208e-02 1.22461811e-01
-5.31895280e-01 4.10870016e-01 2.77002126e-01 1.89960495e-01
7.99661398e-01 1.25906160e-02 -7.67857492e-01 -2.39609450e-01
-7.95463920e-01 2.56004572e-01 1.39465276e-03 2.66616851e-01
4.94786829e-01 -1.27771306e+00 -6.64495766e-01 -4.10714559e-03
3.91362667e-01 5.85982323e-01 -3.25954497e-01 6.99366152e-01
-2.27418646e-01 4.43581879e-01 2.87510425e-01 8.36056620e-02
-1.53645301e+00 7.06730485e-01 1.63506389e-01 -5.60050130e-01
-7.69961953e-01 1.06243932e+00 -1.70966089e-01 -5.41183412e-01
1.54440641e-01 -5.06684542e-01 -5.12590051e-01 -1.63995877e-01
4.60445195e-01 -1.78663522e-01 3.55979383e-01 -4.83322293e-01
-5.35889149e-01 1.76641103e-02 -4.56494004e-01 2.59566814e-01
1.50740528e+00 1.26083747e-01 -6.78862214e-01 4.12902355e-01
1.16132724e+00 2.49469131e-01 -6.18297637e-01 -6.17704749e-01
5.95839143e-01 -1.46428347e-01 -2.03967407e-01 -9.10854280e-01
-1.11639416e+00 3.14590871e-01 -4.76814240e-01 2.99944758e-01
9.10203218e-01 5.75890481e-01 8.61761808e-01 9.70244706e-01
3.04552943e-01 -7.74152696e-01 -5.61879635e-01 7.82443881e-01
7.47982264e-01 -1.27206433e+00 3.63506734e-01 -1.24930084e+00
-3.59073848e-01 1.02550375e+00 9.14385498e-01 2.41376922e-01
8.48308921e-01 9.62226808e-01 -6.86347112e-02 -4.88705426e-01
-8.98158967e-01 -4.14786696e-01 5.17936885e-01 8.72974277e-01
7.63007998e-01 1.73603758e-01 -3.31753343e-01 1.02711511e+00
-5.66936851e-01 -1.94555178e-01 2.50249743e-01 9.90403414e-01
9.39375386e-02 -1.38941097e+00 8.01900700e-02 9.04899478e-01
-5.05107403e-01 -6.48669600e-01 -8.79002333e-01 8.44978273e-01
1.24573372e-01 9.83749330e-01 -3.69356088e-02 -5.30853927e-01
1.04127675e-01 7.02353045e-02 4.38450515e-01 -1.01281607e+00
-5.16109586e-01 -4.20062423e-01 1.10357583e+00 -4.00332659e-01
-3.84265840e-01 -2.61366785e-01 -1.58912873e+00 -2.57976711e-01
-6.25955641e-01 3.74646127e-01 1.20282523e-01 1.35881078e+00
4.58043814e-01 6.87153816e-01 1.20243661e-01 1.02997005e-01
-9.23855510e-03 -9.85168040e-01 -5.01540005e-01 2.51084954e-01
1.33408800e-01 -5.82551003e-01 3.39861959e-02 1.93539336e-01] | [9.30806827545166, 8.538050651550293] |
edd2e6d2-6a3b-4206-901c-5b0c7cc801c0 | compositional-3d-human-object-neural | 2304.14070 | null | https://arxiv.org/abs/2304.14070v1 | https://arxiv.org/pdf/2304.14070v1.pdf | Compositional 3D Human-Object Neural Animation | Human-object interactions (HOIs) are crucial for human-centric scene understanding applications such as human-centric visual generation, AR/VR, and robotics. Since existing methods mainly explore capturing HOIs, rendering HOI remains less investigated. In this paper, we address this challenge in HOI animation from a compositional perspective, i.e., animating novel HOIs including novel interaction, novel human and/or novel object driven by a novel pose sequence. Specifically, we adopt neural human-object deformation to model and render HOI dynamics based on implicit neural representations. To enable the interaction pose transferring among different persons and objects, we then devise a new compositional conditional neural radiance field (or CC-NeRF), which decomposes the interdependence between human and object using latent codes to enable compositionally animation control of novel HOIs. Experiments show that the proposed method can generalize well to various novel HOI animation settings. Our project page is https://zhihou7.github.io/CHONA/ | ['DaCheng Tao', 'Baosheng Yu', 'Zhi Hou'] | 2023-04-27 | null | null | null | null | ['human-object-interaction-detection'] | ['computer-vision'] | [ 2.54395068e-01 1.19234882e-02 1.81165323e-01 -1.36389777e-01
2.27380600e-02 -2.42054060e-01 9.05342877e-01 -4.61930573e-01
1.76882669e-01 4.97947723e-01 4.06807810e-01 2.02176884e-01
8.98507312e-02 -7.01718330e-01 -9.70435619e-01 -5.57865322e-01
1.05108932e-01 4.40093309e-01 -5.60973547e-02 -1.17773890e-01
-1.65763926e-02 6.48854494e-01 -1.69853497e+00 7.50466809e-02
7.56504953e-01 3.78597617e-01 2.79718071e-01 1.03923929e+00
3.85151744e-01 7.28351712e-01 -5.35788596e-01 3.76729779e-02
2.62849510e-01 -7.62062550e-01 -4.80394244e-01 1.25797510e-01
3.99502128e-01 -2.68801242e-01 -3.16060185e-01 7.56680667e-01
5.24286687e-01 7.09568799e-01 8.31500709e-01 -1.58962381e+00
-1.12396693e+00 2.32586607e-01 -6.98637068e-01 -2.98670709e-01
6.88142180e-01 5.51682234e-01 4.73036319e-01 -8.31671178e-01
1.01290357e+00 1.74337399e+00 3.37767631e-01 1.10040891e+00
-1.33663881e+00 -7.11788297e-01 1.50373012e-01 -1.60870142e-02
-1.48858190e+00 -8.85024592e-02 1.16271269e+00 -6.01432443e-01
5.09143949e-01 4.48411316e-01 1.11605847e+00 1.38603139e+00
3.38980049e-01 9.68073905e-01 9.46414828e-01 -3.12302858e-01
-4.97880578e-02 9.32515785e-02 -1.85032323e-01 6.47221744e-01
5.83884791e-02 3.33482832e-01 -5.25126398e-01 2.23100949e-02
1.30236244e+00 8.75220671e-02 -4.96292472e-01 -5.28097093e-01
-1.73245013e+00 6.64992511e-01 5.09237289e-01 5.25063798e-02
-4.26443964e-01 5.48961580e-01 8.83205459e-02 -2.96762228e-01
1.38760358e-01 3.32415581e-01 -1.32164329e-01 1.81222007e-01
-3.85969967e-01 9.51291263e-01 4.07183141e-01 1.08243251e+00
6.06703222e-01 3.52693856e-01 -3.52806330e-01 3.97506356e-01
1.97306693e-01 5.66073895e-01 1.82203561e-01 -1.19551075e+00
-9.28643718e-02 3.74743521e-01 1.28424689e-01 -1.15882492e+00
-4.05096263e-01 -5.31999655e-02 -1.18166339e+00 4.06173289e-01
-5.06896339e-02 -1.70022592e-01 -1.06278729e+00 1.91383052e+00
8.02838504e-01 6.09385669e-01 -1.41664520e-01 1.17869759e+00
8.37140262e-01 9.69547510e-01 3.33459884e-01 -2.52532177e-02
1.24397278e+00 -9.49499369e-01 -9.12330627e-01 2.28540659e-01
4.51639771e-01 -5.74175060e-01 1.30928242e+00 1.91594467e-01
-1.28960037e+00 -8.51910055e-01 -8.22600007e-01 -2.63917714e-01
-2.64511764e-01 -3.56057212e-02 6.81703925e-01 5.17752729e-02
-7.69171000e-01 2.30737165e-01 -7.85241902e-01 -4.52503890e-01
1.20384574e-01 1.31917670e-01 -1.80483803e-01 2.21870691e-01
-1.15650737e+00 7.33821392e-01 5.29880822e-01 3.52809727e-01
-9.44179356e-01 -8.99649084e-01 -1.19185555e+00 -1.75237536e-01
2.25876495e-01 -1.44146192e+00 1.08977842e+00 -9.31445897e-01
-1.85372508e+00 5.72495222e-01 -3.36698368e-02 -1.26045793e-01
5.51432490e-01 -3.36847395e-01 -1.60242185e-01 -7.05153793e-02
-1.43203154e-01 1.21166027e+00 9.41239417e-01 -1.70311368e+00
-2.62490898e-01 -1.13882825e-01 1.21217497e-01 6.14743233e-01
8.32780823e-02 -3.49374861e-02 -2.82823622e-01 -8.61130595e-01
-2.10496232e-01 -1.33184731e+00 -1.64647147e-01 3.38674605e-01
-5.47131836e-01 -2.67148852e-01 1.22583854e+00 -5.22769868e-01
8.19143116e-01 -2.03468013e+00 7.90253043e-01 -1.72871035e-02
2.12929234e-01 1.62584856e-01 1.58220157e-02 2.76229918e-01
-2.07400888e-01 2.30736062e-02 -2.38469735e-01 -2.88049787e-01
5.26565313e-02 1.44961908e-01 -1.55984819e-01 3.01368564e-01
3.06393802e-01 1.39071119e+00 -8.83777499e-01 -5.43483853e-01
5.16559124e-01 9.36511695e-01 -6.65614784e-01 4.82227892e-01
-4.13079292e-01 1.34985185e+00 -3.76644254e-01 4.42903668e-01
6.12421691e-01 -2.32245907e-01 -2.68533230e-01 -3.46231759e-01
-1.48228779e-01 -3.35207433e-01 -1.13212037e+00 1.95203090e+00
-4.40091223e-01 3.42523426e-01 1.29861813e-02 -4.89189655e-01
6.44385636e-01 3.78618836e-01 2.97423184e-01 -3.12881351e-01
2.11154684e-01 -1.78810686e-01 1.12211588e-03 -5.46081126e-01
4.34409678e-01 -1.98077589e-01 -1.16302900e-01 1.90805197e-01
-1.85067788e-01 -4.78751481e-01 -2.15479553e-01 2.25404575e-01
4.48430985e-01 7.44972289e-01 3.10329080e-01 -1.61690906e-01
4.65092063e-01 -1.70957401e-01 4.49853927e-01 4.50331867e-01
-5.93811721e-02 7.80022442e-01 2.20062006e-02 -4.19793874e-01
-1.10382235e+00 -1.19934034e+00 1.35995969e-01 7.09847450e-01
4.85739231e-01 -6.62718294e-03 -6.66540265e-01 -1.92521334e-01
-3.95753905e-02 8.22512269e-01 -7.73286462e-01 -3.59582692e-01
-1.01815820e+00 -2.15481296e-01 1.44320562e-01 5.95745683e-01
4.61872995e-01 -1.71864474e+00 -1.02028525e+00 5.70989586e-02
-1.41526014e-01 -8.79023910e-01 -6.92263901e-01 -5.40156662e-01
-4.90739763e-01 -7.07174122e-01 -1.11588335e+00 -7.69115567e-01
5.86191773e-01 1.87440231e-01 8.44671249e-01 -6.37824982e-02
-6.93439245e-01 7.05283403e-01 -2.09977835e-01 -3.24987531e-01
-3.51478815e-01 -2.79968053e-01 1.60869241e-01 8.10867772e-02
-3.09447020e-01 -6.80819631e-01 -7.09446430e-01 1.67125538e-01
-9.92572963e-01 8.55381310e-01 2.39751190e-01 7.31561065e-01
4.61467713e-01 -3.03701609e-01 2.38954961e-01 -7.57704675e-01
5.11859596e-01 -3.79401922e-01 -1.98139161e-01 2.55218357e-01
1.14390254e-01 -9.42132547e-02 3.55145246e-01 -1.09428596e+00
-1.40559232e+00 1.04585111e-01 2.11157605e-01 -8.24723125e-01
-4.77450222e-01 3.21030527e-01 -3.73897582e-01 1.27103269e-01
5.29288411e-01 6.92062229e-02 -2.01296955e-01 -1.77039027e-01
8.86643112e-01 1.19327329e-01 7.10436165e-01 -8.50115180e-01
1.08664095e+00 4.66952324e-01 -7.29199871e-03 -9.46392596e-01
-2.57051319e-01 -6.55900761e-02 -6.81484282e-01 -5.30840814e-01
1.41688192e+00 -7.05682635e-01 -1.16075456e+00 5.90578496e-01
-1.53048480e+00 -5.72811186e-01 -4.31828648e-01 5.26825726e-01
-8.31891000e-01 2.69943833e-01 -5.55860102e-01 -8.80892098e-01
-2.61619687e-01 -1.15470302e+00 1.37282836e+00 4.59178418e-01
-5.21918595e-01 -7.60892630e-01 3.14255059e-01 1.71928614e-01
8.25515017e-02 9.96571898e-01 9.03074026e-01 2.62299001e-01
-7.91037560e-01 5.59432544e-02 -1.20699160e-01 -1.54557019e-01
7.19541982e-02 8.86702910e-02 -6.04708135e-01 -2.03077957e-01
-2.94373393e-01 -2.57270902e-01 3.04428667e-01 3.43320459e-01
1.16601932e+00 -4.69345391e-01 -2.77353883e-01 7.27345467e-01
1.07329381e+00 4.22474056e-01 6.57863736e-01 -8.40760954e-03
1.31498253e+00 7.06951678e-01 4.67620313e-01 5.57012498e-01
4.51636612e-01 9.37097013e-01 3.22167039e-01 -9.28049907e-02
-3.59907836e-01 -4.95023787e-01 3.14904928e-01 9.53264236e-01
-4.71445590e-01 -2.57783473e-01 -7.82257557e-01 4.01417255e-01
-1.89181566e+00 -8.52111459e-01 -1.22610237e-02 1.87004018e+00
5.42672157e-01 -3.68422419e-01 3.85490768e-02 -3.78243625e-01
8.43919694e-01 3.14522274e-02 -7.11397171e-01 -3.02555233e-01
1.25137061e-01 2.08250374e-01 -1.20451219e-01 4.48243827e-01
-7.06256092e-01 1.08939171e+00 4.72188091e+00 4.54126328e-01
-1.11118782e+00 1.56798854e-01 1.12624891e-01 -6.78014159e-02
-5.26814699e-01 2.31698584e-02 -4.87520486e-01 1.88144788e-01
5.34124434e-01 -3.00693482e-01 4.41268563e-01 6.58783317e-01
3.58893573e-01 2.15880737e-01 -1.07100642e+00 1.14221501e+00
2.32791722e-01 -1.22769451e+00 4.85178918e-01 2.40842775e-02
8.71568322e-01 -7.45430589e-01 2.49964565e-01 3.83153409e-01
2.28040680e-01 -8.37055624e-01 9.81243193e-01 1.04351318e+00
8.44404280e-01 -5.15375018e-01 2.74872463e-02 4.17144477e-01
-1.36531734e+00 2.49070749e-01 2.17884526e-01 5.18237129e-02
5.98848939e-01 -2.24966571e-01 -4.33592170e-01 4.81204093e-01
5.84907889e-01 6.85913205e-01 -2.82128990e-01 7.04084158e-01
-1.66108981e-01 6.17069602e-02 -1.47981435e-01 1.23436883e-01
-1.38772488e-01 -2.76266783e-01 8.24260235e-01 9.94468749e-01
2.97517538e-01 4.76766557e-01 3.59734446e-01 1.20587277e+00
1.29180506e-01 -8.30298066e-02 -8.32652450e-01 1.11520737e-01
2.06817035e-02 9.77945089e-01 -4.47017729e-01 -3.49323064e-01
-3.72425690e-02 1.36652100e+00 2.38053337e-01 8.28932285e-01
-1.21656239e+00 -2.22819135e-01 5.78883290e-01 2.22963080e-01
-5.56956418e-02 -6.10246599e-01 6.36472628e-02 -1.33534455e+00
-1.55344427e-01 -6.79943740e-01 -1.30585909e-01 -1.38815701e+00
-9.78726745e-01 4.62569863e-01 6.48103297e-01 -1.23764181e+00
-2.96520978e-01 -4.06257272e-01 -6.65230811e-01 6.90882504e-01
-9.72407103e-01 -1.81735599e+00 -6.76955879e-01 6.85460448e-01
7.42772400e-01 3.20826888e-01 5.80172122e-01 8.98458809e-02
-5.59662342e-01 4.01790500e-01 -6.00549221e-01 -2.83315241e-01
5.08390963e-01 -8.88306320e-01 3.62701714e-01 6.53819442e-01
1.35810738e-02 7.60502279e-01 7.66694665e-01 -8.51623654e-01
-1.45949686e+00 -1.12509370e+00 2.69216806e-01 -5.16720474e-01
2.43680254e-01 -5.27102947e-01 -9.51647341e-01 9.70585465e-01
2.46553838e-01 2.42852941e-02 1.51896626e-01 -3.35093349e-01
-6.18227050e-02 3.99185538e-01 -9.32814538e-01 1.28594458e+00
1.46205783e+00 -4.62702870e-01 -1.99975625e-01 1.98002860e-01
1.13853252e+00 -6.36707544e-01 -7.35689580e-01 6.35450423e-01
7.22809434e-01 -6.07066453e-01 1.11515474e+00 -6.60073817e-01
3.50062579e-01 -5.56543887e-01 7.84723610e-02 -1.10668004e+00
-5.38766265e-01 -8.15303624e-01 -4.56988543e-01 9.19482648e-01
-2.44829446e-01 -4.66748953e-01 5.65503001e-01 7.31709361e-01
-1.49874970e-01 -5.54905772e-01 -4.97257262e-01 -6.37408912e-01
-2.58289222e-02 -2.15189368e-01 6.77957356e-01 1.02712142e+00
-3.46868366e-01 3.01502347e-01 -9.11549926e-01 3.36745441e-01
6.17077291e-01 -6.33179350e-03 1.37794590e+00 -1.05677402e+00
-4.21183139e-01 -4.24776822e-01 -4.24319476e-01 -9.91364241e-01
3.23334843e-01 -7.37849414e-01 1.85888320e-01 -1.41526413e+00
2.79341280e-01 -2.88627416e-01 1.44412950e-01 3.01166385e-01
-2.98130870e-01 8.89312699e-02 6.72597826e-01 3.48560452e-01
-3.39292079e-01 1.14115298e+00 1.91190946e+00 -7.25350156e-02
-5.08389771e-01 -4.09470588e-01 -1.85241699e-01 8.93657804e-01
7.11520195e-01 -2.17969596e-01 -5.48545122e-01 -3.73403341e-01
-1.60822067e-02 3.22789311e-01 9.55966294e-01 -1.29541409e+00
8.33874643e-02 -5.80103457e-01 3.29271257e-01 -5.33487678e-01
6.28530920e-01 -7.08249032e-01 9.42982674e-01 4.93949741e-01
-4.23026264e-01 1.01268664e-01 8.44779164e-02 6.50934398e-01
5.45262434e-02 2.72862345e-01 6.46105587e-01 -2.01909035e-01
-7.05343664e-01 5.92998207e-01 -2.18275428e-01 -1.12284712e-01
1.16665947e+00 -2.28541061e-01 -2.56240685e-02 -3.74122620e-01
-8.58486474e-01 8.83056000e-02 5.08866906e-01 7.03474581e-01
8.44638586e-01 -1.59060180e+00 -7.28369594e-01 2.66825378e-01
2.08883747e-01 2.87559837e-01 6.91784382e-01 5.25081933e-01
-6.12221956e-01 7.36743286e-02 -3.16497386e-01 -6.71821833e-01
-1.19964218e+00 6.18424058e-01 1.27082154e-01 6.09056186e-03
-1.08712780e+00 6.78239286e-01 1.01305139e+00 -6.38320506e-01
1.55805741e-02 -2.41031438e-01 -1.47488207e-01 -5.73819816e-01
2.33992279e-01 3.20510358e-01 -8.83381784e-01 -1.01460505e+00
3.31084542e-02 8.39793563e-01 1.90543786e-01 -3.71406913e-01
7.90061116e-01 5.67536987e-03 1.00253955e-01 7.69153178e-01
1.12502253e+00 -3.88566941e-01 -1.34573162e+00 1.41460925e-01
-5.52471399e-01 -5.56425452e-01 -3.69622171e-01 -4.42904264e-01
-7.51922488e-01 9.52899754e-01 6.70919836e-01 -4.79816467e-01
9.49930489e-01 -1.37460474e-02 1.07653821e+00 1.30498677e-01
4.70103800e-01 -5.50644279e-01 2.96166509e-01 2.47635499e-01
1.57128596e+00 -8.58855486e-01 -1.67714015e-01 -4.39639509e-01
-8.96115184e-01 7.09082901e-01 9.50300395e-01 -1.84364989e-01
6.87430978e-01 -1.06662303e-01 -2.06803694e-01 -2.37825036e-01
-5.33135951e-01 -1.44892633e-02 4.99201030e-01 7.98749268e-01
2.44798765e-01 3.33808959e-01 -2.73580581e-01 1.38614580e-01
-1.97123155e-01 1.05208747e-01 4.64020133e-01 9.62814033e-01
8.29113498e-02 -7.13962138e-01 -3.26510549e-01 -2.03350503e-02
2.95084596e-01 1.53133854e-01 -3.79682183e-02 9.76912081e-01
4.29796845e-01 2.21253321e-01 8.04532021e-02 -2.61173218e-01
4.44615155e-01 -8.96329954e-02 7.25275755e-01 -7.09417880e-01
-4.32768553e-01 1.60084009e-01 -4.13124233e-01 -6.31877065e-01
-6.14743650e-01 -6.25120938e-01 -1.50395358e+00 -1.74472108e-01
-1.03826240e-01 -1.06102064e-01 5.43878675e-01 5.11455059e-01
3.78359765e-01 7.35271871e-01 3.15367907e-01 -1.65398419e+00
5.46813495e-02 -7.58867323e-01 -3.82436812e-01 8.43738973e-01
2.72071332e-01 -1.04288006e+00 -1.48099959e-01 5.47626257e-01] | [10.927425384521484, -0.7304286956787109] |
d1462e87-76c4-4b97-b34a-3a5ea3f8c32c | p-vectors-a-parallel-coupled-tdnn-transformer | 2305.14778 | null | https://arxiv.org/abs/2305.14778v2 | https://arxiv.org/pdf/2305.14778v2.pdf | P-vectors: A Parallel-Coupled TDNN/Transformer Network for Speaker Verification | Typically, the Time-Delay Neural Network (TDNN) and Transformer can serve as a backbone for Speaker Verification (SV). Both of them have advantages and disadvantages from the perspective of global and local feature modeling. How to effectively integrate these two style features is still an open issue. In this paper, we explore a Parallel-coupled TDNN/Transformer Network (p-vectors) to replace the serial hybrid networks. The p-vectors allows TDNN and Transformer to learn the complementary information from each other through Soft Feature Alignment Interaction (SFAI) under the premise of preserving local and global features. Also, p-vectors uses the Spatial Frequency-channel Attention (SFA) to enhance the spatial interdependence modeling for input features. Finally, the outputs of dual branches of p-vectors are combined by Embedding Aggregation Layer (EAL). Experiments show that p-vectors outperforms MACCIF-TDNN and MFA-Conformer with relative improvements of 11.5% and 13.9% in EER on VoxCeleb1-O. | ['Jing Xiao', 'Liang Xu', 'Bo Xu', 'Fangyuan Wang', 'Xiyuan Wang'] | 2023-05-24 | null | null | null | null | ['speaker-verification'] | ['speech'] | [-1.58410043e-01 -5.45585081e-02 8.78261309e-03 -7.29023874e-01
-7.18461752e-01 -2.54883468e-01 4.38782662e-01 -2.96129346e-01
-1.44794717e-01 3.74330223e-01 4.68668282e-01 -2.33707786e-01
-8.29783529e-02 -3.55977058e-01 -3.80794168e-01 -8.17811131e-01
-1.28692016e-01 -2.61616051e-01 2.08370779e-02 -2.45083243e-01
-1.12924270e-01 4.55683440e-01 -1.28352201e+00 3.21079373e-01
8.63058865e-01 1.29230070e+00 8.61374661e-02 4.49572563e-01
-1.68432325e-01 5.26967704e-01 -5.88798225e-01 -3.63730043e-01
9.67926383e-02 -3.04733783e-01 -5.06430566e-01 -5.16660094e-01
3.13266218e-01 -1.29560083e-01 -6.66924298e-01 9.34910476e-01
1.05535126e+00 1.56454995e-01 3.30288768e-01 -1.58527052e+00
-6.75179183e-01 9.04258966e-01 -5.54375708e-01 3.21168751e-01
8.35604891e-02 8.82493053e-03 1.14489245e+00 -9.96356726e-01
8.37749243e-02 1.49175823e+00 8.51501763e-01 5.80840528e-01
-9.03986871e-01 -1.04083133e+00 3.64188731e-01 6.66691780e-01
-1.47311056e+00 -8.97858500e-01 1.15009403e+00 -6.15076907e-03
1.22710502e+00 4.67469364e-01 5.50515831e-01 1.02400017e+00
5.70033416e-02 9.95649099e-01 7.62922049e-01 -3.11843336e-01
-2.54113317e-01 3.83600920e-01 5.50046384e-01 3.80356133e-01
-4.89254028e-01 4.27529037e-01 -1.07763028e+00 -1.14280581e-01
3.55225503e-01 -7.21292719e-02 -4.91991878e-01 2.65473664e-01
-9.19855416e-01 6.96722567e-01 5.79398274e-01 5.32713115e-01
-7.30445534e-02 -3.80791947e-02 4.73147631e-01 5.85787594e-01
3.50577027e-01 2.43006051e-02 -5.65281153e-01 -1.34446234e-01
-7.50087440e-01 -8.19126610e-03 4.26994115e-01 1.01207209e+00
5.32719672e-01 6.85897708e-01 -4.97416973e-01 1.07243562e+00
6.47556424e-01 6.22374058e-01 6.77235842e-01 -3.94836783e-01
3.74868006e-01 3.30570459e-01 -4.41274703e-01 -9.95820403e-01
-4.37723279e-01 -8.35408807e-01 -9.44387674e-01 -1.98432311e-01
-1.31231561e-01 -1.23116232e-01 -8.51775110e-01 2.13990092e+00
1.16387643e-01 4.42992508e-01 9.32754055e-02 7.68187702e-01
1.26151896e+00 1.03916931e+00 -3.58914360e-02 -1.32999733e-01
1.19294333e+00 -1.14036572e+00 -1.21311641e+00 -7.13144466e-02
3.19469929e-01 -8.51302326e-01 8.42873693e-01 -2.41729170e-02
-1.06593156e+00 -8.90667439e-01 -1.13828540e+00 -2.24728569e-01
-4.17441458e-01 1.72746718e-01 2.91946441e-01 7.73497224e-01
-1.28066659e+00 3.56656432e-01 -5.44606745e-01 1.38777234e-02
2.65103966e-01 7.92178690e-01 -3.47774833e-01 3.16731036e-01
-1.60661876e+00 5.67160189e-01 -1.16729066e-01 5.00153303e-01
-5.63256323e-01 -9.38822508e-01 -9.43077743e-01 2.50186771e-01
-3.24479431e-01 -2.87046552e-01 1.16809475e+00 -8.40906322e-01
-1.80840266e+00 2.93042570e-01 -3.80079746e-01 -2.58144677e-01
1.07391231e-01 -1.40257537e-01 -1.03518641e+00 -2.42177159e-01
-1.93186715e-01 7.30219603e-01 8.26942146e-01 -8.08461547e-01
-6.38379872e-01 -4.92452383e-01 -4.59900022e-01 2.50542819e-01
-5.86193562e-01 2.36467451e-01 -3.70961070e-01 -8.09459388e-01
1.62437931e-01 -6.42872393e-01 2.12225765e-01 -1.17385775e-01
-6.32873416e-01 -3.37798953e-01 1.40899289e+00 -9.62115943e-01
1.48727584e+00 -2.64695239e+00 1.45907924e-01 3.92715275e-01
2.58947343e-01 5.43541908e-01 -3.01283568e-01 2.76388824e-01
-2.08650783e-01 3.19449641e-02 5.90316392e-02 -5.39188862e-01
2.38607571e-01 -1.05482407e-01 -2.97022849e-01 4.87253606e-01
1.39147252e-01 9.56200719e-01 -3.26739877e-01 -4.01063442e-01
1.58885032e-01 9.41314578e-01 -5.61530709e-01 8.86284411e-02
3.41592729e-01 2.70749032e-01 -1.70376837e-01 5.50565004e-01
1.06106567e+00 2.56134391e-01 2.83949934e-02 -5.25359392e-01
-9.49921459e-02 6.02437019e-01 -1.03102326e+00 1.61745036e+00
-3.75062376e-01 7.66833842e-01 1.89134881e-01 -6.67224765e-01
1.03206003e+00 6.46069229e-01 2.33005345e-01 -1.02603245e+00
2.93323964e-01 1.33011147e-01 2.10643977e-01 -1.61512360e-01
3.27765673e-01 -8.28330070e-02 7.43084848e-02 3.02586854e-01
4.48651642e-01 5.98617971e-01 -5.65262675e-01 1.55050820e-02
6.52659416e-01 -3.88585895e-01 -5.10380752e-02 -2.65993744e-01
7.57951677e-01 -1.07839477e+00 1.01823986e+00 3.82790178e-01
-4.11901563e-01 4.27375227e-01 2.89899766e-01 -3.18101197e-01
-6.60717845e-01 -9.70910668e-01 -4.04628515e-01 1.15722454e+00
8.90851114e-03 -4.97866631e-01 -4.88260508e-01 -5.56032896e-01
5.44310734e-02 5.73580444e-01 -5.00132978e-01 -3.45426977e-01
-5.40947080e-01 -4.11445051e-01 8.64750445e-01 7.27818847e-01
5.98692000e-01 -7.60285437e-01 -1.83462759e-03 2.79594183e-01
-1.66808546e-01 -8.24606657e-01 -1.05380952e+00 3.54334265e-01
-4.98069257e-01 -3.57836157e-01 -6.86750531e-01 -9.03068602e-01
3.11377406e-01 2.71034569e-01 5.59791863e-01 -5.13267778e-02
2.15828702e-01 -5.54629005e-02 -4.03257281e-01 -3.58367622e-01
4.60486189e-02 1.14203215e-01 2.64999628e-01 4.58532274e-01
5.06720126e-01 -8.31419826e-01 -4.28429872e-01 4.87351596e-01
-4.86340404e-01 -1.60485849e-01 3.76107097e-01 1.18058383e+00
3.28223586e-01 -2.65371293e-01 7.81859636e-01 -4.49359655e-01
5.65176368e-01 -2.06728891e-01 -3.12981635e-01 2.33727708e-01
-4.43022758e-01 5.25417775e-02 4.91142333e-01 -4.95795459e-01
-1.07065594e+00 -2.05995932e-01 -5.40524542e-01 -7.13842809e-01
2.88294494e-01 2.39277601e-01 -8.42897356e-01 -1.93084717e-01
1.47201553e-01 4.21833545e-01 6.56232685e-02 -6.00478172e-01
1.84394374e-01 1.13248456e+00 3.22991371e-01 -1.91200748e-01
5.69855750e-01 -1.27811767e-02 -5.82085013e-01 -8.54045570e-01
-2.87444502e-01 -3.27524066e-01 -3.08608025e-01 -7.91981742e-02
6.76440656e-01 -8.66984904e-01 -8.17076445e-01 7.83747017e-01
-1.15911961e+00 2.27996528e-01 -2.60538131e-01 6.31086946e-01
-4.69114855e-02 1.93763033e-01 -6.17640138e-01 -7.21427858e-01
-5.67694366e-01 -1.29193091e+00 9.69964802e-01 4.87489015e-01
2.50146743e-02 -7.54209459e-01 -1.51340201e-01 1.13744721e-01
8.79626453e-01 -4.64519173e-01 7.25245833e-01 -8.98240566e-01
-2.80752033e-01 -4.58956044e-03 -1.01263940e-01 5.56946099e-01
1.31824821e-01 -9.31323022e-02 -1.73943508e+00 -4.86555696e-01
1.75819248e-01 1.41287997e-01 8.05735230e-01 6.68931901e-01
1.06206095e+00 -4.96219248e-01 -3.89946729e-01 9.84450996e-01
8.69915962e-01 5.90172052e-01 5.45501828e-01 -1.00991011e-01
5.84453523e-01 3.68370354e-01 1.01426683e-01 2.99474329e-01
4.52641606e-01 1.04703677e+00 1.72162186e-02 -1.71096727e-01
-3.38986367e-01 -3.08025539e-01 6.29813671e-01 1.42911029e+00
1.73486277e-01 -1.57345459e-01 -6.51400626e-01 3.62479866e-01
-1.57246792e+00 -1.06411970e+00 1.67836860e-01 1.92273796e+00
6.40821040e-01 -7.87864402e-02 7.22426102e-02 4.57679451e-01
8.72295320e-01 3.70402545e-01 -4.50529903e-01 -5.97270787e-01
-4.49595004e-01 -5.24972230e-02 4.74988222e-02 6.12415612e-01
-9.83816206e-01 5.78450143e-01 5.65182018e+00 1.05305815e+00
-1.56691146e+00 4.72799510e-01 6.30198538e-01 -1.57590926e-01
-4.98480707e-01 -2.41559535e-01 -9.98393357e-01 6.32651687e-01
1.09049070e+00 8.71158112e-03 3.09233904e-01 5.21750391e-01
-5.31497225e-02 6.88172460e-01 -1.01119816e+00 1.21949565e+00
1.12339184e-01 -9.46707249e-01 -1.11171529e-01 2.18411740e-02
2.76069671e-01 5.00644781e-02 4.63753045e-01 4.60813820e-01
-1.79799557e-01 -1.04956710e+00 8.94516468e-01 3.99591386e-01
8.97266448e-01 -9.57335830e-01 9.00916100e-01 -1.06784016e-01
-1.53986955e+00 -1.83781520e-01 -1.66714460e-01 3.19148481e-01
1.70829758e-01 3.78935784e-01 -3.94330323e-01 6.73056483e-01
8.18775237e-01 7.71847308e-01 -2.68598676e-01 9.49496031e-01
-5.01968572e-03 5.84094048e-01 -5.27551889e-01 1.04797229e-01
1.44391626e-01 1.88105002e-01 6.66148484e-01 1.23018122e+00
2.48627603e-01 -2.21448824e-01 -2.63190538e-01 7.19740689e-01
-4.53882292e-02 -3.05812433e-02 -3.22088480e-01 7.15289488e-02
7.38008142e-01 1.00881863e+00 7.93008730e-02 4.88696853e-03
-4.81671542e-01 7.32899904e-01 1.97853386e-01 3.58433902e-01
-9.29121614e-01 -8.66279304e-01 9.43717599e-01 -1.16367877e-01
4.04734313e-01 1.09190680e-01 -8.26139897e-02 -9.12833095e-01
2.20128223e-01 -9.19541836e-01 2.38350123e-01 -5.58730483e-01
-1.22762001e+00 1.18993974e+00 -4.35599715e-01 -1.42666543e+00
7.70241022e-02 -3.55765432e-01 -7.59984434e-01 1.18248808e+00
-1.81071460e+00 -1.38051569e+00 -3.40927728e-02 9.29551125e-01
3.43898296e-01 -4.79962617e-01 9.18830812e-01 6.99930489e-01
-7.79317677e-01 1.55736792e+00 1.02989830e-01 1.61712363e-01
6.94596648e-01 -9.92145121e-01 2.57545352e-01 8.21970582e-01
1.50318623e-01 6.32057607e-01 2.83175021e-01 -2.00205192e-01
-1.25677490e+00 -1.00345302e+00 1.16252947e+00 1.51671574e-01
1.75611079e-01 -5.32503366e-01 -8.67094398e-01 6.37621105e-01
5.03087461e-01 1.91650569e-01 9.97585714e-01 3.52401704e-01
-7.75833786e-01 -7.13980079e-01 -1.07144320e+00 3.41355473e-01
8.31625044e-01 -9.47791278e-01 -4.25186813e-01 -7.98930079e-02
9.26407576e-01 -2.44495496e-01 -8.71404350e-01 3.88405949e-01
6.52454853e-01 -1.03099656e+00 8.59580040e-01 -2.63762355e-01
-2.43639812e-01 -4.16831821e-01 -4.28438097e-01 -1.38991940e+00
-4.46981370e-01 -8.81478012e-01 7.01905116e-02 1.73675883e+00
6.05868995e-01 -8.81472170e-01 1.97719917e-01 2.00863525e-01
-5.33204734e-01 -7.44394720e-01 -1.45140946e+00 -7.50423133e-01
-1.93389893e-01 -4.85264003e-01 1.00513327e+00 9.57238019e-01
2.89981604e-01 4.97878999e-01 -6.15655899e-01 1.29147545e-01
2.87160099e-01 -1.52060404e-01 2.74329394e-01 -9.90101278e-01
-2.39783615e-01 -5.71300209e-01 -6.29288733e-01 -1.16346896e+00
2.03300327e-01 -9.81428206e-01 -2.11854517e-01 -9.56496954e-01
-2.19207490e-03 -4.98164177e-01 -8.31768870e-01 6.44569159e-01
-1.26765212e-02 -8.65694359e-02 2.90335923e-01 -2.71159448e-02
-2.67715186e-01 9.95334148e-01 9.93717730e-01 -3.32694530e-01
-3.93832594e-01 1.97944902e-02 -6.40236735e-01 3.47409159e-01
8.49156797e-01 -2.23431647e-01 -4.47274953e-01 -5.32351851e-01
-3.73640835e-01 1.04005083e-01 1.77937113e-02 -9.47330892e-01
6.45502031e-01 3.76503199e-01 3.99071395e-01 -6.65416300e-01
7.61830509e-01 -8.13004494e-01 5.60058132e-02 2.05107212e-01
-3.38737845e-01 2.05101162e-01 4.59402770e-01 3.30635011e-01
-7.30404317e-01 3.95978332e-01 8.99700999e-01 5.13972402e-01
-4.15145159e-01 4.04607266e-01 -3.10128242e-01 -3.16103011e-01
7.33556449e-01 -2.33511031e-01 -3.98540437e-01 -4.69657898e-01
-5.44260085e-01 2.83181518e-01 -3.00266743e-01 6.31165922e-01
8.88000190e-01 -1.73090959e+00 -6.33715212e-01 9.47606325e-01
4.03902829e-02 -4.41456079e-01 7.86931932e-01 9.53792810e-01
1.45809650e-01 6.98763311e-01 -3.27279925e-01 -6.16204143e-01
-1.41656435e+00 1.77814811e-01 6.00171804e-01 -3.93126123e-02
-5.65648139e-01 1.32829702e+00 2.37832785e-01 -6.23058558e-01
7.15775371e-01 -2.86970466e-01 -3.31634730e-01 -2.05250848e-02
6.12945437e-01 2.23433301e-01 1.21291101e-01 -9.65050459e-01
-7.18774557e-01 5.89216173e-01 -4.61765677e-01 -1.45751894e-01
1.43902743e+00 -2.08223373e-01 -3.69639769e-02 2.35330224e-01
1.53882289e+00 1.32366106e-01 -9.18607295e-01 -6.44075394e-01
-3.89709443e-01 -2.91854739e-01 3.85656118e-01 -8.64900112e-01
-1.63179100e+00 1.23344791e+00 9.26297545e-01 -1.91074666e-02
1.30820382e+00 -2.07864299e-01 8.43529582e-01 -1.17407814e-01
-1.20024368e-01 -9.33607340e-01 -3.07515979e-01 4.18507308e-01
1.03520000e+00 -7.77956069e-01 -5.30542433e-01 -2.52600938e-01
-6.54015660e-01 9.74291086e-01 5.79716504e-01 3.50853890e-01
1.18295097e+00 5.61672032e-01 2.33523026e-01 1.65086359e-01
-8.79796088e-01 2.40874991e-01 4.92826372e-01 5.80491006e-01
4.75856811e-01 1.09291859e-01 2.04871502e-02 1.13297296e+00
-3.63710344e-01 -4.48874712e-01 -1.37229934e-01 7.71205962e-01
-9.52918008e-02 -1.17546988e+00 -2.36863509e-01 3.60883325e-01
-2.99101084e-01 -2.52025932e-01 -1.51681438e-01 5.01878500e-01
1.32789135e-01 1.05557263e+00 1.02746479e-01 -1.16743135e+00
2.41584197e-01 2.98210859e-01 2.33236924e-02 6.42507430e-03
-9.51785624e-01 4.64045078e-01 -3.71394530e-02 -4.61942375e-01
-1.61473781e-01 -6.66767240e-01 -1.01918602e+00 -4.31509048e-01
-6.36407495e-01 2.90955871e-01 7.33397305e-01 7.16693223e-01
7.02201486e-01 8.81806016e-01 9.06444788e-01 -4.38793778e-01
-4.50351954e-01 -1.22476029e+00 -8.15368176e-01 -9.68142524e-02
7.81867623e-01 -5.45005798e-01 -4.10649002e-01 -2.99585491e-01] | [14.40996265411377, 6.0098395347595215] |
ab475380-80c1-4052-845e-8190b3324351 | recurrent-transformer-for-dynamic-graph | 2304.10079 | null | https://arxiv.org/abs/2304.10079v1 | https://arxiv.org/pdf/2304.10079v1.pdf | Recurrent Transformer for Dynamic Graph Representation Learning with Edge Temporal States | Dynamic graph representation learning is growing as a trending yet challenging research task owing to the widespread demand for graph data analysis in real world applications. Despite the encouraging performance of many recent works that build upon recurrent neural networks (RNNs) and graph neural networks (GNNs), they fail to explicitly model the impact of edge temporal states on node features over time slices. Additionally, they are challenging to extract global structural features because of the inherent over-smoothing disadvantage of GNNs, which further restricts the performance. In this paper, we propose a recurrent difference graph transformer (RDGT) framework, which firstly assigns the edges in each snapshot with various types and weights to illustrate their specific temporal states explicitly, then a structure-reinforced graph transformer is employed to capture the temporal node representations by a recurrent learning paradigm. Experimental results on four real-world datasets demonstrate the superiority of RDGT for discrete dynamic graph representation learning, as it consistently outperforms competing methods in dynamic link prediction tasks. | ['Yixin Chen', 'Bofeng Zhang', 'Chenyang Zhou', 'Liangrui Wu', 'Shiyi Lin', 'Guobing Zou', 'Shengxiang Hu'] | 2023-04-20 | null | null | null | null | ['dynamic-link-prediction'] | ['graphs'] | [ 1.88672066e-01 -8.00918415e-02 -5.51813543e-01 4.02987078e-02
1.91458070e-03 -1.91628516e-01 7.89838254e-01 2.12643668e-01
1.54148862e-01 3.88212740e-01 3.52660626e-01 -5.88083327e-01
-3.94131869e-01 -9.28355932e-01 -4.18180346e-01 -6.35491908e-01
-6.27744019e-01 2.56959140e-01 2.32866451e-01 -2.63739824e-01
-9.45713893e-02 4.32464957e-01 -1.13840866e+00 -9.37116146e-02
7.58899927e-01 7.90036261e-01 1.71710607e-02 4.82946336e-01
-2.39476874e-01 1.24033487e+00 -3.29379857e-01 -1.94697291e-01
2.13940740e-02 -4.65219676e-01 -5.28183520e-01 -1.06935404e-01
-1.63773187e-02 1.98991373e-02 -1.11628342e+00 9.81971323e-01
4.28835303e-01 2.10772857e-01 4.12227482e-01 -1.46047378e+00
-9.23804045e-01 8.83369148e-01 -7.34668434e-01 5.97123802e-01
2.68500894e-01 -1.44554272e-01 1.41717815e+00 -4.72276896e-01
7.83180654e-01 1.15320492e+00 9.03404832e-01 2.69095063e-01
-1.41455615e+00 -6.21781468e-01 7.70146787e-01 4.16439623e-01
-1.11926460e+00 -1.47857055e-01 1.37740242e+00 -2.57919192e-01
1.14303589e+00 -3.26779205e-03 8.58221114e-01 1.29759705e+00
4.74358380e-01 6.81874216e-01 6.92798316e-01 -2.77175587e-02
-1.87263787e-01 -5.73660672e-01 3.21563512e-01 7.97249496e-01
5.05181439e-02 2.04873741e-01 -3.80454808e-01 -1.02948762e-01
8.07526588e-01 4.90274042e-01 -1.23845898e-01 -5.96341550e-01
-9.36743140e-01 7.77845025e-01 8.79774690e-01 3.95053059e-01
-5.51505625e-01 3.92181396e-01 8.53147328e-01 6.45575404e-01
7.37390757e-01 -1.20468706e-01 -2.40102664e-01 -1.98261607e-02
-5.11985362e-01 -2.11999118e-02 6.17909789e-01 7.63666809e-01
5.58203518e-01 4.56059605e-01 -2.39181161e-01 8.81455302e-01
1.64338142e-01 2.30791941e-02 5.84773302e-01 -2.76946187e-01
4.20586586e-01 1.11177051e+00 -5.15449703e-01 -1.57729614e+00
-5.12490153e-01 -7.42607951e-01 -1.34647715e+00 -3.36224556e-01
6.40749931e-02 2.13816673e-01 -9.82991934e-01 1.85364568e+00
1.15219511e-01 7.88589060e-01 -2.09222928e-01 4.84590143e-01
1.05490208e+00 7.30924606e-01 1.17239378e-01 -3.40650707e-01
8.76836300e-01 -9.45003390e-01 -7.84552872e-01 -1.43640995e-01
7.34387815e-01 -8.21665153e-02 8.20034921e-01 -1.20819528e-02
-7.08865047e-01 -2.58553207e-01 -8.44840348e-01 1.31821990e-01
-4.87561882e-01 -2.82744259e-01 9.88257945e-01 2.94190079e-01
-1.21673810e+00 6.81958318e-01 -1.02899265e+00 -4.26247120e-01
1.59298778e-01 1.98727220e-01 -1.70043871e-01 -5.87569028e-02
-1.40923858e+00 5.45711160e-01 2.90253043e-01 5.75154662e-01
-7.24775136e-01 -6.36066079e-01 -1.17796850e+00 3.29009384e-01
6.24510288e-01 -4.57465619e-01 7.87979126e-01 -5.88081896e-01
-1.41025698e+00 4.92497057e-01 8.73910934e-02 -7.28055358e-01
3.67169768e-01 1.95868433e-01 -9.48201358e-01 -5.25528528e-02
-1.76033884e-01 5.00095263e-02 7.99864709e-01 -9.54492033e-01
-1.76233396e-01 -2.20371380e-01 1.18068248e-01 6.17947169e-02
-4.60913837e-01 -2.25743860e-01 -5.48284233e-01 -1.00192547e+00
3.11983943e-01 -9.55339193e-01 -4.68607128e-01 -4.44376618e-01
-6.69827998e-01 -4.66400027e-01 1.12555885e+00 -5.76096416e-01
1.90631819e+00 -1.98075831e+00 4.36326802e-01 3.01584989e-01
6.39113605e-01 2.63908148e-01 -3.44126999e-01 8.36023331e-01
-4.10478979e-01 7.93669447e-02 -5.35466596e-02 -9.82620940e-02
-2.71701273e-02 3.38595748e-01 -4.61539567e-01 3.49590570e-01
6.47384897e-02 1.22823334e+00 -1.04388845e+00 -2.52415806e-01
1.63545117e-01 6.94566548e-01 -2.26102918e-01 1.51903704e-01
-1.36870787e-01 6.72112182e-02 -5.69928467e-01 4.79743361e-01
2.04274610e-01 -7.56075203e-01 7.03592300e-01 -1.57680839e-01
2.36167774e-01 3.89567286e-01 -9.15073156e-01 1.31029737e+00
-3.13944250e-01 5.63749969e-01 -2.52914429e-01 -1.41505754e+00
1.05610311e+00 3.16629112e-01 6.83140993e-01 -9.86852109e-01
-9.41537023e-02 -1.48433805e-01 2.11339429e-01 -2.48355240e-01
3.86381388e-01 1.48504376e-01 -9.25576165e-02 6.32697523e-01
-1.27301350e-01 4.27315146e-01 3.01195472e-01 6.07632279e-01
1.51872981e+00 -8.33957344e-02 1.93753466e-01 -3.40658054e-02
4.35467333e-01 -6.12580180e-01 8.17171872e-01 4.27340776e-01
-1.96675420e-01 2.63221383e-01 1.00466883e+00 -7.00145841e-01
-7.22071409e-01 -9.04617846e-01 4.39436615e-01 1.22518277e+00
6.53292760e-02 -7.31971681e-01 -5.55483662e-02 -7.81374395e-01
7.76298940e-02 2.05903560e-01 -8.65227938e-01 -5.54429352e-01
-7.69672275e-01 -7.61521697e-01 3.46820146e-01 7.12560475e-01
1.32263422e-01 -1.25046563e+00 -9.84964222e-02 4.57800061e-01
-2.81291502e-03 -1.03671765e+00 -5.69393516e-01 9.13569331e-02
-9.99581218e-01 -1.13881624e+00 -5.50336242e-01 -8.65211070e-01
5.45641065e-01 5.95405579e-01 1.23549402e+00 4.62969899e-01
-2.57613182e-01 3.45518857e-01 -3.69394302e-01 2.40259305e-01
-2.58916795e-01 3.82171243e-01 -6.69810772e-02 3.93505841e-02
-6.50792643e-02 -1.14967704e+00 -4.38232839e-01 1.11493461e-01
-7.68510580e-01 1.40099525e-01 5.04000247e-01 1.03484809e+00
4.09844905e-01 1.41926631e-01 8.07622492e-01 -1.32772481e+00
9.69059706e-01 -7.52904236e-01 -4.61576909e-01 4.18388486e-01
-1.02609539e+00 8.07214603e-02 7.13123798e-01 -5.11497498e-01
-6.54887140e-01 -4.57543373e-01 3.77637744e-01 -6.40055001e-01
6.04939759e-01 1.25517535e+00 3.09821695e-01 9.44566205e-02
1.37799576e-01 4.35901433e-01 3.01732458e-02 -2.17711613e-01
1.83070824e-01 -1.20278999e-01 3.46136361e-01 -3.85344118e-01
9.40253615e-01 1.16741046e-01 9.22588259e-02 -6.34053171e-01
-5.81335783e-01 -2.16716900e-01 -4.05594885e-01 -3.50604177e-01
3.31842989e-01 -6.93013906e-01 -6.65307462e-01 4.81317073e-01
-7.47615397e-01 -4.54238355e-01 -3.52668879e-03 1.02662601e-01
-2.70057321e-01 5.70101619e-01 -9.19791520e-01 -8.02257359e-01
-5.00225186e-01 -9.22413647e-01 5.31340361e-01 9.54021811e-02
1.82486236e-01 -1.50535798e+00 1.63587615e-01 -2.36331329e-01
5.61876953e-01 6.83100760e-01 1.28161514e+00 -4.37709093e-01
-5.36390901e-01 -3.69602054e-01 -4.62903053e-01 -1.36606157e-01
3.04977626e-01 2.24731684e-01 -2.89669871e-01 -5.31187654e-01
-5.88511705e-01 -1.15027323e-01 9.80490506e-01 5.18452406e-01
1.23515844e+00 -3.25102866e-01 -6.01322651e-01 5.46701074e-01
1.36831081e+00 2.44737286e-02 5.05715251e-01 -1.04587831e-01
1.04546380e+00 4.20539141e-01 1.23726271e-01 2.71715611e-01
6.71815157e-01 5.75150192e-01 5.55900872e-01 -1.01276614e-01
-2.91957706e-01 -5.36794901e-01 4.39164549e-01 1.32573855e+00
-2.17338726e-01 -4.31211770e-01 -1.00923026e+00 5.87825656e-01
-2.18463039e+00 -1.03738534e+00 -1.09655119e-01 1.85133922e+00
4.03121829e-01 3.75352055e-01 2.97983438e-01 1.40518695e-01
8.92813802e-01 9.05813336e-01 -7.74637163e-01 -1.49954125e-01
-6.15963042e-02 -3.84346731e-02 2.82194078e-01 1.93899065e-01
-8.67617548e-01 9.19893563e-01 6.40754032e+00 5.41933477e-01
-1.16289270e+00 -1.53809234e-01 4.91729110e-01 2.99911499e-01
-4.35890943e-01 1.98417511e-02 -2.24396303e-01 4.37752903e-01
1.06541562e+00 -6.26643956e-01 6.10385299e-01 5.07615566e-01
6.51409253e-02 5.53991199e-01 -9.08727229e-01 8.54873598e-01
-1.59286052e-01 -1.43800175e+00 2.35903621e-01 6.50335848e-02
6.06342316e-01 2.91072845e-01 2.07021803e-01 5.57133138e-01
7.20558226e-01 -1.18954909e+00 1.74076766e-01 5.94913006e-01
6.33249700e-01 -7.62354612e-01 4.22834039e-01 1.40240565e-01
-1.93801391e+00 -2.73712993e-01 -1.74373329e-01 -1.86523855e-01
1.58532724e-01 5.82642674e-01 -6.93083286e-01 9.04641569e-01
5.97371876e-01 1.49374306e+00 -7.09150791e-01 8.04957092e-01
-1.16703443e-01 7.96985507e-01 -1.46003336e-01 -5.90241700e-02
3.14677387e-01 -4.57477212e-01 5.32563329e-01 1.04038572e+00
2.03894451e-01 -1.81933746e-01 4.48325574e-01 6.50631070e-01
-2.95419663e-01 -1.96852218e-02 -9.13311064e-01 -6.24543607e-01
5.47879994e-01 1.30004454e+00 -9.38664496e-01 -9.96239949e-04
-4.83055502e-01 6.27442896e-01 7.56866455e-01 6.66348338e-01
-7.80632436e-01 -3.03801268e-01 4.53891695e-01 -1.23467296e-02
3.13326687e-01 -3.83183599e-01 2.26512030e-01 -1.26347244e+00
1.04412995e-01 -7.12633491e-01 8.92290294e-01 -4.32349294e-01
-1.52240181e+00 8.18990827e-01 -2.68707007e-01 -1.27876735e+00
-4.99038428e-01 -2.59402692e-01 -9.59315836e-01 5.17869771e-01
-1.59682417e+00 -1.51325595e+00 -2.35753730e-01 5.51613271e-01
3.06494713e-01 -1.19710736e-01 6.39824867e-01 3.63395691e-01
-9.97249663e-01 6.60847068e-01 6.17539510e-02 3.68573308e-01
2.37467483e-01 -1.08246505e+00 7.22606003e-01 8.55335712e-01
2.96207964e-01 6.38573587e-01 6.73102796e-01 -7.58859158e-01
-1.67316556e+00 -1.17877460e+00 7.13252306e-01 -1.09704375e-01
1.22003925e+00 -4.97863531e-01 -1.24695575e+00 1.01726329e+00
8.04100931e-02 5.25175869e-01 3.82520318e-01 5.21680653e-01
-6.80472374e-01 -1.89247146e-01 -5.00414729e-01 6.59314275e-01
1.48995519e+00 -8.36985826e-01 -6.53572977e-02 2.83232182e-01
9.05896485e-01 -2.26270586e-01 -1.02193975e+00 5.69948256e-01
4.52918261e-01 -7.93335438e-01 9.83556569e-01 -8.13258231e-01
2.90643632e-01 -1.09834023e-01 2.10196242e-01 -1.40069985e+00
-5.90266943e-01 -7.49597192e-01 -7.77047813e-01 1.39572608e+00
2.78253317e-01 -9.08138514e-01 8.07583213e-01 8.40259418e-02
1.54529205e-02 -1.00151753e+00 -9.22755301e-01 -7.39413917e-01
-2.34274834e-01 -2.66476542e-01 5.98790824e-01 1.25735688e+00
1.03611480e-02 5.39081275e-01 -6.57261610e-01 -1.11022340e-02
3.62157434e-01 3.67688149e-01 5.41680455e-01 -1.47124648e+00
-1.91294298e-01 -7.88843393e-01 -6.18255079e-01 -8.98178816e-01
4.83375877e-01 -1.19549799e+00 -3.71803373e-01 -1.69501281e+00
1.56536117e-01 -4.67396140e-01 -7.72319853e-01 4.41846281e-01
-2.60155886e-01 3.34008299e-02 -7.59488642e-02 1.79777667e-01
-7.46934891e-01 8.81105542e-01 1.04248559e+00 -3.59777123e-01
-2.13116840e-01 4.93819416e-02 -5.23587346e-01 3.92924458e-01
3.84751350e-01 -2.29362056e-01 -7.98257530e-01 -2.93918431e-01
5.71214497e-01 3.88294250e-01 3.27243268e-01 -7.32011139e-01
3.24576467e-01 5.96816232e-03 5.59320003e-02 -7.22551227e-01
4.96332441e-03 -6.50017500e-01 4.13245678e-01 6.42325938e-01
-3.24840277e-01 5.01845062e-01 -9.34469774e-02 1.25006270e+00
-2.51954079e-01 4.68216151e-01 3.53671074e-01 1.74865112e-01
-8.18062186e-01 9.87851560e-01 -7.66174123e-02 -4.70975637e-02
8.14130127e-01 -2.22800314e-01 -3.02358449e-01 -6.58255577e-01
-7.41534412e-01 5.24131358e-01 1.71491101e-01 6.96845114e-01
5.83735287e-01 -1.59937656e+00 -5.23426950e-01 4.43269573e-02
2.06298575e-01 -2.64146000e-01 4.73434836e-01 9.59587753e-01
-1.00786418e-01 1.57655329e-01 -1.10730780e-02 -4.09395188e-01
-1.00058317e+00 8.07016253e-01 3.68083864e-01 -9.81796741e-01
-1.30556858e+00 5.28653204e-01 3.61887552e-02 -4.80518311e-01
2.86955029e-01 -2.46677637e-01 -5.11656404e-01 1.04878411e-01
9.20499340e-02 3.61612707e-01 -8.10533296e-03 -4.99660552e-01
-1.92140028e-01 3.28053504e-01 -3.31858486e-01 4.91281897e-01
1.62541735e+00 -1.08695380e-01 -2.44999811e-01 6.63706362e-01
1.04041672e+00 -4.22138005e-01 -1.18136084e+00 -6.91206872e-01
4.11856622e-01 -9.81352553e-02 -2.13665552e-02 -3.77604246e-01
-1.53623796e+00 6.64266646e-01 1.57001674e-01 8.74195099e-01
1.06294143e+00 -6.71756417e-02 9.96263683e-01 2.80991375e-01
2.75105596e-01 -8.03889513e-01 2.37769961e-01 7.02483833e-01
7.56195605e-01 -9.75450397e-01 4.98494953e-02 -5.12506187e-01
-3.12501281e-01 1.06509054e+00 4.70152408e-01 -4.93723780e-01
9.35888827e-01 6.95771351e-02 -3.04753423e-01 -4.53377187e-01
-1.08850706e+00 -1.18849322e-01 2.93906748e-01 5.38468957e-01
5.02836585e-01 5.43605871e-02 -2.12986916e-01 3.64076346e-01
7.19973668e-02 -3.51778775e-01 5.00198245e-01 7.09420562e-01
9.51734856e-02 -1.11453342e+00 4.80145395e-01 8.01010847e-01
-1.80898428e-01 -6.91645220e-03 -3.61937970e-01 8.19736958e-01
-7.03935981e-01 4.89343107e-01 2.68380763e-03 -6.70657277e-01
1.77298293e-01 -9.03097093e-02 1.67644814e-01 -5.79138279e-01
-5.43914497e-01 -1.16026483e-01 2.55082250e-02 -6.19488597e-01
-3.69077384e-01 -5.71061432e-01 -1.19081509e+00 -4.52045232e-01
-2.90433556e-01 1.42806426e-01 1.29117385e-01 7.59148240e-01
6.40750825e-01 1.10556185e+00 8.25006962e-01 -6.99551940e-01
-2.97501653e-01 -9.30792689e-01 -7.31012881e-01 4.19803768e-01
3.76504511e-01 -9.03414845e-01 -3.30482155e-01 -3.30240011e-01] | [7.199073314666748, 5.973312854766846] |
5fa11ae2-3632-48be-8e26-66d1660b8d84 | fun2vec-a-contrastive-learning-framework-of | 2209.02442 | null | https://arxiv.org/abs/2209.02442v1 | https://arxiv.org/pdf/2209.02442v1.pdf | Fun2Vec:a Contrastive Learning Framework of Function-level Representation for Binary | Function-level binary code similarity detection is essential in the field of cyberspace security. It helps us find bugs and detect patent infringements in released software and plays a key role in the prevention of supply chain attacks. A practical embedding learning framework relies on the robustness of vector representation system of assembly code and the accuracy of the annotation of function pairs. Supervised learning based methods are traditionally emploied. But annotating different function pairs with accurate labels is very difficult. These supervised learning methods are easily overtrained and suffer from vector robustness issues. To mitigate these problems, we propose Fun2Vec: a contrastive learning framework of function-level representation for binary. We take an unsupervised learning approach and formulate the binary code similarity detection as instance discrimination. Fun2Vec works directly on disassembled binary functions, and could be implemented with any encoder. It does not require manual labeled similar or dissimilar information. We use the compiler optimization options and code obfuscation techniques to generate augmented data. Our experimental results demonstrate that our method surpasses the state-of-the-art in accuracy and have great advantage in few-shot settings. | ['Pan ZhiSong', 'Sun Meng', 'Guo JinHong', 'Guo ShiZe', 'Sun RuiJin'] | 2022-09-06 | null | null | null | null | ['compiler-optimization'] | ['computer-code'] | [ 9.91792083e-02 -3.51911724e-01 -4.65602905e-01 -2.14367613e-01
-5.96115768e-01 -1.12602806e+00 2.19217449e-01 5.38112581e-01
-1.30469605e-01 3.26134413e-01 -1.06288463e-01 -9.32031274e-01
3.97504903e-02 -7.55238652e-01 -5.89504600e-01 -4.68472749e-01
-1.43569976e-01 -8.09816718e-02 1.40077427e-01 -3.64167333e-01
4.89157110e-01 2.26738706e-01 -1.15523458e+00 3.55867863e-01
7.59428501e-01 7.25068212e-01 1.42491475e-01 8.42164338e-01
-4.10826236e-01 8.94693911e-01 -9.20909047e-01 -8.08043003e-01
3.18797141e-01 -2.03565493e-01 -5.00703990e-01 -3.41356695e-01
3.28135118e-02 -4.45420407e-02 -3.70965987e-01 1.52907908e+00
3.26356828e-01 -4.84650761e-01 5.28256416e-01 -1.45972705e+00
-8.00275564e-01 8.18066418e-01 -5.03224671e-01 2.03389868e-01
3.31562549e-01 1.27751693e-01 1.17748451e+00 -6.63532794e-01
3.43342364e-01 8.76267672e-01 8.02658081e-01 1.87423766e-01
-1.21305346e+00 -7.30574131e-01 -1.02688894e-01 1.55606866e-01
-1.27906227e+00 -5.47849350e-02 7.75833905e-01 -8.53223026e-01
1.01478195e+00 3.27804506e-01 2.65512079e-01 1.02034140e+00
6.14720345e-01 5.01959622e-01 6.12808943e-01 -3.85406554e-01
3.17177027e-01 3.43869358e-01 5.71388423e-01 1.08353996e+00
4.01919901e-01 6.82043806e-02 6.45278096e-02 -5.71121454e-01
5.08790910e-01 4.50667679e-01 -5.33784509e-01 -4.58391428e-01
-1.13382638e+00 1.06562853e+00 4.40760195e-01 5.33218741e-01
2.62945622e-01 3.85578245e-01 8.51355374e-01 6.87553644e-01
6.50378987e-02 1.03362906e+00 -5.82068264e-01 -2.83723801e-01
-5.86878657e-01 -1.78783387e-01 6.68323636e-01 9.80216742e-01
8.00513804e-01 2.34873697e-01 6.59577549e-02 6.26356781e-01
9.96950716e-02 -2.00135093e-02 7.84680188e-01 -2.67600983e-01
3.90306383e-01 8.59539151e-01 -3.24174345e-01 -1.25828564e+00
-5.63075347e-03 -3.97841275e-01 -5.32990217e-01 4.36526328e-01
6.45175278e-02 -7.87156671e-02 -5.33312023e-01 1.14459741e+00
-6.72705621e-02 1.42554894e-01 -4.28743474e-02 4.47104633e-01
4.84676838e-01 6.80581093e-01 -4.87992436e-01 3.39102820e-02
1.33538485e+00 -1.00761330e+00 -7.90112197e-01 2.63686795e-02
9.64409590e-01 -7.67784953e-01 1.21998787e+00 4.84711975e-01
-8.34346935e-02 -5.62561154e-01 -1.64289844e+00 2.60084540e-01
-6.82902336e-01 2.12301642e-01 9.10776973e-01 1.21831107e+00
-5.12515128e-01 8.90161157e-01 -8.03175509e-01 1.89972445e-01
4.12956744e-01 4.00172234e-01 -5.88780165e-01 1.94104940e-01
-8.09754193e-01 6.42425656e-01 4.18028593e-01 -2.79783577e-01
-8.24186265e-01 -6.03943646e-01 -1.24221480e+00 2.82190770e-01
4.20075536e-01 6.88083395e-02 1.06720543e+00 -8.40908825e-01
-1.16740191e+00 3.38233888e-01 4.23881322e-01 -3.58609408e-01
-6.96272328e-02 4.75039519e-02 -6.00432575e-01 -3.03757906e-01
-1.01376705e-01 -2.22787082e-01 1.08071184e+00 -8.71747673e-01
-1.92822725e-01 -1.95275664e-01 2.78317153e-01 -6.34140432e-01
-6.96195662e-01 1.45534307e-01 -2.98133306e-02 -8.52783382e-01
-2.86240786e-01 -6.96521938e-01 -2.49745205e-01 -2.39713155e-02
-4.42443132e-01 3.08966767e-02 9.27502513e-01 -7.54446149e-01
1.48642576e+00 -2.46941257e+00 -8.58865157e-02 6.23174496e-02
3.69106621e-01 4.98336464e-01 -1.13786288e-01 5.11773229e-01
-3.73539865e-01 3.52437139e-01 -4.15847540e-01 2.06004694e-01
1.76840976e-01 -1.30362228e-01 -4.55225199e-01 7.92281985e-01
4.35936600e-01 7.63854563e-01 -9.91875052e-01 -1.95914835e-01
-2.83249151e-02 2.87551224e-01 -6.93195224e-01 3.39448750e-01
-2.17509001e-01 -1.25086248e-01 -3.46123308e-01 8.03026259e-01
5.04419923e-01 -2.29923978e-01 3.32697123e-01 -3.39610457e-01
6.90719858e-03 3.01371664e-01 -1.04858387e+00 1.60193717e+00
-5.16837001e-01 8.17122102e-01 -2.21391827e-01 -1.08718514e+00
1.03009522e+00 1.08927466e-01 -6.00821078e-02 -4.55297455e-02
3.44749689e-01 1.36393726e-01 3.40803951e-01 -6.47105515e-01
3.25937152e-01 2.56116480e-01 -4.81379569e-01 5.33697248e-01
2.54289180e-01 -1.39308602e-01 -1.73612893e-01 9.03723538e-02
1.40629220e+00 9.82825384e-02 6.75300658e-01 -1.05833285e-01
5.88636935e-01 -1.09669536e-01 6.88822865e-01 4.09364641e-01
-8.70733708e-02 1.55407354e-01 8.26505184e-01 -5.20592749e-01
-9.74285245e-01 -7.14521646e-01 -8.79486732e-04 9.27204907e-01
8.82495046e-02 -9.73844707e-01 -6.33013248e-01 -1.35832310e+00
1.38607070e-01 5.68546176e-01 -4.81759816e-01 -4.51679498e-01
-4.39107776e-01 -5.61923862e-01 8.67810011e-01 6.13992691e-01
-2.16178276e-04 -5.64662993e-01 -3.15819860e-01 1.42463982e-01
4.53709006e-01 -6.01993263e-01 -6.17915809e-01 5.01088858e-01
-4.78342891e-01 -1.46536386e+00 -2.34296411e-01 -1.02211046e+00
7.32018352e-01 2.40519810e-02 9.15305495e-01 1.61645561e-01
-6.30773723e-01 -1.90509632e-01 -5.23192406e-01 -1.79149285e-01
-7.69329548e-01 -1.13983549e-01 5.54430224e-02 -9.18576643e-02
2.79437393e-01 -3.75385046e-01 -1.22626774e-01 3.08585793e-01
-9.25408959e-01 -6.67541444e-01 5.36199093e-01 1.19638777e+00
1.90710977e-01 5.69024324e-01 3.62809479e-01 -9.62830782e-01
7.94052660e-01 -3.66873294e-01 -9.35450554e-01 4.59239572e-01
-5.80932260e-01 5.22043049e-01 1.02279186e+00 -5.64494073e-01
-4.17521954e-01 1.42690614e-01 -1.91781685e-01 -3.54935467e-01
1.98366016e-01 4.31728333e-01 -3.58190715e-01 -5.11366546e-01
7.31789708e-01 1.35256529e-01 4.03701663e-02 -5.85837185e-01
3.65198314e-01 1.06370974e+00 3.56926292e-01 -3.44555676e-01
8.84564817e-01 -1.54660702e-01 -3.40157986e-01 -6.29596889e-01
-4.29730266e-02 -3.35718244e-01 -2.49315977e-01 3.04503322e-01
6.19006455e-01 -5.04125059e-01 -7.78686941e-01 -2.24953936e-03
-1.29449821e+00 2.10837126e-01 8.76656100e-02 2.73753494e-01
-1.23568378e-01 9.22436178e-01 -6.18989766e-01 -5.75559556e-01
-2.09111005e-01 -1.66981947e+00 7.74670959e-01 -2.18952954e-01
-6.63221106e-02 -7.11128414e-01 1.25491798e-01 -1.09762095e-01
2.28162289e-01 2.24129602e-01 1.49483442e+00 -1.10492408e+00
-4.58815008e-01 -6.08485281e-01 7.44702071e-02 6.26672149e-01
3.64288956e-01 1.45055041e-01 -8.01782429e-01 -4.54216212e-01
2.29886964e-01 -1.90028846e-01 6.47029698e-01 -3.36530656e-01
1.47063649e+00 -4.44767654e-01 -3.39198977e-01 6.08924925e-01
1.53123164e+00 3.31268489e-01 5.37048578e-01 8.51232186e-02
7.22130179e-01 4.42324430e-01 6.89502418e-01 3.29568744e-01
-3.24017942e-01 7.08619773e-01 6.96370065e-01 2.20662683e-01
1.43796340e-01 -2.75298476e-01 6.00645661e-01 1.00400531e+00
4.23283100e-01 -4.82862815e-02 -8.35205853e-01 3.33891481e-01
-1.57123339e+00 -8.83840024e-01 5.06305285e-02 2.25587606e+00
9.16736305e-01 2.01155841e-01 -1.72323644e-01 6.34132445e-01
8.82830799e-01 5.80151677e-02 -2.54589468e-01 -5.84534109e-01
4.28859532e-01 4.53677148e-01 7.11148739e-01 3.02629888e-01
-1.20371342e+00 7.33481348e-01 5.61241674e+00 1.04514968e+00
-1.03295898e+00 3.36853981e-01 3.89933854e-01 4.25473392e-01
-4.24552590e-01 4.30871584e-02 -5.94218135e-01 6.99416041e-01
8.48884821e-01 -3.62019576e-02 3.62282187e-01 1.37661743e+00
-6.56496525e-01 4.72470552e-01 -1.40658188e+00 1.24795926e+00
2.10773125e-01 -1.50118113e+00 -4.52475905e-01 -5.61319366e-02
3.09840530e-01 -3.82531196e-01 -1.22096138e-02 6.67887807e-01
4.59660813e-02 -1.18413329e+00 4.31423277e-01 -1.44996457e-02
8.14869702e-01 -9.12885308e-01 1.06755579e+00 1.38350725e-01
-1.33426523e+00 -3.23159993e-01 -6.22062624e-01 3.17396335e-02
-2.23946601e-01 3.98472607e-01 -1.07060194e+00 4.07936513e-01
3.41759413e-01 7.55154550e-01 -7.20646620e-01 1.00969982e+00
-2.91284382e-01 4.04458940e-01 3.99621546e-01 -4.83403444e-01
-1.83404787e-04 1.00235604e-01 2.68418491e-01 1.23384261e+00
2.50521481e-01 -5.96899629e-01 2.02587485e-01 8.90805840e-01
-1.12836041e-01 5.48624285e-02 -9.42761540e-01 -7.33658612e-01
5.71730554e-01 1.14037955e+00 -6.96360767e-01 -1.07930891e-01
-4.74751890e-01 1.15400195e+00 1.88795999e-01 -7.90793374e-02
-1.13627636e+00 -1.27310014e+00 9.85991478e-01 -2.53895879e-01
5.98456085e-01 -3.75065714e-01 -1.10337250e-01 -1.32420552e+00
-2.55800914e-02 -1.31779397e+00 1.00517191e-01 -2.35469848e-01
-1.23696208e+00 8.73308182e-01 -4.43693876e-01 -1.61857355e+00
-8.60424340e-02 -1.08699977e+00 -6.67849481e-01 5.20180285e-01
-1.04286873e+00 -5.69982529e-01 2.58851778e-02 2.31074423e-01
5.75765371e-01 -7.39326298e-01 1.08589697e+00 3.20780545e-01
-6.23347580e-01 1.16486168e+00 1.09802015e-01 5.30290067e-01
5.83553374e-01 -1.23939729e+00 7.59021103e-01 9.23855484e-01
2.96997249e-01 1.03332245e+00 4.86879379e-01 -6.62783802e-01
-1.90872669e+00 -1.15548432e+00 5.97144842e-01 -3.80940855e-01
1.01468503e+00 -7.70530343e-01 -7.16709495e-01 4.35368270e-01
1.63326532e-01 1.57207981e-01 1.14897645e+00 -2.00305030e-01
-8.39955807e-01 -9.37428996e-02 -1.15455937e+00 4.60639745e-01
6.06385589e-01 -9.11162198e-01 -5.65002680e-01 4.96400863e-01
1.24368989e+00 -2.25289725e-02 -8.69149864e-01 2.58566290e-01
3.59750926e-01 -7.58769810e-01 9.27725732e-01 -8.35484028e-01
4.02005166e-01 -6.89596951e-01 -3.11991692e-01 -1.15165913e+00
-4.18007582e-01 -6.63856745e-01 -1.16695173e-01 1.18450916e+00
5.66771805e-01 -6.49040580e-01 5.32520473e-01 5.64457104e-03
-2.01337457e-01 -5.93190610e-01 -5.16004086e-01 -1.15872240e+00
-3.12861532e-01 -4.12323385e-01 8.18463504e-01 1.38810706e+00
4.89372253e-01 2.97720641e-01 -3.04392338e-01 3.35142642e-01
2.72158951e-01 2.11608395e-01 7.24227369e-01 -9.22146320e-01
-1.06530118e+00 -3.54001254e-01 -1.17466223e+00 -6.46361947e-01
3.21012557e-01 -1.20739508e+00 -3.11794952e-02 -7.16085434e-01
-1.45065356e-02 -3.11415255e-01 -4.11139190e-01 6.90587580e-01
-1.52785271e-01 1.87992126e-01 -8.49586073e-03 -2.55758632e-02
-2.26306647e-01 3.56398553e-01 4.46857840e-01 -6.44433975e-01
1.08718045e-01 -1.79107860e-01 -6.98839009e-01 4.80782211e-01
7.61730134e-01 -6.45729065e-01 -4.41810131e-01 -1.64029866e-01
3.18832725e-01 -5.65843843e-02 8.81097242e-02 -8.52549434e-01
-4.72047292e-02 1.86999097e-01 1.38765750e-02 -1.29579276e-01
-8.76233801e-02 -9.82462525e-01 2.42787730e-02 8.75787139e-01
-2.39396960e-01 4.85594451e-01 9.11522582e-02 8.35637212e-01
-1.56025290e-01 -8.02770436e-01 4.64337498e-01 -4.10426334e-02
-7.64001548e-01 6.58473149e-02 -3.57341409e-01 -3.56451064e-01
1.09064043e+00 -6.81508482e-02 -5.44214904e-01 4.26342599e-02
-1.78856462e-01 -3.19151700e-01 7.50580609e-01 5.75781226e-01
7.55105376e-01 -1.39592898e+00 -2.57126182e-01 5.45251369e-01
5.98279893e-01 -7.64842808e-01 -4.40064967e-01 5.46971202e-01
-7.17225552e-01 2.22151518e-01 -1.42409503e-01 -3.70131373e-01
-1.35992062e+00 1.04500079e+00 4.16416302e-03 1.86329223e-02
-3.23544323e-01 9.34414685e-01 -3.27780955e-02 -5.93438387e-01
4.60098624e-01 -5.62528253e-01 -2.14810610e-01 -2.83074230e-01
8.01305592e-01 1.78200081e-01 3.12469244e-01 -1.95695132e-01
-5.26723027e-01 3.05081129e-01 -2.66988486e-01 4.59613621e-01
1.17029870e+00 7.62646735e-01 -4.25138116e-01 2.64776051e-01
1.70724320e+00 2.41106123e-01 -6.18767202e-01 2.14612231e-01
1.49643719e-01 -6.87337279e-01 6.43610358e-02 -5.56763887e-01
-8.87924850e-01 9.19593453e-01 7.14040577e-01 2.57566929e-01
7.39938080e-01 -2.17733100e-01 6.88227534e-01 6.08081341e-01
5.94896615e-01 -5.77442884e-01 2.15823516e-01 3.66541535e-01
4.78245944e-01 -1.23015904e+00 -3.75641771e-02 -5.85517824e-01
-4.15592909e-01 1.31245542e+00 4.25091147e-01 -2.42345661e-01
4.72127408e-01 6.83339119e-01 -1.85000136e-01 -1.46691635e-01
-4.29725349e-01 1.58436298e-01 2.51389027e-01 6.11168385e-01
6.91989005e-01 -1.27639556e-02 -4.17631209e-01 8.75337303e-01
-3.70886214e-02 -5.30266941e-01 5.99085152e-01 1.25874710e+00
-3.37717086e-01 -1.55138314e+00 -3.49523723e-01 4.31440264e-01
-4.53513116e-01 -3.64415377e-01 -4.66625631e-01 4.63490158e-01
2.39292905e-01 8.77377093e-01 -1.72438905e-01 -1.07211506e+00
3.50455493e-02 1.18907792e-02 3.33863795e-01 -8.40572417e-01
-8.17527950e-01 -1.92292869e-01 2.42358930e-02 -5.66530466e-01
2.03095779e-01 -1.25522017e-01 -9.14827943e-01 -4.38905433e-02
-7.78960705e-01 3.58675331e-01 7.99769342e-01 3.70474994e-01
4.76207554e-01 7.88044035e-01 9.56315756e-01 -6.46468401e-01
-9.89671409e-01 -7.09734797e-01 -4.13602382e-01 1.62116840e-01
4.75238889e-01 -6.49621964e-01 -2.92043149e-01 -4.98764999e-02] | [7.172255992889404, 7.806069374084473] |
c0da5dd2-3831-412f-a256-5b39781dd8e6 | caponimage-context-driven-dense-captioning-on | 2204.12974 | null | https://arxiv.org/abs/2204.12974v1 | https://arxiv.org/pdf/2204.12974v1.pdf | CapOnImage: Context-driven Dense-Captioning on Image | Existing image captioning systems are dedicated to generating narrative captions for images, which are spatially detached from the image in presentation. However, texts can also be used as decorations on the image to highlight the key points and increase the attractiveness of images. In this work, we introduce a new task called captioning on image (CapOnImage), which aims to generate dense captions at different locations of the image based on contextual information. To fully exploit the surrounding visual context to generate the most suitable caption for each location, we propose a multi-modal pre-training model with multi-level pre-training tasks that progressively learn the correspondence between texts and image locations from easy to difficult. Since the model may generate redundant captions for nearby locations, we further enhance the location embedding with neighbor locations as context. For this new task, we also introduce a large-scale benchmark called CapOnImage2M, which contains 2.1 million product images, each with an average of 4.8 spatially localized captions. Compared with other image captioning model variants, our model achieves the best results in both captioning accuracy and diversity aspects. We will make code and datasets public to facilitate future research. | ['Peng Wang', 'Yuning Jiang', 'Tiezheng Ge', 'Yuanmeng Zhang', 'Xinglin Hou', 'Yiqi Gao'] | 2022-04-27 | null | null | null | null | ['dense-captioning'] | ['computer-vision'] | [ 4.03583258e-01 2.47089684e-01 -1.49031326e-01 -3.70129198e-01
-8.68967474e-01 -6.51789904e-01 7.03513861e-01 3.98147218e-02
-1.24530882e-01 6.21644378e-01 5.56192160e-01 -8.39026272e-02
5.66776633e-01 -6.33421361e-01 -1.20846021e+00 -4.65553880e-01
4.00838435e-01 1.66127741e-01 1.88130841e-01 -3.38519186e-01
1.12439826e-01 2.11131319e-01 -1.32959676e+00 8.28241706e-01
6.65343285e-01 7.96726048e-01 6.50447130e-01 3.95215809e-01
-1.08709238e-01 5.78979135e-01 -7.64501095e-01 -5.37073731e-01
3.75977084e-02 -5.74996829e-01 -5.14803708e-01 3.57952505e-01
6.33243322e-01 -3.10474306e-01 -3.88568908e-01 8.87853801e-01
3.18969607e-01 -2.04522852e-02 6.02919996e-01 -1.31739652e+00
-1.55274844e+00 7.29674399e-01 -9.95048285e-01 3.80833559e-02
3.63640904e-01 3.20882529e-01 8.92836988e-01 -1.05705416e+00
7.53017604e-01 1.11259520e+00 2.88935333e-01 6.50879145e-01
-1.02774644e+00 -5.94421685e-01 3.66619885e-01 2.30055973e-01
-1.27267289e+00 -2.18416050e-01 9.64107454e-01 -2.89049268e-01
2.53084123e-01 3.48239988e-01 2.42616296e-01 1.33214533e+00
-1.85223252e-01 1.05022049e+00 8.89386237e-01 -3.27705532e-01
-9.45023075e-02 3.00853670e-01 -3.54280323e-01 3.35158736e-01
5.52129783e-02 -2.25723222e-01 -5.18936813e-01 9.90329385e-02
9.09921110e-01 4.76066284e-02 -4.59135741e-01 -3.30557168e-01
-1.51315367e+00 9.09770548e-01 9.01120603e-01 1.48855701e-01
-5.03385007e-01 3.38699400e-01 1.28868103e-01 -4.32988077e-01
3.20746124e-01 6.97831094e-01 7.57029057e-02 1.19240351e-01
-7.54734933e-01 1.02358483e-01 9.35797393e-02 1.02712286e+00
7.18601942e-01 -2.62674928e-01 -7.29268014e-01 8.26696277e-01
1.75795823e-01 6.59455240e-01 3.23946267e-01 -5.23006856e-01
9.05474246e-01 5.97604394e-01 5.22119343e-01 -1.24421418e+00
3.18761803e-02 -3.71045291e-01 -7.17657506e-01 -2.81671762e-01
1.35152563e-01 -2.90764064e-01 -1.16794181e+00 1.68846452e+00
1.66863829e-01 3.81761134e-01 1.71202406e-01 1.24371409e+00
1.06226909e+00 1.27232862e+00 3.40528667e-01 3.49950135e-01
1.52344060e+00 -1.50456691e+00 -9.18640137e-01 -7.69666791e-01
4.95603889e-01 -9.65842605e-01 1.35782683e+00 -3.35000038e-01
-1.06254566e+00 -5.11927545e-01 -9.85296667e-01 -2.08150432e-01
-5.27238250e-01 4.35937166e-01 4.71211046e-01 1.83556885e-01
-9.70574319e-01 -3.22201774e-02 -1.37048706e-01 -6.65023625e-02
5.47783613e-01 -2.30438277e-01 -3.49676192e-01 -3.95706862e-01
-1.32132196e+00 7.86481857e-01 3.96142930e-01 2.84423828e-01
-7.96411157e-01 -7.14871049e-01 -1.17352605e+00 5.83525561e-02
5.47736213e-02 -4.52131063e-01 1.05689752e+00 -1.29778469e+00
-9.40074205e-01 8.18286896e-01 -1.33762717e-01 -2.10975707e-01
4.21453148e-01 -1.27934828e-01 -4.24680501e-01 3.44273567e-01
4.01670158e-01 1.48648453e+00 7.93006241e-01 -1.87223625e+00
-4.59972918e-01 1.16756856e-01 1.79532394e-01 4.61151630e-01
-2.69455254e-01 -3.39982137e-02 -8.28618288e-01 -9.48400259e-01
-3.17687213e-01 -8.65894914e-01 -1.62011817e-01 7.54576782e-03
-6.12066686e-01 1.71606198e-01 9.82778311e-01 -7.36739874e-01
1.11854529e+00 -2.31571245e+00 2.80034095e-02 -1.87290430e-01
-4.06873450e-02 2.11070016e-01 -5.94194174e-01 4.54019636e-01
-1.53712243e-01 1.98625132e-01 -3.29724222e-01 -5.01524508e-01
-1.83500499e-01 1.08148582e-01 -5.34713268e-01 7.16336370e-02
5.65378845e-01 1.60015821e+00 -1.12962174e+00 -5.51112235e-01
2.07326800e-01 5.75239182e-01 -2.22234964e-01 2.05368191e-01
-4.63465184e-01 3.36211354e-01 -4.46497798e-01 5.71549356e-01
8.81067991e-01 -6.61329865e-01 -1.79529339e-01 -3.22373033e-01
2.18411297e-01 -2.10097089e-01 -6.91555679e-01 1.71670246e+00
-6.98918819e-01 9.66217041e-01 -2.36767411e-01 -5.12225330e-01
9.04754639e-01 1.70827672e-01 1.80840239e-01 -8.06598842e-01
-3.62506211e-02 -3.01179178e-02 -4.25488383e-01 -5.38846731e-01
8.84554982e-01 1.78694516e-01 -4.38533872e-01 3.97822797e-01
-5.41186213e-01 1.11025639e-01 4.02755328e-02 3.55431914e-01
5.96142948e-01 1.28935650e-01 -2.85364658e-01 1.77163631e-01
2.05077380e-01 2.90783376e-01 -2.53448598e-02 5.05045354e-01
-5.71545288e-02 1.55483913e+00 4.94325399e-01 -4.04779136e-01
-1.27364910e+00 -1.13363028e+00 3.85029577e-02 9.90990579e-01
7.82935083e-01 -1.60051495e-01 -7.29053438e-01 -7.25224257e-01
-2.07629770e-01 8.01452398e-01 -1.02333248e+00 -1.90411791e-01
-5.86787999e-01 -5.09424806e-01 1.49557859e-01 5.33824861e-01
7.10794330e-01 -1.46984732e+00 -4.08323407e-01 -5.61705455e-02
-6.84272826e-01 -1.25757992e+00 -1.14107037e+00 -3.01731408e-01
-2.30942845e-01 -7.49914348e-01 -1.43341506e+00 -1.34947085e+00
1.08780444e+00 5.00671685e-01 1.21884942e+00 -2.37332620e-02
-7.31740892e-02 1.99359044e-01 -6.63663924e-01 -3.24900836e-01
-2.44806007e-01 1.76839933e-01 -6.10596657e-01 3.07387263e-01
1.46313369e-01 -8.11225325e-02 -8.82858813e-01 5.13834357e-01
-1.27059627e+00 7.55398512e-01 8.38211715e-01 9.42472219e-01
5.79689980e-01 -3.68145436e-01 6.48172557e-01 -7.56675243e-01
6.79435909e-01 -9.03143644e-01 -1.27264649e-01 4.39378500e-01
9.90985632e-02 4.18107491e-03 5.09783924e-01 -6.97187185e-01
-9.52593923e-01 4.37425762e-01 2.70494550e-01 -6.30320966e-01
3.02056633e-02 4.65035826e-01 -2.22429261e-01 2.54498661e-01
5.76209426e-01 5.19353449e-01 1.57550499e-02 -2.14097172e-01
7.49920547e-01 7.47564077e-01 6.70901775e-01 -3.62368494e-01
8.18309009e-01 3.76729697e-01 -4.03558761e-01 -3.55728894e-01
-1.03999662e+00 -2.25526646e-01 -2.96174109e-01 -1.99495956e-01
1.09261668e+00 -1.06622052e+00 -2.55964518e-01 1.93920702e-01
-1.48461890e+00 -3.75076711e-01 -1.40202805e-01 5.09314910e-02
-4.42326367e-01 2.30172545e-01 -2.82402605e-01 -3.39341551e-01
-2.42986500e-01 -1.21685445e+00 1.61613095e+00 4.54344124e-01
-1.77752067e-04 -8.64467800e-01 -2.07567319e-01 4.86526072e-01
3.50870341e-01 5.93068182e-01 4.59194392e-01 -1.64839283e-01
-7.36718237e-01 -2.21026286e-01 -7.40724027e-01 6.80863559e-02
1.06169283e-01 -3.15933824e-01 -7.89755464e-01 -8.19056928e-02
-5.46979249e-01 -3.56763273e-01 8.02275658e-01 2.04307780e-01
1.26707685e+00 -5.01132786e-01 -4.83770460e-01 3.61825198e-01
1.36567092e+00 7.76001289e-02 1.04024124e+00 6.13596737e-01
9.86392736e-01 7.44094551e-01 9.45108652e-01 2.05012619e-01
3.17536741e-01 7.94471681e-01 7.53894329e-01 -7.47233987e-01
-3.33953738e-01 -7.96616197e-01 2.35206574e-01 1.40496701e-01
5.72240233e-01 -5.89138508e-01 -6.75162196e-01 9.36690032e-01
-2.04474926e+00 -8.36479068e-01 -2.37041507e-02 1.87745988e+00
7.06258833e-01 -2.59492397e-01 -1.79044325e-02 -5.57928681e-01
1.28368831e+00 3.75477761e-01 -6.48525417e-01 -2.64205933e-01
-4.11666453e-01 -3.42820317e-01 6.28387451e-01 3.53191257e-01
-1.09594488e+00 9.30096269e-01 6.19764566e+00 9.47552919e-01
-1.12599909e+00 9.08770785e-02 1.27864993e+00 -2.24627703e-01
-6.98403120e-01 -2.73757815e-01 -5.84189355e-01 1.00502717e+00
5.90377510e-01 7.73526877e-02 1.47164240e-01 8.94893050e-01
2.00017482e-01 4.27236501e-03 -8.64939094e-01 1.19727468e+00
4.33454603e-01 -1.65273786e+00 4.94601518e-01 -3.92371193e-02
1.15084708e+00 -2.55262285e-01 6.04925931e-01 7.21614510e-02
3.34028751e-02 -1.26715839e+00 8.83235574e-01 2.31537953e-01
1.00453353e+00 -7.14855254e-01 7.31209397e-01 -6.30567819e-02
-1.13535213e+00 -1.56517327e-02 -2.85065502e-01 2.22447306e-01
7.06769407e-01 2.92699724e-01 -7.75907815e-01 3.50718230e-01
4.96550024e-01 7.53010690e-01 -8.54292393e-01 1.00243962e+00
-3.68196964e-01 9.45272073e-02 -1.22208513e-01 -1.94784582e-01
7.19569504e-01 -1.16812950e-02 1.61389336e-01 1.06892312e+00
5.63244164e-01 -8.84252936e-02 -1.82333872e-01 1.02889705e+00
-2.89022267e-01 1.10135645e-01 -6.52626276e-01 5.94028719e-02
5.72988629e-01 1.28007770e+00 -6.32830679e-01 -3.23095143e-01
-3.70971948e-01 1.40166008e+00 2.19196945e-01 4.90393251e-01
-1.36635125e+00 -4.23306972e-01 4.53799337e-01 4.78492767e-01
4.69552636e-01 8.20897054e-03 -1.34582058e-01 -9.20788229e-01
2.91919142e-01 -6.56378806e-01 1.62417248e-01 -1.60800529e+00
-1.17963219e+00 8.88975680e-01 7.77926967e-02 -1.42555761e+00
-3.64378542e-02 -4.87311989e-01 -8.02975178e-01 9.78330910e-01
-1.64301050e+00 -1.61872172e+00 -5.96873701e-01 2.73615122e-01
5.85016608e-01 1.90213665e-01 3.13360214e-01 3.22662771e-01
-4.43123221e-01 6.81508005e-01 1.23944096e-01 1.46552175e-01
9.27344859e-01 -9.90328252e-01 5.62033057e-01 8.09944272e-01
1.96571067e-01 2.60405421e-01 6.75859451e-01 -6.40934110e-01
-9.29050565e-01 -1.49340141e+00 8.50827992e-01 -6.59055650e-01
4.12452698e-01 -5.50132990e-01 -7.06109107e-01 6.14166975e-01
5.79012215e-01 -2.23795185e-03 3.95823985e-01 -3.42114955e-01
-3.07878673e-01 1.25089481e-01 -9.29893255e-01 1.02244854e+00
9.01927769e-01 -2.69550800e-01 -3.48853230e-01 5.99992871e-01
1.05349028e+00 -5.34420669e-01 -4.68158633e-01 -1.04806926e-02
1.38223305e-01 -6.65145934e-01 1.17246211e+00 -2.34321907e-01
1.08850884e+00 -4.92738813e-01 1.54576972e-01 -1.42122161e+00
-2.24790707e-01 -5.93343318e-01 2.11878732e-01 1.39722025e+00
8.53033483e-01 -2.59930730e-01 8.80626142e-01 6.43399894e-01
-1.55906051e-01 -6.86611652e-01 -6.23714209e-01 -4.39965546e-01
-6.50122017e-02 3.39186415e-02 9.39492881e-01 8.66264105e-01
-1.42870709e-01 5.12142420e-01 -7.82390654e-01 2.02480987e-01
2.25404486e-01 3.71641487e-01 6.71017826e-01 -4.06143486e-01
-2.56923903e-02 -3.60535443e-01 -2.50773311e-01 -1.33953142e+00
-4.45907228e-02 -6.85908496e-01 2.06316724e-01 -1.77956092e+00
4.49112982e-01 -3.34109932e-01 -1.13348149e-01 5.59050381e-01
-5.78937888e-01 8.42602909e-01 2.32380167e-01 2.35423267e-01
-8.06472600e-01 6.66685283e-01 1.89419246e+00 -3.26489627e-01
-9.39200222e-02 -5.12402594e-01 -1.09076345e+00 6.00542650e-02
7.60457575e-01 -2.21232653e-01 -5.50356269e-01 -7.04021752e-01
1.29390627e-01 -1.45186067e-01 4.91757244e-01 -7.19910145e-01
-1.57585908e-02 -2.77096808e-01 6.83695853e-01 -7.06486762e-01
5.17523348e-01 -6.87398732e-01 1.85574457e-01 1.19249351e-01
-6.27371490e-01 2.76558429e-01 3.09266359e-01 6.11733794e-01
-3.83254856e-01 -3.01432282e-01 6.07714415e-01 -3.84413358e-03
-8.98660421e-01 3.50754827e-01 4.25687263e-04 -8.84066597e-02
1.53185940e+00 -1.84123859e-01 -6.55505121e-01 -7.82202661e-01
-4.58352238e-01 5.07760644e-01 7.74532199e-01 9.34587717e-01
6.65589333e-01 -1.74248087e+00 -8.23854864e-01 -1.32848173e-01
5.66642821e-01 1.65163487e-01 6.20175302e-01 1.96672916e-01
-6.32868648e-01 4.28456575e-01 -3.33534211e-01 -4.86119926e-01
-1.00878763e+00 1.02796614e+00 3.90158892e-02 1.81213021e-02
-5.24914622e-01 7.00724542e-01 7.96371043e-01 8.48413184e-02
-6.08862527e-02 -9.31743998e-03 -3.11709613e-01 -9.25802812e-02
7.71953106e-01 -4.09760803e-01 -4.45162296e-01 -1.00847912e+00
-6.15629042e-03 4.97196168e-01 -2.79970616e-01 -8.34301859e-02
1.00242960e+00 -4.20072317e-01 2.70154744e-01 -7.51138618e-03
1.33788300e+00 -2.60946244e-01 -1.65090048e+00 -1.08558498e-01
-3.97719175e-01 -7.09454536e-01 -6.90430626e-02 -9.48946238e-01
-1.22007048e+00 7.95031786e-01 5.01457870e-01 -4.55929711e-02
1.14355028e+00 2.79167891e-01 9.76080596e-01 -2.39460275e-01
1.85730569e-02 -8.66335690e-01 6.91759348e-01 -6.03392208e-03
1.30052459e+00 -1.50379372e+00 -2.99490452e-01 -5.60876429e-01
-1.22502100e+00 6.80747747e-01 9.55855310e-01 8.70985985e-02
1.09610744e-01 -1.89139307e-01 2.40951911e-01 -1.53573707e-01
-3.86362851e-01 -1.73588946e-01 3.73275459e-01 9.52767730e-01
2.29809880e-01 -2.10747439e-02 -4.86841463e-02 5.60786128e-01
3.75519656e-02 -3.91684830e-01 6.43912733e-01 6.96448743e-01
-1.96205899e-01 -9.11747694e-01 -5.77313304e-01 2.00485051e-01
-1.13783367e-01 -3.46182704e-01 -4.90059733e-01 6.86748326e-01
-4.33922373e-02 8.81369829e-01 3.42532396e-01 -2.00047955e-01
2.04637244e-01 -3.64767551e-01 6.52114674e-02 -6.82687819e-01
-2.46082395e-01 -2.18179688e-01 4.88013029e-02 -2.50071973e-01
-2.10007295e-01 -3.66280615e-01 -9.93169010e-01 -1.91650689e-02
-4.19256054e-02 2.08916500e-01 7.55081952e-01 4.00945038e-01
8.13994646e-01 5.21204531e-01 7.79294431e-01 -1.10267627e+00
1.76828653e-02 -8.57454598e-01 -3.69000345e-01 8.33867967e-01
1.51379257e-01 -5.56596398e-01 -2.03104988e-01 2.64251351e-01] | [10.97808837890625, 0.9605631828308105] |
06e42eb6-538a-438c-b62c-1461c3a0f6e9 | the-text-anonymization-benchmark-tab-a | 2202.00443 | null | https://arxiv.org/abs/2202.00443v2 | https://arxiv.org/pdf/2202.00443v2.pdf | The Text Anonymization Benchmark (TAB): A Dedicated Corpus and Evaluation Framework for Text Anonymization | We present a novel benchmark and associated evaluation metrics for assessing the performance of text anonymization methods. Text anonymization, defined as the task of editing a text document to prevent the disclosure of personal information, currently suffers from a shortage of privacy-oriented annotated text resources, making it difficult to properly evaluate the level of privacy protection offered by various anonymization methods. This paper presents TAB (Text Anonymization Benchmark), a new, open-source annotated corpus developed to address this shortage. The corpus comprises 1,268 English-language court cases from the European Court of Human Rights (ECHR) enriched with comprehensive annotations about the personal information appearing in each document, including their semantic category, identifier type, confidential attributes, and co-reference relations. Compared to previous work, the TAB corpus is designed to go beyond traditional de-identification (which is limited to the detection of predefined semantic categories), and explicitly marks which text spans ought to be masked in order to conceal the identity of the person to be protected. Along with presenting the corpus and its annotation layers, we also propose a set of evaluation metrics that are specifically tailored towards measuring the performance of text anonymization, both in terms of privacy protection and utility preservation. We illustrate the use of the benchmark and the proposed metrics by assessing the empirical performance of several baseline text anonymization models. The full corpus along with its privacy-oriented annotation guidelines, evaluation scripts and baseline models are available on: https://github.com/NorskRegnesentral/text-anonymisation-benchmark | ['Montserrat Batet', 'David Sánchez', 'Anthi Papadopoulou', 'Lilja Øvrelid', 'Pierre Lison', 'Ildikó Pilán'] | 2022-01-25 | null | null | null | null | ['text-anonymization'] | ['natural-language-processing'] | [ 4.01855081e-01 3.50806952e-01 -1.69698521e-01 -4.07290846e-01
-7.91610122e-01 -1.14122796e+00 7.25083351e-01 8.28125954e-01
-6.52822733e-01 7.53243268e-01 7.18881130e-01 -2.84218520e-01
-3.80395800e-01 -5.29415727e-01 -2.19670549e-01 -2.75885373e-01
2.93806076e-01 5.96721590e-01 -3.77703190e-01 2.65762955e-02
1.33695424e-01 5.90929449e-01 -1.08603084e+00 4.75682020e-01
1.03621042e+00 7.27841914e-01 -5.97945452e-01 1.44396335e-01
-9.75943133e-02 2.52599418e-01 -7.22928643e-01 -1.38646936e+00
5.51444948e-01 -1.41500667e-01 -1.10352767e+00 -4.17222768e-01
4.17244673e-01 -1.76986024e-01 -4.68679011e-01 1.23938262e+00
6.20343268e-01 -1.68135628e-01 4.28108811e-01 -1.14763701e+00
-6.36808276e-01 8.14228415e-01 -2.23206446e-01 5.16603366e-02
5.41421294e-01 -1.10232487e-01 1.00760591e+00 -3.94619823e-01
1.05888307e+00 8.30682814e-01 8.32007587e-01 7.98588693e-01
-1.30299056e+00 -5.24899364e-01 -2.31170535e-01 -1.03225388e-01
-1.42007709e+00 -6.80848360e-01 3.08306992e-01 -5.85154593e-01
6.72878504e-01 7.31820762e-01 2.13808864e-01 1.32589281e+00
-1.99490026e-01 5.89144170e-01 7.09397197e-01 -4.12031412e-01
1.09013192e-01 3.05222601e-01 3.24848622e-01 1.11532047e-01
8.73655260e-01 -1.41354665e-01 -4.42069799e-01 -7.64184713e-01
-1.09479681e-01 -1.89322099e-01 -3.57375711e-01 -6.45709634e-01
-1.09041262e+00 4.63848263e-01 6.44110814e-02 4.38576609e-01
-9.69077498e-02 -3.98783088e-01 1.14315438e+00 1.32107094e-01
5.28169453e-01 5.94605148e-01 -3.28966916e-01 -9.26506370e-02
-9.56784546e-01 4.25344646e-01 8.41277957e-01 1.23350835e+00
6.67023659e-01 -5.25281787e-01 -6.35054886e-01 8.39408100e-01
-2.39171516e-02 1.07349493e-01 6.22905612e-01 -7.26953745e-01
9.72125173e-01 9.65277612e-01 2.87051350e-01 -9.23546135e-01
-1.87530637e-01 -1.62636563e-01 -9.60010648e-01 -1.01108871e-01
5.21241426e-01 -2.60386020e-01 -3.90942305e-01 1.55035436e+00
2.24137321e-01 -5.88056982e-01 3.11749190e-01 5.23548722e-01
8.17602336e-01 1.94787569e-02 3.73598933e-01 -8.24042335e-02
1.57019353e+00 -2.21599787e-01 -9.87390995e-01 -1.05805276e-02
9.49155688e-01 -5.43464541e-01 8.96580100e-01 -1.33509725e-01
-7.24114954e-01 4.91139293e-02 -7.81975806e-01 -3.95856410e-01
-8.28398883e-01 2.37077184e-04 4.87117648e-01 1.50431335e+00
-7.04056978e-01 5.43313205e-01 -5.42708457e-01 -6.27831995e-01
8.39458704e-01 2.23896697e-01 -1.06622005e+00 1.35288954e-01
-1.40735757e+00 5.48199177e-01 6.34236693e-01 -5.88088594e-02
-9.02822241e-02 -8.71415913e-01 -9.26224113e-01 1.87466413e-01
2.81883717e-01 -6.50524497e-01 7.47759938e-01 -7.50086784e-01
-7.73896754e-01 1.45826137e+00 -1.42461538e-01 -7.42608666e-01
1.09976482e+00 -2.73423523e-01 -7.47752309e-01 -3.85515019e-02
5.39596736e-01 3.36444557e-01 3.40072393e-01 -1.01739037e+00
-7.21431077e-01 -6.57449067e-01 -1.32282853e-01 -2.65581273e-02
-7.57791460e-01 3.32500994e-01 -5.06085575e-01 -1.18873119e+00
-3.57652664e-01 -7.25239933e-01 -1.58471271e-01 -3.07112277e-01
-7.87432253e-01 1.57929063e-01 7.83692241e-01 -1.05571556e+00
1.67051160e+00 -2.22810078e+00 -2.61976898e-01 2.28464976e-01
4.03628498e-01 6.13832414e-01 1.92153022e-01 8.04699242e-01
-3.54399025e-01 8.07915449e-01 -7.80477047e-01 -6.78766429e-01
4.00247306e-01 -1.31452903e-01 -5.13415933e-01 6.99072897e-01
-2.65404612e-01 1.03388274e+00 -6.68031573e-01 -3.90006810e-01
1.66957349e-01 3.16211581e-01 -3.55560899e-01 -2.12342381e-01
1.00872353e-01 4.19586957e-01 -3.95956486e-01 6.35734916e-01
7.99496114e-01 3.31683546e-01 4.17878032e-01 1.59528814e-02
1.54483825e-01 2.88701981e-01 -1.03753483e+00 1.70228338e+00
-1.87383834e-02 6.44760668e-01 -3.11368201e-02 -4.27316010e-01
7.36072600e-01 4.43503201e-01 6.47407651e-01 -4.87797201e-01
1.70624182e-02 2.28752553e-01 -4.10474300e-01 -3.88194472e-01
8.91079068e-01 3.22707862e-01 -5.81875145e-01 5.98183990e-01
-2.47740671e-01 3.95139366e-01 4.35502857e-01 3.83815140e-01
1.19083512e+00 -7.68608525e-02 5.31010926e-01 -2.51685351e-01
8.57187390e-01 -1.87390327e-01 7.66715527e-01 5.87577164e-01
-3.00743818e-01 4.62830454e-01 6.80560887e-01 -5.08150876e-01
-1.18447447e+00 -7.02419162e-01 -3.52246225e-01 5.99078417e-01
-1.67908490e-01 -8.96999776e-01 -1.16319942e+00 -1.10585952e+00
3.34894300e-01 1.00461042e+00 -8.53155255e-01 -1.09201081e-01
-4.67649609e-01 -7.75833428e-01 1.19777262e+00 6.22228123e-02
5.72861373e-01 -9.09109473e-01 -4.86405969e-01 -1.40084654e-01
-4.74980801e-01 -1.28798234e+00 -8.04920316e-01 -3.79650950e-01
-5.16947269e-01 -1.22412908e+00 -2.16773301e-01 -1.43522397e-01
6.21884167e-01 -3.56766045e-01 8.76750708e-01 -5.93163483e-02
-1.94129437e-01 3.82185131e-01 -5.28491139e-01 -3.26373756e-01
-6.92200303e-01 5.10047376e-01 -1.78607479e-01 4.04876918e-01
6.76616490e-01 -4.51617002e-01 -4.06978309e-01 1.96934119e-01
-1.38997579e+00 -4.37448025e-01 1.71946973e-01 5.25640309e-01
4.08636332e-01 -8.04618746e-02 3.49452436e-01 -1.58120167e+00
1.01138973e+00 -3.90502781e-01 -4.94539976e-01 5.39792001e-01
-6.90025270e-01 -1.10638678e-01 6.92744076e-01 2.17203677e-01
-1.06783664e+00 3.45306136e-02 -1.06588945e-01 1.43353706e-02
-4.67705667e-01 2.54638910e-01 -9.26168144e-01 8.10721517e-02
6.40776932e-01 -1.49783082e-02 -4.06195581e-01 -7.65591979e-01
5.53175747e-01 9.46168542e-01 7.37546563e-01 -5.95674336e-01
7.40574956e-01 5.88969052e-01 -1.38441533e-01 -4.43027288e-01
-4.52643842e-01 -4.64582145e-01 -1.05984819e+00 2.57572442e-01
7.09816813e-01 -5.85995913e-01 -7.07882702e-01 2.63042659e-01
-1.03302836e+00 2.23070815e-01 -6.47148192e-01 6.77909106e-02
-3.14937860e-01 9.45031106e-01 -2.00739533e-01 -5.92181146e-01
-6.81036472e-01 -7.76417196e-01 9.15696383e-01 -4.98539448e-01
-7.81363666e-01 -9.04603124e-01 6.00469001e-02 6.37438416e-01
1.26626408e-02 6.37715280e-01 1.06781697e+00 -1.17507052e+00
5.58147244e-02 -8.25096726e-01 -4.29681353e-02 2.15985239e-01
2.58688301e-01 -2.93039888e-01 -9.18440282e-01 -3.88660431e-01
-2.80827680e-03 3.01164389e-01 6.27453983e-01 -2.14549050e-01
1.15257597e+00 -8.86703134e-01 -5.00262499e-01 8.44763517e-01
1.37918711e+00 -9.13902447e-02 9.44601238e-01 6.93521082e-01
7.72139132e-01 9.70897317e-01 2.95773089e-01 5.57681262e-01
1.80515587e-01 9.62274790e-01 2.74628550e-01 1.83443725e-01
1.25798196e-01 -5.16269565e-01 -1.69418436e-02 1.51449829e-01
1.60135418e-01 -6.98082030e-01 -9.62979913e-01 5.79493225e-01
-1.82949412e+00 -1.05669355e+00 -1.90367877e-01 2.53912568e+00
6.13375366e-01 -3.13088506e-01 2.06754476e-01 2.13536412e-01
9.09850657e-01 6.79597855e-02 -3.55275512e-01 -5.59442997e-01
-4.22957242e-01 -3.52201797e-02 8.64081562e-01 3.84093463e-01
-1.14835167e+00 8.70519161e-01 5.84083366e+00 7.97573507e-01
-5.64360738e-01 1.68306887e-01 5.28923392e-01 -1.48226485e-01
-5.63979745e-01 1.40813023e-01 -8.62927735e-01 8.34046960e-01
9.04387951e-01 -7.16899157e-01 1.88625664e-01 5.71703851e-01
1.53610379e-01 3.12972963e-01 -1.33284187e+00 8.42948377e-01
-3.12136263e-02 -1.29254973e+00 5.03913641e-01 3.94451529e-01
4.48634833e-01 -4.34512973e-01 1.51957408e-01 -1.40647560e-01
2.21119598e-01 -9.23050642e-01 9.29440796e-01 3.36926430e-01
1.21273017e+00 -7.21038222e-01 8.50414634e-01 -3.49853784e-02
-7.87811637e-01 -2.05829293e-01 -1.26401976e-01 4.37430322e-01
1.18588552e-01 5.82371831e-01 -4.85247642e-01 1.15092063e+00
8.31967115e-01 7.08701849e-01 -7.89003432e-01 7.46890903e-01
-2.34609202e-01 3.11974198e-01 -5.35954982e-02 4.19820368e-01
-9.00858194e-02 -2.89835244e-01 9.12567258e-01 1.50652766e+00
2.62615412e-01 -9.28909853e-02 -3.78666192e-01 8.20230007e-01
-7.12290525e-01 4.89491373e-01 -5.35756826e-01 -7.16957822e-02
8.53123307e-01 1.11812413e+00 -4.00303304e-01 -1.34483904e-01
-3.43593419e-01 1.07848132e+00 2.91216820e-01 2.16061234e-01
-4.80910927e-01 -6.55104756e-01 9.37199593e-01 4.78308529e-01
1.57674421e-02 2.39808306e-01 -5.76531410e-01 -1.25480068e+00
4.93630111e-01 -1.00464141e+00 8.62653732e-01 -2.52630651e-01
-1.25113535e+00 6.40262604e-01 3.09792776e-02 -1.19400644e+00
-1.42357484e-01 -2.40028188e-01 -1.94321111e-01 1.04478419e+00
-8.87767732e-01 -1.29595149e+00 -1.08914897e-01 4.76604223e-01
-1.32582292e-01 -3.61856222e-01 1.12868452e+00 6.44382536e-01
-8.40606570e-01 1.24672699e+00 3.91932338e-01 6.00175381e-01
8.78303707e-01 -1.25958431e+00 9.62226331e-01 1.18051672e+00
1.73851382e-02 9.43315625e-01 6.41051412e-01 -9.87918377e-01
-8.47689390e-01 -1.31139028e+00 1.33629429e+00 -1.03463769e+00
4.79624897e-01 -8.62911999e-01 -1.04238081e+00 1.07709312e+00
2.38467380e-02 -2.10606724e-01 1.04612410e+00 -3.83655205e-02
-5.82023740e-01 -1.34327427e-01 -1.70496070e+00 5.91709673e-01
1.31256652e+00 -6.66140258e-01 -5.33132911e-01 3.46008480e-01
5.47166109e-01 -2.98012435e-01 -1.02297783e+00 1.88517734e-01
4.14430380e-01 -9.81775045e-01 7.35039711e-01 -7.12096214e-01
2.71523874e-02 -1.94623813e-01 -4.83795367e-02 -7.54065394e-01
-1.32751480e-01 -1.03051412e+00 1.80339605e-01 1.89940763e+00
3.98035944e-01 -1.17533886e+00 9.57098901e-01 1.15974796e+00
7.21902326e-02 -8.10823590e-02 -1.31482112e+00 -6.59896493e-01
1.52040705e-01 -2.56204009e-01 1.24576771e+00 1.32202458e+00
1.13441043e-01 -3.31996717e-02 -3.39757413e-01 2.01025501e-01
6.65246844e-01 -2.84802943e-01 7.85459578e-01 -1.26212811e+00
4.26157504e-01 -4.21594501e-01 -5.96512735e-01 -1.52119353e-01
4.81454611e-01 -1.23290431e+00 -7.90510714e-01 -1.37323666e+00
7.28817955e-02 -3.06127876e-01 1.34062752e-01 6.07536077e-01
-1.15476593e-01 -2.96262465e-02 1.52611554e-01 4.56177622e-01
-4.07981783e-01 4.24491704e-01 3.45071167e-01 -9.92331579e-02
1.38669182e-02 9.06253979e-02 -1.07759273e+00 4.66777533e-01
7.38051534e-01 -7.22028911e-01 -1.15918510e-01 -5.19638598e-01
2.38345802e-01 -3.96526426e-01 4.01966631e-01 -8.80750775e-01
3.89074028e-01 3.42630893e-01 3.00215155e-01 -3.07755381e-01
-8.34257081e-02 -1.16544235e+00 5.88742316e-01 2.63741553e-01
-5.84183276e-01 1.42190829e-01 3.69469404e-01 5.28813958e-01
-2.40787596e-01 -3.66095096e-01 4.81036693e-01 4.17293457e-04
-4.54873562e-01 5.22519708e-01 -2.76603132e-01 2.97519952e-01
1.01522589e+00 -5.06201625e-01 -3.56298000e-01 -1.21639304e-01
-7.22535670e-01 2.38956809e-01 1.06608009e+00 4.65908736e-01
1.27247140e-01 -1.15975893e+00 -9.08545434e-01 1.20211445e-01
4.81503725e-01 -4.84504282e-01 3.71118337e-01 2.12165788e-01
-4.39494461e-01 5.05334258e-01 -3.15127105e-01 1.71024874e-01
-1.45161390e+00 8.20983887e-01 3.37755382e-01 -3.83379012e-01
-6.98525786e-01 1.04432955e-01 2.21288130e-02 -6.72884881e-01
8.89637470e-02 2.96050608e-02 -1.97195202e-01 6.68139383e-02
4.76237833e-01 7.19842076e-01 4.54120755e-01 -9.41028714e-01
-4.55383003e-01 1.97895154e-01 -1.38291731e-01 -5.56606017e-02
1.14443588e+00 -3.96188408e-01 -5.93995214e-01 -2.87008613e-01
1.23719788e+00 6.05506301e-01 -5.57397246e-01 -1.54917687e-01
5.63063264e-01 -7.74347544e-01 -5.37977517e-01 -8.74593019e-01
-8.68243814e-01 4.78431642e-01 4.61099356e-01 2.18130842e-01
1.03130436e+00 -3.23637635e-01 5.31971335e-01 1.41420260e-01
5.56612350e-02 -8.92413557e-01 -9.29794967e-01 1.90857291e-01
8.66505504e-01 -8.46545458e-01 2.33278319e-01 -6.89594984e-01
-6.52375996e-01 6.99640930e-01 2.78628618e-01 6.27971113e-01
2.81794190e-01 3.85075212e-02 9.38310102e-02 3.89578044e-02
-4.19015773e-02 2.64771938e-01 2.53066003e-01 8.90445411e-01
2.66111344e-01 2.32768022e-02 -6.21788442e-01 7.60565460e-01
-6.35900557e-01 -2.75173157e-01 3.67228150e-01 6.32376552e-01
3.61935467e-01 -1.60566068e+00 -2.62495875e-01 5.70240915e-01
-1.07922769e+00 -2.52642959e-01 -9.49411809e-01 7.00367391e-01
2.56497771e-01 7.61773467e-01 -7.65448585e-02 -5.65645918e-02
6.96196139e-01 3.04767098e-02 -1.18528649e-01 -5.99972844e-01
-1.13405335e+00 -7.33793139e-01 4.47753906e-01 -4.27086830e-01
-1.46744341e-01 -1.01284623e+00 -7.71170497e-01 -7.20452905e-01
1.47605509e-01 5.01049519e-01 3.67339104e-01 5.04187226e-01
6.81692898e-01 3.67951058e-02 1.79414913e-01 -1.00393683e-01
-1.64941534e-01 -7.24726319e-01 -5.86480439e-01 9.62597907e-01
1.02109395e-01 2.06607506e-02 -2.39258692e-01 1.14164136e-01] | [6.180400371551514, 6.981301784515381] |
fe913b34-68d8-4e82-a5f0-2f3b13d157f2 | hcld-a-hierarchical-framework-for-zero-shot | null | null | https://aclanthology.org/2022.coling-1.396 | https://aclanthology.org/2022.coling-1.396.pdf | HCLD: A Hierarchical Framework for Zero-shot Cross-lingual Dialogue System | Recently, many task-oriented dialogue systems need to serve users in different languages. However, it is time-consuming to collect enough data of each language for training. Thus, zero-shot adaptation of cross-lingual task-oriented dialog systems has been studied. Most of existing methods consider the word-level alignments to conduct two main tasks for task-oriented dialogue system, i.e., intent detection and slot filling, and they rarely explore the dependency relations among these two tasks. In this paper, we propose a hierarchical framework to classify the pre-defined intents in the high-level and fulfill slot filling under the guidance of intent in the low-level. Particularly, we incorporate sentence-level alignment among different languages to enhance the performance of intent detection. The extensive experiments report that our proposed method achieves the SOTA performance on a public task-oriented dialog dataset. | ['Jianfeng Liu', 'Xurui Yang', 'Jian Ye', 'Zhanyu Ma'] | null | null | null | null | coling-2022-10 | ['intent-detection', 'slot-filling', 'task-oriented-dialogue-systems'] | ['natural-language-processing', 'natural-language-processing', 'natural-language-processing'] | [-1.91148907e-01 -1.05514698e-01 -2.53188640e-01 -6.10694051e-01
-7.75750995e-01 -4.22172129e-01 5.44454873e-01 -3.53077576e-02
-5.37357688e-01 6.73219323e-01 5.77498078e-01 -4.54304338e-01
3.87449712e-01 -4.49568927e-01 3.82493258e-01 -3.47815096e-01
3.37632686e-01 7.99339831e-01 5.45901299e-01 -7.89949894e-01
2.24509209e-01 -3.11908305e-01 -7.95932710e-01 3.62126589e-01
1.28297317e+00 6.23657465e-01 5.61021924e-01 3.96174014e-01
-8.27821434e-01 6.49247050e-01 -7.11989582e-01 -4.16590750e-01
-1.05548687e-01 -5.02196491e-01 -1.09092951e+00 3.01634312e-01
-3.16699982e-01 -3.69673640e-01 -1.20273098e-01 9.87844765e-01
5.31116962e-01 3.08857620e-01 5.51179826e-01 -1.12891090e+00
-4.36762363e-01 7.05778658e-01 -2.70201087e-01 2.17460155e-01
3.73215914e-01 2.00752169e-01 1.23821151e+00 -9.34319854e-01
4.42395806e-01 1.47866774e+00 2.25639343e-01 8.62275362e-01
-7.98396051e-01 -3.76255304e-01 2.52237439e-01 5.70001230e-02
-1.02758586e+00 -5.65958083e-01 9.29887414e-01 -4.99109834e-01
1.01446748e+00 5.17862663e-02 1.99509785e-01 1.00579655e+00
2.74736006e-02 9.25177693e-01 1.12257111e+00 -5.99353552e-01
2.02766851e-01 4.36465651e-01 7.67349422e-01 5.79349041e-01
-3.47891718e-01 -6.20705843e-01 -6.35223210e-01 -3.74191642e-01
3.41581255e-01 -2.23666951e-01 -1.32143227e-02 3.45445275e-02
-9.61241305e-01 1.13937044e+00 -2.32086122e-01 5.35700560e-01
2.50204392e-02 -8.11379611e-01 1.08044088e+00 2.86761969e-01
7.00082481e-01 3.20938438e-01 -6.65680647e-01 -2.28396162e-01
-3.59591097e-01 -8.08008760e-02 1.20102990e+00 1.16727889e+00
6.94328070e-01 -2.13188514e-01 -6.60089791e-01 1.40484548e+00
4.82690275e-01 7.05641061e-02 8.75654161e-01 -5.37592232e-01
9.37763274e-01 9.04902101e-01 9.60434079e-02 -4.68700290e-01
-4.19663668e-01 2.77950734e-01 -7.40314364e-01 -5.75717986e-01
3.26157212e-01 -7.18297064e-01 -5.15607834e-01 1.68483400e+00
4.66349304e-01 -4.41541940e-01 5.96450448e-01 8.68467033e-01
1.02065778e+00 8.71629536e-01 5.75264096e-01 -6.75013661e-01
1.92423725e+00 -1.27357340e+00 -1.15792835e+00 -4.86278564e-01
9.86237407e-01 -9.64431524e-01 1.57017231e+00 -1.73653185e-01
-6.99612319e-01 -5.80410779e-01 -8.44033718e-01 -3.75565380e-01
-2.63774484e-01 3.45794797e-01 5.79950988e-01 6.33657813e-01
-6.00686729e-01 -2.38362670e-01 -3.33257407e-01 -6.12571597e-01
-3.24850529e-01 -5.72404414e-02 2.05065757e-02 2.34792605e-01
-1.83870327e+00 8.90001953e-01 4.09726262e-01 -1.66417152e-01
-5.07413626e-01 -1.58095270e-01 -8.83736968e-01 3.26959640e-02
4.71962869e-01 -1.59723029e-01 1.66171432e+00 -2.70861536e-01
-1.65341771e+00 9.16658580e-01 -4.71199393e-01 -2.79200703e-01
3.82993639e-01 -2.56540805e-01 -2.97464103e-01 -4.24185902e-01
3.19894195e-01 3.60221177e-01 2.96818972e-01 -7.23795712e-01
-9.69125390e-01 -4.29521590e-01 2.77886361e-01 9.34381664e-01
-7.60926008e-01 2.65432090e-01 -7.93380916e-01 -2.67587841e-01
2.87683755e-02 -7.01232195e-01 -2.81774044e-01 -5.48072636e-01
-5.69976330e-01 -1.11978626e+00 8.81952941e-01 -7.00553477e-01
1.47948754e+00 -1.99064827e+00 -6.38257116e-02 -4.96226788e-01
2.04369556e-02 2.82307744e-01 1.91916883e-01 6.14500344e-01
6.36685431e-01 -6.83308095e-02 -1.84142321e-01 -5.89933515e-01
1.20766819e-01 2.31214538e-01 -3.94624859e-01 -7.62116089e-02
-1.72123998e-01 5.12240708e-01 -7.44153500e-01 -1.03401446e+00
3.11138570e-01 -1.98529795e-01 -2.85902977e-01 8.71710598e-01
-3.91224325e-01 4.82009202e-01 -7.63032734e-01 3.36429387e-01
3.33443105e-01 -5.92990071e-02 5.24558425e-01 -2.28445396e-01
-2.14134023e-01 8.86467755e-01 -5.91859639e-01 1.94338119e+00
-7.21498013e-01 2.86849409e-01 1.39467597e-01 -7.64432669e-01
1.19664001e+00 7.48309553e-01 4.16675776e-01 -6.90157712e-01
1.68730140e-01 1.99903473e-01 3.08826119e-01 -7.04111397e-01
6.99885309e-01 -2.87139475e-01 -6.95900142e-01 6.36852562e-01
2.33971715e-01 -1.26212742e-02 3.26668262e-01 1.97142228e-01
7.00555563e-01 -2.95387387e-01 5.20506918e-01 -4.04749572e-01
1.04484200e+00 2.97147125e-01 6.83188617e-01 4.70155269e-01
-5.51638424e-01 8.71950835e-02 4.29478884e-01 -2.36082360e-01
-9.15633738e-01 -5.12939870e-01 -1.27488405e-01 1.81386697e+00
2.22024396e-01 -3.91771585e-01 -9.82869446e-01 -1.01961350e+00
-3.87457848e-01 7.96279669e-01 -6.12322874e-02 -6.33760393e-02
-6.48932576e-01 -6.49156630e-01 5.89615405e-01 1.13481889e-03
1.02007008e+00 -1.14108694e+00 -8.80467370e-02 4.10764545e-01
-5.50529420e-01 -1.28094554e+00 -7.54423499e-01 9.89853367e-02
-5.69426358e-01 -8.38265300e-01 -7.16973126e-01 -1.14944375e+00
3.20843756e-01 3.92292231e-01 9.09842014e-01 -8.92799199e-02
2.27936953e-01 -1.09035242e-03 -4.88393903e-01 -7.35088438e-02
-4.54679251e-01 5.99877000e-01 3.25598806e-01 -1.21356994e-01
9.37895894e-01 -1.25540704e-01 -1.06588848e-01 6.61547542e-01
-4.07133609e-01 2.55221069e-01 2.65397668e-01 8.68897855e-01
-9.35040414e-02 -2.26815969e-01 8.28543663e-01 -1.13367963e+00
1.42762423e+00 -3.07434678e-01 -2.35115141e-01 6.87173069e-01
-3.79432768e-01 1.19658276e-01 8.60379159e-01 -3.34343582e-01
-1.66422880e+00 -2.53280010e-02 -3.67603421e-01 2.08869919e-01
-2.78622091e-01 5.65763235e-01 -5.06673813e-01 3.98505211e-01
4.84636039e-01 5.06602943e-01 -1.95541278e-01 -6.92504048e-01
2.57971466e-01 1.41802061e+00 1.24749117e-01 -7.36537993e-01
3.80034894e-01 -1.82256415e-01 -8.22044969e-01 -9.02299404e-01
-1.07837117e+00 -9.08300996e-01 -7.78316379e-01 -1.34608284e-01
9.83671129e-01 -8.10524762e-01 -5.41783452e-01 4.08847213e-01
-1.50390971e+00 -2.95474142e-01 1.95490584e-01 2.77611077e-01
-3.77813101e-01 5.32282054e-01 -8.65567267e-01 -1.09949088e+00
-7.60286510e-01 -1.29038727e+00 8.30066919e-01 4.79448557e-01
-2.74735272e-01 -1.12612426e+00 2.88742930e-01 5.60882628e-01
2.33925223e-01 -9.79474425e-01 1.09168172e+00 -1.32099569e+00
3.44062112e-02 2.98100680e-01 -1.37388974e-01 4.11908589e-02
5.28429210e-01 -5.84796309e-01 -9.68306661e-01 -9.78882164e-02
5.89644611e-01 -7.19686031e-01 2.19081923e-01 -6.60514692e-03
6.04682386e-01 -5.20527244e-01 -1.54668942e-01 -2.90043093e-02
6.60373569e-01 5.76914012e-01 2.45468527e-01 1.25042051e-01
4.47919428e-01 1.05268061e+00 1.31647897e+00 4.38378423e-01
6.10416472e-01 1.06289279e+00 -5.10066926e-01 5.51730394e-02
2.54209012e-01 -2.40600288e-01 4.23497647e-01 1.35794151e+00
4.01339203e-01 -2.51959175e-01 -1.01259267e+00 4.35855478e-01
-2.15285826e+00 -5.05132258e-01 1.48932664e-02 1.84728456e+00
1.43544090e+00 2.54854083e-01 1.04737617e-01 -5.15033543e-01
1.01265669e+00 4.29841548e-01 -5.69556534e-01 -3.37050140e-01
2.08728939e-01 -4.84396338e-01 -8.11861679e-02 7.15616107e-01
-1.16057014e+00 1.52306855e+00 5.52412462e+00 9.82291996e-01
-8.02013457e-01 4.36258644e-01 7.09108055e-01 4.77568001e-01
-7.87223876e-02 1.26527101e-01 -1.41156018e+00 6.27356291e-01
6.85107589e-01 -5.09147406e-01 -4.55025248e-02 1.12833357e+00
2.47901917e-01 -4.01574820e-02 -7.87652194e-01 7.51568496e-01
-4.37541232e-02 -7.86480486e-01 -1.12171680e-01 -2.48128653e-01
2.74204910e-01 -2.84473538e-01 -3.65350842e-01 9.21694100e-01
5.59012532e-01 -4.96474355e-01 -5.38843870e-02 -4.36414890e-02
5.77882171e-01 -4.34442490e-01 6.85755789e-01 8.69051635e-01
-1.36238837e+00 3.42744768e-01 -6.11685395e-01 -7.09056994e-03
5.65552294e-01 1.65052474e-01 -1.03352618e+00 3.34222287e-01
2.54238427e-01 4.32411104e-01 -2.54439741e-01 5.18329263e-01
-8.57961103e-02 6.03809416e-01 3.11524537e-03 -3.85141343e-01
3.99569482e-01 -4.02803481e-01 4.42758292e-01 1.12112844e+00
4.79062609e-02 2.75099218e-01 7.60323346e-01 3.54503602e-01
1.26928419e-01 5.93856037e-01 -5.66223681e-01 -1.57015733e-02
6.18367910e-01 1.33642697e+00 -3.76223922e-01 -4.39285666e-01
-6.76604271e-01 1.00908518e+00 4.70391363e-01 7.26340115e-02
-6.33688271e-01 -4.09290612e-01 6.02428675e-01 -3.69389266e-01
-4.51205641e-01 -4.99767512e-01 -3.66649330e-01 -1.28167975e+00
-1.06863100e-02 -8.92612338e-01 6.51895225e-01 -3.20015132e-01
-1.58411932e+00 7.55047381e-01 -1.19631156e-01 -1.15164649e+00
-4.72068816e-01 -4.43710804e-01 -8.54925931e-01 1.08261526e+00
-1.16001689e+00 -8.76590371e-01 -1.41619682e-01 5.96325815e-01
1.57142758e+00 -5.66475332e-01 9.98465359e-01 2.24567741e-01
-9.10286188e-01 4.86525297e-01 -1.12912789e-01 4.81155366e-01
9.96926308e-01 -9.61089909e-01 4.62167293e-01 5.53435445e-01
-1.29535422e-01 7.99009025e-01 6.83452427e-01 -8.27281654e-01
-1.04028237e+00 -7.37945616e-01 1.30314040e+00 -5.55640981e-02
7.04178095e-01 -6.57898664e-01 -1.12403023e+00 5.52541912e-01
6.34724736e-01 -5.67209959e-01 7.81062067e-01 5.94515502e-01
-5.05806580e-02 1.03807963e-01 -9.07049656e-01 8.61053884e-01
8.71455431e-01 -7.17660844e-01 -1.05390847e+00 7.33137131e-01
9.97641981e-01 -3.91518593e-01 -6.44585073e-01 1.79490522e-01
3.86403613e-02 -7.10300684e-01 7.29523718e-01 -7.56607831e-01
1.56462342e-01 -7.10533047e-03 -3.20075750e-02 -1.09963095e+00
1.00804558e-02 -5.34406841e-01 1.57773092e-01 1.66819715e+00
4.25764114e-01 -5.12012064e-01 6.14739418e-01 1.01635897e+00
-2.67589778e-01 -4.24899161e-01 -1.00535178e+00 -5.46331704e-01
-3.91132534e-02 -4.16908376e-02 2.79366195e-01 1.05281556e+00
7.68639922e-01 1.41134059e+00 -6.81460023e-01 -2.74205893e-01
1.86254993e-01 1.91651165e-01 8.45333934e-01 -1.17813468e+00
-7.64423013e-02 -9.77672935e-02 3.20053905e-01 -1.60252857e+00
3.72117639e-01 -4.72315282e-01 4.52180058e-01 -1.43138874e+00
2.16875479e-01 -9.08802569e-01 8.58395919e-02 4.24080580e-01
-5.66100359e-01 -5.36419690e-01 -9.97799188e-02 7.30198145e-01
-8.83964300e-01 9.10175800e-01 9.45368409e-01 -1.57963276e-01
-6.39782012e-01 1.59829915e-01 -4.38204348e-01 7.47022808e-01
1.01802683e+00 -2.55405456e-01 -5.68351805e-01 -3.36064547e-01
-5.48878789e-01 4.40481842e-01 -5.53659916e-01 -8.19502771e-01
5.21084428e-01 -5.38991332e-01 -4.09404188e-01 -6.05083168e-01
4.83899504e-01 -5.32626271e-01 -6.05725765e-01 4.78083402e-01
-7.54559696e-01 -1.22534066e-01 4.50314097e-02 2.96606421e-01
-2.00312600e-01 -6.35062754e-01 7.03010976e-01 -2.78912961e-01
-1.05374885e+00 1.80061787e-01 -6.36553288e-01 5.36565721e-01
8.83883357e-01 4.72316831e-01 -5.56848049e-01 -5.13053358e-01
-3.44106436e-01 5.82939744e-01 1.55750364e-01 6.45707905e-01
2.32135922e-01 -1.16800427e+00 -5.08674562e-01 -4.27209176e-02
3.86432379e-01 -1.98478401e-01 4.34139043e-01 5.84923208e-01
3.73508409e-02 8.53090286e-01 -2.52345145e-01 -5.07041991e-01
-1.44831526e+00 1.25404939e-01 2.70065129e-01 -7.91501820e-01
-2.97187597e-01 7.13166118e-01 4.89943653e-01 -9.31931257e-01
3.69131207e-01 4.23889190e-01 -7.42565095e-01 1.79048449e-01
4.79894131e-01 -2.10130543e-01 -2.59333491e-01 -7.36705720e-01
-2.48808235e-01 2.30572410e-02 -5.22197843e-01 -4.76395339e-01
6.66588068e-01 -5.31541109e-01 -1.67040303e-01 7.09528387e-01
8.58183622e-01 -1.91639826e-01 -9.55179036e-01 -7.60365427e-01
3.57985407e-01 -1.88581467e-01 -3.21240276e-01 -4.07795519e-01
-4.82748657e-01 1.16372693e+00 2.64715612e-01 5.24694383e-01
6.75674677e-01 -7.45370910e-02 1.08530366e+00 7.63322473e-01
5.76167405e-01 -1.51276803e+00 9.75664556e-02 1.24323952e+00
5.43097913e-01 -1.46130383e+00 -2.39866495e-01 -5.21193266e-01
-1.17228329e+00 8.31725895e-01 1.35463643e+00 5.09045124e-01
3.78036231e-01 -3.29354629e-02 4.30478990e-01 -4.19151969e-02
-8.58444214e-01 -4.19517428e-01 2.02217832e-01 3.85695517e-01
8.98116231e-01 -9.63687226e-02 -9.55581903e-01 8.03395569e-01
-6.14942797e-02 -4.51096147e-01 1.81250200e-01 9.29373324e-01
-8.40688109e-01 -1.51277995e+00 -5.34619903e-03 3.32694143e-01
-2.49535501e-01 -3.67536426e-01 -5.61171532e-01 4.14290518e-01
-4.31809813e-01 1.27997303e+00 -3.82443443e-02 -4.57292050e-01
3.29034090e-01 6.39475942e-01 -3.61071885e-01 -1.09688830e+00
-6.25350237e-01 1.53998747e-01 5.48107982e-01 1.23670381e-02
-1.72604203e-01 -3.71941805e-01 -1.16393709e+00 -9.63335112e-02
-3.67662400e-01 7.82424331e-01 2.46046051e-01 1.18883383e+00
1.66587472e-01 1.58815473e-01 7.57704973e-01 -3.43915373e-01
-6.68108463e-01 -1.47396469e+00 -5.11504769e-01 3.45829666e-01
-2.40188122e-01 -6.00844145e-01 -1.16961449e-01 -2.27787837e-01] | [12.721749305725098, 7.754916667938232] |
a2bd64d2-ca0b-4cce-b71b-86ba5baf9374 | ilgnet-inception-modules-with-connected-local | 1610.02256 | null | http://arxiv.org/abs/1610.02256v3 | http://arxiv.org/pdf/1610.02256v3.pdf | ILGNet: Inception Modules with Connected Local and Global Features for Efficient Image Aesthetic Quality Classification using Domain Adaptation | In this paper, we address a challenging problem of aesthetic image
classification, which is to label an input image as high or low aesthetic
quality. We take both the local and global features of images into
consideration. A novel deep convolutional neural network named ILGNet is
proposed, which combines both the Inception modules and an connected layer of
both Local and Global features. The ILGnet is based on GoogLeNet. Thus, it is
easy to use a pre-trained GoogLeNet for large-scale image classification
problem and fine tune our connected layers on an large scale database of
aesthetic related images: AVA, i.e. \emph{domain adaptation}. The experiments
reveal that our model achieves the state of the arts in AVA database. Both the
training and testing speeds of our model are higher than those of the original
GoogLeNet. | ['Xiao-Dong Li', 'Geng Zhao', 'Xin Jin', 'Xiaokun Zhang', 'Le Wu', 'Siwei Peng', 'Shuying Li', 'Shiming Ge', 'Jingying Chi'] | 2016-10-07 | null | null | null | null | ['image-quality-estimation'] | ['computer-vision'] | [-1.54692784e-01 -8.45957324e-02 -1.28777400e-02 -4.90164161e-01
-4.09116775e-01 -3.54668140e-01 2.76570112e-01 -2.40247205e-01
-2.10877851e-01 1.68831781e-01 -8.94445255e-02 -2.42320728e-02
-6.52597025e-02 -1.20472300e+00 -6.06230795e-01 -4.11292374e-01
3.14429671e-01 1.62615836e-01 -3.20796743e-02 -3.74279171e-01
2.82634318e-01 2.65467763e-01 -1.51909578e+00 4.51764166e-01
8.90743852e-01 1.88927031e+00 1.15331836e-01 4.64603186e-01
-3.01874071e-01 6.41431451e-01 -5.61839938e-01 -7.25544870e-01
4.85618293e-01 -3.38703156e-01 -1.07695019e+00 2.50731289e-01
2.63889909e-01 -1.15157008e-01 -6.98019890e-03 1.13155687e+00
5.94730139e-01 -1.13547944e-01 5.87921619e-01 -1.54858935e+00
-1.46384883e+00 3.03366065e-01 -6.21814609e-01 -1.11918353e-01
-1.21784456e-01 4.13811296e-01 1.25655890e+00 -6.45260334e-01
4.05072749e-01 1.09816575e+00 7.53859162e-01 3.12549323e-01
-7.11178243e-01 -8.22156250e-01 1.25153646e-01 4.39505279e-01
-1.42973888e+00 -3.04196533e-02 1.05726326e+00 -3.88451010e-01
6.74023211e-01 9.46139321e-02 1.08608603e+00 1.10418844e+00
6.77520782e-02 7.75852680e-01 1.27588105e+00 -2.77145952e-01
2.66450137e-01 1.00744016e-01 -2.35185876e-01 7.06567407e-01
-1.09160885e-01 -2.55621403e-01 -4.48543161e-01 1.78292811e-01
7.87980556e-01 -8.80050287e-02 6.65839091e-02 -2.78862804e-01
-6.75564170e-01 1.11032724e+00 1.07585096e+00 1.15675919e-01
-3.96178395e-01 3.73252571e-01 5.66619277e-01 3.85722607e-01
5.18658757e-01 7.21130550e-01 -4.48587686e-01 -1.03079580e-01
-4.94225919e-01 -3.00768614e-01 4.23083127e-01 9.56401646e-01
8.13347697e-01 -3.48710828e-02 -4.03894298e-02 1.25633943e+00
1.54687002e-01 3.60807776e-01 7.47967958e-01 -7.80893266e-01
1.25130028e-01 1.07037246e+00 -4.39900815e-01 -1.36672902e+00
-1.69833124e-01 -5.11441708e-01 -1.05294681e+00 4.52793151e-01
-5.86117432e-02 2.10073769e-01 -1.00070488e+00 1.62863147e+00
1.58690959e-02 -1.43123433e-01 -1.39722852e-02 9.58006561e-01
1.26622534e+00 7.45353460e-01 2.95122236e-01 2.99249917e-01
1.24065518e+00 -1.48282981e+00 -3.52403820e-01 -3.95791888e-01
2.68845171e-01 -9.10354614e-01 1.72900474e+00 5.82538605e-01
-9.79775786e-01 -8.70516777e-01 -1.07951617e+00 -1.85177922e-01
-8.50474417e-01 3.36852163e-01 8.34617198e-01 6.07415736e-01
-1.29701471e+00 6.29463196e-01 -1.10959485e-01 -5.38356483e-01
9.21451926e-01 2.34132111e-01 -4.17884350e-01 -5.66684157e-02
-9.69866097e-01 5.75332046e-01 3.93625349e-01 -1.60044134e-01
-7.03891337e-01 -2.27353185e-01 -6.60563350e-01 9.44637731e-02
-4.22840239e-03 -6.13845050e-01 1.04907382e+00 -1.79505968e+00
-1.54066730e+00 1.23080778e+00 4.81678218e-01 -6.40465394e-02
2.94400573e-01 -9.74983051e-02 -5.77995956e-01 6.31971583e-02
2.37438977e-01 8.36873174e-01 7.92170465e-01 -9.75072920e-01
-3.84854287e-01 -2.91761551e-02 3.55320990e-01 1.30339608e-01
-1.02519095e+00 -1.44104153e-01 -7.60843396e-01 -7.14758277e-01
-2.50762969e-01 -7.45905042e-01 -1.49282306e-01 3.84351045e-01
-4.46783692e-01 -3.25149894e-01 7.55196035e-01 -4.15690064e-01
1.08246660e+00 -2.24899316e+00 -2.18899623e-01 4.22465771e-01
2.87009984e-01 1.22348137e-01 -6.49939835e-01 3.91483642e-02
-3.90482306e-01 4.03305531e-01 6.79670200e-02 -1.10075334e-02
-1.70370992e-02 8.13914388e-02 1.36964932e-01 8.02212395e-03
2.89168537e-01 1.37460029e+00 -7.10485339e-01 -5.79228103e-01
1.19258888e-01 2.51554489e-01 -5.70670426e-01 2.36285567e-01
-4.89766523e-02 1.31904393e-01 -4.50884461e-01 8.64334047e-01
7.69566059e-01 -7.25947082e-01 -2.10661769e-01 -3.66051942e-01
2.53993064e-01 5.37019596e-02 -8.15569580e-01 1.88103461e+00
-7.63601899e-01 6.10266447e-01 -2.49014810e-01 -7.96266437e-01
1.22632062e+00 -5.72101511e-02 4.32819575e-01 -1.20389879e+00
4.85989213e-01 1.42719910e-01 -2.67479867e-01 -4.90591496e-01
4.44470406e-01 6.03975030e-03 -2.00406387e-01 4.52724427e-01
3.16649109e-01 3.60316746e-02 -1.46079317e-01 1.33256182e-01
9.56272423e-01 6.57515526e-02 3.90924811e-01 -3.81898433e-01
3.21587205e-01 1.58969779e-02 2.63024539e-01 4.18720126e-01
-2.02100113e-01 7.79158294e-01 5.10399938e-01 -7.41896868e-01
-1.17247617e+00 -9.02604282e-01 -3.56585197e-02 1.17027688e+00
4.20728773e-01 -8.25839758e-01 -8.36847663e-01 -8.29815865e-01
-2.56662011e-01 1.17259778e-01 -9.92771804e-01 -5.30447841e-01
-4.86095846e-02 -6.77779257e-01 3.08367759e-01 4.82950211e-01
1.22961116e+00 -1.40526402e+00 -3.40933740e-01 -2.00838238e-01
-1.75626382e-01 -7.52965569e-01 -4.08879280e-01 -1.56952478e-02
-5.59012890e-01 -1.15723288e+00 -5.72244167e-01 -9.17910576e-01
6.16257668e-01 2.55475402e-01 1.47289073e+00 1.94371402e-01
-2.07176477e-01 1.22893505e-01 -5.90327919e-01 -3.14074457e-01
-3.81592885e-02 2.17705771e-01 -4.57116723e-01 2.61782110e-01
4.26873177e-01 -6.41242921e-01 -1.01941705e+00 4.62504119e-01
-8.74638319e-01 2.60297894e-01 9.28654969e-01 7.91914821e-01
7.32416332e-01 2.75652796e-01 4.85444129e-01 -8.20507467e-01
9.64233577e-01 -3.68442953e-01 -1.35176316e-01 3.15519333e-01
-7.46157467e-01 -3.10704023e-01 5.31533539e-01 -4.61259097e-01
-5.99191844e-01 1.33979499e-01 -3.69128257e-01 -4.98498350e-01
-1.51136920e-01 6.13092005e-01 -4.75050271e-01 -4.93124843e-01
7.23376870e-01 -3.02924053e-03 -1.47838965e-01 -4.63975698e-01
5.27577877e-01 8.44830811e-01 3.90291452e-01 -5.06856799e-01
5.69707036e-01 2.87972344e-03 -2.49367833e-01 -4.11334604e-01
-9.13420916e-01 -3.21775019e-01 -4.73329961e-01 -5.60176432e-01
8.85529280e-01 -9.85549331e-01 -6.48288369e-01 5.54484963e-01
-9.54981208e-01 -3.27641368e-01 -5.18360674e-01 -2.58643210e-01
-5.63394904e-01 3.00558433e-02 -4.35013622e-01 -3.08903158e-01
-6.68974400e-01 -1.19826913e+00 9.85457301e-01 3.21172535e-01
-1.54174820e-01 -8.44193399e-01 -1.47263378e-01 4.12915826e-01
6.34396255e-01 3.78625333e-01 8.10838640e-01 -4.12221044e-01
-4.52352196e-01 -3.24704111e-01 -6.88724518e-01 6.56497061e-01
6.52929023e-02 -7.16646109e-03 -1.10803664e+00 3.29163186e-02
-1.40329391e-01 -9.25930977e-01 9.44279909e-01 2.24685207e-01
1.90942526e+00 -5.06731749e-01 8.14666748e-02 9.53217089e-01
1.62303221e+00 -6.04531653e-02 1.01126862e+00 8.76957953e-01
7.48153865e-01 2.90269226e-01 6.63442016e-01 4.43299145e-01
3.73790592e-01 3.91920477e-01 7.25699604e-01 -4.97866273e-01
-1.80011705e-01 -2.79079109e-01 -2.29658466e-02 7.91363299e-01
-2.54506916e-01 -2.55242288e-01 -9.01507258e-01 5.94671845e-01
-1.82921076e+00 -6.78079486e-01 4.67358008e-02 1.69427669e+00
7.91000187e-01 7.89349452e-02 1.41694888e-01 2.81543890e-03
6.67343259e-01 2.09539443e-01 -5.95679700e-01 -7.13725328e-01
-2.46862352e-01 4.19254869e-01 2.73852557e-01 -6.68903589e-02
-1.31519890e+00 1.01784623e+00 6.41065359e+00 1.31313109e+00
-1.28073156e+00 3.26621890e-01 1.23959351e+00 2.78645847e-02
-1.81378543e-01 -3.46278906e-01 2.21015210e-03 5.91203809e-01
8.25053096e-01 4.48127650e-02 4.69729900e-01 1.55002475e+00
-2.32324392e-01 2.80428320e-01 -6.02194011e-01 1.41594708e+00
1.96342707e-01 -1.15235233e+00 3.34395856e-01 -3.27696204e-02
8.58999670e-01 -1.19765781e-01 5.53944528e-01 3.58523577e-01
3.94440621e-01 -1.27232659e+00 7.69489408e-01 2.91425914e-01
1.26513076e+00 -8.76347303e-01 7.65998185e-01 -1.24980278e-01
-1.22699833e+00 -1.16159447e-01 -4.97844487e-01 3.61330844e-02
-2.73709178e-01 4.81030792e-01 -2.48237565e-01 2.73857445e-01
1.28888130e+00 1.01583087e+00 -1.08133864e+00 1.14045167e+00
-2.46811360e-01 3.60038012e-01 -2.87054852e-02 -5.91336079e-02
4.24360126e-01 -1.49709061e-01 -1.43822923e-01 9.02812600e-01
4.23701763e-01 5.13429306e-02 1.44817894e-02 9.60871637e-01
-5.99254310e-01 4.51579660e-01 -5.06455898e-01 -1.90769911e-01
-8.51918384e-02 1.71446967e+00 -7.57966757e-01 -2.77660161e-01
-2.62230128e-01 1.31491435e+00 5.00202119e-01 1.16498180e-01
-9.09058690e-01 -5.53574860e-01 5.50970256e-01 2.31286008e-02
2.96821952e-01 3.22308272e-01 -6.21571302e-01 -1.06701708e+00
-4.60187346e-02 -8.12667787e-01 4.59151864e-01 -1.28014898e+00
-1.65392280e+00 9.96333361e-01 -6.16030037e-01 -1.40539086e+00
3.86112630e-01 -9.02373016e-01 -1.02841580e+00 7.00519919e-01
-1.55044878e+00 -1.59317493e+00 -9.11934674e-01 9.14789915e-01
5.03049195e-01 -3.41334790e-01 8.25139463e-01 3.01841557e-01
-2.40295306e-01 6.84839666e-01 -9.60772336e-02 2.70785719e-01
6.83669627e-01 -1.23381627e+00 3.49366426e-01 3.75014991e-01
-7.87447765e-02 3.12355399e-01 1.54289201e-01 -4.22127068e-01
-7.29134023e-01 -1.27367771e+00 6.25905275e-01 -1.71380624e-01
7.03413963e-01 -3.31802249e-01 -4.64431882e-01 4.03480709e-01
5.35950303e-01 6.79844767e-02 9.30958152e-01 3.17644149e-01
-6.13586903e-01 -3.51620644e-01 -1.20589900e+00 3.38006079e-01
1.23262107e+00 -4.93352622e-01 -1.08372256e-01 5.15863061e-01
6.76559567e-01 -5.20597324e-02 -8.60357821e-01 2.14090943e-01
4.63289320e-01 -1.05891180e+00 1.06441092e+00 -5.13174117e-01
1.17173445e+00 4.63593565e-02 -1.99643791e-01 -1.41349804e+00
-7.37693012e-01 -1.85211286e-01 3.03727001e-01 1.34678733e+00
4.12006319e-01 -4.01497990e-01 7.70737827e-01 4.87266183e-01
-7.11751357e-02 -1.09286869e+00 -6.61381841e-01 -6.59787118e-01
-1.38558866e-02 -4.50807393e-01 7.73163438e-01 1.00451672e+00
-2.32897058e-01 5.93797445e-01 -4.51817751e-01 -3.05244714e-01
2.14067340e-01 4.84646171e-01 7.11304963e-01 -1.36057425e+00
-1.61794633e-01 -7.67163813e-01 -6.57285631e-01 -7.04043806e-01
-1.59338593e-01 -9.25370693e-01 -1.35786518e-01 -1.58807409e+00
5.26953459e-01 -5.93224883e-01 -8.08615506e-01 7.41235256e-01
-1.77524518e-02 8.27975154e-01 1.09916285e-01 2.72663981e-01
-1.06726515e+00 7.96889544e-01 1.48729694e+00 -5.97603619e-01
-1.40082687e-02 -1.80972874e-01 -1.15807891e+00 7.43025422e-01
1.04841340e+00 -2.27200106e-01 -4.12321925e-01 -5.13557613e-01
4.97006923e-01 -6.81162417e-01 3.76713008e-01 -1.10987699e+00
-2.90324725e-02 -2.67951995e-01 6.73379838e-01 -2.55499899e-01
3.83572757e-01 -7.79413402e-01 4.79289666e-02 7.62965456e-02
-3.87665272e-01 2.22368747e-01 1.54561207e-01 4.32778597e-01
-3.27822864e-01 2.34246887e-02 9.71096396e-01 -3.98026854e-01
-1.09785295e+00 6.15627110e-01 2.64692843e-01 -1.37087151e-01
1.04673564e+00 -1.69260040e-01 -4.56570357e-01 -5.21282434e-01
-5.90242326e-01 6.31754175e-02 6.10324919e-01 6.64479017e-01
7.46718109e-01 -1.95105696e+00 -2.34352872e-01 2.03444343e-02
6.10078216e-01 -2.14710429e-01 2.74792761e-01 4.83930260e-01
-7.76476502e-01 -7.36887529e-02 -7.77526140e-01 -4.00032401e-01
-1.17038798e+00 6.73811972e-01 3.57450455e-01 -7.61210844e-02
-3.81739825e-01 1.09322429e+00 4.78894144e-01 -6.28408968e-01
7.98757225e-02 5.09821065e-02 -3.81433398e-01 -3.86536717e-02
3.20860267e-01 -1.30719528e-01 -7.14231879e-02 -7.24851489e-01
-1.78239614e-01 7.87306607e-01 1.99488789e-01 2.63835073e-01
1.60471988e+00 -7.86128044e-02 -3.49977404e-01 2.19501957e-01
1.47086763e+00 -3.34642559e-01 -8.83751750e-01 -5.18497117e-02
-4.94216114e-01 -7.12132931e-01 1.96544006e-01 -1.20464897e+00
-1.78350174e+00 9.99810457e-01 9.13409114e-01 2.17962563e-01
1.67193604e+00 6.78354800e-02 9.55460548e-01 2.94344604e-01
1.51207119e-01 -1.43360913e+00 7.69322455e-01 2.15070218e-01
1.13348389e+00 -1.36716354e+00 -1.64827809e-01 -2.16929615e-01
-8.08236301e-01 1.07031965e+00 1.03585052e+00 -4.31354940e-01
6.42313004e-01 -3.14754546e-01 2.53596693e-01 -4.28020537e-01
-3.55554014e-01 -2.17304870e-01 4.19314951e-01 5.25032878e-01
1.21350124e-01 9.08758044e-02 -1.92442730e-01 1.04553521e+00
-5.42155325e-01 3.03116571e-02 1.46539837e-01 5.91983318e-01
-4.52781886e-01 -9.45650935e-01 -2.98178680e-02 4.74859774e-01
-3.54024261e-01 -2.23097950e-01 -7.63246775e-01 7.66265810e-01
5.81221044e-01 7.32103646e-01 9.53789055e-02 -1.04043126e+00
4.07796949e-01 -1.99922413e-01 -9.72025562e-04 -2.18145609e-01
-6.11007631e-01 -1.49272636e-01 -1.63758341e-02 -8.55111122e-01
-3.21739078e-01 -1.49129719e-01 -6.44316554e-01 -6.56947017e-01
-3.00605506e-01 1.60631463e-02 7.80581653e-01 5.32302618e-01
4.40546364e-01 6.52823269e-01 7.33847380e-01 -6.74600720e-01
9.81375575e-02 -8.28290999e-01 -7.17614889e-01 7.53146768e-01
-4.58808206e-02 -4.90523189e-01 -1.41686022e-01 -1.87513558e-03] | [11.504947662353516, -1.052346110343933] |
9101afd7-701a-4500-bb92-63fc8610d036 | efficient-anomaly-detection-with-budget | 2306.03492 | null | https://arxiv.org/abs/2306.03492v1 | https://arxiv.org/pdf/2306.03492v1.pdf | Efficient Anomaly Detection with Budget Annotation Using Semi-Supervised Residual Transformer | Anomaly Detection is challenging as usually only the normal samples are seen during training and the detector needs to discover anomalies on-the-fly. The recently proposed deep-learning-based approaches could somehow alleviate the problem but there is still a long way to go in obtaining an industrial-class anomaly detector for real-world applications. On the other hand, in some particular AD tasks, a few anomalous samples are labeled manually for achieving higher accuracy. However, this performance gain is at the cost of considerable annotation efforts, which can be intractable in many practical scenarios. In this work, the above two problems are addressed in a unified framework. Firstly, inspired by the success of the patch-matching-based AD algorithms, we train a sliding vision transformer over the residuals generated by a novel position-constrained patch-matching. Secondly, the conventional pixel-wise segmentation problem is cast into a block-wise classification problem. Thus the sliding transformer can attain even higher accuracy with much less annotation labor. Thirdly, to further reduce the labeling cost, we propose to label the anomalous regions using only bounding boxes. The unlabeled regions caused by the weak labels are effectively exploited using a highly-customized semi-supervised learning scheme equipped with two novel data augmentation methods. The proposed method outperforms all the state-of-the-art approaches using all the evaluation metrics in both the unsupervised and supervised scenarios. On the popular MVTec-AD dataset, our SemiREST algorithm obtains the Average Precision (AP) of 81.2% in the unsupervised condition and 84.4% AP for supervised anomaly detection. Surprisingly, with the bounding-box-based semi-supervisions, SemiREST still outperforms the SOTA methods with full supervision (83.8% AP) on MVTec-AD. | ['Chunhua Shen', 'Mingwen Wang', 'Hao Chen', 'Jingqi Wu', 'Hanxi Li'] | 2023-06-06 | null | null | null | null | ['supervised-anomaly-detection', 'unsupervised-anomaly-detection'] | ['computer-vision', 'methodology'] | [ 3.97955775e-01 1.32194057e-01 1.27322719e-01 -3.22254449e-01
-9.10362065e-01 -1.81144387e-01 4.31564450e-01 2.40472749e-01
-3.41960460e-01 4.47467357e-01 -6.86521113e-01 -1.37423500e-01
-3.48724760e-02 -6.34388149e-01 -5.22701502e-01 -1.13559484e+00
2.13090181e-01 3.95548254e-01 6.84117556e-01 -4.00042944e-02
1.37302056e-01 5.39711177e-01 -1.83395386e+00 3.51322964e-02
1.28689837e+00 1.51713419e+00 1.57300346e-02 2.62300432e-01
-2.20270813e-01 4.64024544e-01 -4.77972537e-01 -2.74868459e-01
4.92315859e-01 -3.32811236e-01 -4.72615212e-01 5.29081702e-01
7.25207508e-01 -3.20229828e-01 -8.31319094e-02 1.23097539e+00
2.39357471e-01 5.95787391e-02 4.10157710e-01 -1.24085951e+00
-8.26045796e-02 1.83338244e-02 -1.04689825e+00 2.16128066e-01
-5.94055913e-02 1.65233627e-01 8.89484406e-01 -8.50331366e-01
3.00152570e-01 5.89013159e-01 4.90913182e-01 2.99056709e-01
-1.08289754e+00 -4.46522743e-01 3.38128835e-01 3.81300658e-01
-1.21491289e+00 -1.91933170e-01 1.01893282e+00 -4.56203431e-01
5.83991945e-01 2.93325990e-01 4.94848698e-01 8.72387171e-01
-1.66303739e-01 8.35489511e-01 9.21676457e-01 -3.51345927e-01
2.78452575e-01 5.75494319e-02 2.41891950e-01 6.96124554e-01
2.67710268e-01 -4.37481068e-02 -1.37855947e-01 -1.27463162e-01
4.60736275e-01 3.19399953e-01 -3.42977166e-01 -4.74662811e-01
-9.45302784e-01 6.08963907e-01 3.27314615e-01 4.25900489e-01
-3.75401676e-01 -3.67519647e-01 5.27837515e-01 2.66657889e-01
7.03216136e-01 3.90533715e-01 -4.76429492e-01 1.08001404e-01
-1.31819427e+00 6.54205754e-02 2.98638910e-01 6.68041825e-01
7.99480975e-01 2.82114148e-01 -1.04687341e-01 9.57061052e-01
2.28520259e-01 2.50148982e-01 3.77893895e-01 -5.77559471e-01
3.60371172e-01 8.29627633e-01 1.64169129e-02 -9.82539952e-01
-3.57067764e-01 -8.18327308e-01 -9.80910540e-01 3.61795217e-01
8.45475614e-01 1.28901586e-01 -1.18092442e+00 1.33532643e+00
4.90915745e-01 3.25406402e-01 -1.27735704e-01 9.00951326e-01
2.36768469e-01 5.38132608e-01 -1.37738705e-01 -4.59793121e-01
1.14670789e+00 -1.16774416e+00 -7.06544161e-01 -2.41724312e-01
7.64726579e-01 -6.90854013e-01 1.10692906e+00 7.70908177e-01
-7.13551521e-01 -4.51208979e-01 -1.20462143e+00 4.42337811e-01
-2.16432393e-01 3.25320274e-01 2.71011353e-01 5.07762969e-01
-6.31551683e-01 5.48209131e-01 -9.98016715e-01 -4.14384007e-01
6.14641070e-01 1.63966984e-01 -3.44240040e-01 -1.66198939e-01
-7.75013745e-01 4.67112750e-01 2.53851891e-01 3.76879245e-01
-7.81824708e-01 -4.68516827e-01 -6.98519289e-01 -8.66764784e-02
8.26147079e-01 1.97199695e-02 9.82583940e-01 -1.10812259e+00
-1.21693003e+00 9.01081860e-01 -2.32482352e-03 -5.44416606e-01
7.65006483e-01 -4.03228790e-01 -5.71070850e-01 2.71315336e-01
1.96748033e-01 2.21222833e-01 1.14985001e+00 -1.17604280e+00
-9.34287131e-01 -5.95622778e-01 -2.53998399e-01 -1.02867559e-01
-4.81666088e-01 -2.19363123e-01 -6.09213412e-01 -7.62913287e-01
5.93948424e-01 -9.28357124e-01 -3.55679780e-01 1.64378792e-01
-5.59901297e-01 -1.31444782e-01 1.31400108e+00 -5.91554046e-01
1.14852715e+00 -2.40266299e+00 -1.58335000e-01 3.98607433e-01
1.44866318e-01 6.00643754e-01 2.84300834e-01 -1.21388055e-01
-1.59907639e-01 -4.05626774e-01 -7.75520325e-01 -3.90203387e-01
-3.07023227e-01 3.21013153e-01 -3.52574855e-01 6.13856852e-01
2.87151814e-01 4.03153062e-01 -7.42327154e-01 -5.44227362e-01
3.55883569e-01 -1.69411730e-02 -4.15769458e-01 2.37284437e-01
-2.99433500e-01 6.54322803e-01 -4.06861752e-01 9.72440660e-01
8.76635909e-01 -2.17317194e-02 -2.23648369e-01 -5.29534975e-03
-4.81659621e-02 -3.13228607e-01 -1.24562716e+00 1.71221936e+00
-1.16284564e-01 5.05333543e-01 6.38581393e-03 -1.51681519e+00
1.05780411e+00 1.77917853e-01 6.16917610e-01 -6.39506102e-01
-1.05218627e-01 6.68504953e-01 -3.88562679e-02 -5.47513008e-01
2.16942966e-01 2.37089284e-02 8.85927379e-02 1.82831913e-01
-6.97138309e-02 2.16580167e-01 5.67068942e-02 -4.28693704e-02
1.28084946e+00 2.44139925e-01 1.51031002e-01 -1.76630482e-01
8.99827242e-01 1.59502164e-01 8.51817131e-01 7.25428104e-01
-3.76315624e-01 9.20323133e-01 4.54586297e-01 -4.88632172e-01
-8.14677835e-01 -9.41197813e-01 -4.08300042e-01 6.86330080e-01
1.64154887e-01 -9.44734365e-02 -8.71119916e-01 -1.15525746e+00
-1.26664817e-01 6.23023272e-01 -3.10326457e-01 -1.73634425e-01
-5.88431776e-01 -9.87664044e-01 5.83355129e-01 6.08869970e-01
8.17029715e-01 -9.29171979e-01 -5.84449708e-01 1.40641630e-01
-1.13107795e-02 -1.25128412e+00 -9.50586796e-02 2.04255328e-01
-1.00225055e+00 -1.08875811e+00 -7.82790244e-01 -5.61113775e-01
9.12336171e-01 1.41710758e-01 6.59353137e-01 2.01760948e-01
-3.20339113e-01 9.04877484e-02 -5.30492246e-01 -2.27924004e-01
-9.38156769e-02 -1.32788613e-01 3.36015075e-02 5.39558709e-01
3.92957002e-01 -6.78283930e-01 -5.95198095e-01 5.87998986e-01
-1.00966907e+00 -3.08047503e-01 7.60038376e-01 9.67259169e-01
9.54040468e-01 3.17505240e-01 6.56882167e-01 -9.16332424e-01
-7.07559884e-02 -2.26844504e-01 -7.67427623e-01 2.96367239e-02
-6.87750578e-01 -1.27936304e-01 7.81975985e-01 -4.03005093e-01
-1.12180912e+00 3.84035885e-01 -2.84720361e-01 -6.23119175e-01
-5.37420630e-01 7.38064200e-02 -4.46826011e-01 -3.58454697e-02
7.10720122e-01 3.88688087e-01 -1.61623850e-01 -6.83591485e-01
-2.33283527e-02 6.32458448e-01 7.35802889e-01 -4.33157742e-01
9.85992849e-01 6.83213770e-01 4.97645624e-02 -9.52621996e-01
-1.10281456e+00 -6.19575918e-01 -7.53452659e-01 -1.69396117e-01
8.62949789e-01 -6.36620343e-01 -7.67881982e-03 9.05725121e-01
-8.01783502e-01 -1.76290378e-01 -5.25650442e-01 2.82243937e-01
-3.04207623e-01 7.98545241e-01 -3.06441784e-01 -8.90269220e-01
-2.79197603e-01 -1.19817841e+00 1.13101912e+00 1.19808502e-01
1.10165127e-01 -4.88908201e-01 -1.50590479e-01 3.58416706e-01
1.46296933e-01 3.74256313e-01 7.80253351e-01 -1.04685986e+00
-5.00832915e-01 -5.00783384e-01 -3.99151802e-01 7.16309190e-01
7.65348375e-02 -1.18913271e-01 -1.25447929e+00 -1.64421409e-01
2.60002494e-01 2.64620557e-02 8.46639812e-01 1.90105900e-01
1.50387371e+00 4.53555137e-02 -2.40806282e-01 5.31577945e-01
1.30652189e+00 2.91314185e-01 7.38293409e-01 5.65438390e-01
8.56478274e-01 6.30537629e-01 1.16401792e+00 3.87201041e-01
-1.09356478e-01 8.14110398e-01 7.39690363e-01 -1.62330434e-01
1.48826644e-01 1.24559499e-01 2.38809720e-01 3.31245631e-01
-8.11760724e-02 -2.02762991e-01 -9.39016163e-01 6.64451003e-01
-2.09575963e+00 -6.94347084e-01 -5.62649131e-01 2.35928512e+00
4.36492085e-01 4.43831950e-01 8.79742578e-02 7.50595868e-01
6.58521295e-01 1.82895273e-01 -5.56100428e-01 -5.40470369e-02
-8.79968926e-02 5.23173921e-02 4.08495128e-01 -2.10110769e-02
-1.42440522e+00 6.82339907e-01 4.74076509e+00 9.74486291e-01
-1.11884320e+00 1.24786027e-01 6.48108721e-01 5.66651374e-02
3.22280735e-01 9.65859089e-03 -6.93862855e-01 5.63454330e-01
6.34437323e-01 5.32029510e-01 -1.54550374e-01 1.08845937e+00
1.10470422e-01 -3.83309245e-01 -8.85252357e-01 9.88917768e-01
9.40530971e-02 -8.13877285e-01 -2.15696514e-01 1.64179221e-01
7.09896624e-01 -1.62767038e-01 -1.93090327e-02 1.39248937e-01
-5.20119488e-01 -5.22997737e-01 4.68253821e-01 3.39603335e-01
6.82171226e-01 -7.13831186e-01 1.06687784e+00 4.74527806e-01
-9.02765274e-01 -2.33234704e-01 -2.89750546e-01 1.33638963e-01
1.58854797e-01 1.17850673e+00 -4.05971825e-01 7.22795904e-01
9.31812048e-01 7.05649555e-01 -6.43515229e-01 1.10763586e+00
-2.12884858e-01 7.84333169e-01 -5.18442333e-01 6.12562120e-01
4.26575571e-01 -3.97066802e-01 8.34434330e-01 9.03430939e-01
4.02047157e-01 -1.64688483e-01 1.51801020e-01 6.54368579e-01
2.57362723e-01 2.01504230e-01 -4.63228673e-01 2.39038512e-01
-6.27661422e-02 1.39783943e+00 -8.20740163e-01 -2.33885944e-01
-5.67361414e-01 1.14897418e+00 4.52542631e-03 2.14094535e-01
-8.52223992e-01 -4.29595679e-01 3.51984262e-01 4.28197592e-01
4.44270313e-01 7.11922301e-03 -3.77024561e-01 -1.22056043e+00
5.34682810e-01 -8.36916983e-01 4.98131543e-01 -3.76343071e-01
-1.25608099e+00 6.67065442e-01 -1.49231613e-01 -1.54738343e+00
-1.01931311e-01 -6.68762922e-01 -8.81600320e-01 3.36039811e-01
-1.41886091e+00 -1.01094353e+00 -6.56027555e-01 5.62553048e-01
6.17744386e-01 -2.28628263e-01 6.58366740e-01 6.86626256e-01
-1.07537806e+00 6.91511393e-01 9.85767394e-02 1.64958566e-01
8.24652255e-01 -1.31884074e+00 5.98300658e-02 1.44724703e+00
1.69147491e-01 2.68426705e-02 6.55631125e-01 -5.61255872e-01
-7.61384785e-01 -1.15599704e+00 3.92856598e-01 -1.96721375e-01
6.13609135e-01 -1.59007505e-01 -1.49210024e+00 4.97499764e-01
-1.92363441e-01 4.71640199e-01 3.31694841e-01 -2.27128759e-01
-1.23333476e-01 -3.61004651e-01 -1.31834519e+00 4.01083171e-01
1.00184131e+00 -1.78763613e-01 -4.74824905e-01 3.54360789e-01
4.51150209e-01 -2.45598450e-01 -6.02793753e-01 6.63379252e-01
1.23660997e-01 -1.13352036e+00 6.73649073e-01 -1.97085530e-01
1.90630630e-01 -6.87805235e-01 -2.23326162e-01 -9.62713063e-01
2.46162191e-01 -3.90095025e-01 -3.02384406e-01 1.49402630e+00
2.93946743e-01 -8.55503857e-01 9.23041582e-01 3.70288253e-01
-4.13730890e-01 -9.18602705e-01 -9.91335928e-01 -7.96488822e-01
-3.36935610e-01 -6.49738848e-01 3.66415411e-01 8.17697346e-01
-4.07456219e-01 -8.69174525e-02 -2.40447864e-01 3.63592356e-01
7.67163873e-01 9.73776281e-02 6.78871512e-01 -1.55228984e+00
-2.60125279e-01 -1.07464001e-01 -6.90612435e-01 -9.93256211e-01
-8.03445205e-02 -4.65618908e-01 1.49375811e-01 -9.60274696e-01
-1.36217818e-01 -6.67085111e-01 -4.12485868e-01 5.42335153e-01
-1.37963638e-01 3.96716118e-01 -2.34330416e-01 3.85080159e-01
-6.21344388e-01 5.17174900e-01 8.72997463e-01 -1.88629348e-02
-3.03865045e-01 3.19607705e-01 -3.61024290e-01 1.09831798e+00
8.63919199e-01 -4.48585212e-01 -3.42249602e-01 -5.67919016e-02
-6.78190067e-02 -2.72283614e-01 4.08184826e-01 -1.22163570e+00
2.08595321e-01 1.20156646e-01 2.27329776e-01 -9.07866836e-01
1.41401842e-01 -1.10730994e+00 -2.93343484e-01 2.45662913e-01
1.93748295e-01 -2.12104931e-01 1.60993963e-01 8.84316623e-01
-5.05503416e-01 -2.71973163e-01 1.06002390e+00 1.77653462e-01
-8.46187949e-01 2.86541164e-01 -2.73205370e-01 -2.52307225e-02
1.32497323e+00 -3.26365620e-01 -1.44954771e-01 -1.23789199e-01
-7.30489194e-01 1.29491091e-01 4.83943492e-01 1.21812016e-01
4.64495093e-01 -1.10807049e+00 -3.18849862e-01 5.10748625e-01
4.36913520e-01 5.99342525e-01 4.40487891e-01 1.31304824e+00
-4.57303911e-01 1.47388533e-01 -1.46976531e-01 -1.08782196e+00
-1.11448944e+00 5.38124502e-01 4.25270677e-01 -4.12169844e-01
-9.42944288e-01 5.11807263e-01 2.44874775e-01 -2.65127182e-01
2.94367343e-01 -1.28484935e-01 -1.55973807e-01 9.64882672e-02
4.40676957e-01 5.37076771e-01 4.13403958e-01 -6.34007692e-01
-2.45727077e-01 6.70558691e-01 -2.54021436e-01 2.27325916e-01
1.29929769e+00 -5.58144487e-02 1.91582385e-02 1.96503162e-01
8.93636882e-01 -3.99504527e-02 -1.32612264e+00 -4.54311818e-01
2.95891196e-01 -5.53341448e-01 1.22903302e-01 -5.01643896e-01
-1.48245084e+00 9.97202516e-01 8.65543365e-01 1.92430168e-01
1.47775996e+00 -7.69826323e-02 9.49755430e-01 2.71888822e-01
2.21052945e-01 -1.16195643e+00 1.04255892e-01 8.43579918e-02
5.62406063e-01 -1.42389822e+00 -2.77588442e-02 -5.71133554e-01
-6.70181870e-01 1.00270915e+00 9.47122633e-01 -1.54201135e-01
4.81860548e-01 9.28392448e-03 8.45654383e-02 -2.58806914e-01
-2.15060517e-01 -1.94978759e-01 4.49339181e-01 3.78010064e-01
-1.03005655e-01 -2.23674312e-01 -1.64564490e-01 5.85564196e-01
2.96091467e-01 -4.40468878e-01 3.41627389e-01 8.13871741e-01
-4.93652165e-01 -1.00602603e+00 -5.09781539e-01 6.54149950e-01
-5.86819649e-01 2.83771306e-01 -1.52123928e-01 7.44040072e-01
2.68101990e-01 8.86928320e-01 1.46889597e-01 -2.96273828e-01
4.59999263e-01 2.62512982e-01 1.46055510e-02 -4.98880893e-01
-2.55210876e-01 2.89331943e-01 -1.62560269e-01 -7.78945208e-01
-3.82798582e-01 -7.74396658e-01 -1.33214271e+00 1.56863675e-01
-6.02623761e-01 -5.85584268e-02 3.69654626e-01 1.12690008e+00
1.90821156e-01 5.56575954e-01 5.85383654e-01 -7.24574804e-01
-5.17921925e-01 -9.10176516e-01 -8.18107069e-01 4.71551120e-01
3.33252251e-01 -8.13544095e-01 -7.25727618e-01 -2.37641618e-01] | [7.626263618469238, 2.0777740478515625] |
bb9aad67-db13-4218-b5ac-43d3d5036fed | multi-scale-prototypical-transformer-for | 2307.02308 | null | https://arxiv.org/abs/2307.02308v1 | https://arxiv.org/pdf/2307.02308v1.pdf | Multi-Scale Prototypical Transformer for Whole Slide Image Classification | Whole slide image (WSI) classification is an essential task in computational pathology. Despite the recent advances in multiple instance learning (MIL) for WSI classification, accurate classification of WSIs remains challenging due to the extreme imbalance between the positive and negative instances in bags, and the complicated pre-processing to fuse multi-scale information of WSI. To this end, we propose a novel multi-scale prototypical Transformer (MSPT) for WSI classification, which includes a prototypical Transformer (PT) module and a multi-scale feature fusion module (MFFM). The PT is developed to reduce redundant instances in bags by integrating prototypical learning into the Transformer architecture. It substitutes all instances with cluster prototypes, which are then re-calibrated through the self-attention mechanism of the Trans-former. Thereafter, an MFFM is proposed to fuse the clustered prototypes of different scales, which employs MLP-Mixer to enhance the information communication between prototypes. The experimental results on two public WSI datasets demonstrate that the proposed MSPT outperforms all the compared algorithms, suggesting its potential applications. | ['Jun Shi', 'Juncheng Li', 'Jun Wang', 'Saisai Ding'] | 2023-07-05 | null | null | null | null | ['classification-1', 'multiple-instance-learning'] | ['methodology', 'methodology'] | [ 2.92386889e-01 -8.63205492e-02 -7.33595863e-02 -3.11585397e-01
-1.14686203e+00 4.37339582e-02 3.71256053e-01 3.92525345e-01
-2.91280270e-01 6.11146450e-01 6.71394691e-02 1.07754529e-01
-4.84106421e-01 -6.04591191e-01 -5.76368570e-01 -1.28819025e+00
2.10129961e-01 4.77796316e-01 3.16968948e-01 -1.33833051e-01
9.30689648e-02 3.55505824e-01 -1.49346685e+00 8.18531752e-01
1.06474495e+00 1.28082550e+00 3.47627014e-01 3.84547859e-01
-5.22125542e-01 7.04898775e-01 -6.53913677e-01 -2.85706013e-01
-1.82279184e-01 -1.40938535e-01 -6.10349774e-01 1.76699057e-01
8.05663597e-03 2.16709226e-01 -2.91633606e-02 9.06779230e-01
6.93130851e-01 -1.23996042e-01 5.50651610e-01 -1.41443563e+00
-3.05701971e-01 5.33951819e-01 -9.26946223e-01 2.37513438e-01
-2.72579074e-01 -6.29767356e-03 8.69966507e-01 -8.42627347e-01
4.34727877e-01 1.30320370e+00 7.03688323e-01 1.72303006e-01
-9.52011824e-01 -8.84345472e-01 2.14406982e-01 6.36348546e-01
-1.57855749e+00 -3.08858395e-01 9.11160111e-01 -1.75488412e-01
7.59604812e-01 5.61182380e-01 7.53289342e-01 5.65078259e-01
2.11453781e-01 1.04153562e+00 1.22570384e+00 -3.20231915e-01
-7.49979615e-02 5.25667846e-01 4.49192375e-01 4.22972381e-01
-3.42394970e-03 -4.35603917e-01 -3.06691855e-01 -7.94668049e-02
4.44233000e-01 3.40842932e-01 -2.41667524e-01 -5.61402142e-02
-1.20736456e+00 5.42706370e-01 8.02622080e-01 6.27856374e-01
-2.49280632e-01 -4.02729154e-01 5.79371095e-01 8.94833878e-02
6.17877543e-01 1.08797692e-01 -3.78979653e-01 5.14318049e-01
-8.21611941e-01 -1.11092940e-01 3.28706980e-01 6.13436580e-01
4.98134285e-01 -5.21912277e-01 -4.36356544e-01 1.27402687e+00
2.21461743e-01 1.32906660e-01 1.03292382e+00 -1.67909458e-01
4.93553400e-01 1.29872417e+00 -3.78646165e-01 -1.07030189e+00
-6.32786930e-01 -8.75930429e-01 -1.07058942e+00 -3.13348472e-01
1.38430037e-02 3.91461611e-01 -7.86800742e-01 1.29746902e+00
7.07492352e-01 6.91730559e-01 1.91084296e-02 8.83987606e-01
1.28833675e+00 5.80974460e-01 2.25854456e-01 -2.82461882e-01
1.41256082e+00 -9.50256348e-01 -5.19865274e-01 4.56426963e-02
5.50012290e-01 -5.70626974e-01 7.98980117e-01 1.12763219e-01
-8.21789563e-01 -5.95657349e-01 -1.06698966e+00 -3.42307379e-03
-5.56773901e-01 5.01259089e-01 3.28900725e-01 1.27938107e-01
-7.55573094e-01 2.34806865e-01 -6.12965286e-01 -3.31056148e-01
7.51162767e-01 5.30836582e-01 -5.05607903e-01 -3.44680659e-02
-9.98634994e-01 6.57282889e-01 4.80038196e-01 4.48709965e-01
-3.91191840e-01 -9.80405390e-01 -6.15076959e-01 1.71242639e-01
2.18352318e-01 -3.98462921e-01 7.18902051e-01 -9.17749465e-01
-1.17836845e+00 8.94246221e-01 -1.66554242e-01 -2.66348571e-02
3.42189223e-01 4.53773230e-01 -5.16227663e-01 1.03519171e-01
5.10364212e-03 3.89902115e-01 7.18850136e-01 -1.35607386e+00
-1.05874372e+00 -7.80615747e-01 -2.14580283e-01 4.48979855e-01
-6.28042638e-01 -1.30012721e-01 -4.74264175e-01 -4.85929459e-01
3.43733460e-01 -7.59939551e-01 -8.13829899e-02 -2.87185460e-01
-6.73050523e-01 -4.53172863e-01 7.68547595e-01 -7.71612883e-01
1.27958143e+00 -2.24549651e+00 3.70970309e-01 3.70312691e-01
3.40838820e-01 4.58494425e-01 -1.97138101e-01 4.72052135e-02
-4.02961582e-01 -1.68084949e-01 -2.17735618e-02 -5.33952296e-01
-3.42548907e-01 1.77490577e-01 9.36006084e-02 3.84508431e-01
3.88010025e-01 7.87092090e-01 -7.41441429e-01 -1.00231421e+00
3.63260835e-01 4.22418863e-01 -2.06171960e-01 3.12595278e-01
2.41741478e-01 4.05050367e-01 -2.37913832e-01 9.19127405e-01
7.78476477e-01 -6.64058208e-01 6.83161542e-02 -7.03473389e-01
9.96590629e-02 -2.24089786e-01 -1.22859085e+00 1.40718544e+00
-2.55105406e-01 -8.15101489e-02 1.87030748e-01 -1.55030680e+00
7.74941742e-01 2.59417742e-01 6.88749015e-01 -5.37971318e-01
1.75535724e-01 2.76441813e-01 -5.85496649e-02 -7.82783687e-01
-6.66964203e-02 -2.63005823e-01 7.30389059e-02 -1.44848734e-01
1.89515218e-01 2.48681054e-01 2.17550948e-01 1.81669280e-01
9.12314713e-01 -3.03761512e-01 3.20162326e-01 -1.74691796e-01
1.08098292e+00 3.53436805e-02 9.16407287e-01 2.87609369e-01
-2.43244529e-01 2.90560514e-01 3.28617275e-01 -3.87976229e-01
-7.79007673e-01 -9.01103556e-01 -4.07469809e-01 8.64444435e-01
3.77753705e-01 -3.85662448e-03 -4.27757561e-01 -7.45585322e-01
2.34441966e-01 6.38339967e-02 -8.50152016e-01 -2.68347949e-01
-3.65829200e-01 -1.38117146e+00 1.69131398e-01 5.80761969e-01
4.47341621e-01 -8.07245612e-01 -2.47732433e-03 2.52177417e-01
-4.72123742e-01 -6.63540781e-01 -3.32967132e-01 1.27530545e-01
-6.84375048e-01 -1.20340598e+00 -7.50474691e-01 -9.91576612e-01
9.51349020e-01 5.79593897e-01 6.33438945e-01 2.61644453e-01
-8.03571165e-01 -4.60440665e-03 -3.27163845e-01 -6.11969054e-01
-2.72298694e-01 6.73974156e-02 -3.06132883e-01 5.63441098e-01
4.95478511e-01 -4.06700850e-01 -6.34708226e-01 2.27580935e-01
-9.44349587e-01 2.97051549e-01 1.01391542e+00 1.26507068e+00
7.84806728e-01 1.46987483e-01 7.69126236e-01 -8.31050038e-01
3.66573364e-01 -5.92024565e-01 -2.10477218e-01 7.61756182e-01
-2.97255218e-01 -3.77629340e-01 5.57851136e-01 -4.94866133e-01
-1.15074062e+00 -2.92446911e-01 -1.20597132e-01 -3.26149583e-01
-4.59323376e-02 5.69965065e-01 -3.08497638e-01 -2.23177880e-01
2.64363706e-01 4.26890403e-01 2.92366952e-01 -1.99813068e-01
-1.88352048e-01 9.92497861e-01 2.14265838e-01 -1.14512593e-01
4.35673445e-01 3.93368393e-01 -1.62126482e-01 -7.33279705e-01
-9.32224274e-01 -7.37811744e-01 -5.01809776e-01 -2.45992124e-01
6.30746722e-01 -9.59757507e-01 -7.68959701e-01 6.98793411e-01
-8.55189562e-01 1.47543341e-01 -2.83930123e-01 3.40864748e-01
-1.67557448e-01 1.76779002e-01 -9.11806822e-01 -5.87316394e-01
-6.45838737e-01 -1.39204013e+00 1.28117836e+00 6.81927860e-01
1.86623752e-01 -6.92291558e-01 -4.92533185e-02 7.62152016e-01
2.74751097e-01 1.26425862e-01 1.02270734e+00 -8.91290009e-01
-2.85117596e-01 -4.24620003e-01 -6.18891954e-01 3.12057763e-01
2.83111900e-01 3.57359881e-03 -9.70647395e-01 -3.74091804e-01
-9.98006761e-02 -2.47830957e-01 9.41501558e-01 4.42671031e-01
1.41558373e+00 -1.13707855e-01 -7.96766341e-01 6.58273697e-01
1.56897223e+00 3.82434279e-01 2.68417835e-01 3.51706058e-01
6.70588613e-01 6.91286623e-01 6.70325220e-01 3.49090785e-01
6.70216441e-01 3.13169658e-01 3.10190111e-01 -4.13759708e-01
-7.05874115e-02 2.33858675e-01 -2.13451147e-01 1.14178264e+00
3.00308578e-02 5.58663234e-02 -7.13215947e-01 5.43450952e-01
-1.82027817e+00 -8.28247368e-01 2.24536136e-01 1.86714411e+00
8.57855499e-01 -4.19366406e-03 -1.30603507e-01 5.55494189e-01
9.10147429e-01 -8.51915255e-02 -6.44629657e-01 2.10546792e-01
-1.55817062e-01 -3.51470858e-02 1.18293881e-01 6.97800368e-02
-1.20163667e+00 3.43918264e-01 4.66383505e+00 1.40346444e+00
-1.14633131e+00 3.07663709e-01 9.94741678e-01 -1.54189110e-01
1.85708720e-02 -4.83810216e-01 -1.01570797e+00 8.53677928e-01
4.06966120e-01 4.31451276e-02 1.34626359e-01 5.80290794e-01
-2.58037541e-02 4.99157608e-02 -5.64645231e-01 8.85654509e-01
1.60531655e-01 -1.38507688e+00 8.55386071e-03 -1.90526336e-01
3.40046257e-01 -1.47546902e-01 -1.53144756e-02 4.50119883e-01
-1.61704004e-01 -5.94802380e-01 3.20306480e-01 7.29132414e-01
5.98630250e-01 -8.13389421e-01 1.16026950e+00 3.16952944e-01
-1.39174294e+00 -6.29008830e-01 -5.01734734e-01 6.53145373e-01
-3.56907278e-01 9.21187878e-01 -9.72427189e-01 9.79579866e-01
7.96901345e-01 7.40173221e-01 -7.69939780e-01 1.25294816e+00
3.52868915e-01 3.81846398e-01 -4.19529766e-01 -2.25125290e-02
-5.34049347e-02 -2.39416897e-01 2.43562743e-01 1.22370255e+00
3.12251329e-01 2.09026426e-01 2.37694204e-01 3.17237705e-01
2.45070830e-01 4.17848557e-01 6.29770532e-02 2.70548284e-01
5.85811853e-01 1.80577755e+00 -8.38719964e-01 -5.95348656e-01
-3.85724276e-01 5.86485147e-01 5.99078298e-01 1.00584231e-01
-7.71364987e-01 -4.64079142e-01 3.04918587e-01 -6.78491816e-02
1.83695570e-01 5.27152002e-01 -1.68040752e-01 -1.10747612e+00
5.95709197e-02 -6.50230587e-01 8.59449446e-01 -5.65728247e-01
-1.66450715e+00 5.57803094e-01 -2.89110810e-01 -1.50640094e+00
5.30551732e-01 -5.12943447e-01 -8.18018854e-01 6.45524263e-01
-1.79108334e+00 -1.74824989e+00 -6.36754215e-01 6.48051977e-01
3.52372676e-01 -1.15119062e-01 6.29545033e-01 5.31219184e-01
-1.21182978e+00 8.44107091e-01 2.92653710e-01 5.69735374e-03
8.98988128e-01 -1.10040843e+00 -4.73672301e-01 3.31987560e-01
-4.33251143e-01 4.41928118e-01 3.79521430e-01 -5.78665555e-01
-1.21223855e+00 -1.29112017e+00 6.22403085e-01 1.44527793e-01
6.01144552e-01 3.12171746e-02 -1.05296481e+00 4.80201066e-01
-2.04985812e-01 3.95569056e-01 1.25863385e+00 -9.95761156e-02
-8.72305855e-02 -7.72568524e-01 -1.28201413e+00 2.34076560e-01
6.25138342e-01 -1.79761514e-01 -3.34348440e-01 6.27734721e-01
6.68470800e-01 -4.75005925e-01 -1.33552849e+00 6.36717916e-01
4.75753069e-01 -7.39223957e-01 1.12712610e+00 -3.94090116e-01
2.06724212e-01 -4.61296320e-01 4.16750424e-02 -1.43342793e+00
-4.81946975e-01 2.94831246e-01 2.22308278e-01 1.57288468e+00
2.21986800e-01 -7.52698481e-01 6.31333232e-01 3.85292709e-01
-3.77260864e-01 -1.14993250e+00 -1.35805655e+00 -4.54902172e-01
-2.51326025e-01 7.34176040e-02 8.02059650e-01 1.02926207e+00
2.67791003e-01 1.02730744e-01 -5.13194799e-02 2.66326189e-01
7.23649383e-01 4.17760015e-01 3.95041108e-01 -1.27360761e+00
-2.68898547e-01 -4.65327471e-01 -5.23983777e-01 -2.27678657e-01
-9.81581062e-02 -1.15400267e+00 2.08401494e-02 -1.39146757e+00
5.75438082e-01 -7.31503248e-01 -8.83269846e-01 4.03149009e-01
-7.08382308e-01 3.90884548e-01 3.47294025e-02 3.60917032e-01
-8.24657857e-01 5.66153407e-01 1.27232122e+00 -4.76958811e-01
-1.88832223e-01 -4.08767611e-02 -8.17525327e-01 5.93749702e-01
6.80096745e-01 -2.21021280e-01 -2.17927650e-01 1.69219822e-02
-2.98896134e-01 1.04004115e-01 2.69436151e-01 -1.24072850e+00
5.09942770e-01 -5.50414845e-02 7.77071178e-01 -9.91093576e-01
4.05783474e-01 -6.97357893e-01 1.22752436e-01 4.91079688e-01
-2.16674834e-01 -4.94902521e-01 1.05239987e-01 5.88422656e-01
-4.23588485e-01 1.86960194e-02 8.70568156e-01 -1.55979171e-01
-4.33376193e-01 5.17732739e-01 1.21652678e-01 -3.25960428e-01
1.46515703e+00 -2.02340826e-01 -5.19940138e-01 4.93229002e-01
-8.69181514e-01 6.89117014e-01 -3.33852544e-02 3.14312935e-01
6.60351634e-01 -1.43124104e+00 -7.55172133e-01 4.15103614e-01
4.25930202e-01 3.11162114e-01 9.86757994e-01 1.33277583e+00
-1.50539517e-01 2.51601607e-01 -1.48555309e-01 -9.37339008e-01
-1.51406789e+00 6.22631907e-01 2.52665073e-01 -8.04816604e-01
-2.91757286e-01 1.15162754e+00 3.41151386e-01 -5.93033552e-01
1.44569263e-01 -1.63804471e-01 -6.02325499e-01 3.77713263e-01
7.53823519e-01 2.31540889e-01 3.69100153e-01 -5.42273760e-01
-3.34330112e-01 6.20076299e-01 -5.97041965e-01 6.03857100e-01
1.48876524e+00 -1.39792547e-01 -5.35890162e-01 4.97799695e-01
1.25780535e+00 -5.38079083e-01 -8.61643434e-01 -6.54286146e-01
-2.32460216e-01 -2.47204021e-01 -1.15541860e-01 -6.85383797e-01
-1.16732037e+00 8.70529234e-01 7.94663906e-01 6.75409064e-02
1.31197357e+00 -8.43209028e-03 9.24833596e-01 5.39513379e-02
1.76674649e-01 -1.08752775e+00 -3.24438438e-02 -4.83510643e-02
7.63236046e-01 -1.33617294e+00 -8.91636387e-02 -4.14563835e-01
-5.01270115e-01 1.00075781e+00 7.58170307e-01 -1.03615738e-01
7.87171781e-01 3.09688658e-01 -9.64736566e-02 4.42266650e-02
-6.30104184e-01 1.26004383e-01 2.63723314e-01 3.17748725e-01
9.78257135e-02 4.97584455e-02 -4.27773893e-01 1.07607996e+00
2.50019312e-01 1.15097627e-01 -8.95401184e-03 7.90383875e-01
-4.67465490e-01 -8.27143192e-01 -5.50405025e-01 1.11150372e+00
-2.40283668e-01 2.31668547e-01 8.03054348e-02 6.41354203e-01
4.83522266e-01 8.71222198e-01 1.53924108e-01 -5.91814160e-01
3.24695021e-01 2.65969569e-03 2.91254759e-01 -2.66675651e-01
-8.39751005e-01 1.96653858e-01 -3.27566832e-01 -4.55248386e-01
-5.96813381e-01 -5.57237566e-01 -1.17160296e+00 9.70890820e-02
-5.99684894e-01 2.13756323e-01 4.84517217e-01 1.12049675e+00
3.77522409e-01 8.57852221e-01 7.89387643e-01 -8.84693384e-01
-5.12826622e-01 -1.13557839e+00 -6.64695501e-01 5.85740447e-01
3.58299226e-01 -7.51389802e-01 -3.81728470e-01 -1.50722653e-01] | [15.091320991516113, -2.8070013523101807] |
83c39907-a4fe-436e-b486-2c23c12441f2 | fine-grained-visual-categorization-via-multi | 1402.0453 | null | http://arxiv.org/abs/1402.0453v2 | http://arxiv.org/pdf/1402.0453v2.pdf | Fine-Grained Visual Categorization via Multi-stage Metric Learning | Fine-grained visual categorization (FGVC) is to categorize objects into
subordinate classes instead of basic classes. One major challenge in FGVC is
the co-occurrence of two issues: 1) many subordinate classes are highly
correlated and are difficult to distinguish, and 2) there exists the large
intra-class variation (e.g., due to object pose). This paper proposes to
explicitly address the above two issues via distance metric learning (DML). DML
addresses the first issue by learning an embedding so that data points from the
same class will be pulled together while those from different classes should be
pushed apart from each other; and it addresses the second issue by allowing the
flexibility that only a portion of the neighbors (not all data points) from the
same class need to be pulled together. However, feature representation of an
image is often high dimensional, and DML is known to have difficulty in dealing
with high dimensional feature vectors since it would require $\mathcal{O}(d^2)$
for storage and $\mathcal{O}(d^3)$ for optimization. To this end, we proposed a
multi-stage metric learning framework that divides the large-scale high
dimensional learning problem to a series of simple subproblems, achieving
$\mathcal{O}(d)$ computational complexity. The empirical study with FVGC
benchmark datasets verifies that our method is both effective and efficient
compared to the state-of-the-art FGVC approaches. | ['Yuanqing Lin', 'Shenghuo Zhu', 'Qi Qian', 'Rong Jin'] | 2014-02-03 | fine-grained-visual-categorization-via-multi-1 | http://openaccess.thecvf.com/content_cvpr_2015/html/Qian_Fine-Grained_Visual_Categorization_2015_CVPR_paper.html | http://openaccess.thecvf.com/content_cvpr_2015/papers/Qian_Fine-Grained_Visual_Categorization_2015_CVPR_paper.pdf | cvpr-2015-6 | ['fine-grained-visual-categorization'] | ['computer-vision'] | [-8.70048851e-02 -3.69627297e-01 6.06552437e-02 -4.22496378e-01
-6.13783896e-01 -5.04851580e-01 2.61977553e-01 3.93285900e-01
-4.26480860e-01 5.00935555e-01 -1.75177559e-01 4.05642949e-02
-5.21517515e-01 -8.95657420e-01 -3.31119925e-01 -8.88904870e-01
-2.74274379e-01 3.16544980e-01 4.63578671e-01 -1.77559648e-02
4.67174351e-01 6.67076468e-01 -1.89975059e+00 1.41678482e-01
9.02557969e-01 1.51112187e+00 2.21027404e-01 3.27232659e-01
-4.28525835e-01 4.36189294e-01 -6.35020852e-01 -1.56170845e-01
3.99040639e-01 -2.55236894e-01 -6.74896121e-01 1.70341939e-01
6.42545938e-01 -5.86286075e-02 -6.47008196e-02 1.24820805e+00
3.87519091e-01 3.26470852e-01 8.61008406e-01 -1.53839302e+00
-7.28034914e-01 -4.31916583e-03 -8.26501310e-01 2.26537347e-01
-1.44455343e-01 -2.43155092e-01 1.15112841e+00 -9.82524574e-01
3.22677225e-01 1.22653031e+00 3.72970164e-01 2.22926542e-01
-1.23175323e+00 -7.05063224e-01 4.87767428e-01 3.85514349e-01
-1.73791981e+00 -8.35897923e-02 8.07923257e-01 -7.95178294e-01
5.68019271e-01 4.12746102e-01 4.11273628e-01 4.35985178e-01
7.93490093e-03 4.52285558e-01 1.05840445e+00 -2.76973546e-01
5.32809496e-01 1.61851019e-01 4.40277398e-01 6.08238876e-01
3.10572624e-01 -4.43870462e-02 -1.64682940e-01 -8.15036595e-02
4.67001319e-01 3.13573241e-01 -1.45537138e-01 -7.25003004e-01
-9.98040080e-01 1.11065078e+00 6.98966920e-01 3.40505302e-01
-3.13834287e-02 -7.73253245e-03 3.68402511e-01 3.42142552e-01
1.60207063e-01 4.29821432e-01 -3.42410207e-01 2.46738177e-02
-7.58123875e-01 1.05363481e-01 5.14616847e-01 1.02975106e+00
1.12322545e+00 -2.53640324e-01 -1.23935662e-01 1.03853083e+00
2.56224424e-01 2.07271934e-01 5.42057216e-01 -6.92051291e-01
6.65017843e-01 9.57663417e-01 -8.58847201e-02 -1.60651517e+00
-3.35710108e-01 -3.97248685e-01 -1.15742838e+00 4.33292121e-01
4.09973741e-01 1.15262002e-01 -5.66851676e-01 1.67162025e+00
3.78277391e-01 1.09461378e-02 -2.34922767e-01 9.51338530e-01
7.83161640e-01 6.34881437e-01 -1.19552426e-01 -2.43588120e-01
1.25642920e+00 -8.63345385e-01 -3.21741402e-01 -2.24820822e-01
4.33148056e-01 -7.36858904e-01 1.12380278e+00 1.46245152e-01
-7.40766525e-01 -8.28389168e-01 -1.29550755e+00 -7.52199888e-02
-5.78665912e-01 4.49382439e-02 6.92227006e-01 4.85435545e-01
-6.99857354e-01 5.28773069e-01 -5.20981014e-01 1.34287570e-02
4.93693978e-01 3.92520577e-01 -3.89854074e-01 -1.04130842e-01
-8.52564573e-01 4.66371536e-01 3.43874216e-01 4.42958549e-02
-5.64524412e-01 -6.35762095e-01 -7.86191583e-01 1.23312145e-01
3.42863262e-01 -4.80321646e-02 5.73425770e-01 -7.16888309e-01
-9.51144278e-01 7.46169508e-01 -1.24024063e-01 5.27873710e-02
4.58068907e-01 8.72071162e-02 -4.41510469e-01 1.59573406e-02
2.09203288e-01 4.71950144e-01 8.37843776e-01 -1.23046756e+00
-8.75860572e-01 -7.28797853e-01 8.21967795e-02 1.88439965e-01
-7.06481159e-01 -1.20855466e-01 -6.15172267e-01 -8.43787491e-01
4.17384386e-01 -9.64278698e-01 -1.86216179e-02 3.42821270e-01
-2.91995078e-01 -4.70902085e-01 1.01657748e+00 -1.58061072e-01
1.49588048e+00 -2.24600601e+00 2.70576239e-01 3.62065643e-01
5.49778283e-01 3.90338361e-01 -7.52221569e-02 1.98988438e-01
1.19409915e-02 9.27114189e-02 -1.02281876e-01 -3.51421162e-02
-4.22044210e-02 1.49507925e-01 -5.73216602e-02 5.36412239e-01
-4.06493992e-02 5.62354088e-01 -7.59528756e-01 -5.33279359e-01
1.89287767e-01 2.75484651e-01 -3.99297118e-01 1.68754116e-01
2.24875659e-01 8.94387141e-02 -5.22953510e-01 6.58169091e-01
1.02926433e+00 -4.36893433e-01 1.65669024e-02 -5.33253372e-01
-9.38172042e-02 -1.68550551e-01 -1.86144662e+00 1.29411483e+00
-1.54433385e-01 3.38862360e-01 -1.65497378e-01 -1.19097698e+00
1.01126075e+00 -1.24497280e-01 5.89961708e-01 -6.41977012e-01
1.66181952e-01 1.16604023e-01 -3.18944044e-02 -2.69375533e-01
2.08857745e-01 -2.27330737e-02 -1.66241646e-01 5.39884984e-01
-6.59129098e-02 3.87295932e-02 2.71349400e-01 -1.30749702e-01
8.01648915e-01 -5.14404297e-01 2.60640919e-01 -4.60239500e-01
7.00582743e-01 -1.99968621e-01 8.84032905e-01 5.33870339e-01
-4.89891708e-01 4.25278604e-01 5.37787914e-01 -4.54897374e-01
-7.16227710e-01 -1.17264009e+00 -2.89996088e-01 1.03159380e+00
6.21891856e-01 -4.91269976e-01 -6.17363155e-01 -8.51119220e-01
2.27258489e-01 2.42507666e-01 -7.93535531e-01 -3.17774057e-01
-4.11914289e-01 -7.09427357e-01 1.39547661e-01 7.10096657e-01
4.25581843e-01 -6.46573961e-01 -5.89873970e-01 7.23957419e-02
8.86785462e-02 -7.46259749e-01 -6.47938967e-01 3.33311141e-01
-8.06709111e-01 -1.25896823e+00 -5.45889735e-01 -9.14645255e-01
8.99738491e-01 7.29747295e-01 8.08618903e-01 6.62840605e-02
-5.41724265e-01 -8.17541480e-02 -4.65994835e-01 -3.14510345e-01
9.30130035e-02 -7.66198039e-02 1.79203600e-01 2.75271714e-01
7.23834991e-01 -3.91633272e-01 -7.12670624e-01 7.88929522e-01
-9.07591581e-01 -2.49208167e-01 4.11138833e-01 9.28235292e-01
1.06281757e+00 5.10292470e-01 4.38294798e-01 -6.17988765e-01
4.26062882e-01 -2.96397775e-01 -7.94230938e-01 3.42853516e-01
-6.83345199e-01 -4.92864400e-02 8.07469487e-01 -5.51956832e-01
-4.84925121e-01 1.03686722e-02 2.92822540e-01 -4.98757601e-01
3.39427292e-02 2.00620845e-01 -3.23647320e-01 -2.28262305e-01
5.18903613e-01 1.45380288e-01 -1.85811836e-02 -5.90462506e-01
2.06804320e-01 8.33592474e-01 2.17943653e-01 -5.48623919e-01
1.02229524e+00 4.02683496e-01 1.11914255e-01 -7.84999192e-01
-8.02599728e-01 -4.26753163e-01 -8.79273474e-01 -4.32422534e-02
8.57367396e-01 -6.69364750e-01 -7.77470827e-01 2.90025562e-01
-5.76040745e-01 1.15662321e-01 -2.21186042e-01 4.45731521e-01
-3.08576554e-01 4.60358173e-01 -2.47607812e-01 -4.90466356e-01
-1.04679331e-01 -1.34106445e+00 7.27707803e-01 2.60096371e-01
-1.36893481e-01 -7.14908063e-01 -2.21862063e-01 3.54581386e-01
2.04143971e-01 2.09695354e-01 1.28363836e+00 -4.61572081e-01
-4.72483367e-01 -2.81540215e-01 -5.16569972e-01 4.99906152e-01
5.58519125e-01 -2.63084490e-02 -6.34033203e-01 -7.12022483e-01
2.26765443e-02 -4.29854751e-01 6.51365161e-01 7.22440183e-02
1.48952019e+00 -1.80260271e-01 -3.01608801e-01 6.92542195e-01
1.46711516e+00 2.71043658e-01 3.59897077e-01 2.48998910e-01
8.54670525e-01 6.11215949e-01 8.43771338e-01 4.44677651e-01
3.49044442e-01 8.70451272e-01 4.46226716e-01 -1.07910007e-01
-8.61708298e-02 -5.24316877e-02 -1.41035840e-01 7.75294662e-01
2.14020431e-01 -1.78499743e-02 -7.21095264e-01 4.35943663e-01
-1.77070487e+00 -7.69039392e-01 -1.22444890e-01 2.37183166e+00
5.62621295e-01 -4.63477857e-02 6.83827773e-02 5.17905891e-01
7.71964610e-01 2.18969733e-01 -6.98046803e-01 -2.98416793e-01
-9.22641009e-02 4.07280922e-02 1.92744061e-01 2.05998316e-01
-1.25297594e+00 6.53722048e-01 5.17150831e+00 1.04117322e+00
-1.15811276e+00 -1.35968059e-01 5.86314678e-01 -1.32071346e-01
2.44650338e-02 -1.98503062e-01 -8.80863965e-01 7.14327931e-01
2.85933942e-01 -1.94913596e-01 3.20772499e-01 8.74856591e-01
-2.17793033e-01 -6.09742887e-02 -1.22524667e+00 1.33305907e+00
7.17549920e-02 -1.18796301e+00 1.27035156e-01 1.45014450e-01
6.46375299e-01 -2.33573988e-01 7.77983963e-02 2.30468452e-01
-1.24923000e-03 -1.12874055e+00 6.43104911e-01 2.64714193e-02
8.71007979e-01 -8.83169532e-01 5.34003437e-01 3.24575931e-01
-1.70249653e+00 -3.67387861e-01 -8.49367440e-01 1.73813730e-01
-2.30551556e-01 6.86456263e-01 -1.39897659e-01 4.85025942e-01
1.02770901e+00 6.49409890e-01 -7.04653680e-01 9.49993849e-01
9.23368558e-02 -5.64070195e-02 -1.08898662e-01 1.19266484e-03
2.88774908e-01 -3.06017846e-01 2.63967037e-01 9.01415229e-01
3.88555497e-01 1.23535804e-01 4.82100129e-01 5.49990177e-01
1.71567872e-02 2.62802720e-01 -2.90441900e-01 1.73713222e-01
6.03381932e-01 1.19227087e+00 -7.23575771e-01 -1.67975441e-01
-5.24123073e-01 9.94104326e-01 5.92183948e-01 8.49495158e-02
-7.25546598e-01 -7.56888092e-01 9.74935412e-01 8.21891725e-02
6.09710217e-01 -2.01451585e-01 -2.88100511e-01 -1.11896205e+00
3.45670879e-01 -7.04358578e-01 7.94161558e-01 -2.53882110e-01
-1.59140956e+00 7.37851322e-01 -2.57061899e-01 -1.70362031e+00
1.33788571e-01 -7.68520415e-01 -1.96403891e-01 9.20659661e-01
-1.50621045e+00 -7.48260796e-01 -7.00270355e-01 8.54054391e-01
3.21795106e-01 -2.04111949e-01 8.96207929e-01 5.28786182e-01
-5.55294573e-01 9.76232767e-01 3.54177475e-01 2.39754871e-01
6.43475413e-01 -1.17453384e+00 -1.81585759e-01 5.33346236e-01
8.63916129e-02 4.58938837e-01 2.42904365e-01 -2.89588094e-01
-1.24604332e+00 -1.27100945e+00 7.86576748e-01 -2.14168176e-01
3.97073269e-01 -4.87262994e-01 -1.01775599e+00 1.57931313e-01
-4.90837902e-01 4.70021158e-01 9.70099330e-01 -5.60154505e-02
-6.91574275e-01 -6.25700474e-01 -1.31476068e+00 4.65274692e-01
9.33947325e-01 -5.30290842e-01 -5.38294733e-01 1.48662642e-01
3.73855650e-01 4.64446060e-02 -1.02517915e+00 4.17295814e-01
4.55164880e-01 -1.08515203e+00 1.02809572e+00 -4.91263241e-01
1.48820505e-01 -6.79669559e-01 -6.40774190e-01 -1.23597431e+00
-5.46586275e-01 -1.59316197e-01 -3.20249677e-01 1.21518743e+00
1.19903937e-01 -6.31203592e-01 6.55061603e-01 5.46603799e-01
1.58394888e-01 -9.47056174e-01 -1.03675699e+00 -1.07023847e+00
2.61836380e-01 -1.85612425e-01 6.75499141e-01 9.77361321e-01
-2.14309245e-01 1.91491127e-01 -1.99919105e-01 1.53615966e-01
6.70927167e-01 5.82429469e-01 6.14175260e-01 -1.60274196e+00
-1.71354204e-01 -5.19791126e-01 -8.43420684e-01 -1.00799894e+00
-1.19110957e-01 -8.13394904e-01 -8.32428187e-02 -1.33810878e+00
3.03678453e-01 -1.03738260e+00 -5.32362819e-01 4.23180312e-01
-2.72679746e-01 2.18272284e-01 3.37585896e-01 3.50369036e-01
-4.24832225e-01 5.48757136e-01 1.15757859e+00 -2.13201836e-01
-1.77968234e-01 1.29489508e-02 -9.32065189e-01 6.51220560e-01
5.84607780e-01 -4.28066343e-01 -6.36476994e-01 -4.95989919e-01
-1.67182997e-01 -1.27232850e-01 1.83182314e-01 -1.07530046e+00
2.58119613e-01 -3.20893735e-01 5.33717811e-01 -6.60247087e-01
3.75321865e-01 -1.02061689e+00 -9.78192613e-02 3.59961957e-01
-6.30901605e-02 1.41446933e-01 -2.14044400e-03 6.31098688e-01
-3.87348562e-01 -8.28220844e-02 1.18968749e+00 7.69632775e-03
-8.27707350e-01 5.89943886e-01 -1.64060798e-02 2.30793521e-01
1.50319421e+00 -4.91602689e-01 -3.35242689e-01 4.45265882e-02
-4.16597754e-01 2.80405045e-01 4.63705659e-01 6.61067724e-01
5.21419644e-01 -1.57151926e+00 -3.95252615e-01 3.36078912e-01
4.37872231e-01 7.02561289e-02 3.68375003e-01 5.96559107e-01
-2.36298263e-01 3.64361703e-01 -2.66199261e-01 -7.37629116e-01
-1.33929396e+00 8.97484064e-01 1.99702114e-01 -1.14609003e-01
-4.67072427e-01 1.04224062e+00 4.30842072e-01 -3.81015688e-01
4.63178456e-01 -8.34830925e-02 -2.85527140e-01 4.68074322e-01
7.02670932e-01 7.15546668e-01 3.63511927e-02 -7.68503547e-01
-5.79521537e-01 1.08998799e+00 -2.73847044e-01 4.16730851e-01
1.24551976e+00 -1.25759959e-01 -1.92192495e-01 4.10197228e-01
1.80029535e+00 -2.64373839e-01 -1.27842307e+00 -5.09248793e-01
4.51257117e-02 -8.25130284e-01 6.76645637e-02 -4.99232680e-01
-1.20568967e+00 1.02439058e+00 9.47551072e-01 2.99520165e-01
1.26038980e+00 2.69479211e-03 6.30285144e-01 2.88568914e-01
5.45314908e-01 -1.42949200e+00 3.27097684e-01 3.70147616e-01
7.72502303e-01 -1.21956205e+00 1.27917290e-01 -5.19266307e-01
-6.34971559e-01 1.04078650e+00 7.95024693e-01 -1.60988256e-01
9.81744111e-01 -5.25374077e-02 4.76000160e-02 -1.94856584e-01
-2.68130690e-01 8.69859010e-03 5.50684988e-01 4.53976393e-01
2.13047057e-01 1.70717344e-01 -3.22414786e-01 5.09927750e-01
-4.44436781e-02 -2.93738514e-01 3.89440283e-02 1.03595996e+00
-4.61396247e-01 -1.14197373e+00 -2.51995623e-01 5.83543837e-01
-1.50813028e-01 2.38648534e-01 -2.68631816e-01 6.64407372e-01
4.55570787e-01 1.11453056e+00 2.51734614e-01 -4.63183343e-01
3.71108830e-01 -2.59032905e-01 2.45285526e-01 -5.79006493e-01
-2.39176735e-01 -1.18300095e-01 -4.37209398e-01 -6.33228123e-01
-1.63970038e-01 -4.89538908e-01 -1.24476635e+00 -2.92838097e-01
-2.11339816e-01 2.35809773e-01 4.32135999e-01 7.05449700e-01
3.51742715e-01 4.25205410e-01 1.07841873e+00 -6.75104082e-01
-7.24470615e-01 -4.95667398e-01 -8.97781491e-01 5.98131895e-01
3.63171667e-01 -9.82444584e-01 -4.65613365e-01 -3.36176217e-01] | [9.635170936584473, 2.2147717475891113] |
a107f398-7a9e-46ce-b0e8-bbcb7b4b9cc9 | analyzing-assumptions-in-conversation | 1810.11118 | null | https://arxiv.org/abs/1810.11118v2 | https://arxiv.org/pdf/1810.11118v2.pdf | A Large-Scale Corpus for Conversation Disentanglement | Disentangling conversations mixed together in a single stream of messages is a difficult task, made harder by the lack of large manually annotated datasets. We created a new dataset of 77,563 messages manually annotated with reply-structure graphs that both disentangle conversations and define internal conversation structure. Our dataset is 16 times larger than all previously released datasets combined, the first to include adjudication of annotation disagreements, and the first to include context. We use our data to re-examine prior work, in particular, finding that 80% of conversations in a widely used dialogue corpus are either missing messages or contain extra messages. Our manually-annotated data presents an opportunity to develop robust data-driven methods for conversation disentanglement, which will help advance dialogue research. | ['Walter S. Lasecki', 'Chulaka Gunasekara', 'Sai R. Gouravajhala', 'Lazaros Polymenakos', 'Jonathan K. Kummerfeld', 'Siva Sankalp Patel', 'Vignesh Athreya', 'Joseph Peper', 'Jatin Ganhotra'] | 2018-10-25 | a-large-scale-corpus-for-conversation | https://aclanthology.org/P19-1374 | https://aclanthology.org/P19-1374.pdf | acl-2019-7 | ['conversation-disentanglement'] | ['natural-language-processing'] | [ 4.11847413e-01 6.77732527e-01 -3.54947001e-01 -5.20593047e-01
-9.97397661e-01 -1.15919673e+00 1.11895108e+00 4.65586960e-01
-1.51280984e-01 1.09172869e+00 1.34411037e+00 -7.25615025e-01
2.15499736e-02 -3.35695982e-01 2.86551006e-02 -2.26641297e-01
8.49128589e-02 7.53944337e-01 -9.06881616e-02 -5.15063763e-01
3.72943819e-01 -5.58182597e-02 -1.03148735e+00 6.98675334e-01
7.60586977e-01 2.26832330e-01 -2.02297986e-01 1.01899779e+00
-2.38410458e-01 1.07745540e+00 -9.60052907e-01 -6.78851306e-01
-1.30119830e-01 -7.49542356e-01 -1.36877680e+00 7.49131665e-02
3.74498367e-01 -3.12563211e-01 -4.18439835e-01 3.55360389e-01
5.16481817e-01 7.80992303e-03 4.00415897e-01 -1.41481686e+00
-1.94123194e-01 1.04815722e+00 -1.85131803e-01 4.70577955e-01
8.72636318e-01 2.46711969e-01 1.40899265e+00 -2.47830003e-01
9.09448147e-01 1.45562291e+00 6.46745741e-01 6.24647021e-01
-1.51537037e+00 -7.35800505e-01 -1.20513856e-01 -2.05714792e-01
-6.12521827e-01 -9.54963267e-01 6.14315331e-01 -6.49640441e-01
9.48054135e-01 8.45358014e-01 5.73927402e-01 1.54527712e+00
-3.66950065e-01 5.52138329e-01 1.29302824e+00 -4.15847778e-01
-1.96233928e-01 -9.83356535e-02 4.95920360e-01 3.52467537e-01
2.62988567e-01 -2.97592878e-01 -5.63045502e-01 -9.09600735e-01
1.22867048e-01 -3.80210489e-01 -5.09697795e-01 2.27572083e-01
-1.44972944e+00 1.02947330e+00 -2.80177563e-01 5.12602329e-01
1.96041718e-01 -3.73950839e-01 7.28223979e-01 5.75529933e-01
6.00050807e-01 8.87475252e-01 -5.56420505e-01 -1.08558547e+00
-4.41779464e-01 5.26230514e-01 1.63240647e+00 8.89837265e-01
4.92584944e-01 -4.57516313e-01 -5.62099516e-02 9.12196457e-01
1.68094501e-01 1.20753087e-02 2.79977918e-01 -1.27271986e+00
1.11066091e+00 8.17617118e-01 3.93981069e-01 -1.07828975e+00
-6.48102462e-01 3.40817213e-01 -4.53898340e-01 -4.99147445e-01
1.12336802e+00 -6.21483028e-01 -5.13711013e-03 1.87097323e+00
1.60433501e-01 -3.52702528e-01 1.25714630e-01 4.44308102e-01
1.08024240e+00 1.92303747e-01 -1.65376440e-01 -4.12644058e-01
1.48731828e+00 -6.19441092e-01 -8.93282831e-01 -1.49756804e-01
1.01540434e+00 -1.06895494e+00 1.00724113e+00 1.97329253e-01
-9.45670843e-01 4.09355164e-02 -8.77246559e-01 -2.77600586e-01
1.08629338e-01 -5.22471070e-01 8.04455221e-01 8.61046851e-01
-9.29051280e-01 2.09145412e-01 -3.86730313e-01 -4.78803515e-01
1.03296198e-01 5.31356782e-02 -5.51494062e-01 2.93610781e-01
-1.35902488e+00 9.39418316e-01 8.99436101e-02 -1.64288536e-01
-4.04033244e-01 -5.14316320e-01 -9.59657609e-01 -2.68070549e-01
5.42132914e-01 -3.47157985e-01 1.57687509e+00 -5.59084237e-01
-1.47600520e+00 9.81792212e-01 -3.66266370e-01 -1.56355977e-01
5.86462379e-01 4.08667028e-02 -3.35134476e-01 -1.14359036e-01
5.80275580e-02 1.05329335e-01 1.95664585e-01 -1.23430717e+00
-5.30605614e-01 -1.50982857e-01 3.25361907e-01 2.41178140e-01
-9.25613418e-02 2.41430685e-01 9.06014873e-04 -5.34754574e-01
-2.10184623e-02 -1.02882457e+00 1.08484156e-01 -6.04272068e-01
-6.62958741e-01 -3.66777033e-01 7.02671826e-01 -8.07383716e-01
1.33669400e+00 -1.82074749e+00 1.88445613e-01 -2.14163393e-01
9.11673188e-01 -1.04616798e-01 -1.10520646e-01 9.91432309e-01
4.48783375e-02 6.63349211e-01 -1.45563986e-02 -6.76475286e-01
5.56022376e-02 2.73380786e-01 -2.89864123e-01 4.88049060e-01
-1.51157051e-01 8.40714574e-01 -1.00173664e+00 -4.90355194e-01
-8.03024620e-02 -7.77710006e-02 -5.59217930e-01 3.02978396e-01
-2.03861922e-01 7.46652722e-01 -1.71809763e-01 2.58808583e-01
3.85330915e-01 -1.45273998e-01 7.05654860e-01 1.68787956e-01
-3.50881785e-01 1.30725169e+00 -5.58542430e-01 1.49661040e+00
-3.57420236e-01 1.21467233e+00 4.13047165e-01 -3.40798825e-01
7.28104830e-01 6.06432855e-01 2.99031019e-01 8.98907632e-02
4.53576118e-01 -6.73530772e-02 5.41759729e-01 -7.29719698e-01
7.48328149e-01 -5.46319559e-02 -5.05585551e-01 1.13099802e+00
-2.39948958e-01 -5.06830812e-01 2.52274394e-01 6.53945029e-01
1.35686278e+00 -7.25632906e-01 3.98960829e-01 -1.01288579e-01
2.28377610e-01 8.37645531e-02 6.07849658e-01 7.79024482e-01
-4.68545020e-01 4.81347919e-01 1.20439339e+00 -2.22747460e-01
-1.05837095e+00 -6.51550114e-01 -1.04859553e-01 1.02174330e+00
-3.02484989e-01 -7.67597497e-01 -5.85219085e-01 -9.40403819e-01
-5.48811220e-02 7.32477725e-01 -5.45925975e-01 2.91508377e-01
-6.35571122e-01 -5.84962368e-01 1.10272348e+00 1.78027034e-01
2.48328149e-01 -6.90550566e-01 4.52777632e-02 1.41294032e-01
-1.20646000e+00 -1.31355953e+00 -6.41675532e-01 -1.12291664e-01
-4.67522979e-01 -1.57695758e+00 -2.36381385e-02 -4.30424541e-01
3.53678107e-01 3.21108431e-01 1.33797026e+00 4.21274722e-01
1.41087621e-01 2.07334027e-01 -4.50183451e-01 -1.66019723e-01
-1.16481972e+00 3.12435508e-01 -8.14499557e-02 -3.42083395e-01
5.80253243e-01 -5.02593696e-01 -1.71376377e-01 5.85172951e-01
-5.50322592e-01 1.82045013e-01 -1.14221841e-01 1.01799774e+00
-6.88040435e-01 -5.53935885e-01 8.28905821e-01 -1.23457038e+00
1.39753175e+00 -7.15360343e-01 -1.73873588e-01 -5.53529933e-02
-3.37443948e-01 -1.32730767e-01 3.17661315e-01 -2.68186539e-01
-1.18288946e+00 -6.34111941e-01 -1.23240553e-01 7.63112307e-01
-2.37870812e-01 3.34076941e-01 1.42236669e-02 2.96485543e-01
7.24350572e-01 -3.21294308e-01 3.64994615e-01 -3.69080424e-01
3.50523978e-01 1.28841543e+00 2.49075800e-01 -6.92608178e-01
4.24729913e-01 1.68666258e-01 -5.98239005e-01 -8.78880978e-01
-5.39990008e-01 -5.72439909e-01 -6.37213767e-01 -1.36992916e-01
6.86252415e-01 -8.03647876e-01 -1.24151087e+00 8.54908526e-02
-1.26760745e+00 -5.52040875e-01 2.01296076e-01 4.21996862e-01
-2.94359028e-01 7.67540812e-01 -1.00320745e+00 -7.72417903e-01
-1.07452609e-01 -1.02750039e+00 6.16531312e-01 -1.89572394e-01
-1.42144096e+00 -1.07489479e+00 4.43194628e-01 1.18184900e+00
2.07379922e-01 2.17284530e-01 9.47111785e-01 -1.22377920e+00
-1.69862881e-01 -1.20189287e-01 -1.45180836e-01 -8.38795006e-02
4.80100244e-01 1.11275859e-01 -9.17116582e-01 -1.38032198e-01
1.19202890e-01 -7.04554439e-01 4.83958632e-01 -2.93030500e-01
4.98309970e-01 -7.72166908e-01 -3.28340054e-01 -2.42772236e-01
2.87965000e-01 2.97400113e-02 4.37575310e-01 7.88529664e-02
3.77559841e-01 1.12508368e+00 1.50372818e-01 5.44186473e-01
1.03384268e+00 6.78755999e-01 -3.57046723e-02 3.44877511e-01
3.25753465e-02 -1.96169302e-01 1.27858460e-01 1.27727783e+00
4.76947576e-02 -6.91795170e-01 -9.93299723e-01 4.99970853e-01
-1.84042645e+00 -1.07593822e+00 -6.30466759e-01 1.77434015e+00
1.24449944e+00 2.21224591e-01 3.82097632e-01 2.15388402e-01
7.44322717e-01 4.85826343e-01 6.05475381e-02 -4.71538097e-01
-3.52160148e-02 -1.12342156e-01 6.78383261e-02 1.11305654e+00
-6.89311981e-01 6.26977861e-01 7.46005440e+00 1.55013815e-01
-3.77266258e-01 1.18628956e-01 4.12485659e-01 -8.53416845e-02
-7.59545445e-01 3.69110614e-01 -6.43296778e-01 5.77177227e-01
9.44300354e-01 -1.52565762e-01 4.87269700e-01 2.95736045e-01
4.60766584e-01 -2.82425642e-01 -1.41899741e+00 7.03253031e-01
1.13962255e-01 -1.38670635e+00 -4.03886586e-01 1.01766720e-01
4.32927519e-01 -8.54122639e-03 -5.85307121e-01 2.76561916e-01
9.39632952e-01 -9.85620499e-01 2.25168452e-01 1.25554249e-01
5.88107884e-01 -2.36100852e-01 6.38473630e-01 5.18379271e-01
-6.79800868e-01 1.22734860e-01 2.65220791e-01 -7.46039867e-01
4.12098676e-01 3.60433787e-01 -1.34132016e+00 1.46278486e-01
2.26213783e-01 5.21269560e-01 -3.15465152e-01 3.38982642e-01
-1.98250026e-01 9.06430185e-01 -2.12915435e-01 -4.44222808e-01
-1.19128697e-01 8.63241330e-02 8.95496905e-01 1.21216559e+00
-4.51650053e-01 3.32943738e-01 1.78401262e-01 6.04751766e-01
-4.26131427e-01 -2.57527918e-01 -6.88014686e-01 -3.61275345e-01
8.78152132e-01 1.34349871e+00 -4.28492546e-01 -3.73907506e-01
-4.61978346e-01 5.20696044e-01 5.06521285e-01 2.02451900e-01
-3.95962924e-01 -2.62870044e-01 9.08028066e-01 -1.59570321e-01
-4.85819727e-01 -5.71063936e-01 -4.99749750e-01 -1.34506524e+00
-1.29932255e-01 -1.25885773e+00 4.69278216e-01 -4.24843997e-01
-1.48270714e+00 5.24844110e-01 3.30346897e-02 -6.31230712e-01
-6.35048032e-01 -6.72500581e-02 -5.18401444e-01 8.14403057e-01
-7.73281872e-01 -7.02169895e-01 -5.08217990e-01 1.85176760e-01
7.33345866e-01 4.25920077e-02 9.71743464e-01 1.95515439e-01
-6.54191017e-01 6.33807361e-01 -2.29149103e-01 4.01762813e-01
1.02255034e+00 -1.11617923e+00 6.27072930e-01 2.47233763e-01
-7.05324188e-02 8.62126648e-01 9.46374416e-01 -7.07395852e-01
-1.18370414e+00 -2.10772127e-01 1.41711831e+00 -1.28039110e+00
8.64992559e-01 -1.03178144e+00 -8.89512956e-01 1.09263158e+00
6.60302222e-01 -6.50734603e-01 1.31492162e+00 6.22901261e-01
-4.64681894e-01 5.26192904e-01 -1.12977934e+00 8.21300507e-01
1.08163691e+00 -7.18870103e-01 -1.15265191e+00 1.84916660e-01
9.14493620e-01 -6.66458607e-01 -8.66461515e-01 8.55665281e-02
7.22988248e-01 -1.00059056e+00 3.81721705e-01 -7.54686773e-01
5.39770901e-01 3.33815694e-01 8.98059830e-03 -1.38595605e+00
2.96777897e-02 -1.27658963e+00 8.24223757e-02 1.58141506e+00
7.33087599e-01 -9.11569476e-01 6.61509097e-01 1.29935002e+00
-4.13844734e-02 -3.31291318e-01 -7.43335724e-01 -3.02650899e-01
2.79182754e-02 -3.51996958e-01 5.94519258e-01 1.45452666e+00
1.07012379e+00 7.18390048e-01 -3.43735278e-01 -5.08089960e-01
1.32531673e-01 -6.18140958e-02 1.30912125e+00 -1.33096290e+00
-2.59645879e-01 -5.54072917e-01 7.79622002e-03 -1.27210402e+00
3.02617162e-01 -8.00770879e-01 -2.26367377e-02 -1.42472208e+00
4.17632580e-01 -6.20575845e-01 8.48064423e-01 4.16813463e-01
-1.83458328e-01 -4.17760499e-02 -2.68883798e-02 2.72899270e-01
-4.12720561e-01 4.67500329e-01 1.10282910e+00 -7.34937191e-02
-3.97936106e-01 -3.75000760e-02 -1.21768284e+00 6.91568613e-01
8.57381642e-01 -4.87058222e-01 -2.82139391e-01 -8.91000479e-02
2.33027130e-01 3.19322139e-01 -7.85022154e-02 1.38433499e-03
2.25390773e-02 -2.58067191e-01 -1.78230986e-01 -2.72534311e-01
3.46371680e-01 -2.76329577e-01 -1.93770871e-01 1.70580328e-01
-9.42879915e-01 6.82106540e-02 1.68091998e-01 5.66777110e-01
-2.05286145e-02 -1.49952397e-01 9.59198549e-02 -1.71398088e-01
4.74974737e-02 -3.88619483e-01 -9.59413886e-01 8.77004027e-01
7.47106969e-01 -4.87925904e-03 -9.88396883e-01 -1.01578200e+00
-5.26504040e-01 2.89138466e-01 3.22991282e-01 6.59355819e-01
1.49598107e-01 -9.65375960e-01 -8.41015875e-01 -1.94746241e-01
1.25931472e-01 -1.20688356e-01 -5.90243675e-02 8.41874838e-01
-2.61770368e-01 3.38706106e-01 1.58837482e-01 -3.55246246e-01
-1.62945271e+00 -5.52450120e-02 -6.99180588e-02 -2.55318761e-01
-4.91401762e-01 5.98148346e-01 -3.28462988e-01 -8.31196010e-01
1.50614813e-01 -4.01282281e-01 -1.34055793e-01 3.39964032e-01
5.24285376e-01 5.40713847e-01 -1.95195049e-01 -5.41622937e-01
-2.76735991e-01 -3.83284897e-01 -2.15002626e-01 -4.54170465e-01
9.96021748e-01 -5.79301238e-01 -5.03550828e-01 7.63196468e-01
1.30433059e+00 7.14571536e-01 -8.95064652e-01 -2.17398509e-01
1.63861826e-01 -5.98989844e-01 -5.80211401e-01 -7.52107441e-01
-1.44323543e-01 2.79469639e-01 -5.22346735e-01 9.50335681e-01
2.41855115e-01 2.10882783e-01 7.48002648e-01 3.42588156e-01
8.61475989e-02 -7.05173552e-01 2.92160869e-01 8.67639244e-01
1.16608381e+00 -1.51188588e+00 -5.55452667e-02 -6.26434982e-01
-7.38408744e-01 9.72519815e-01 6.08862698e-01 4.35024768e-01
4.13948774e-01 5.23777068e-01 1.98164850e-01 -3.45680237e-01
-1.09485209e+00 1.78332403e-01 -4.18140665e-02 3.73816967e-01
9.07981455e-01 8.28190707e-03 -8.77789080e-01 7.58117974e-01
-7.15613782e-01 -4.12649125e-01 1.02001405e+00 8.09295893e-01
-9.05737430e-02 -1.39475632e+00 -1.95612669e-01 5.57377219e-01
-3.84113193e-01 -1.64130479e-01 -1.26708591e+00 8.47606301e-01
-4.99501944e-01 1.73475099e+00 5.10365367e-02 -5.95627904e-01
9.39107910e-02 2.69100934e-01 3.17855626e-01 -7.70931244e-01
-9.38866317e-01 -4.12961036e-01 1.38166726e+00 -7.22162426e-02
-5.02048671e-01 -8.44556272e-01 -9.94439006e-01 -8.75354946e-01
-5.96691370e-01 4.14840609e-01 3.19475830e-01 1.20085096e+00
5.44797659e-01 1.81073189e-01 5.96080124e-01 -5.14835119e-01
-3.26600671e-01 -1.44368219e+00 6.06462322e-02 6.60750031e-01
4.86191183e-01 -4.47074831e-01 -8.19296300e-01 -1.76864326e-01] | [12.501896858215332, 8.000479698181152] |
e322565d-e5d7-4fba-bb35-fb9abee156ff | adversarially-robust-neural-architecture | 2304.04168 | null | https://arxiv.org/abs/2304.04168v1 | https://arxiv.org/pdf/2304.04168v1.pdf | Adversarially Robust Neural Architecture Search for Graph Neural Networks | Graph Neural Networks (GNNs) obtain tremendous success in modeling relational data. Still, they are prone to adversarial attacks, which are massive threats to applying GNNs to risk-sensitive domains. Existing defensive methods neither guarantee performance facing new data/tasks or adversarial attacks nor provide insights to understand GNN robustness from an architectural perspective. Neural Architecture Search (NAS) has the potential to solve this problem by automating GNN architecture designs. Nevertheless, current graph NAS approaches lack robust design and are vulnerable to adversarial attacks. To tackle these challenges, we propose a novel Robust Neural Architecture search framework for GNNs (G-RNA). Specifically, we design a robust search space for the message-passing mechanism by adding graph structure mask operations into the search space, which comprises various defensive operation candidates and allows us to search for defensive GNNs. Furthermore, we define a robustness metric to guide the search procedure, which helps to filter robust architectures. In this way, G-RNA helps understand GNN robustness from an architectural perspective and effectively searches for optimal adversarial robust GNNs. Extensive experimental results on benchmark datasets show that G-RNA significantly outperforms manually designed robust GNNs and vanilla graph NAS baselines by 12.1% to 23.4% under adversarial attacks. | ['Wenwu Zhu', 'Rex Ying', 'Zhiqiang Zhang', 'Daixin Wang', 'Xin Wang', 'Ziwei Zhang', 'Heng Chang', 'Beini Xie'] | 2023-04-09 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Xie_Adversarially_Robust_Neural_Architecture_Search_for_Graph_Neural_Networks_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Xie_Adversarially_Robust_Neural_Architecture_Search_for_Graph_Neural_Networks_CVPR_2023_paper.pdf | cvpr-2023-1 | ['architecture-search', 'robust-design'] | ['methodology', 'miscellaneous'] | [ 9.77715552e-02 -7.23290443e-02 -4.49450351e-02 -5.80831952e-02
-5.12965083e-01 -1.20446730e+00 5.69342375e-01 -5.94851049e-03
-7.04438984e-02 1.52713448e-01 3.77803333e-02 -7.31136918e-01
-2.71372229e-01 -1.14187825e+00 -9.58752394e-01 -5.41226447e-01
-2.31044620e-01 1.54637024e-01 3.87451440e-01 -7.08820522e-01
4.19464290e-01 8.78572762e-01 -8.99436772e-01 -7.12254867e-02
7.29273319e-01 9.02511716e-01 -4.98406112e-01 6.49243593e-01
4.16529278e-04 6.35637403e-01 -8.39784741e-01 -7.75762737e-01
7.43189216e-01 5.90290036e-03 -7.60220885e-01 -7.29257286e-01
5.76843441e-01 -1.79329917e-01 -8.10192823e-01 1.42923641e+00
5.91124058e-01 1.86000884e-01 3.25853348e-01 -1.45546687e+00
-8.16424906e-01 1.14303041e+00 -3.10464919e-01 2.61929542e-01
-1.17722070e-02 5.58197379e-01 1.10471976e+00 -3.23118985e-01
4.38996673e-01 1.53473914e+00 4.82417703e-01 9.28876996e-01
-1.13531077e+00 -8.13722730e-01 4.08623278e-01 2.49062981e-02
-1.15942848e+00 -3.75927001e-01 9.57485795e-01 -1.64935842e-01
9.46281135e-01 7.04626262e-01 1.99727282e-01 1.56757390e+00
2.80659944e-01 5.27922750e-01 5.03846526e-01 1.94027185e-01
4.03862685e-01 -5.53715467e-01 4.96429242e-02 7.10812151e-01
4.17512447e-01 3.04297328e-01 -3.54235917e-01 -3.43466222e-01
4.91287947e-01 -8.40833709e-02 -2.35165253e-01 -2.13270977e-01
-6.58759177e-01 7.98004627e-01 1.12487388e+00 4.13536467e-02
8.00581053e-02 3.50036770e-01 7.70328462e-01 5.47778487e-01
-9.32593793e-02 1.13741469e+00 -4.51600075e-01 4.60718244e-01
-2.12776914e-01 6.84651956e-02 7.04896986e-01 6.78055525e-01
3.30822706e-01 6.74693882e-01 8.49767867e-03 5.43419063e-01
3.17929745e-01 2.51489341e-01 2.76069865e-02 -4.89300996e-01
6.11713827e-01 1.07183421e+00 -9.04418945e-01 -1.34382129e+00
-5.54799497e-01 -7.71687627e-01 -1.15691733e+00 1.20740227e-01
2.53679603e-01 -7.01255165e-03 -1.07806349e+00 1.92649066e+00
3.43390226e-01 1.08186767e-01 2.85169274e-01 7.18213320e-01
7.90327966e-01 5.08518100e-01 -6.10829219e-02 2.23097429e-01
1.16167569e+00 -6.75850511e-01 -1.39132917e-01 -5.00982702e-01
4.74597186e-01 -3.11742514e-01 1.17463291e+00 4.94784266e-01
-7.68778145e-01 -1.28926516e-01 -1.36507463e+00 2.09570572e-01
-5.20563304e-01 -6.67407274e-01 5.66934407e-01 1.17251539e+00
-1.00940287e+00 5.22316277e-01 -7.05751657e-01 -4.92646322e-02
4.57900524e-01 6.60287499e-01 -3.17293376e-01 6.61001652e-02
-1.48690021e+00 6.13355398e-01 6.26517773e-01 2.11757928e-01
-1.40207171e+00 -5.60732841e-01 -8.47306669e-01 1.77208275e-01
8.33937287e-01 -5.32412887e-01 6.70218289e-01 -3.93394619e-01
-1.42075038e+00 2.87084132e-01 8.83852363e-01 -7.68018663e-01
3.27188522e-01 -7.72951692e-02 -6.29525900e-01 1.12507746e-01
-5.30079067e-01 1.35614768e-01 9.76216197e-01 -1.18224645e+00
1.02172792e-01 -2.43522733e-01 5.39738715e-01 -1.58371776e-01
-5.27316332e-01 2.58347839e-01 -4.13309008e-01 -1.08483255e+00
7.08330348e-02 -1.13654280e+00 -5.47923744e-01 -3.63857001e-01
-1.09335554e+00 8.42813775e-02 5.73782980e-01 -3.41087341e-01
1.44710779e+00 -2.07304478e+00 4.54745263e-01 8.03292572e-01
6.36723220e-01 7.20789611e-01 -5.39297163e-01 2.69843310e-01
-2.26444766e-01 6.36867881e-01 -9.95732695e-02 3.38892668e-01
1.81965277e-01 -4.77134548e-02 -6.30340993e-01 2.54957557e-01
2.41840005e-01 1.09889758e+00 -5.27514219e-01 4.50538136e-02
-1.22508541e-01 1.62158996e-01 -7.84842610e-01 1.50539517e-01
-4.91850525e-01 -1.22121451e-02 -6.07544184e-01 1.04484344e+00
6.59219623e-01 3.82790864e-02 3.47469479e-01 -1.83021575e-01
4.57445651e-01 4.05743532e-02 -9.13088024e-01 1.14461482e+00
-1.68462649e-01 1.72806472e-01 1.19649619e-01 -7.20236123e-01
1.11871672e+00 -2.38767698e-01 -1.30135477e-01 -6.30384922e-01
4.49160367e-01 7.69612342e-02 2.59023160e-01 8.61092657e-02
1.97043478e-01 2.67665833e-01 -6.46424174e-01 2.49665692e-01
-6.51386604e-02 6.87557608e-02 -1.33805990e-01 5.00199556e-01
1.78217161e+00 -6.19179308e-01 1.41507834e-02 -1.36134744e-01
7.24718928e-01 -4.10973430e-01 7.69664168e-01 8.53580236e-01
-1.75182015e-01 2.29090944e-01 9.09664571e-01 -7.83431053e-01
-6.76706195e-01 -1.31949389e+00 5.46603262e-01 1.19654632e+00
-1.13241924e-02 -6.59142733e-01 -9.90434408e-01 -1.22440469e+00
-1.22848593e-01 5.76350212e-01 -7.03881562e-01 -1.05416751e+00
-8.41336012e-01 -6.99724138e-01 1.35737300e+00 3.81226540e-01
3.86580110e-01 -9.61253583e-01 1.91361662e-02 6.88591681e-04
7.50745013e-02 -9.03471410e-01 -5.85564971e-01 2.58741498e-01
-6.22294724e-01 -1.23657751e+00 4.41719741e-02 -5.06954193e-01
6.06238425e-01 1.39843345e-01 1.02501476e+00 5.72815537e-01
-3.16695809e-01 9.40553285e-03 -1.15248926e-01 -3.40923101e-01
-7.85783172e-01 3.46103877e-01 4.33421880e-01 -1.77346706e-01
-1.82525069e-01 -6.81963205e-01 -4.53089267e-01 6.57227874e-01
-1.30751526e+00 -6.86465383e-01 6.93273544e-01 5.35460532e-01
3.91180634e-01 2.17617959e-01 6.40373886e-01 -8.59917641e-01
8.63924801e-01 -5.10320723e-01 -9.86010432e-01 5.27930677e-01
-7.73025393e-01 3.88464749e-01 1.25347519e+00 -5.62209249e-01
-4.50509012e-01 -1.99456424e-01 -2.92460173e-01 -7.46461749e-01
2.08108678e-01 4.81854111e-01 -8.53729546e-01 -5.93027174e-01
1.23286259e+00 -3.35629843e-02 -3.51333804e-02 -3.17168295e-01
5.33133328e-01 -1.12018973e-01 5.86729825e-01 -7.23981678e-01
1.57253313e+00 7.45234117e-02 4.24235731e-01 -4.33583111e-01
-3.84644836e-01 1.30012691e-01 -8.55547413e-02 -4.59584519e-02
6.00475848e-01 -4.44392622e-01 -1.08594763e+00 5.89496374e-01
-7.18920708e-01 -2.96798319e-01 2.25189090e-01 -3.59284133e-01
-1.68890581e-01 5.51755905e-01 -7.95085728e-01 -4.73081023e-01
-7.95445204e-01 -1.43546951e+00 3.64333630e-01 1.51052073e-01
1.35348931e-01 -7.21991956e-01 -1.11657128e-01 3.84621322e-01
6.43863201e-01 6.19631171e-01 1.32592368e+00 -1.14141369e+00
-9.42928135e-01 -3.76228213e-01 -9.67373177e-02 3.79050553e-01
-2.60137916e-01 1.94123536e-01 -7.73083091e-01 -4.64702427e-01
-1.39300838e-01 -4.37980771e-01 9.74794030e-01 -5.54324761e-02
1.39347827e+00 -8.51371586e-01 -1.19226187e-01 1.04476547e+00
1.37664962e+00 1.75767064e-01 8.43471527e-01 6.38661563e-01
1.20917368e+00 4.88173991e-01 2.11522266e-01 4.40576710e-02
-1.74345989e-02 4.75868851e-01 1.30389905e+00 1.65700480e-01
3.47363025e-01 -2.47198120e-01 5.01731634e-01 4.34111595e-01
1.48612767e-01 -6.62627220e-01 -1.10270441e+00 5.81261553e-02
-1.56912434e+00 -6.24815345e-01 2.13894650e-01 1.83519661e+00
6.04556739e-01 7.41264522e-01 9.39930305e-02 2.17527896e-01
5.52434027e-01 3.65768820e-01 -8.97748649e-01 -6.09984994e-01
-2.45042518e-01 6.59339949e-02 6.80416167e-01 1.61650777e-01
-1.11923480e+00 1.16451681e+00 5.57869959e+00 9.90954638e-01
-9.91754174e-01 -5.23973763e-01 5.51625431e-01 -1.59755304e-01
-5.30660093e-01 2.53414698e-02 -6.46943629e-01 1.15984485e-01
1.04722738e+00 -1.76789701e-01 8.18385720e-01 9.61263657e-01
-4.68564719e-01 8.75680745e-01 -8.25495660e-01 5.71893275e-01
-9.66146812e-02 -1.57972014e+00 6.08100951e-01 1.25431359e-01
3.40273410e-01 -1.56196626e-02 4.63275224e-01 4.98092175e-01
8.33335817e-01 -1.24480474e+00 6.04136050e-01 2.16709226e-01
4.88272458e-01 -1.22901809e+00 4.55061376e-01 3.94470915e-02
-1.22503567e+00 -4.28550780e-01 -4.16299760e-01 4.96313691e-01
-2.40990147e-01 1.29862770e-01 -9.56393421e-01 6.25581980e-01
6.56562150e-01 -4.54715490e-02 -1.02626050e+00 7.21118093e-01
-3.44749629e-01 6.07414365e-01 -2.11777329e-01 -9.58339125e-03
3.61138880e-01 1.47056431e-01 8.74463022e-01 7.77164042e-01
-3.04111447e-02 -2.01329246e-01 9.10282284e-02 8.26260090e-01
-5.29509544e-01 -1.11678235e-01 -8.17772806e-01 -3.23330879e-01
6.04665160e-01 1.20562434e+00 -8.43403280e-01 4.26250964e-01
6.99688420e-02 7.03254700e-01 5.75338483e-01 3.00688922e-01
-8.53615522e-01 -6.11606658e-01 1.05836141e+00 -8.76865387e-02
-1.19127318e-01 -1.12472937e-01 -6.53738901e-02 -9.29251969e-01
-1.15879327e-01 -1.62468171e+00 8.65041435e-01 -1.51748791e-01
-1.43117440e+00 1.01861620e+00 -3.05556834e-01 -1.04187083e+00
8.63106623e-02 -7.42940843e-01 -8.95002127e-01 4.61690634e-01
-9.23697472e-01 -1.25831342e+00 1.03749610e-01 7.36064255e-01
2.33088672e-01 -8.16668272e-01 6.37989938e-01 4.21308242e-02
-1.16218781e+00 1.33418310e+00 -2.59723067e-01 4.73201901e-01
5.78103840e-01 -1.11719954e+00 1.40913141e+00 1.45372629e+00
1.20590314e-01 1.30021060e+00 7.01118648e-01 -8.62255335e-01
-1.91820252e+00 -1.37012851e+00 -1.17282793e-01 -6.25309169e-01
9.97254610e-01 -7.44186163e-01 -1.17383254e+00 5.73264778e-01
-3.02733004e-01 5.12161776e-02 7.35182405e-01 6.99763745e-02
-1.17043126e+00 -1.28146172e-01 -1.22615039e+00 1.19163346e+00
1.27533257e+00 -6.45644784e-01 -1.09905541e-01 2.05336571e-01
1.46487570e+00 -3.78123283e-01 -7.30247676e-01 6.57826960e-01
2.15127140e-01 -8.47465873e-01 1.43869305e+00 -8.89903605e-01
8.52834731e-02 -3.92983347e-01 -2.74691045e-01 -1.17903697e+00
-4.52166140e-01 -1.23001814e+00 -3.63824606e-01 1.14324963e+00
6.33858383e-01 -7.36577451e-01 8.88385832e-01 5.11801481e-01
-2.90178448e-01 -8.50994229e-01 -8.65533769e-01 -9.96785462e-01
8.10837671e-02 -4.83903408e-01 8.98215473e-01 9.99357939e-01
-3.94838214e-01 1.65655479e-01 -4.40578789e-01 4.98359501e-01
7.61974096e-01 -4.84212995e-01 8.63527596e-01 -1.11592054e+00
-3.08954239e-01 -8.88804495e-01 -6.24676704e-01 -3.34926158e-01
3.33626240e-01 -1.04909027e+00 -2.09968761e-01 -1.06788886e+00
-3.21611524e-01 -3.47663820e-01 -6.00477993e-01 6.16367757e-01
-3.02425832e-01 1.23380587e-01 2.17803895e-01 -5.48337922e-02
-4.96956110e-01 3.94534320e-01 6.85112357e-01 -4.09759879e-01
-2.01872855e-01 5.15268147e-02 -1.22594905e+00 8.14917386e-01
1.08709967e+00 -3.93332332e-01 -6.81507289e-01 -3.98288906e-01
6.50998116e-01 -3.29470187e-01 3.18696439e-01 -9.25423205e-01
1.66534498e-01 -2.68755823e-01 3.42397019e-02 -2.90148407e-01
-1.84022322e-01 -6.24113202e-01 2.12808177e-01 6.41054630e-01
-3.17790210e-01 3.65652382e-01 4.01250780e-01 9.53679800e-01
2.26945281e-01 1.47535339e-01 8.84819448e-01 1.29562005e-01
-6.57087505e-01 6.93194270e-01 -1.66104659e-01 8.00124332e-02
8.66240382e-01 -1.13233671e-01 -8.89915526e-01 -3.01580895e-02
-4.84370977e-01 5.73271155e-01 5.06386757e-01 8.13455105e-01
9.62445498e-01 -1.19810390e+00 -4.67314333e-01 3.60435218e-01
2.64501810e-01 -4.38863412e-02 3.28879774e-01 1.96391359e-01
-7.19734550e-01 1.28116524e-02 -4.09007907e-01 7.20341923e-03
-1.36257625e+00 1.18020225e+00 4.98199999e-01 -3.74135971e-01
-3.46077740e-01 1.33022356e+00 3.51383239e-01 -6.80676699e-01
5.82240283e-01 -6.53926358e-02 -1.75790772e-01 -2.60122389e-01
4.40442055e-01 4.65971857e-01 1.89739972e-01 -3.62616390e-01
-5.78664184e-01 1.35796860e-01 -4.62968081e-01 4.08163041e-01
1.20996249e+00 3.08012009e-01 -2.78250158e-01 -3.38764220e-01
1.03661942e+00 1.02200858e-01 -8.54221284e-01 -2.57574528e-01
2.21205339e-01 -2.40486264e-01 -1.75312847e-01 -7.81439483e-01
-1.28713322e+00 6.60458803e-01 2.84611106e-01 3.67562830e-01
1.33428490e+00 -1.55095458e-01 8.48716378e-01 7.34499216e-01
2.59039134e-01 -6.76442564e-01 5.36772847e-01 6.80177033e-01
1.04779589e+00 -7.12197006e-01 -5.63134663e-02 -3.69894803e-01
-4.51955706e-01 9.75890160e-01 1.14802361e+00 -1.76739365e-01
4.30397987e-01 2.97820330e-01 -1.33256819e-02 -4.31399167e-01
-8.18034947e-01 2.42370307e-01 5.64202607e-01 6.36159599e-01
-3.87355536e-01 -8.64407048e-02 3.55854094e-01 7.66147554e-01
-3.36714745e-01 -1.07180667e+00 4.75801557e-01 6.14292443e-01
-3.39409143e-01 -1.17432404e+00 -3.02477270e-01 3.09062034e-01
-6.58729494e-01 -2.08721817e-01 -1.11188185e+00 4.50550109e-01
-3.50875556e-01 7.84713566e-01 -7.69454300e-01 -1.25369358e+00
6.11322641e-01 -1.79494649e-01 -7.44028762e-02 -4.25909162e-01
-1.12739527e+00 -1.63244620e-01 1.78989708e-01 -7.80806363e-01
6.68032467e-01 -8.60507414e-02 -1.04425716e+00 -5.88378727e-01
-3.91305625e-01 2.72258129e-02 4.90891635e-01 4.66772437e-01
4.49842006e-01 8.34096372e-01 7.66441405e-01 -4.03213829e-01
-9.26889837e-01 -2.87983924e-01 -2.34309465e-01 1.83284372e-01
2.77882904e-01 -4.27197158e-01 -5.00594258e-01 -7.96989381e-01] | [6.072975158691406, 7.380462646484375] |
9cab8284-8403-4764-a135-99b93dc367f7 | adapting-deep-learning-for-sentiment | 2001.01047 | null | https://arxiv.org/abs/2001.01047v1 | https://arxiv.org/pdf/2001.01047v1.pdf | Adapting Deep Learning for Sentiment Classification of Code-Switched Informal Short Text | Nowadays, an abundance of short text is being generated that uses nonstandard writing styles influenced by regional languages. Such informal and code-switched content are under-resourced in terms of labeled datasets and language models even for popular tasks like sentiment classification. In this work, we (1) present a labeled dataset called MultiSenti for sentiment classification of code-switched informal short text, (2) explore the feasibility of adapting resources from a resource-rich language for an informal one, and (3) propose a deep learning-based model for sentiment classification of code-switched informal short text. We aim to achieve this without any lexical normalization, language translation, or code-switching indication. The performance of the proposed models is compared with three existing multilingual sentiment classification models. The results show that the proposed model performs better in general and adapting character-based embeddings yield equivalent performance while being computationally more efficient than training word-based domain-specific embeddings. | ['Asim Karim', 'Muhammad Haroon Shakeel'] | 2020-01-04 | null | null | null | null | ['lexical-normalization'] | ['natural-language-processing'] | [-6.49534911e-02 -3.40732068e-01 -4.33352232e-01 -5.02193987e-01
-6.74768150e-01 -7.57788122e-01 6.42165840e-01 5.37954628e-01
-7.87394524e-01 6.79203808e-01 4.61410999e-01 -5.91071129e-01
2.59863228e-01 -5.18793583e-01 -2.74800032e-01 -2.96680391e-01
4.32749301e-01 4.27223802e-01 -9.65550989e-02 -5.47106981e-01
5.98204613e-01 2.94121593e-01 -1.22780228e+00 2.99654514e-01
9.70362663e-01 7.23644674e-01 3.91963154e-01 6.32393301e-01
-7.02623367e-01 9.09009397e-01 -4.71116215e-01 -6.77870274e-01
6.50150403e-02 -2.24043950e-01 -9.16496396e-01 1.62779482e-03
3.56705189e-01 2.35322610e-01 4.79635932e-02 1.05947363e+00
4.30453300e-01 4.59554896e-04 8.97694647e-01 -7.89604247e-01
-1.09205258e+00 7.88521707e-01 -3.88297051e-01 2.74100527e-02
2.58980274e-01 -5.55322051e-01 1.00164950e+00 -1.13820601e+00
8.60150754e-01 9.76265252e-01 8.11862469e-01 4.28446174e-01
-7.03955710e-01 -5.73004544e-01 2.95222312e-01 -1.10421345e-01
-1.36024904e+00 -3.20268273e-01 7.89623857e-01 -8.13735485e-01
1.11427617e+00 -1.83966160e-01 3.61334443e-01 1.12347960e+00
5.92346013e-01 4.19410467e-01 9.72698927e-01 -9.52075064e-01
6.79838285e-02 9.03409958e-01 2.62936652e-01 7.37589300e-01
4.01534766e-01 -5.33043265e-01 -6.27306163e-01 3.64243947e-02
-3.66147682e-02 1.39348701e-01 -7.75602181e-03 -3.41838926e-01
-1.09804761e+00 1.06082618e+00 -6.88182786e-02 8.17335904e-01
-2.78540161e-02 -1.05449714e-01 9.23935950e-01 5.32635927e-01
9.48523641e-01 4.99744982e-01 -9.67402816e-01 -2.92496741e-01
-1.08554733e+00 -4.57958169e-02 8.60635161e-01 1.24744833e+00
6.75450861e-01 2.21211940e-01 1.28524005e-01 1.28723156e+00
4.34684604e-01 5.31442642e-01 1.21931446e+00 -1.45732164e-01
7.77152121e-01 7.12951720e-01 -5.18384650e-02 -9.03663456e-01
-5.76653659e-01 -3.11369598e-01 -5.02843022e-01 -2.01372623e-01
1.86880082e-01 -4.10002708e-01 -5.97133279e-01 1.42227030e+00
-2.18074247e-01 -7.05150545e-01 4.47036892e-01 3.31373304e-01
6.41202092e-01 7.54072428e-01 1.12010408e-02 1.04210503e-01
1.24316406e+00 -1.28271043e+00 -6.45668328e-01 -4.18547481e-01
1.16011930e+00 -1.15382576e+00 1.29916620e+00 2.65190482e-01
-4.94847983e-01 -6.77250385e-01 -1.08614600e+00 -2.06314251e-01
-1.19230270e+00 5.28684676e-01 4.50939208e-01 1.20380139e+00
-1.10950291e+00 1.15089618e-01 -4.75913256e-01 -8.04361343e-01
2.32715234e-01 5.22135645e-02 -5.82364857e-01 -1.16213933e-01
-1.04815423e+00 1.01479304e+00 1.20785728e-01 -1.31532639e-01
-4.17294919e-01 -4.87747192e-01 -1.10335934e+00 -2.84316272e-01
-1.70041233e-01 -5.05448319e-02 9.56032217e-01 -1.40490782e+00
-1.54257846e+00 1.08006990e+00 -2.85826355e-01 -8.24868232e-02
2.33874485e-01 -2.14432985e-01 -8.79953325e-01 -2.53375083e-01
3.30928355e-01 2.62117922e-01 8.02992404e-01 -9.53577876e-01
-6.51107609e-01 -1.80339932e-01 1.37952700e-01 1.59618676e-01
-1.24740028e+00 4.42584932e-01 -3.48915935e-01 -9.60903645e-01
-3.03120583e-01 -1.05245662e+00 -2.07212985e-01 -1.34170488e-01
-1.91600263e-01 -2.29519568e-02 6.41121328e-01 -6.80617273e-01
1.41875720e+00 -2.08297110e+00 1.81444641e-02 3.83342919e-03
-2.05526561e-01 2.27179915e-01 -3.24739993e-01 8.22147906e-01
-1.25121534e-01 4.06251252e-01 -3.44528198e-01 -5.91616690e-01
6.90454245e-02 1.30865738e-01 -2.24728972e-01 4.75597709e-01
1.74902797e-01 5.41577935e-01 -8.12135994e-01 -4.18727100e-01
-5.71694337e-02 3.32426012e-01 -7.03691006e-01 -8.03281888e-02
9.06854123e-02 1.43021429e-02 -4.76804614e-01 7.55405188e-01
4.46221471e-01 1.82364568e-01 4.05186862e-01 9.72061753e-02
-5.31247020e-01 2.17433706e-01 -8.18760633e-01 1.98664141e+00
-1.34490621e+00 9.10406172e-01 -2.00952441e-01 -9.11165833e-01
1.11348212e+00 2.79424250e-01 5.34297042e-02 -5.51097989e-01
2.86347061e-01 5.90704441e-01 -1.65345907e-01 -5.95617712e-01
1.06168509e+00 -2.57430822e-01 -6.04648709e-01 5.23475230e-01
3.57096344e-01 -3.26069444e-01 3.51446331e-01 7.91965201e-02
7.34530926e-01 2.95867443e-01 3.23362052e-01 -7.77432919e-01
1.06909680e+00 1.00541130e-01 2.66059697e-01 3.63485426e-01
-3.74820232e-02 5.69929838e-01 4.74078119e-01 -3.76509935e-01
-9.15461719e-01 -5.18618166e-01 -3.66264105e-01 1.37471986e+00
-7.53523260e-02 -4.39658970e-01 -5.49294889e-01 -9.37109590e-01
-1.86862171e-01 5.00924230e-01 -5.41883528e-01 8.31449926e-02
-3.24125528e-01 -7.89975166e-01 5.12518525e-01 4.52201813e-01
1.11759389e-02 -8.79878223e-01 -1.57716170e-01 2.85433203e-01
4.83612381e-02 -9.37304616e-01 -5.42650819e-01 6.59798324e-01
-6.90821171e-01 -8.30191016e-01 -8.19749534e-01 -1.28655124e+00
8.87523770e-01 -1.08431660e-01 1.02868629e+00 -8.54379162e-02
1.51075795e-01 2.20348701e-01 -9.57304776e-01 -5.62414706e-01
-6.23514116e-01 7.34027684e-01 2.29110315e-01 4.31899838e-02
6.94002628e-01 -1.28777266e-01 8.84637088e-02 -5.12663200e-02
-1.13474000e+00 -3.79261643e-01 5.23969471e-01 8.47629845e-01
1.10561490e-01 -1.47333175e-01 8.95135105e-01 -1.42577302e+00
9.82979238e-01 -6.19003117e-01 -3.74748707e-01 2.68634111e-01
-7.07104862e-01 2.46114597e-01 1.19498515e+00 -3.59490931e-01
-1.19182682e+00 -5.70755713e-02 -4.18009162e-01 4.08248752e-01
8.76891799e-03 8.09923172e-01 9.95548815e-02 -7.42038488e-02
7.49266922e-01 7.46076107e-02 -3.64863455e-01 -6.55517638e-01
3.09694409e-01 1.32259321e+00 -1.01844445e-02 -8.76872659e-01
7.31509328e-01 3.24829251e-01 -6.22087836e-01 -9.01130974e-01
-7.67391384e-01 -6.50724649e-01 -1.05926228e+00 -8.86697322e-02
9.03831542e-01 -9.21665490e-01 3.97204608e-01 5.84066451e-01
-1.10591435e+00 -2.88175553e-01 -6.53787032e-02 5.71819067e-01
-1.43104121e-01 3.83643419e-01 -6.24562621e-01 -4.93553191e-01
-2.29203969e-01 -1.26370108e+00 9.49037075e-01 -2.71529630e-02
-3.51243317e-01 -1.59256792e+00 5.12433112e-01 2.65240312e-01
6.15295053e-01 -2.39917189e-01 1.12457716e+00 -9.43757534e-01
3.92126769e-01 -6.02276802e-01 -9.83741805e-02 7.75032818e-01
2.88294971e-01 1.50061101e-01 -9.46286380e-01 -2.60643601e-01
-1.99786514e-01 -5.72527528e-01 5.38114548e-01 -1.61077902e-01
8.42774332e-01 2.28508394e-02 6.26968592e-02 5.62845945e-01
1.68036914e+00 2.17509657e-01 1.34656250e-01 7.12164819e-01
7.19238818e-01 7.85495579e-01 4.83984083e-01 4.69447315e-01
5.35454869e-01 2.53847718e-01 -7.18449205e-02 -3.50948051e-02
4.73179258e-02 -1.05772659e-01 6.42762363e-01 1.84392238e+00
3.18668962e-01 -2.71253765e-01 -1.14281368e+00 8.66980851e-01
-1.51335907e+00 -4.62831795e-01 -2.34427661e-01 1.71123159e+00
9.53369200e-01 1.65074497e-01 -3.73696834e-01 1.89229567e-02
5.15528202e-01 2.84717023e-01 -2.06517637e-01 -1.26067877e+00
-1.83246076e-01 5.62864840e-01 6.69591248e-01 4.58607107e-01
-1.03104746e+00 1.09536219e+00 5.85185719e+00 7.92300701e-01
-1.23267984e+00 4.43781018e-01 4.14656371e-01 2.07498759e-01
-4.22517657e-01 -9.93673801e-02 -8.84870291e-01 3.62828791e-01
1.03897631e+00 -1.87685460e-01 -1.00503303e-02 1.10363722e+00
-2.19327100e-02 9.30630565e-02 -9.00716841e-01 7.62400746e-01
6.49337471e-01 -1.16147041e+00 1.94533765e-01 -2.94155419e-01
1.17322981e+00 2.71542400e-01 1.17369760e-02 6.82979286e-01
1.41742244e-01 -7.44602799e-01 1.18076873e+00 2.61962682e-01
9.28117990e-01 -7.92983711e-01 1.18072522e+00 1.38493448e-01
-1.15973639e+00 -5.75214736e-02 -4.95599926e-01 -2.15778172e-01
-1.77879453e-01 4.76271629e-01 -1.73365951e-01 6.38051450e-01
5.86051404e-01 1.02992523e+00 -1.05099201e+00 4.72289771e-01
-6.40201792e-02 4.32705671e-01 2.00076863e-01 -4.16031092e-01
5.39853930e-01 -3.12212020e-01 -3.10315248e-02 1.81288671e+00
5.31816840e-01 -7.80690253e-01 2.11768616e-02 2.43012682e-01
-2.52107739e-01 7.61961937e-01 -7.67390013e-01 -4.10206497e-01
1.65430084e-01 1.33656514e+00 -8.52863371e-01 -3.66058499e-01
-9.87074077e-01 1.09828782e+00 2.50756532e-01 2.43970856e-01
-4.67730910e-01 -9.16763783e-01 5.63745618e-01 -2.46024176e-01
2.61859089e-01 -2.82371432e-01 -2.20412850e-01 -1.44172502e+00
3.11105214e-02 -8.52634251e-01 8.56734291e-02 -6.43579483e-01
-1.40437913e+00 1.09728754e+00 -3.36849242e-01 -1.38643134e+00
-1.13785835e-02 -1.13043952e+00 -2.25897297e-01 9.66872633e-01
-1.81793034e+00 -1.18874311e+00 1.27986893e-01 4.68840599e-01
8.79105330e-01 -8.71024549e-01 1.08899128e+00 5.87395370e-01
-5.81243694e-01 6.92291737e-01 7.44279325e-01 2.46532097e-01
1.06768560e+00 -1.19410086e+00 8.82762074e-02 1.05283499e+00
1.25539124e-01 8.56624126e-01 4.99809474e-01 -3.82973462e-01
-1.18052268e+00 -1.28590965e+00 1.51880515e+00 -6.23762012e-01
1.14903450e+00 -7.39413500e-01 -4.37931806e-01 5.12053192e-01
5.30872226e-01 -1.61034971e-01 1.19268024e+00 1.72288492e-01
-5.94998062e-01 -2.05291942e-01 -1.01053786e+00 6.30118310e-01
5.56507945e-01 -9.59890246e-01 -6.55876577e-01 3.47641319e-01
4.94995594e-01 2.78630536e-02 -7.35157013e-01 -1.01617128e-01
5.76563001e-01 -6.93114460e-01 3.52620721e-01 -4.00001675e-01
7.65409172e-01 -1.69822097e-01 -3.86422992e-01 -1.41896081e+00
-6.84308782e-02 -1.67233601e-01 5.34447610e-01 1.25600076e+00
8.60469818e-01 -5.92714131e-01 5.36538661e-01 1.71232000e-01
-3.99022937e-01 -5.08841634e-01 -6.03346407e-01 -6.28133714e-01
6.62503242e-01 -5.60578883e-01 2.66746670e-01 1.29869330e+00
2.85148472e-01 2.91060448e-01 -1.76850408e-01 -2.14012682e-01
-3.96570750e-02 7.14996681e-02 5.35824537e-01 -1.22719419e+00
6.43906593e-02 -5.38087845e-01 -3.83348167e-01 -7.18558013e-01
5.58992267e-01 -1.20559025e+00 -2.57410761e-02 -1.47744381e+00
1.87883899e-02 -5.16256571e-01 -2.97299117e-01 3.58006209e-01
1.31022155e-01 4.37718868e-01 -9.34610292e-02 -6.57336637e-02
-5.46881258e-01 3.35098892e-01 6.79142773e-01 -2.17443556e-01
2.64675934e-02 -4.11035895e-01 -9.84207213e-01 8.32895398e-01
8.49407613e-01 -7.51227558e-01 -3.70806903e-01 -5.35344779e-01
7.67730236e-01 -4.76446509e-01 -5.91217220e-01 -9.26145852e-01
-1.88035984e-02 -5.78332925e-03 1.14722058e-01 -1.89562798e-01
-2.11401910e-01 -9.03419137e-01 -5.39710462e-01 2.84110278e-01
-4.51417923e-01 3.96569848e-01 3.18173587e-01 2.31272757e-01
-5.04212379e-01 -8.91290367e-01 7.11767018e-01 1.80932805e-02
-8.27787697e-01 8.83237123e-02 -1.03524125e+00 1.78122982e-01
7.23159850e-01 -1.35237798e-01 -9.16311443e-02 1.31065631e-02
-1.47059172e-01 -1.60855845e-01 6.89208448e-01 9.91235912e-01
2.28436962e-01 -1.09614336e+00 -4.80755180e-01 2.78494596e-01
7.42716253e-01 -5.31462908e-01 -2.02156715e-02 4.98301208e-01
-1.18671668e+00 5.88705897e-01 -1.79329842e-01 -4.03370196e-03
-9.29035246e-01 3.76483738e-01 3.12649682e-02 -3.99080336e-01
-1.55067131e-01 8.68707240e-01 -1.02550879e-01 -1.15436721e+00
4.41561304e-02 -5.34071982e-01 -5.21503925e-01 5.27137935e-01
3.62446398e-01 -1.60430018e-02 4.57222104e-01 -1.02526653e+00
-4.24288690e-01 7.37669468e-01 -1.42440975e-01 -6.19220883e-02
1.41755855e+00 -1.94288179e-01 -3.66366148e-01 9.32713032e-01
1.54496586e+00 6.29748881e-01 -5.42484283e-01 -1.12560995e-01
3.43077898e-01 -2.46933773e-01 6.76814169e-02 -6.25402451e-01
-1.00702906e+00 9.13878977e-01 4.95078117e-01 1.43646123e-02
7.98411310e-01 -3.80345643e-01 6.40899062e-01 6.21317685e-01
3.83165181e-01 -1.71584904e+00 -9.03736949e-02 9.40264583e-01
5.60890377e-01 -1.20012808e+00 -4.11762036e-02 2.87024885e-01
-7.79393315e-01 1.42804933e+00 3.73844296e-01 -9.44753066e-02
1.05295801e+00 2.68060952e-01 5.83558977e-01 -1.01960199e-02
-4.82963055e-01 -9.75255221e-02 9.20606032e-02 6.98858559e-01
1.09270740e+00 -1.56581923e-01 -6.18931651e-01 5.83944499e-01
-3.10171247e-01 -1.89129844e-01 9.26814914e-01 1.15703976e+00
-2.05517828e-01 -1.52224338e+00 -1.98287919e-01 4.86817956e-01
-8.21259201e-01 -5.47186494e-01 -3.02729905e-01 5.97685337e-01
2.82847285e-01 9.97245431e-01 -7.78488144e-02 -2.26068601e-01
1.69771910e-01 3.81703198e-01 2.60497862e-03 -1.17082739e+00
-8.60870540e-01 -2.67240256e-01 1.91239432e-01 2.18575429e-02
-7.60282516e-01 -4.91366386e-01 -8.65025520e-01 -1.72981948e-01
-3.23005915e-01 3.72818470e-01 1.13732231e+00 8.35524082e-01
1.14824452e-01 5.13952672e-01 5.85807383e-01 -7.23676324e-01
-1.82257101e-01 -1.00320566e+00 -7.84702063e-01 3.58727574e-01
2.02986389e-01 -3.40567976e-01 -4.34942186e-01 3.84069204e-01] | [9.8646821975708, 10.052085876464844] |
07619254-5fbd-4bf8-856b-a4b629624fdd | external-knowledge-selection-with-weighted | 2209.02251 | null | https://arxiv.org/abs/2209.02251v1 | https://arxiv.org/pdf/2209.02251v1.pdf | External Knowledge Selection with Weighted Negative Sampling in Knowledge-grounded Task-oriented Dialogue Systems | Constructing a robust dialogue system on spoken conversations bring more challenge than written conversation. In this respect, DSTC10-Track2-Task2 is proposed, which aims to build a task-oriented dialogue (TOD) system incorporating unstructured external knowledge on a spoken conversation, extending DSTC9-Track1. This paper introduces our system containing four advanced methods: data construction, weighted negative sampling, post-training, and style transfer. We first automatically construct a large training data because DSTC10-Track2 does not release the official training set. For the knowledge selection task, we propose weighted negative sampling to train the model more fine-grained manner. We also employ post-training and style transfer for the response generation task to generate an appropriate response with a similar style to the target response. In the experiment, we investigate the effect of weighted negative sampling, post-training, and style transfer. Our model ranked 7 out of 16 teams in the objective evaluation and 6 in human evaluation. | ['Stanley Jungkyu Choi', 'Yireun Kim', 'Gyeonghun Kim', 'Hyunjik Jo', 'Hosung Song', 'Joongbo Shin', 'Janghoon Han'] | 2022-09-06 | null | null | null | null | ['task-oriented-dialogue-systems'] | ['natural-language-processing'] | [ 1.95701480e-01 5.68022311e-01 4.67588186e-01 -7.76864648e-01
-8.23854089e-01 -5.49284220e-01 9.10787582e-01 -3.60217065e-01
-6.94948673e-01 1.07347536e+00 7.79351175e-01 -4.34229560e-02
3.12716186e-01 -3.61047834e-01 1.08652189e-01 -3.17758054e-01
4.06694859e-01 9.64553595e-01 2.48055160e-01 -8.07579219e-01
4.74216312e-01 -6.46747872e-02 -1.00304973e+00 8.44240010e-01
9.89812553e-01 7.67936945e-01 2.69178957e-01 9.92803037e-01
-4.82784033e-01 9.42829072e-01 -1.16043055e+00 -3.99563760e-01
-3.42064798e-02 -8.27953517e-01 -1.41366506e+00 2.66155824e-02
-4.32051904e-03 -2.63614565e-01 2.08252206e-01 4.92244333e-01
1.04509985e+00 5.08108974e-01 6.25304282e-01 -1.05536187e+00
-6.00210249e-01 9.29634094e-01 3.72773558e-02 -2.46130869e-01
7.06120431e-01 2.81706423e-01 5.70216417e-01 -1.13398695e+00
7.13404894e-01 1.58571243e+00 3.40154201e-01 1.43375921e+00
-9.84577000e-01 -3.82254332e-01 1.07041392e-02 -2.00986251e-01
-7.92875051e-01 -7.51999080e-01 6.40616179e-01 -3.70929062e-01
1.01744676e+00 4.48640257e-01 3.65909219e-01 1.33127975e+00
-3.86062533e-01 9.06465709e-01 1.33579373e+00 -6.81673288e-01
1.30904943e-01 6.64114356e-01 2.64159560e-01 1.45202830e-01
-8.55114579e-01 -1.71322629e-01 -7.61721671e-01 -3.61310422e-01
4.78718191e-01 -7.15182900e-01 -3.57367814e-01 1.16188705e-01
-1.25704098e+00 7.76453018e-01 -1.81984887e-01 2.47151762e-01
-3.10457498e-01 -5.36296248e-01 8.66913617e-01 9.02778029e-01
6.89737260e-01 8.58310461e-01 -9.17067111e-01 -6.29023194e-01
-1.96956307e-01 3.79326463e-01 1.35241175e+00 1.17348933e+00
4.18590158e-01 2.35426612e-02 -8.81775260e-01 1.67914128e+00
-6.53749555e-02 3.61487895e-01 7.07988262e-01 -9.65119183e-01
7.85478830e-01 6.66088998e-01 3.70662510e-01 -2.71339506e-01
-4.72359926e-01 5.71722426e-02 -9.44670260e-01 -1.57584652e-01
2.94454068e-01 -9.16548371e-01 -4.91907924e-01 1.70241451e+00
2.59562463e-01 -6.87514961e-01 6.43524528e-01 8.05235803e-01
1.07094347e+00 7.45870590e-01 -9.36913416e-02 -5.31129479e-01
1.06905890e+00 -1.45807207e+00 -9.62597072e-01 -1.97575092e-02
1.02824736e+00 -1.01151848e+00 1.42695880e+00 5.58721781e-01
-1.00446069e+00 -6.83878124e-01 -5.55618763e-01 1.38391480e-01
-1.25483930e-01 1.13747567e-01 1.97106346e-01 4.54229921e-01
-1.13681459e+00 1.53068766e-01 1.55312479e-01 -3.91387939e-01
-4.56577271e-01 1.47675052e-01 -3.35072011e-01 1.26672328e-01
-1.51364291e+00 1.16435397e+00 2.84368783e-01 -3.34305763e-02
-5.28083801e-01 -5.04265428e-01 -7.41366446e-01 -4.99691308e-01
2.88103223e-01 -6.61524117e-01 1.85790694e+00 -1.30213499e+00
-2.37782598e+00 8.24954510e-01 9.63576958e-02 -2.05372542e-01
8.02686155e-01 -3.29922706e-01 -3.54381472e-01 -3.34110945e-01
-6.81237057e-02 7.32151806e-01 4.88024324e-01 -1.17531276e+00
-6.59283578e-01 2.28841063e-02 7.75688291e-02 5.67751288e-01
-3.67842555e-01 3.69069189e-01 -2.06386298e-01 -7.44148314e-01
-4.85350639e-01 -1.07774162e+00 -2.56854922e-01 -7.84087956e-01
-3.86139005e-01 -9.31521893e-01 5.30130506e-01 -5.11540294e-01
1.33974123e+00 -1.97272813e+00 3.14459950e-01 -8.77748504e-02
-4.60095778e-02 6.58208370e-01 -4.15990680e-01 7.73808062e-01
2.30008319e-01 6.44594878e-02 -1.15033604e-01 -4.53631431e-01
7.06082731e-02 -1.18063860e-01 -1.15671605e-01 -4.24281150e-01
3.90971005e-01 5.56522727e-01 -9.83594298e-01 -2.31590956e-01
-1.79968148e-01 -3.11969500e-02 -5.59056580e-01 9.58586514e-01
-5.00447750e-01 5.89093328e-01 -4.32773948e-01 -4.30833921e-02
4.67179269e-01 2.76551127e-01 1.06413983e-01 9.15959924e-02
-2.44678155e-01 5.75526834e-01 -8.83192420e-01 1.71738851e+00
-6.50175452e-01 3.22454184e-01 1.53577417e-01 -3.39294881e-01
1.41696966e+00 6.95659161e-01 2.31057382e-03 -7.07902908e-01
7.56243169e-02 1.36430219e-01 2.11665079e-01 -8.27299118e-01
1.02857685e+00 -2.58052051e-01 -4.97178882e-01 9.76154268e-01
1.79943696e-01 -5.09894192e-01 4.22347449e-02 3.05896729e-01
8.67204726e-01 -3.11318394e-02 -4.73056845e-02 -2.04161152e-01
7.10683763e-01 -1.18229970e-01 3.90012980e-01 7.37492383e-01
-2.71022439e-01 6.68719411e-01 5.01532078e-01 -2.66604125e-01
-7.59770989e-01 -4.11331862e-01 3.45738381e-01 1.61216688e+00
-4.07123268e-01 -3.53117585e-01 -1.06087935e+00 -1.14453936e+00
-4.55099523e-01 9.26731110e-01 -6.34653807e-01 -2.11075187e-01
-4.49039549e-01 -3.93021941e-01 7.79160738e-01 2.96615243e-01
6.36711657e-01 -1.59961915e+00 2.32599825e-02 4.81204003e-01
-5.79048157e-01 -8.73698056e-01 -9.08619821e-01 5.73427156e-02
-3.56522441e-01 -7.08923280e-01 -9.21312392e-01 -1.04490817e+00
4.49498743e-01 1.01578034e-01 1.26127172e+00 7.03743771e-02
3.71297151e-01 3.17712426e-01 -8.34643245e-01 -5.21765053e-01
-9.43896174e-01 3.23053598e-01 1.24452531e-01 -1.14596523e-02
3.30902517e-01 1.99194183e-03 -2.32843325e-01 7.00240612e-01
-5.48673570e-01 5.01568317e-01 2.67808050e-01 1.21012259e+00
-2.15411320e-01 -6.12477899e-01 1.26735055e+00 -1.09385884e+00
1.59255588e+00 -3.91300954e-02 9.94026214e-02 5.47828019e-01
-2.51435429e-01 8.90424550e-02 5.65384507e-01 -5.57729363e-01
-1.67586577e+00 -1.22980878e-01 -4.10697699e-01 3.90221417e-01
-2.52152473e-01 3.64145637e-01 -1.48289055e-01 3.19104522e-01
9.27136004e-01 2.33307958e-01 2.13393942e-01 -4.66529995e-01
2.73089379e-01 1.34731722e+00 1.66374281e-01 -9.08815324e-01
2.91493237e-01 -4.01516348e-01 -1.04975879e+00 -8.87874424e-01
-8.87430251e-01 -4.11225826e-01 -5.58248997e-01 -3.39910269e-01
7.28072822e-01 -6.78739488e-01 -5.32131732e-01 7.61593938e-01
-1.57310116e+00 -1.03895533e+00 -7.45409429e-02 4.03714001e-01
-5.86714387e-01 5.07943816e-02 -6.92430735e-01 -1.01726294e+00
-8.51502001e-01 -8.89578938e-01 6.86790645e-01 7.62177780e-02
-7.76200891e-01 -8.77440333e-01 4.70344156e-01 5.55257261e-01
6.78969562e-01 -3.04515958e-01 6.84292078e-01 -1.16934800e+00
3.44471276e-01 7.23140910e-02 -7.14074299e-02 7.37264872e-01
1.22292176e-01 -1.60320848e-01 -1.04500091e+00 -1.22767165e-01
8.49285498e-02 -1.17906785e+00 3.61378700e-01 -2.96399504e-01
5.29166281e-01 -3.92035335e-01 3.25682372e-01 -1.34849072e-01
4.61326987e-01 3.31913829e-01 5.78738987e-01 2.37643704e-01
5.24033964e-01 1.34388459e+00 1.03364658e+00 5.39754629e-01
5.86719215e-01 8.71544063e-01 -4.01551425e-01 1.18689612e-02
6.47121295e-02 -9.21682343e-02 5.39695978e-01 1.33958745e+00
2.81752758e-02 -3.21136981e-01 -7.98323989e-01 4.13350523e-01
-1.84937119e+00 -7.60906696e-01 -1.66578978e-01 2.07417750e+00
1.50790620e+00 8.75046700e-02 3.82682413e-01 -3.85809094e-01
7.27033377e-01 -1.54234260e-01 -1.78166345e-01 -9.14622784e-01
-1.36122927e-01 -1.18576162e-01 -3.53608906e-01 8.55388463e-01
-6.15265727e-01 1.18102300e+00 5.82257223e+00 7.09349573e-01
-1.00700092e+00 9.63067263e-02 6.40708089e-01 1.38145193e-01
-4.46061790e-01 -2.28757724e-01 -8.38518202e-01 2.91158706e-01
9.31451559e-01 -3.07523072e-01 1.83824718e-01 5.90732217e-01
4.63242382e-01 1.13792531e-01 -1.11560404e+00 6.10034883e-01
2.47545436e-01 -8.54486108e-01 1.53591290e-01 -2.90814906e-01
6.77816093e-01 -2.74718106e-01 -4.36076880e-01 9.81650531e-01
6.09335780e-01 -8.05325031e-01 4.58540797e-01 5.61912179e-01
6.22889757e-01 -7.96286345e-01 1.09698820e+00 6.44488394e-01
-4.88159269e-01 2.85711646e-01 -1.56854481e-01 -8.85346681e-02
2.96136379e-01 1.01222046e-01 -1.37932432e+00 3.86652410e-01
2.91827023e-01 1.06879540e-01 -2.95955688e-01 5.45798779e-01
-3.22073102e-01 4.95096803e-01 3.71735916e-02 -5.14833629e-01
2.04241186e-01 7.28725940e-02 3.74861449e-01 1.54748392e+00
-2.34767720e-02 3.05762112e-01 4.50339764e-01 2.83639669e-01
-1.79302678e-01 5.00766516e-01 -5.60652316e-01 1.45712569e-01
6.97537661e-01 1.14465559e+00 7.49692544e-02 -5.58768928e-01
-8.05481449e-02 1.07452488e+00 4.97840613e-01 1.86809927e-01
-3.05059552e-01 -6.48968458e-01 2.77461737e-01 -2.86484689e-01
-2.70273179e-01 3.17257196e-02 -1.01077825e-01 -9.08585370e-01
-1.36885509e-01 -1.44466925e+00 3.62837873e-02 -7.18450546e-01
-1.60406804e+00 1.15443921e+00 -1.80200502e-01 -1.08612168e+00
-7.32398987e-01 -3.02639574e-01 -6.14872873e-01 1.10766315e+00
-9.47814882e-01 -9.96106923e-01 -1.98599532e-01 5.70578218e-01
1.15022433e+00 -6.30605459e-01 1.29569781e+00 3.36581767e-02
-4.91606444e-01 8.19530010e-01 -3.03761184e-01 2.62521237e-01
1.50091732e+00 -1.33715916e+00 5.14386237e-01 3.87020595e-02
-6.45265639e-01 6.12188697e-01 6.08850002e-01 -6.65430367e-01
-9.06966090e-01 -8.94007564e-01 1.19764853e+00 -5.43527842e-01
5.54613709e-01 -5.13093531e-01 -9.28662241e-01 2.92838216e-01
8.38406980e-01 -7.93697715e-01 9.17278588e-01 2.93427736e-01
-1.48899615e-01 1.01932734e-01 -1.12025344e+00 6.60351872e-01
8.03229511e-01 -4.45554644e-01 -8.32680345e-01 2.34722599e-01
1.11884475e+00 -4.20763314e-01 -9.99697864e-01 2.51271546e-01
4.72204208e-01 -7.60685205e-01 3.57614487e-01 -5.42406857e-01
4.19557601e-01 6.83225393e-02 6.98015690e-02 -2.01210141e+00
-7.46940523e-02 -1.13844347e+00 4.04199779e-01 1.71869862e+00
8.94893944e-01 -4.97646868e-01 3.94843072e-01 9.08954978e-01
-3.88827682e-01 -4.63347703e-01 -4.89523470e-01 -4.87986982e-01
1.76037043e-01 -9.33841690e-02 5.30193925e-01 1.12760401e+00
4.80492324e-01 1.19893062e+00 -7.42802858e-01 -6.71957076e-01
7.88711086e-02 -3.47706109e-01 1.33575773e+00 -1.00409770e+00
-3.12015623e-01 -2.51277536e-01 3.56575966e-01 -1.32460690e+00
1.26912072e-01 -3.07602435e-01 5.70648670e-01 -1.33980024e+00
-4.80068028e-02 -4.72296029e-01 1.35716215e-01 3.96163851e-01
-3.87993902e-01 -1.95319191e-01 1.78726375e-01 -3.70968617e-02
-8.12508523e-01 9.88076806e-01 1.57222986e+00 3.24524008e-02
-6.88383281e-01 2.97350794e-01 -7.64564753e-01 3.99819881e-01
1.02438474e+00 -2.82403082e-01 -5.93319654e-01 -3.54979068e-01
-6.53384775e-02 3.52258295e-01 -3.37807000e-01 -3.95892650e-01
2.22695366e-01 -2.69603312e-01 -8.53794590e-02 -4.72488254e-01
3.01194340e-01 -2.68360376e-01 -4.94274288e-01 9.65738073e-02
-9.95452523e-01 1.79655150e-01 1.60422057e-01 8.71347263e-02
-3.52768481e-01 -3.26841354e-01 5.60873628e-01 -2.90856898e-01
-5.33404529e-01 -2.18258947e-01 -7.47694254e-01 5.26536345e-01
6.05653048e-01 6.81246258e-03 -4.91030186e-01 -7.26267993e-01
-6.76452458e-01 6.95426345e-01 1.03965588e-01 7.42358983e-01
6.66791201e-01 -1.28001070e+00 -1.21250820e+00 2.66348310e-02
5.24065197e-01 -1.04071461e-02 2.95269966e-01 6.36514485e-01
-1.76282063e-01 2.89272964e-01 -1.60821691e-01 -2.69285381e-01
-1.41023493e+00 -5.50834835e-02 2.63428509e-01 -4.48731393e-01
-1.22559108e-01 1.06363583e+00 1.34183079e-01 -1.39774501e+00
5.18971264e-01 1.99421614e-01 -4.96138692e-01 1.42899424e-01
7.71481037e-01 2.11384758e-01 3.05382032e-02 -2.81870395e-01
8.43046159e-02 -2.16638893e-02 -5.31100810e-01 -7.17237711e-01
1.08542347e+00 -2.87717193e-01 -2.25563377e-01 6.49792075e-01
8.74236166e-01 -6.83194548e-02 -9.93990481e-01 -5.35117567e-01
2.03181610e-01 -5.60017191e-02 -4.69028562e-01 -1.31421971e+00
-4.34530020e-01 6.79316103e-01 1.72358394e-01 4.00917888e-01
6.95478678e-01 -3.26742619e-01 7.37056911e-01 8.86050999e-01
2.97060996e-01 -1.68481350e+00 4.80196089e-01 1.38964081e+00
1.59359920e+00 -1.28765631e+00 -6.05366170e-01 -1.92710608e-01
-1.51271021e+00 1.05080760e+00 1.40951872e+00 3.36835891e-01
2.47362554e-01 1.13973901e-01 7.60518193e-01 -1.68916509e-02
-1.34750569e+00 7.76086599e-02 9.87488180e-02 7.44794011e-01
9.61980581e-01 1.25343472e-01 -5.14770269e-01 7.55401552e-01
-4.75888848e-01 -1.12230089e-02 5.54524720e-01 9.39017355e-01
-4.75688130e-01 -1.36050367e+00 -1.13564417e-01 1.55863240e-01
-4.04026173e-02 -2.20858976e-01 -1.27259219e+00 3.60238731e-01
-4.18461621e-01 1.36860931e+00 -1.88511893e-01 -8.12233150e-01
9.75724041e-01 3.37466538e-01 1.82272002e-01 -1.13666677e+00
-1.28758025e+00 6.26340657e-02 1.16257083e+00 1.58953473e-01
-2.98164457e-01 -3.03544164e-01 -9.80822384e-01 -1.78624719e-01
-4.24203694e-01 8.09921205e-01 4.30020988e-01 9.23973799e-01
2.20888749e-01 4.33559984e-01 1.14967024e+00 -7.87922800e-01
-8.27584863e-01 -1.90471041e+00 -1.90313518e-01 4.20425117e-01
2.62669902e-02 -1.59041882e-01 -2.41976649e-01 -2.82282978e-02] | [12.812052726745605, 8.064651489257812] |
4b998537-c4a6-4f2f-ae21-4b6a09272a63 | on-the-identifiability-of-markov-switching | 2305.15925 | null | https://arxiv.org/abs/2305.15925v2 | https://arxiv.org/pdf/2305.15925v2.pdf | On the Identifiability of Markov Switching Models | Identifiability of latent variable models has recently gained interest in terms of its applications to interpretability or out of distribution generalisation. In this work, we study identifiability of Markov Switching Models as a first step towards extending recent results to sequential latent variable models. We present identifiability conditions within first-order Markov dependency structures, and parametrise the transition distribution via non-linear Gaussians. Our experiments showcase the applicability of our approach for regime-dependent causal discovery and high-dimensional time series segmentation. | ['Yingzhen Li', 'Yixin Wang', 'Carles Balsells-Rodas'] | 2023-05-25 | null | null | null | null | ['causal-discovery'] | ['knowledge-base'] | [ 4.12689805e-01 8.74642730e-02 -3.70560229e-01 -3.90046299e-01
-5.57997525e-01 -6.76594436e-01 8.92862737e-01 -8.76774415e-02
9.10854563e-02 6.92332864e-01 2.52281785e-01 -7.66784906e-01
-7.60474682e-01 -3.41706097e-01 -3.65084261e-01 -8.15167904e-01
-4.93794918e-01 1.07032681e+00 -6.78938031e-02 3.60453367e-01
1.87331632e-01 7.38369286e-01 -8.49593759e-01 -1.40420794e-01
5.20914555e-01 4.14690286e-01 -1.68930247e-01 9.11994576e-01
1.08869478e-01 3.91458333e-01 -4.27529246e-01 8.75615031e-02
-1.52867839e-01 -4.93341982e-01 -8.93904448e-01 2.46976927e-01
-3.80400896e-01 -1.88247357e-02 -1.50688156e-01 8.39896977e-01
3.47137749e-02 3.87941413e-02 1.15325058e+00 -1.52632594e+00
-5.56143343e-01 7.22547293e-01 -4.78573740e-01 5.44196844e-01
2.08882317e-01 2.57223733e-02 8.82112980e-01 -1.49417654e-01
5.25780916e-01 1.84144187e+00 5.03201723e-01 2.72922367e-01
-1.85874760e+00 -5.41025519e-01 4.04827654e-01 6.04517125e-02
-1.05739057e+00 -5.14204502e-01 5.20686507e-01 -9.13879633e-01
9.35845315e-01 9.07941982e-02 2.58623838e-01 1.65827978e+00
4.87311929e-01 6.03277683e-01 1.44074416e+00 -4.88747030e-01
2.87063777e-01 -3.42460960e-01 6.24033928e-01 2.38478705e-01
2.06295535e-01 4.52683628e-01 1.32957911e-02 -6.01346135e-01
1.25663245e+00 1.02258101e-01 2.92058170e-01 -2.67678648e-01
-1.08686638e+00 1.09884012e+00 -3.56793016e-01 4.03869063e-01
-5.14198482e-01 4.68813419e-01 1.99329302e-01 4.44171429e-01
6.28323138e-01 1.02084205e-01 -4.18112874e-01 -3.93553644e-01
-8.72197688e-01 5.82231022e-03 7.14568555e-01 1.00488400e+00
2.43692964e-01 2.86531985e-01 -2.58890241e-01 4.07407135e-01
6.83694422e-01 8.33631933e-01 2.17952263e-02 -9.34740901e-01
2.24306151e-01 -6.60802820e-04 3.15326095e-01 -5.38203657e-01
-5.72387576e-01 -1.39281258e-01 -9.32617486e-01 -3.15493941e-01
3.98764282e-01 -2.23358288e-01 -1.10091054e+00 1.63191712e+00
1.48162663e-01 4.13210779e-01 1.18256457e-01 5.03123045e-01
-5.57855964e-02 7.42512047e-01 2.20676303e-01 -8.03690732e-01
1.09242117e+00 -1.90817326e-01 -1.00010335e+00 2.16786474e-01
3.44238281e-01 -5.81605732e-01 5.16708434e-01 4.10618126e-01
-9.71535385e-01 -2.65975565e-01 -1.23627037e-01 4.38793093e-01
9.16866809e-02 -3.25726897e-01 8.19406688e-01 7.29441464e-01
-9.58351433e-01 4.51358020e-01 -1.53175592e+00 -4.49510098e-01
9.30230916e-02 4.26000684e-01 -2.37617400e-02 1.41482919e-01
-1.23763275e+00 4.28600013e-01 2.82124072e-01 1.17450319e-01
-1.32698309e+00 -4.76570994e-01 -6.13581836e-01 7.16636032e-02
3.64183128e-01 -6.16144121e-01 1.30887294e+00 -6.81109130e-01
-1.47255480e+00 4.64317560e-01 -3.59506428e-01 -4.67218965e-01
4.76338714e-01 2.31723431e-02 -8.07424486e-01 3.41325291e-02
8.18150118e-02 -1.21451281e-01 1.15338886e+00 -1.14455986e+00
-3.50016057e-01 -2.86034942e-01 -1.15946636e-01 -4.40822512e-01
3.48991632e-01 4.09767836e-01 -1.00185089e-01 -6.94269538e-01
2.42226765e-01 -1.17871332e+00 -4.74659860e-01 -8.23314011e-01
-6.77452564e-01 -1.78881288e-01 6.96538746e-01 -4.88641053e-01
1.47226250e+00 -2.02012491e+00 4.91115183e-01 5.32785654e-01
9.62222442e-02 -3.05262208e-01 6.02577962e-02 8.53514850e-01
-4.05819237e-01 2.94324458e-01 -3.80168647e-01 -3.63230914e-01
1.84644267e-01 6.20598376e-01 -6.27925098e-01 8.09545100e-01
2.71638185e-01 7.65460610e-01 -8.45457494e-01 -3.53451222e-01
2.90448457e-01 3.66240561e-01 -2.55627900e-01 -2.02219039e-02
-2.06805155e-01 1.08415532e+00 -7.80183911e-01 5.03311515e-01
3.45859140e-01 -4.15364504e-01 3.70551169e-01 5.42842567e-01
-3.62525195e-01 1.21351890e-01 -1.25268841e+00 1.18447900e+00
-3.56589228e-01 5.46003878e-01 -7.44717494e-02 -1.17406571e+00
4.75639880e-01 5.98774195e-01 7.29135394e-01 -2.10764304e-01
1.59137234e-01 -2.52284110e-01 -2.46919837e-04 -4.91327971e-01
2.32615899e-02 -5.84679544e-01 -3.27410728e-01 5.76453209e-01
-6.24006577e-02 2.06323579e-01 7.83182755e-02 1.90671772e-01
1.04067409e+00 -6.72535822e-02 3.31010103e-01 -3.71087551e-01
1.14901207e-01 -3.04708213e-01 3.04329157e-01 1.03350580e+00
1.81935996e-01 9.71922576e-02 9.46730137e-01 1.01540156e-01
-9.53285396e-01 -1.28535366e+00 -6.11158133e-01 6.83307290e-01
-3.29023033e-01 -2.29770198e-01 -4.03081715e-01 -3.95287603e-01
-2.96804383e-02 9.17316198e-01 -8.42150092e-01 2.48871595e-02
-4.35034961e-01 -9.69636142e-01 4.89042699e-01 5.72026074e-01
-2.66811639e-01 -7.60513365e-01 -5.27199924e-01 5.08663654e-01
1.65609864e-03 -8.50907147e-01 1.10860318e-01 5.42768180e-01
-1.12441361e+00 -9.42550659e-01 -7.35259235e-01 -1.15767457e-01
4.48058665e-01 -1.71857759e-01 9.03021693e-01 -5.91446459e-01
1.75463408e-02 8.21029842e-01 -1.94735169e-01 -9.73481387e-02
-8.48960400e-01 7.26846084e-02 1.75753683e-01 3.18878628e-02
1.55446127e-01 -6.38818502e-01 -2.51792610e-01 4.51951355e-01
-1.20581222e+00 -2.37754375e-01 3.54638457e-01 6.48768544e-01
4.91768003e-01 1.42649993e-01 1.79667488e-01 -1.01951706e+00
6.79564476e-01 -8.57963562e-01 -1.00040734e+00 2.19393045e-01
-7.27109849e-01 3.27551037e-01 1.45829260e-01 -6.44093990e-01
-1.27586186e+00 -1.38888344e-01 1.65294603e-01 -5.18935621e-01
-6.17378533e-01 8.04196894e-01 -1.33679405e-01 5.14923751e-01
-5.85672329e-04 -1.44753501e-01 -2.63688773e-01 -6.92179799e-01
5.21594524e-01 4.59288031e-01 1.93480432e-01 -7.82304525e-01
6.61623657e-01 6.80556059e-01 4.71378744e-01 -7.44272292e-01
-3.80785167e-01 -5.76443195e-01 -9.17021632e-01 -9.05262977e-02
9.82114136e-01 -6.58993244e-01 -6.59934461e-01 3.05781752e-01
-1.13566351e+00 -6.49165928e-01 -2.87651986e-01 6.74368441e-01
-8.68970692e-01 2.82788008e-01 -7.88742542e-01 -1.26421964e+00
5.47356248e-01 -1.06166673e+00 1.19771457e+00 -3.03119898e-01
-5.00009894e-01 -1.73003232e+00 6.21334434e-01 -2.27751568e-01
1.82950981e-02 3.18748862e-01 1.09786713e+00 -5.82311690e-01
-4.06487226e-01 -1.41619012e-01 5.96986376e-02 -1.40153617e-01
1.67693794e-01 3.43500018e-01 -7.48380005e-01 -2.68053472e-01
-6.39498085e-02 3.76817286e-01 7.40457296e-01 1.18058765e+00
8.03554356e-01 -1.76114246e-01 -7.25320876e-01 2.68735886e-01
1.26757765e+00 3.14181715e-01 4.59806353e-01 1.07733950e-01
5.78342795e-01 7.09507823e-01 3.67071033e-01 7.48629391e-01
7.68145770e-02 6.27855480e-01 2.32773095e-01 -1.15431152e-01
3.75230223e-01 -2.84672212e-02 2.69514561e-01 5.74998200e-01
-1.04456194e-01 -5.20950854e-01 -1.25507390e+00 7.20863640e-01
-1.95808768e+00 -1.23174572e+00 -9.57658589e-01 2.09653330e+00
4.16908026e-01 1.32370830e-01 4.74886060e-01 -6.55457303e-02
9.46258366e-01 -1.22754134e-01 -4.74377424e-01 -4.48166579e-01
5.09633720e-02 7.85190836e-02 6.80748284e-01 8.32708299e-01
-1.05973625e+00 8.58746171e-01 8.25697803e+00 6.24277592e-01
-6.86061084e-01 3.27230036e-01 3.74634802e-01 2.06192300e-01
-4.93098170e-01 4.92183387e-01 -9.03159797e-01 3.94095331e-01
1.51334810e+00 1.66612461e-01 1.88104436e-01 1.73409045e-01
7.48438597e-01 -1.80046838e-02 -1.11742902e+00 4.97217149e-01
-5.52576244e-01 -7.83713520e-01 -1.83688074e-01 5.35163045e-01
8.97365212e-01 -2.02537015e-01 3.03683430e-01 -2.50300057e-02
9.30356443e-01 -1.04378521e+00 5.45985520e-01 6.64051235e-01
6.92794621e-01 -6.09052956e-01 2.08726883e-01 2.56081104e-01
-9.39505041e-01 -1.46527231e-01 6.36261553e-02 -3.73552650e-01
9.19825733e-01 6.48112893e-01 -6.80473506e-01 4.35782909e-01
2.85291225e-01 7.48641074e-01 -3.37536074e-03 8.04613590e-01
-2.86247790e-01 1.29978490e+00 -4.28841084e-01 5.89084208e-01
4.40616369e-01 -2.86324590e-01 8.89882028e-01 1.22549224e+00
1.65339395e-01 2.86918059e-02 1.87805668e-01 8.44421625e-01
6.60769939e-01 -5.21320045e-01 -4.10893798e-01 -3.95700574e-01
2.31751874e-01 5.91729283e-01 -1.21955812e+00 -4.61077005e-01
-2.94808596e-01 8.26613843e-01 -2.70029098e-01 7.65762508e-01
-1.02727938e+00 2.01889679e-01 3.95421356e-01 3.99510823e-02
4.58462209e-01 -5.94564676e-01 -1.64733320e-01 -1.27048898e+00
-5.32711506e-01 -5.18282473e-01 6.60630047e-01 -3.50425333e-01
-1.33573580e+00 1.51444763e-01 8.84060562e-01 -7.83014953e-01
-8.99007082e-01 -1.57776818e-01 -3.21399152e-01 9.87620533e-01
-1.15413857e+00 -1.17091525e+00 4.50973362e-01 6.01568460e-01
5.10782719e-01 2.01528594e-01 8.13922584e-01 -6.76144799e-03
-6.58028126e-01 -1.65452380e-02 6.91982865e-01 -3.54139507e-01
3.00240397e-01 -1.35031700e+00 7.23251104e-01 9.82950687e-01
1.81644008e-01 9.12875831e-01 1.30656672e+00 -1.01810515e+00
-1.02135301e+00 -7.46611059e-01 7.06299782e-01 -6.24671280e-01
1.21942210e+00 -4.01799887e-01 -9.97388780e-01 1.19862616e+00
1.69202350e-02 -5.72370172e-01 7.08356738e-01 6.19253218e-01
-4.63419743e-02 6.05473340e-01 -7.51095951e-01 2.32212022e-01
8.74861896e-01 -4.20270205e-01 -3.91556531e-01 3.00035864e-01
5.43074667e-01 8.50410759e-02 -1.03185940e+00 1.88697100e-01
4.72798377e-01 -3.86837751e-01 9.52996492e-01 -1.12588227e+00
8.73591825e-02 7.17437193e-02 6.53507095e-03 -1.10408187e+00
-7.08633602e-01 -1.09640193e+00 -2.03678995e-01 1.47654009e+00
1.93496555e-01 -6.96500003e-01 2.92549521e-01 7.20908046e-01
8.06385279e-02 1.22765213e-01 -1.19137049e+00 -1.08686221e+00
2.94097334e-01 -7.33919084e-01 5.21221638e-01 1.12091398e+00
-1.84442177e-01 3.55924010e-01 -5.92994332e-01 4.52332318e-01
7.78873384e-01 2.87421793e-01 4.97346550e-01 -1.55083704e+00
-6.60766006e-01 -3.57339174e-01 -1.31879345e-01 -1.20370424e+00
5.39306939e-01 -4.43788081e-01 -1.35322316e-02 -1.27038467e+00
2.83841044e-01 -6.07421696e-01 -3.23296040e-01 2.10908219e-01
4.25842293e-02 -2.87625700e-01 -1.61457404e-01 5.08976221e-01
-4.71452296e-01 3.38949949e-01 7.12208271e-01 8.72226134e-02
-4.36827868e-01 5.84680498e-01 -2.23533928e-01 5.11562407e-01
7.92443395e-01 -9.20148909e-01 -6.11926734e-01 -1.35340303e-01
2.29997858e-01 7.68330455e-01 7.00912297e-01 -3.79488736e-01
-2.63699651e-01 -7.35820532e-01 -7.90217593e-02 -6.71799004e-01
1.52109832e-01 -9.40058708e-01 8.68694842e-01 4.13355261e-01
-5.41172206e-01 2.25661561e-01 1.73295826e-01 1.08491707e+00
9.54512954e-02 -1.08973727e-01 4.71265227e-01 1.30279139e-02
-4.07267243e-01 1.13400161e-01 -1.08954954e+00 -1.17353059e-01
8.07183981e-01 -3.66606675e-02 1.91417903e-01 -3.79213780e-01
-1.33086789e+00 9.67092514e-02 2.09865659e-01 2.75645137e-01
2.29177400e-01 -9.75099802e-01 -4.69555050e-01 6.89952448e-02
-1.25948936e-01 -5.29292047e-01 3.60881418e-01 1.21247900e+00
2.09102094e-01 8.47890377e-01 9.69858095e-02 -8.91906559e-01
-1.21559954e+00 8.72927487e-01 -3.06144860e-02 -4.41127658e-01
-6.05485141e-01 4.11072373e-01 2.69447207e-01 9.41492915e-02
-1.15013145e-01 -5.41880012e-01 -7.24554881e-02 7.67491460e-02
1.36571974e-01 6.06687784e-01 -4.48089331e-01 -6.85309410e-01
-3.56936574e-01 3.60782921e-01 4.25704755e-02 -7.22008824e-01
1.15544939e+00 -5.80665112e-01 -4.17510346e-02 1.26493061e+00
9.52468395e-01 -3.83956313e-01 -1.47083330e+00 1.17848972e-02
3.76361459e-01 -2.59747088e-01 7.06270412e-02 -4.93495822e-01
-4.37720180e-01 1.00230062e+00 6.02620542e-01 7.72905529e-01
8.28787506e-01 4.71202254e-01 1.85370371e-01 -1.99088752e-01
1.04280211e-01 -7.31036961e-01 -4.82874483e-01 3.67375642e-01
4.90738928e-01 -9.63289678e-01 -2.37258688e-01 -3.27487856e-01
-3.57298017e-01 1.01464283e+00 -3.76299232e-01 -5.51311299e-02
9.10833895e-01 2.58228123e-01 -2.68753111e-01 -4.40170377e-01
-8.32536757e-01 -3.06319207e-01 2.64408827e-01 7.98170924e-01
3.54979157e-01 4.53432113e-01 -2.06919163e-01 3.40574831e-01
2.46946827e-01 -6.83844015e-02 4.21831012e-01 7.00571179e-01
-9.09718722e-02 -1.28657055e+00 -6.22686863e-01 2.69509524e-01
-5.74640393e-01 -1.78447425e-01 -1.53267339e-01 7.95584857e-01
-4.77098048e-01 1.20954037e+00 1.86526954e-01 2.95258284e-01
1.51355773e-01 1.01748280e-01 3.93585622e-01 -4.95937049e-01
-1.90183982e-01 8.04455400e-01 1.04365041e-02 -3.16651583e-01
-5.99935293e-01 -1.42495155e+00 -9.40047145e-01 -3.20566356e-01
-4.64543134e-01 2.98083305e-01 5.75968027e-01 1.09850132e+00
1.43695563e-01 7.05559075e-01 4.57319677e-01 -4.79577750e-01
-5.61874926e-01 -1.05828357e+00 -6.54639542e-01 6.80482686e-02
5.38270593e-01 -8.70001078e-01 -4.73529637e-01 4.29177672e-01] | [7.097782611846924, 3.935403347015381] |
eafd190c-715c-4ded-b012-30a6fd4b4a34 | using-meta-knowledge-mined-from-identifiers | 2012.09005 | null | https://arxiv.org/abs/2012.09005v1 | https://arxiv.org/pdf/2012.09005v1.pdf | Using Meta-Knowledge Mined from Identifiers to Improve Intent Recognition in Neuro-Symbolic Algorithms | In this paper we explore the use of meta-knowledge embedded in intent identifiers to improve intent recognition in conversational systems. As evidenced by the analysis of thousands of real-world chatbots and in interviews with professional chatbot curators, developers and domain experts tend to organize the set of chatbot intents by identifying them using proto-taxonomies, i.e., meta-knowledge connecting high-level, symbolic concepts shared across different intents. By using neuro-symbolic algorithms able to incorporate such proto-taxonomies to expand intent representation, we show that such mined meta-knowledge can improve accuracy in intent recognition. In a dataset with intents and example utterances from hundreds of professional chatbots, we saw improvements of more than 10% in the equal error rate (EER) in almost a third of the chatbots when we apply those algorithms in comparison to a baseline of the same algorithms without the meta-knowledge. The meta-knowledge proved to be even more relevant in detecting out-of-scope utterances, decreasing the false acceptance rate (FAR) in more than 20\% in about half of the chatbots. The experiments demonstrate that such symbolic meta-knowledge structures can be effectively mined and used by neuro-symbolic algorithms, apparently by incorporating into the learning process higher-level structures of the problem being solved. Based on these results, we also discuss how the use of mined meta-knowledge can be an answer for the challenge of knowledge acquisition in neuro-symbolic algorithms. | ['Gabriel Malfatti', 'Henrique Ferreira', 'Melina Guerra', 'Maira Gatti de Bayser', 'Mauro Pichiliani', 'Ana Appel', 'Julio Nogima', 'Heloisa Candello', 'Victor Ribeiro', 'Paulo Cavalin', 'Claudio Pinhanez'] | 2020-12-16 | null | null | null | null | ['intent-recognition'] | ['natural-language-processing'] | [ 2.24163562e-01 8.06163371e-01 1.54854983e-01 -4.19024020e-01
-2.04755932e-01 -3.82031262e-01 6.83902383e-01 -9.26983505e-02
-3.83154333e-01 7.76382685e-01 8.15305933e-02 -2.34553367e-01
-3.14510703e-01 -5.76049924e-01 -3.32010180e-01 -2.69814461e-01
-6.79386705e-02 7.12989569e-01 2.45825365e-01 -7.03279972e-01
5.02721608e-01 7.02870414e-02 -1.73997676e+00 8.50698531e-01
7.33731568e-01 5.54795325e-01 4.14362729e-01 3.64425600e-01
-4.75114256e-01 1.38897789e+00 -1.13613915e+00 -5.71112990e-01
-7.33914748e-02 -5.03903806e-01 -1.53588080e+00 -2.24809423e-01
-1.26412004e-01 -9.21248645e-02 2.22158749e-02 9.52695787e-01
-1.70696408e-01 1.17562205e-01 5.16272902e-01 -1.49346447e+00
-1.21737085e-01 1.34898317e+00 1.48294240e-01 8.48085210e-02
7.49704003e-01 5.85891530e-02 1.14105999e+00 -5.75901330e-01
8.82396102e-01 1.31719804e+00 9.51766551e-01 7.94338465e-01
-1.13565719e+00 -4.54425812e-01 -3.41033973e-02 4.00115699e-01
-1.19351959e+00 -2.70948082e-01 6.36191249e-01 -4.16758418e-01
1.60765553e+00 3.50628257e-01 5.00357687e-01 1.11093652e+00
-2.13254094e-01 6.19516492e-01 9.46663976e-01 -8.48912597e-01
1.96769178e-01 6.58442438e-01 5.87794900e-01 7.93593049e-01
-4.91422452e-02 -1.10824771e-01 -5.65350771e-01 -3.41130078e-01
6.78086579e-01 -2.55201161e-01 9.55966860e-02 2.88932204e-01
-1.00254798e+00 8.80858302e-01 9.72961783e-02 1.12065935e+00
-3.08235973e-01 9.31857452e-02 7.29553103e-01 5.79446912e-01
4.06511039e-01 1.05647349e+00 -5.59238970e-01 -7.76425123e-01
-4.37428147e-01 1.46296352e-01 1.55044878e+00 9.65554714e-01
9.93458688e-01 -1.64817631e-01 1.64094511e-02 9.92168367e-01
-1.41182035e-01 -9.91723612e-02 8.38757873e-01 -1.26382065e+00
3.68969262e-01 1.11749005e+00 2.91662123e-02 -9.12892282e-01
-5.83171487e-01 -1.09801963e-01 -3.89737993e-01 8.25248882e-02
5.68361461e-01 -2.46023327e-01 -4.26267654e-01 1.60411251e+00
6.26276359e-02 1.53233349e-01 -2.69111712e-02 4.23509508e-01
5.77055871e-01 2.45488286e-01 -1.61411792e-01 -4.02612448e-01
1.43874645e+00 -7.36848056e-01 -6.94358528e-01 -2.00590536e-01
1.24029171e+00 -3.61628354e-01 8.70507717e-01 3.73082399e-01
-6.49400830e-01 -3.84143114e-01 -8.31658602e-01 3.66087496e-01
-5.42682171e-01 -1.42077222e-01 9.32224691e-01 7.97517300e-01
-9.10705984e-01 7.82428741e-01 -5.49026072e-01 -3.87736797e-01
7.88192227e-02 6.83654189e-01 -3.58069152e-01 3.24083507e-01
-1.36847699e+00 1.20545113e+00 7.44225502e-01 -1.70036241e-01
-5.40956855e-01 -7.82563567e-01 -7.21977055e-01 7.46776611e-02
6.66012585e-01 -3.23295504e-01 1.47748530e+00 -1.19970012e+00
-1.61345267e+00 8.44058037e-01 4.77201492e-02 -6.23699486e-01
1.71360254e-01 4.52415012e-02 -1.58832416e-01 7.10538253e-02
2.86800228e-02 3.85625660e-01 6.78953946e-01 -9.24755812e-01
-7.67224491e-01 6.59480970e-03 7.24913657e-01 -9.27997231e-02
-4.04796332e-01 2.44736254e-01 2.66315788e-03 -2.41242126e-01
-7.31859803e-02 -9.57630038e-01 -1.25800252e-01 -6.15596712e-01
-1.77381411e-01 -6.61860287e-01 8.61960292e-01 -3.60133380e-01
1.08930230e+00 -1.77477181e+00 1.66424632e-01 2.57128388e-01
5.07920921e-01 3.89246017e-01 -1.00555778e-01 6.31429613e-01
-1.67202011e-01 3.03312093e-01 4.50155102e-02 -2.03755543e-01
8.79309624e-02 5.30101418e-01 -1.28793716e-01 -1.21039316e-01
1.81412727e-01 7.52993226e-01 -1.08440387e+00 -4.11924332e-01
3.12953413e-01 6.17549308e-02 -3.99997801e-01 2.79518247e-01
-4.09586757e-01 -7.05129132e-02 -4.65536237e-01 2.20991611e-01
-9.66774002e-02 -2.06986561e-01 5.24937570e-01 2.01867938e-01
5.86464144e-02 8.98142993e-01 -9.06541049e-01 1.42992473e+00
-8.79772007e-01 8.03090572e-01 -2.23208189e-01 -1.27853644e+00
1.02133381e+00 7.61342645e-01 2.59300262e-01 -2.54658461e-01
2.28597209e-01 2.46877819e-01 3.77613157e-01 -6.13249421e-01
4.53202635e-01 -3.72277290e-01 -9.60123837e-02 9.38065827e-01
4.91833031e-01 -9.04810056e-02 4.85205799e-01 6.22209273e-02
1.34820175e+00 -1.18934326e-01 5.21302760e-01 -8.39303359e-02
8.02243710e-01 3.92634809e-01 2.21836552e-01 1.01953518e+00
1.10628761e-01 8.00304785e-02 8.84194016e-01 -6.76923156e-01
-7.61631548e-01 -1.66930482e-01 7.15605915e-02 1.50806856e+00
-3.87240946e-01 -5.58269918e-01 -1.05451977e+00 -1.02856588e+00
-3.07671309e-01 8.70786309e-01 -6.48295999e-01 -2.35987425e-01
-7.93051004e-01 -4.80429649e-01 1.21703899e+00 3.80163163e-01
3.29128832e-01 -1.65316904e+00 -9.02141094e-01 5.36929905e-01
-2.69302100e-01 -1.28853238e+00 3.06853116e-01 6.35044217e-01
-8.54560018e-01 -1.23224127e+00 -1.53603643e-01 -6.18447423e-01
3.87554467e-01 -1.49254560e-01 1.17837608e+00 7.10770965e-01
-2.63638765e-01 2.97343343e-01 -8.61053944e-01 -4.36194152e-01
-1.25795805e+00 3.85808200e-01 7.49063119e-02 -4.16698277e-01
6.37354076e-01 -9.93120909e-01 5.22373259e-01 4.11027044e-01
-4.54128027e-01 3.99658494e-02 3.23810875e-01 1.01557779e+00
-5.70213258e-01 -5.12815528e-02 8.65635276e-01 -1.16783559e+00
9.51933146e-01 -3.43743384e-01 -4.97960806e-01 2.22881928e-01
-4.36786354e-01 3.48968148e-01 7.07804620e-01 -6.83998704e-01
-1.08878374e+00 -1.25894129e-01 -1.80793777e-01 -1.93112165e-01
-5.02952695e-01 7.17132449e-01 2.62504578e-01 -2.36348689e-01
9.70495045e-01 8.74603763e-02 2.15934783e-01 -4.54785407e-01
2.56644547e-01 8.71643841e-01 2.49977037e-01 -8.17824364e-01
5.25877297e-01 -2.70532891e-02 -1.40086249e-01 -9.89920139e-01
-7.67204881e-01 -7.34759510e-01 -7.46228456e-01 -2.41064668e-01
5.10099471e-01 -2.77568728e-01 -1.16987491e+00 2.54457295e-01
-1.64160991e+00 -4.62656617e-01 -2.32772574e-01 2.91640788e-01
-7.26743639e-01 3.71838212e-01 -6.06363773e-01 -1.11989772e+00
-1.04667306e-01 -1.00476015e+00 5.96386254e-01 -1.30517587e-01
-8.53538334e-01 -1.10832202e+00 3.63266817e-03 3.43161255e-01
3.56137127e-01 -1.11055665e-01 1.15897596e+00 -1.50316751e+00
-1.31525666e-01 -1.56087592e-01 7.25005716e-02 4.93373305e-01
1.67217016e-01 -4.03562725e-01 -1.23031092e+00 4.81445909e-01
3.70982826e-01 -3.51289928e-01 2.37154722e-01 -3.75908852e-01
5.64951599e-01 -5.61420560e-01 -3.59640330e-01 -1.70097381e-01
8.28563094e-01 4.73842382e-01 5.17257631e-01 3.95795554e-01
4.42680299e-01 1.01262200e+00 6.00560725e-01 3.86655778e-01
3.30172516e-02 9.19743598e-01 2.91507274e-01 5.46788216e-01
1.62944540e-01 2.43953422e-01 3.46029848e-01 7.62386084e-01
-4.91146356e-01 1.81778848e-01 -1.25812817e+00 5.35230994e-01
-1.96807718e+00 -1.28339708e+00 -1.72649488e-01 1.62406921e+00
1.21367860e+00 3.23235869e-01 3.57573926e-01 4.12263274e-01
7.97570348e-01 -1.76511317e-01 -4.76576313e-02 -6.44380629e-01
4.46847081e-01 1.83414072e-01 -4.47439700e-02 5.57500422e-01
-7.97712982e-01 9.82040465e-01 6.33500338e+00 7.50104070e-01
-8.10570419e-01 3.26759398e-01 1.65032730e-01 3.47355366e-01
1.98936820e-01 3.25440355e-02 -5.99404037e-01 9.35679600e-02
1.23313510e+00 -8.84796083e-02 7.30300188e-01 1.27112925e+00
-2.20092282e-01 -2.01007888e-01 -1.50959432e+00 6.29811406e-01
1.88932970e-01 -1.62606251e+00 -4.19534475e-01 -1.14537530e-01
4.30361539e-01 -1.14597447e-01 -6.68450117e-01 8.29739869e-01
3.72105569e-01 -9.87207413e-01 5.33654571e-01 2.49100000e-01
2.59612232e-01 -4.35990363e-01 1.14720523e+00 9.79010344e-01
-8.47838581e-01 -1.99954674e-01 -2.91969359e-01 -7.55761325e-01
-3.18098739e-02 2.36082897e-01 -1.63522434e+00 5.16394198e-01
3.05966854e-01 4.85470086e-01 -4.19291168e-01 6.08922660e-01
-2.56699115e-01 6.10858917e-01 -4.22269791e-01 -6.60670698e-01
3.31157565e-01 1.07249439e-01 5.69765091e-01 1.46809387e+00
-3.09598863e-01 1.95816740e-01 -1.80200666e-01 1.24032545e+00
1.98517576e-01 -3.19579095e-01 -6.74199402e-01 -1.99812725e-01
4.31805015e-01 1.17027521e+00 -6.83275044e-01 -5.39520204e-01
-1.15108266e-01 6.74546361e-01 3.61782789e-01 1.31569162e-01
-6.92629814e-01 -5.72752237e-01 5.76506555e-01 -2.59888679e-01
1.02609150e-01 1.69804931e-01 -3.50139052e-01 -7.79552400e-01
-2.75478363e-02 -1.14665008e+00 5.93161061e-02 -7.26294577e-01
-1.06815338e+00 8.66499543e-01 4.65230107e-01 -8.82243037e-01
-1.01941514e+00 -8.67703557e-01 -7.32212305e-01 8.51958156e-01
-6.65967941e-01 -9.95426416e-01 -6.94305375e-02 2.94431448e-01
5.33952355e-01 -3.03509176e-01 1.20859194e+00 -1.77347399e-02
-1.56889722e-01 3.84089261e-01 -5.74504673e-01 4.26117986e-01
1.57631487e-01 -1.18842006e+00 2.29923233e-01 3.58234048e-01
4.57747728e-01 1.18131828e+00 8.99585724e-01 -4.40080673e-01
-1.13451242e+00 -6.21113300e-01 1.11197329e+00 -7.90307343e-01
8.14119995e-01 -4.85318422e-01 -1.15644145e+00 8.64454269e-01
9.53287110e-02 -5.53634763e-01 4.28748459e-01 6.89339161e-01
-6.30501449e-01 3.88265520e-01 -1.12994659e+00 4.03974891e-01
1.11960101e+00 -8.84891927e-01 -1.45846677e+00 4.69324797e-01
8.34845006e-01 -3.03354293e-01 -7.17545509e-01 1.03395015e-01
5.55088401e-01 -1.20296979e+00 8.42172563e-01 -8.07619393e-01
3.99425864e-01 7.95269310e-02 1.59199908e-01 -1.02088642e+00
8.05386305e-02 -5.62318921e-01 -7.55948201e-02 1.21602321e+00
4.97544676e-01 -6.37402296e-01 7.70874083e-01 6.97406828e-01
-3.36039186e-01 -5.29583097e-01 -1.16645002e+00 -8.74235690e-01
-2.78508902e-01 -8.70978475e-01 4.74200249e-01 1.09510994e+00
1.03635979e+00 5.08886814e-01 -1.13117948e-01 -2.67977029e-01
1.25348680e-02 -1.81357473e-01 6.63132071e-01 -1.43828321e+00
-4.41023707e-01 -5.24101555e-01 -6.37844324e-01 -5.87271094e-01
6.85322762e-01 -9.71132517e-01 2.88815975e-01 -1.28372335e+00
1.64867416e-02 -4.47543949e-01 7.21184313e-02 1.05010867e+00
2.04739630e-01 9.32622254e-02 2.16382220e-01 6.37080967e-02
-6.43669665e-01 -7.08417818e-02 6.16772771e-01 -2.74110157e-02
-3.37742150e-01 3.13329473e-02 -4.44632232e-01 1.17092907e+00
7.23065495e-01 -9.44201648e-01 -3.23381186e-01 1.52097389e-01
3.86716783e-01 9.64718461e-02 3.48528862e-01 -9.38012183e-01
2.81238645e-01 -2.19871193e-01 -5.62012613e-01 -1.09620176e-01
4.85577822e-01 -9.71077263e-01 5.25836386e-02 6.55700684e-01
-4.13446039e-01 -3.70222062e-01 2.26860806e-01 2.20530629e-01
-1.90461889e-01 -1.00731242e+00 2.22619310e-01 -4.45810348e-01
-7.83247650e-01 -8.05155516e-01 -9.25776005e-01 1.34549811e-01
7.75470018e-01 -4.58096147e-01 -3.27011883e-01 -4.50960487e-01
-6.10253692e-01 -2.88318574e-01 8.79692566e-03 4.60673422e-01
2.66018838e-01 -6.90507472e-01 -2.77808458e-01 6.94725439e-02
7.22415522e-02 -4.51918632e-01 -9.34380367e-02 9.64323640e-01
-3.96509796e-01 5.10551333e-01 -2.73947716e-01 -5.60820520e-01
-1.46533084e+00 2.33898863e-01 4.49097872e-01 -4.83433396e-01
-5.42848766e-01 8.58561575e-01 -2.02948511e-01 -6.19520664e-01
4.35367435e-01 -6.99843109e-01 -3.37675780e-01 7.61450604e-02
5.36721706e-01 3.39439213e-01 1.64581031e-01 -3.14646959e-01
-5.64851642e-01 7.10218996e-02 -9.03356075e-02 -2.79234171e-01
1.38019001e+00 4.18742895e-01 -4.58185941e-01 7.66688764e-01
8.67121339e-01 -2.51095921e-01 -4.38475758e-01 7.05930367e-02
7.42933512e-01 2.96798144e-02 -5.45300782e-01 -9.37566757e-01
-3.58917415e-01 7.60572433e-01 -1.49707764e-01 7.36795068e-01
6.73397720e-01 1.84341297e-01 2.67589837e-01 1.27158117e+00
8.10135186e-01 -9.60558772e-01 2.42417574e-01 1.16762710e+00
9.07251656e-01 -9.46172535e-01 -2.30250582e-01 -5.24248540e-01
-8.18419099e-01 1.43127298e+00 5.57116091e-01 -3.15690748e-02
2.85339117e-01 4.30963755e-01 8.64634812e-02 -4.88683850e-01
-9.75737453e-01 -2.97340993e-02 4.44803461e-02 7.86737025e-01
4.74137485e-01 -1.37954026e-01 -1.99095815e-01 9.31725860e-01
-2.12048426e-01 1.97421443e-02 7.78070211e-01 9.47765291e-01
-5.89631200e-01 -1.31705511e+00 -4.14578438e-01 4.71118271e-01
-3.32682133e-01 -1.32843211e-01 -6.07094646e-01 1.09799230e+00
2.05510780e-01 1.11002147e+00 -1.20400414e-01 -7.53946424e-01
1.94963738e-01 7.69299746e-01 2.13522434e-01 -1.19148874e+00
-1.46536815e+00 -5.61299860e-01 8.26078236e-01 -5.32076955e-01
-7.59584785e-01 -3.27019513e-01 -1.28985608e+00 -9.66066048e-02
-6.57157600e-01 4.31311756e-01 5.07806957e-01 1.63885641e+00
3.75022963e-02 4.90313768e-01 2.63198197e-01 -7.64519155e-01
-6.90861523e-01 -1.36512530e+00 -2.84613907e-01 2.36439869e-01
-3.87106091e-02 -7.61556506e-01 -6.26463771e-01 1.70238778e-01] | [12.44571590423584, 7.813089847564697] |
28777994-0b53-4e7f-bc50-0390d601261c | are-natural-language-inference-models | 2004.03066 | null | https://arxiv.org/abs/2004.03066v2 | https://arxiv.org/pdf/2004.03066v2.pdf | Are Natural Language Inference Models IMPPRESsive? Learning IMPlicature and PRESupposition | Natural language inference (NLI) is an increasingly important task for natural language understanding, which requires one to infer whether a sentence entails another. However, the ability of NLI models to make pragmatic inferences remains understudied. We create an IMPlicature and PRESupposition diagnostic dataset (IMPPRES), consisting of >25k semiautomatically generated sentence pairs illustrating well-studied pragmatic inference types. We use IMPPRES to evaluate whether BERT, InferSent, and BOW NLI models trained on MultiNLI (Williams et al., 2018) learn to make pragmatic inferences. Although MultiNLI appears to contain very few pairs illustrating these inference types, we find that BERT learns to draw pragmatic inferences. It reliably treats scalar implicatures triggered by "some" as entailments. For some presupposition triggers like "only", BERT reliably recognizes the presupposition as an entailment, even when the trigger is embedded under an entailment canceling operator like negation. BOW and InferSent show weaker evidence of pragmatic reasoning. We conclude that NLI training encourages models to learn some, but not all, pragmatic inferences. | ['Paloma Jeretic', 'Suvrat Bhooshan', 'Alex Warstadt', 'Adina Williams'] | 2020-04-07 | are-natural-language-inference-models-1 | https://aclanthology.org/2020.acl-main.768 | https://aclanthology.org/2020.acl-main.768.pdf | acl-2020-6 | ['implicatures'] | ['natural-language-processing'] | [ 3.45650733e-01 1.02567863e+00 -2.62844115e-01 -6.49825931e-01
-8.07906568e-01 -7.53134012e-01 9.44128811e-01 6.90057799e-02
-2.69602567e-01 7.47047603e-01 9.46052432e-01 -8.31058443e-01
-3.80683213e-01 -4.83763844e-01 -7.91129589e-01 -2.57987171e-01
1.33963495e-01 8.45926940e-01 -7.76600242e-02 -2.97768414e-01
1.46077052e-01 -8.80700201e-02 -9.00701225e-01 1.00081158e+00
4.81299073e-01 6.33344948e-01 1.93679348e-01 4.28912818e-01
9.32806171e-03 1.47519732e+00 -5.71634352e-01 -6.44750178e-01
-8.23706910e-02 -4.20769602e-01 -1.39304161e+00 -6.08791411e-01
2.93932647e-01 -4.77184325e-01 4.13287617e-02 7.51115561e-01
2.73204923e-01 -2.70137787e-01 8.48362207e-01 -1.08395612e+00
-1.04418898e+00 1.64752841e+00 -7.23681748e-02 5.87191701e-01
7.01050878e-01 5.82787275e-01 2.03375721e+00 -6.14793420e-01
8.58971715e-01 1.77630377e+00 7.18190432e-01 6.08948708e-01
-1.36545658e+00 -1.45851165e-01 6.37898892e-02 2.93706834e-01
-7.91743755e-01 -5.04929483e-01 7.78948963e-01 -3.61584961e-01
1.44020557e+00 4.02140468e-01 6.15673363e-01 1.36583281e+00
-1.41304225e-01 1.24970078e+00 1.45612490e+00 -5.05633831e-01
1.60260573e-01 -9.11873579e-02 4.61800814e-01 1.44930795e-01
-1.14594176e-02 5.72557980e-03 -4.94882464e-01 -1.72860309e-01
1.41544700e-01 -7.72491336e-01 -3.28253210e-01 4.44629967e-01
-1.55165279e+00 9.22217965e-01 4.38574791e-01 3.17923158e-01
-4.28333461e-01 1.98548436e-01 2.92558998e-01 4.16616678e-01
-2.81546444e-01 1.02695692e+00 -8.50784004e-01 -4.55633730e-01
-1.12987690e-01 5.09942114e-01 1.07155991e+00 9.21665549e-01
3.47346574e-01 -6.76401436e-01 -2.24192828e-01 8.46574783e-01
5.35137892e-01 4.50546473e-01 4.32185531e-01 -1.53190815e+00
5.83407700e-01 5.17148197e-01 3.59371752e-01 -7.76340902e-01
-4.79079604e-01 5.80830038e-01 -1.87787786e-01 -3.27177763e-01
7.39580274e-01 -2.26433977e-01 2.52971739e-01 2.13182592e+00
2.91414708e-01 -2.16500059e-01 6.26359522e-01 7.73553073e-01
1.28055000e+00 6.26905620e-01 4.16482180e-01 -4.85489927e-02
1.78691041e+00 -5.57387173e-01 -7.00739980e-01 -7.26869881e-01
8.26883256e-01 -7.56380200e-01 1.72478259e+00 8.87445435e-02
-1.34382510e+00 -3.89945023e-02 -9.32127357e-01 -5.17172813e-01
1.90672889e-01 -5.12556061e-02 8.69105220e-01 -1.60734169e-02
-8.00668538e-01 5.08810639e-01 -3.72496217e-01 -1.62712201e-01
4.53759462e-01 -3.46062705e-02 -3.85733753e-01 1.23119339e-01
-1.48552203e+00 1.32084394e+00 5.86910307e-01 3.39825988e-01
-4.72309828e-01 -7.29837596e-01 -9.98869836e-01 8.94722119e-02
3.49425405e-01 -6.41602933e-01 1.95745850e+00 -8.32971215e-01
-1.65057802e+00 1.25293505e+00 -8.10859799e-02 -8.78342748e-01
2.96062708e-01 -1.55780047e-01 -2.07531139e-01 3.42806250e-01
2.33499914e-01 1.03572416e+00 5.54178178e-01 -7.25311220e-01
-6.20762289e-01 -1.39138803e-01 7.75643885e-01 4.32839900e-01
3.24272007e-01 5.61510742e-01 5.61269701e-01 8.40662196e-02
2.05992222e-01 -7.10379958e-01 4.88388538e-01 3.19322236e-02
-8.43697071e-01 -1.10776842e+00 6.72870874e-01 -4.93421435e-01
9.31041241e-01 -2.25314736e+00 6.81539327e-02 -4.97675776e-01
1.28008693e-01 -1.24129660e-01 -2.06085555e-02 3.74697536e-01
-6.00078609e-03 3.83815676e-01 -2.86959499e-01 -1.98699713e-01
6.67218328e-01 8.01637828e-01 -5.52791893e-01 2.13031739e-01
6.57007515e-01 1.28910005e+00 -9.73847926e-01 -4.39683557e-01
1.75818741e-01 -4.60900404e-02 -9.72135544e-01 4.82387617e-02
-8.84821177e-01 3.16517591e-01 2.08424474e-03 3.42534393e-01
2.66591936e-01 -4.38168406e-01 7.29485869e-01 -2.35499412e-01
-8.17649141e-02 1.55036902e+00 -8.62563133e-01 1.17048025e+00
-3.74950588e-01 8.28692257e-01 2.75575280e-01 -1.02249539e+00
3.27284694e-01 5.83970428e-01 -3.20868492e-01 -2.54756391e-01
1.81023479e-01 4.36519176e-01 7.82321513e-01 -8.45655024e-01
5.38169853e-02 -7.31856346e-01 -4.46207762e-01 5.25958955e-01
-3.28034699e-01 -5.42866349e-01 7.29044154e-02 2.61065185e-01
1.13715029e+00 -8.94507486e-03 6.90135598e-01 -5.17395973e-01
4.60985124e-01 2.34193817e-01 8.17105472e-01 7.51169980e-01
-4.33099061e-01 1.25834391e-01 1.00140357e+00 -3.13819975e-01
-7.25751400e-01 -1.43445766e+00 -6.32582009e-01 8.19143474e-01
-6.18023500e-02 -2.74233222e-01 -4.73426491e-01 -6.32539570e-01
3.93369012e-02 1.35377586e+00 -5.01231790e-01 1.52826563e-01
-1.04552627e+00 -5.06847024e-01 6.31296813e-01 5.10544777e-01
3.89273375e-01 -1.85124934e+00 -6.64803982e-01 1.59769461e-01
-6.37862027e-01 -1.34635484e+00 -5.95358619e-03 -3.55101153e-02
-4.68000621e-01 -1.20277953e+00 3.84116411e-01 -8.51499319e-01
2.01522082e-01 -4.06023115e-01 1.30717707e+00 1.83196411e-01
2.56923437e-01 1.81568414e-01 -2.34604761e-01 -4.28816140e-01
-1.04817796e+00 -3.95246327e-01 -5.82437627e-02 -5.41503191e-01
8.86797428e-01 -5.06204784e-01 -4.32334036e-01 -7.00926350e-04
-4.71876562e-01 2.08254695e-01 3.11849147e-01 9.43678200e-01
2.39497989e-01 -3.60851556e-01 6.77383423e-01 -9.09848928e-01
8.06435704e-01 -4.76367801e-01 -2.09333524e-01 -3.65625583e-02
3.46786864e-02 6.22398145e-02 5.54264903e-01 -4.08193797e-01
-1.53373575e+00 -7.32350171e-01 -5.19370437e-01 5.61251223e-01
-3.79192948e-01 5.70014238e-01 -2.48387754e-01 8.11153829e-01
7.10369527e-01 -3.79381061e-01 2.11856440e-01 -4.62086260e-01
5.46296835e-01 9.52450216e-01 3.78246605e-01 -1.19156170e+00
3.47179115e-01 4.59678799e-01 -4.12786692e-01 -9.25071120e-01
-1.62577498e+00 -1.29449129e-01 -4.48139757e-01 2.36863479e-01
1.04890621e+00 -8.20770442e-01 -1.16717291e+00 1.42052367e-01
-1.38046134e+00 -8.86189163e-01 -5.62698185e-01 4.18902099e-01
-8.43264937e-01 4.50072557e-01 -1.13314486e+00 -4.44092453e-01
-3.64915013e-01 -1.11749971e+00 9.41064775e-01 -1.69400916e-01
-1.40234315e+00 -1.03223336e+00 -2.60946423e-01 7.07631886e-01
1.79072157e-01 -6.44997135e-02 1.52340555e+00 -9.55308497e-01
-2.58827180e-01 2.55498171e-01 -2.70500720e-01 4.72613782e-01
1.84772968e-01 2.05471084e-01 -9.22541142e-01 2.50132710e-01
3.08981359e-01 -5.83045483e-01 1.53162017e-01 3.15768212e-01
5.87627769e-01 -9.71097171e-01 -4.90924381e-02 3.00002061e-02
8.74343753e-01 -6.45249113e-02 5.93269944e-01 1.76925734e-01
7.15033859e-02 8.17126274e-01 6.30967736e-01 -5.24121709e-02
7.37751245e-01 4.05667186e-01 -3.59900892e-02 7.71791875e-01
-2.74858445e-01 -3.70511621e-01 4.85053062e-01 6.81467474e-01
5.70912719e-01 1.49278134e-01 -9.71692383e-01 8.09099853e-01
-1.67683780e+00 -1.18748689e+00 -3.25983942e-01 1.56579208e+00
1.60714698e+00 4.87190336e-01 -2.82144785e-01 2.30065614e-01
3.83978635e-01 1.33097172e-01 -3.49304676e-01 -5.73741198e-01
-3.60221386e-01 1.54396951e-01 -5.88260531e-01 1.08771312e+00
-8.08850765e-01 1.22603238e+00 5.93207836e+00 3.54687244e-01
-7.88797855e-01 1.58242494e-01 4.41412181e-01 1.15748920e-01
-6.46145165e-01 3.03925633e-01 -6.17883801e-01 4.46923852e-01
9.03554738e-01 1.63340688e-01 4.15086269e-01 7.21257031e-01
2.10880518e-01 -3.12254250e-01 -1.79789913e+00 3.54520202e-01
-3.56541455e-01 -1.65897393e+00 -1.81005433e-01 -3.57417792e-01
3.94119859e-01 1.01248026e-01 -2.81997502e-01 6.04191720e-01
4.87198949e-01 -9.19211030e-01 9.80992436e-01 2.02460005e-03
4.74480420e-01 2.44948901e-02 6.01508915e-01 5.39061904e-01
-4.60172951e-01 -1.65586382e-01 -3.07903320e-01 -9.77093101e-01
4.54088688e-01 2.06024259e-01 -1.00509274e+00 -4.45292860e-01
6.02602839e-01 6.22401118e-01 -1.39132187e-01 2.32200652e-01
-1.35528791e+00 9.90751147e-01 -7.15314925e-01 -5.27863145e-01
2.54570067e-01 -1.33107871e-01 7.68812001e-01 9.18578029e-01
-3.43715966e-01 3.99412453e-01 -2.66673207e-01 1.30620193e+00
-6.51509385e-04 -5.26138693e-02 -6.40256405e-01 -1.90521315e-01
8.70978832e-01 9.80468869e-01 -2.57597536e-01 -4.90291417e-01
-4.18397874e-01 6.41198575e-01 6.98463917e-01 3.16509455e-02
-6.12825096e-01 2.59346336e-01 7.91238546e-01 -1.88324824e-01
1.70884803e-02 1.50031954e-01 -4.28397059e-01 -9.24925685e-01
2.65662491e-01 -9.18364942e-01 5.03110886e-01 -1.25416851e+00
-1.60944581e+00 -6.05666265e-02 3.41308832e-01 -5.52153528e-01
-5.18007815e-01 -1.02559197e+00 -9.42388356e-01 5.70955694e-01
-1.44678330e+00 -8.66077363e-01 2.28052005e-01 7.59047121e-02
4.53132361e-01 5.06380260e-01 9.77300048e-01 -3.55373949e-01
-3.72898370e-01 1.67768613e-01 -6.43113673e-01 3.68470758e-01
4.83309090e-01 -1.46963561e+00 4.22366232e-01 3.77501220e-01
1.03699796e-01 1.14871895e+00 9.63810444e-01 -3.24471384e-01
-1.22303140e+00 -3.96917582e-01 1.46848512e+00 -9.16565418e-01
1.28362203e+00 -6.42520860e-02 -8.87758911e-01 1.43334115e+00
6.87698960e-01 -2.03838065e-01 8.27360392e-01 5.26231349e-01
-5.23545027e-01 1.76925287e-01 -1.31437790e+00 1.04428327e+00
1.16608298e+00 -7.88777471e-01 -1.91331100e+00 8.07246923e-01
1.22281528e+00 -4.67248082e-01 -9.64058757e-01 3.98889661e-01
1.64247274e-01 -8.04759204e-01 8.38991702e-01 -9.99496818e-01
9.44856882e-01 -1.09243460e-01 -3.12984556e-01 -1.00783455e+00
3.67069244e-02 -7.01471925e-01 -1.86331198e-01 1.12706852e+00
9.12022054e-01 -7.67904222e-01 2.54865348e-01 9.90714371e-01
-1.72123328e-01 -6.01688743e-01 -1.33641529e+00 -4.29438263e-01
5.63748360e-01 -6.88636303e-01 5.92232764e-01 8.76222849e-01
1.04158974e+00 1.19909227e+00 2.89742500e-01 -8.14275164e-03
2.22365499e-01 5.01468718e-01 5.48918545e-01 -1.18174052e+00
-6.59089327e-01 -6.80329323e-01 1.62954167e-01 -1.20260429e+00
5.59821665e-01 -1.15570641e+00 2.09681481e-01 -1.67857325e+00
4.55981940e-02 -6.36773705e-01 4.38296288e-01 6.68724239e-01
-1.68142363e-01 -1.27844095e-01 2.64932327e-02 2.17020258e-01
-3.42665941e-01 3.95523250e-01 1.23350787e+00 -1.75704584e-01
6.18778765e-02 -2.19619825e-01 -1.14552033e+00 1.43190098e+00
9.13531780e-01 -1.73047051e-01 -3.28579605e-01 -5.51252246e-01
7.99444020e-01 1.90816559e-02 4.78505969e-01 -4.62018847e-01
-1.88267410e-01 -2.83823311e-01 -3.67264062e-01 -1.08321592e-01
5.16651511e-01 -3.91780585e-01 -4.20020312e-01 2.72749931e-01
-7.74250090e-01 -1.65237207e-02 2.26276949e-01 2.19043583e-01
1.27185434e-01 -6.20461106e-01 5.78423202e-01 -2.93956816e-01
-6.49053633e-01 -5.86649716e-01 -5.67040205e-01 8.72174561e-01
5.97687602e-01 4.96609837e-01 -7.69245267e-01 -2.38561630e-01
-8.58956814e-01 1.62187532e-01 2.93721668e-02 1.56426013e-01
5.88086426e-01 -9.52076614e-01 -8.01500559e-01 -2.77734548e-01
1.88932464e-01 3.39169741e-01 3.51130515e-02 1.22741592e+00
-5.25985003e-01 2.85242289e-01 3.59654188e-01 -6.06208980e-01
-8.80609035e-01 4.21908349e-01 2.81436592e-01 -9.89842638e-02
-1.01191664e+00 1.23264873e+00 1.75972432e-02 -1.02542245e+00
-5.89471236e-02 -1.02074111e+00 -5.49559831e-04 -1.53884858e-01
5.66615760e-01 -3.22999865e-01 -4.74832237e-01 -4.72196490e-01
-3.96582574e-01 -1.51115447e-01 -8.02859887e-02 -1.09587289e-01
1.12989819e+00 -4.01209556e-02 -6.27097726e-01 6.30755246e-01
9.55023408e-01 2.52445400e-01 -9.37940180e-01 -2.75416881e-01
2.66391188e-01 7.80861974e-02 -4.71281737e-01 -1.10410571e+00
4.83041108e-02 7.42191076e-01 -6.42557144e-01 2.06565350e-01
2.80527264e-01 5.73535979e-01 7.86372066e-01 8.80323589e-01
2.58206844e-01 -1.15852666e+00 -7.43304566e-02 6.83106184e-01
1.27725089e+00 -1.34819746e+00 -2.20044598e-01 -4.89920646e-01
-6.28398836e-01 6.41151369e-01 6.03677809e-01 3.67174996e-03
5.50541997e-01 4.76251215e-01 8.78033936e-02 -5.09452999e-01
-1.00854397e+00 1.24026693e-01 -1.66767791e-01 3.27690989e-01
5.19863367e-01 5.48032582e-01 -6.77249610e-01 7.28550136e-01
-9.98309791e-01 -4.40789938e-01 6.08899951e-01 6.21516764e-01
-1.46786779e-01 -6.92568660e-01 -1.56523660e-01 4.72232461e-01
-2.19832495e-01 -6.14883125e-01 -5.25435567e-01 1.07429838e+00
-3.13760787e-01 1.12709010e+00 2.88092524e-01 1.82231501e-01
5.82876839e-02 2.84192055e-01 5.87479711e-01 -8.22171390e-01
-4.38050687e-01 -5.67009985e-01 8.81406128e-01 -2.76023537e-01
-5.91483593e-01 -1.11109388e+00 -1.76738882e+00 -2.36118421e-01
2.51387879e-02 1.13068894e-01 5.33250086e-02 1.62608886e+00
2.04120338e-01 1.68373749e-01 -1.63092569e-01 -2.34052435e-01
-1.13160014e+00 -1.49165821e+00 -6.24974594e-02 8.84759545e-01
2.16009393e-01 -3.98458123e-01 -8.77103806e-01 -3.46228480e-01] | [10.489751815795898, 8.693599700927734] |
a846f1fb-cae4-4e94-a75b-d7e57f3039ef | human-eyes-inspired-recurrent-neural-networks | 2206.07282 | null | https://arxiv.org/abs/2206.07282v1 | https://arxiv.org/pdf/2206.07282v1.pdf | Human Eyes Inspired Recurrent Neural Networks are More Robust Against Adversarial Noises | Compared to human vision, computer vision based on convolutional neural networks (CNN) are more vulnerable to adversarial noises. This difference is likely attributable to how the eyes sample visual input and how the brain processes retinal samples through its dorsal and ventral visual pathways, which are under-explored for computer vision. Inspired by the brain, we design recurrent neural networks, including an input sampler that mimics the human retina, a dorsal network that guides where to look next, and a ventral network that represents the retinal samples. Taking these modules together, the models learn to take multiple glances at an image, attend to a salient part at each glance, and accumulate the representation over time to recognize the image. We test such models for their robustness against a varying level of adversarial noises with a special focus on the effect of different input sampling strategies. Our findings suggest that retinal foveation and sampling renders a model more robust against adversarial noises, and the model may correct itself from an attack when it is given a longer time to take more glances at an image. In conclusion, robust visual recognition can benefit from the combined use of three brain-inspired mechanisms: retinal transformation, attention guided eye movement, and recurrent processing, as opposed to feedforward-only CNNs. | ['Zhongming Liu', 'Xiaokai Wang', 'Kuan Han', 'Yizhen Zhang', 'Minkyu Choi'] | 2022-06-15 | null | null | null | null | ['foveation'] | ['computer-vision'] | [ 4.19613123e-01 1.31249741e-01 2.84556150e-01 7.70599395e-02
3.16518806e-02 -6.26099586e-01 6.42379642e-01 -3.81221235e-01
-5.91341555e-01 1.55559897e-01 2.37162754e-01 -3.30445170e-01
2.51870722e-01 -7.96934366e-01 -8.99339497e-01 -8.09025824e-01
3.52112919e-01 -7.25878417e-01 2.67345846e-01 -1.02086775e-01
4.58846599e-01 8.73823762e-01 -1.59351051e+00 2.44543493e-01
5.73634267e-01 1.01683092e+00 -4.97804917e-02 9.62381423e-01
2.27166250e-01 1.15418649e+00 -6.90441787e-01 -4.01069075e-01
4.34259981e-01 -5.76898754e-01 -4.53479856e-01 1.70385733e-01
6.85787261e-01 -2.59196699e-01 -6.22351348e-01 1.32920325e+00
5.75772941e-01 5.73870866e-03 4.52866971e-01 -8.80296826e-01
-1.09930897e+00 1.07457936e-01 -3.44147772e-01 4.33423698e-01
1.22743644e-01 1.03842533e+00 4.67033029e-01 -7.44885981e-01
4.15553033e-01 1.30850029e+00 4.56331789e-01 1.00366664e+00
-1.31235027e+00 -4.52740192e-01 2.26692393e-01 -4.16419394e-02
-9.30664182e-01 -7.89635897e-01 6.23840451e-01 -3.70917499e-01
8.38872969e-01 2.59759992e-01 7.14801967e-01 1.28855824e+00
7.20618188e-01 4.51266766e-01 1.18236613e+00 -2.51681268e-01
1.78395718e-01 -1.07506044e-01 -8.33999179e-03 4.68041897e-01
1.86115175e-01 5.48683584e-01 -3.80734831e-01 4.45757210e-02
8.95681024e-01 1.45133346e-01 -5.32760262e-01 1.43341884e-01
-8.73360932e-01 4.41477239e-01 9.15289521e-01 -6.45769089e-02
-6.52054727e-01 4.52807784e-01 1.40681922e-01 3.37234855e-01
2.53468212e-02 7.65171945e-01 -2.21012682e-02 2.94903100e-01
-6.87923074e-01 4.91042435e-02 5.15238583e-01 4.55817521e-01
5.83901882e-01 4.88685608e-01 -5.12061954e-01 5.01715362e-01
3.92218828e-01 6.74487412e-01 4.14223701e-01 -1.05689490e+00
-8.76672715e-02 5.69134772e-01 -2.38531724e-01 -1.00593925e+00
-1.69883326e-01 -5.30998230e-01 -8.84531617e-01 1.06919169e+00
5.63537002e-01 -2.28973925e-01 -1.10865319e+00 1.75057936e+00
-4.41571139e-02 1.64877430e-01 5.97299077e-02 1.06793797e+00
6.94193482e-01 2.53063947e-01 7.45701194e-02 1.08094610e-01
1.50785041e+00 -7.03139663e-01 -4.98089612e-01 -5.51020861e-01
-2.48590484e-02 -6.77803755e-01 9.50547338e-01 3.63762170e-01
-1.30211508e+00 -7.34383225e-01 -1.09892070e+00 -1.90790996e-01
-3.44834656e-01 4.09846706e-03 1.63412318e-01 5.29380322e-01
-1.14591646e+00 5.02208233e-01 -6.81226254e-01 -3.82636726e-01
8.09584022e-01 3.35447490e-03 -2.27264076e-01 1.66004241e-01
-8.58542800e-01 1.01213968e+00 -2.13892207e-01 4.22712296e-01
-9.50755656e-01 -5.42586446e-01 -7.22065985e-01 7.12329010e-03
-1.04443930e-01 -1.14286101e+00 1.00576425e+00 -1.41475022e+00
-1.67933095e+00 8.85757446e-01 -2.85304248e-01 -8.88159871e-01
3.23804021e-01 -9.59044546e-02 -2.95283198e-01 3.19642186e-01
-4.60416883e-01 7.26790130e-01 1.43733501e+00 -8.76838565e-01
-1.10121690e-01 -4.47033614e-01 8.22030604e-02 -2.63135940e-01
9.04087871e-02 4.76390213e-01 -1.81170985e-01 -7.59642124e-01
-9.63451937e-02 -8.55284691e-01 -2.76921690e-01 5.29530108e-01
-4.30764079e-01 2.53416479e-01 6.82545841e-01 -2.89657801e-01
9.53978419e-01 -2.34356332e+00 -1.89752597e-02 8.06806386e-02
4.88837779e-01 8.57685208e-01 -3.80095154e-01 1.76818609e-01
-3.82033318e-01 3.49343508e-01 2.00914353e-01 -7.47431889e-02
-4.83338177e-01 -1.83947787e-01 -6.66784406e-01 4.61827338e-01
6.48328364e-01 1.21615243e+00 -7.31766760e-01 1.91310465e-01
-5.80547750e-02 6.10471129e-01 -4.22177225e-01 1.68919966e-01
1.64651964e-03 3.71653825e-01 -1.36993200e-01 5.15893519e-01
5.77134192e-01 -1.16402522e-01 -5.30125380e-01 -3.37784588e-01
-2.26079196e-01 1.71811014e-01 -5.31056106e-01 1.00752413e+00
-2.23347321e-01 1.16806364e+00 -2.99735628e-02 -4.46983933e-01
8.93594861e-01 1.79709330e-01 -5.04421234e-01 -8.08830857e-01
4.87965077e-01 -1.32928073e-01 5.65387666e-01 -5.83958447e-01
1.02923982e-01 2.08151355e-01 5.19024611e-01 5.14857173e-01
-2.00115323e-01 -2.49488465e-02 -2.33097956e-01 1.74637362e-01
1.23209274e+00 -1.74800918e-01 1.24978796e-01 1.93992794e-01
2.58049905e-01 -4.51543063e-01 3.61312330e-01 8.51137996e-01
-4.77770358e-01 8.52337420e-01 5.89703143e-01 -5.49379468e-01
-9.62580740e-01 -9.88979638e-01 1.66041389e-01 8.09240580e-01
-4.21263166e-02 4.38625701e-02 -1.00787783e+00 -1.85860589e-01
-2.94172794e-01 7.16340005e-01 -8.07574391e-01 -8.38939428e-01
-2.20860288e-01 -1.64707303e-01 9.67330277e-01 3.81174654e-01
7.44125247e-01 -1.56904793e+00 -1.30021417e+00 -2.21134704e-02
3.30811828e-01 -7.49221563e-01 -5.93993068e-01 -9.94309485e-02
-6.04949832e-01 -1.17047524e+00 -6.38299584e-01 -4.35333312e-01
9.24069941e-01 3.86445105e-01 8.07423711e-01 8.49651322e-02
-5.71378946e-01 5.30291736e-01 -5.28660640e-02 -8.56993318e-01
-4.40617293e-01 -3.93380821e-01 1.17051564e-02 4.75937694e-01
3.82528812e-01 -6.72085941e-01 -8.21373343e-01 1.51838303e-01
-1.03047132e+00 -1.27641797e-01 7.44950593e-01 6.87279940e-01
4.03049469e-01 -2.90304631e-01 9.65048745e-02 -7.24371970e-01
7.50011981e-01 -2.92508155e-01 -4.05255705e-01 1.61780655e-01
-1.31064162e-01 -4.23539206e-02 5.65898776e-01 -8.27877343e-01
-7.08255470e-01 -1.23879068e-01 2.09565341e-01 -8.97526443e-01
-2.35282674e-01 1.21953264e-01 -9.89472643e-02 -4.48896915e-01
1.19577849e+00 4.09809828e-01 5.50996423e-01 -3.24798748e-02
4.19448972e-01 3.33618104e-01 6.61265075e-01 -5.27593605e-02
8.52542162e-01 5.76557040e-01 -9.58001092e-02 -8.94219279e-01
-6.14874482e-01 1.61984161e-01 -3.61957431e-01 -4.47567523e-01
9.89504397e-01 -6.69937253e-01 -9.84536171e-01 9.94644403e-01
-1.40935314e+00 -4.92322564e-01 -4.90203023e-01 2.04304427e-01
-4.31941122e-01 2.15629041e-02 -4.72022831e-01 -8.32695842e-01
-4.28380489e-01 -9.88319993e-01 4.44002002e-01 7.19854176e-01
-1.74248427e-01 -6.63823187e-01 -1.14261478e-01 -7.39306882e-02
7.73411214e-01 2.08492607e-01 7.90710211e-01 -5.35631955e-01
-8.78595412e-01 -1.95509598e-01 -3.12533051e-01 7.25825131e-01
6.75213188e-02 4.86585408e-01 -1.32200408e+00 -1.71613291e-01
2.01729938e-01 -3.46370935e-01 1.16265893e+00 5.08185029e-01
1.08993900e+00 -5.62339008e-01 3.93926762e-02 7.71235466e-01
1.11692214e+00 4.69405763e-02 1.35034227e+00 2.26000801e-01
4.15634930e-01 5.64883530e-01 -1.00481458e-01 -1.05830751e-01
-1.50323823e-01 1.07146546e-01 9.17209923e-01 -4.76449430e-01
-4.99734491e-01 -3.24172378e-02 5.39524376e-01 1.61199242e-01
-1.19935490e-01 -4.66573462e-02 -6.29294634e-01 3.53034675e-01
-1.39860594e+00 -1.21902668e+00 1.52093768e-01 2.26931500e+00
6.81671560e-01 2.92747289e-01 -2.76862849e-02 -7.96824247e-02
5.65277815e-01 1.31775960e-01 -1.01621211e+00 -6.64914072e-01
-3.64692152e-01 2.65642852e-01 4.99185741e-01 1.29148945e-01
-7.52182603e-01 9.43619967e-01 7.08257961e+00 2.73358464e-01
-1.72372460e+00 -3.75488877e-01 7.25255549e-01 -2.46911809e-01
-1.04688279e-01 -8.86381567e-02 -4.63723511e-01 5.10625124e-01
1.02936387e+00 1.96911003e-02 6.83043718e-01 4.65201437e-01
3.97033125e-01 -5.13784699e-02 -8.77706110e-01 6.95819914e-01
4.70772348e-02 -1.43747711e+00 3.92199218e-01 2.62654513e-01
3.89431894e-01 1.94943681e-01 7.16017067e-01 -2.21189648e-01
2.58403838e-01 -1.32926846e+00 6.82363153e-01 1.40722966e+00
5.92497230e-01 -4.84916389e-01 2.35036016e-01 2.76279241e-01
-5.92396796e-01 -2.94458300e-01 -4.93949056e-01 -1.78589851e-01
-4.98559624e-01 1.80840239e-01 -4.98541623e-01 -1.60443559e-01
6.71521366e-01 3.75818580e-01 -8.68913949e-01 1.30935335e+00
-4.64324772e-01 4.96625990e-01 1.92421690e-01 1.36057548e-02
6.87504858e-02 1.52215615e-01 8.91444266e-01 7.94726193e-01
9.96752158e-02 -3.23993526e-02 -4.64500755e-01 1.40142405e+00
-5.07756844e-02 -4.28263217e-01 -9.71345961e-01 -7.78290704e-02
6.21053100e-01 1.27024019e+00 -2.86120474e-01 4.79189940e-02
-3.00191760e-01 9.28306162e-01 1.75818726e-01 9.55621898e-01
-5.06017148e-01 -6.75611258e-01 1.16728354e+00 2.95381665e-01
1.53791517e-01 -1.58281758e-01 -4.33371723e-01 -6.75212085e-01
-1.69184178e-01 -1.04858267e+00 -2.60047317e-01 -1.32517672e+00
-1.12726319e+00 7.97504246e-01 -8.48755002e-01 -1.25863338e+00
-4.59957719e-02 -7.78456688e-01 -1.13543046e+00 1.30198121e+00
-1.46769273e+00 -9.78153288e-01 -1.99426159e-01 7.98987567e-01
3.50920051e-01 -4.00807351e-01 6.79842353e-01 -3.57316315e-01
-7.98085392e-01 7.35574901e-01 -5.33964515e-01 5.35652280e-01
7.40269959e-01 -8.67441058e-01 7.08494544e-01 1.35521412e+00
1.14630081e-01 1.12054074e+00 5.87383270e-01 -2.92524904e-01
-1.31292236e+00 -1.29988956e+00 5.05477965e-01 -5.88943243e-01
5.16312659e-01 -7.49495700e-02 -1.14996362e+00 5.36124110e-01
3.66440296e-01 3.31519783e-01 4.86128837e-01 -4.37222123e-01
-8.46566617e-01 -2.27947742e-01 -1.12224054e+00 1.05909872e+00
7.27131069e-01 -8.51213157e-01 -5.58017433e-01 -2.97930896e-01
7.67111540e-01 -1.58904791e-01 -2.39487529e-01 -4.46159542e-02
5.28715193e-01 -1.17518640e+00 9.52117503e-01 -7.10513771e-01
3.88437063e-01 -5.55191755e-01 2.82621533e-01 -1.46328628e+00
-4.85290051e-01 -9.76386189e-01 1.02251343e-01 8.90628755e-01
3.36360693e-01 -1.07754230e+00 1.97973892e-01 6.89334631e-01
1.13934167e-02 -5.14590502e-01 -6.63518488e-01 -4.14305598e-01
-8.18048045e-02 -2.68444747e-01 4.04760331e-01 4.29757357e-01
-4.59328383e-01 7.39642009e-02 -1.96182832e-01 4.17343020e-01
4.50701565e-01 -4.16605830e-01 8.40036631e-01 -9.89204347e-01
7.67763285e-03 -7.87640631e-01 -6.78074718e-01 -7.93269873e-01
-1.07156001e-01 -5.80970049e-01 1.91504881e-02 -1.05340266e+00
-2.12839484e-01 2.78340727e-01 -3.98798078e-01 6.30235732e-01
-1.31358817e-01 4.02454793e-01 4.98565078e-01 2.26650596e-01
-2.57768124e-01 2.06404790e-01 1.53377616e+00 -7.44930729e-02
-1.33115619e-01 2.40521207e-01 -1.09345829e+00 9.02592719e-01
6.21123075e-01 -2.03483373e-01 -2.61393130e-01 -5.76906383e-01
2.57606298e-01 -2.88562506e-01 1.07212389e+00 -1.08247352e+00
9.37175095e-01 1.50276935e-02 6.87979639e-01 1.35913668e-02
1.30336359e-01 -6.06162727e-01 -3.40614393e-02 5.51580071e-01
-4.75170463e-01 -1.23866610e-01 3.88072968e-01 6.29076660e-01
-1.52973449e-02 2.69920193e-02 1.19120765e+00 -2.66485304e-01
-4.26414520e-01 2.85510182e-01 -9.10727859e-01 -9.10004824e-02
8.47644031e-01 -6.59961641e-01 -6.46257162e-01 -3.37590486e-01
-7.92995572e-01 -7.49135688e-02 4.94103938e-01 3.45563680e-01
8.93574417e-01 -1.04430532e+00 -5.77221811e-01 8.03666532e-01
-1.47440299e-01 -2.60189712e-01 1.91848427e-01 7.68871307e-01
-3.52620035e-01 1.02465130e-01 -5.42751670e-01 -5.24249971e-01
-1.14805913e+00 6.66674733e-01 8.87065172e-01 4.78443116e-01
-4.68147427e-01 9.83976543e-01 2.41549760e-01 1.55047208e-01
3.28254700e-01 -3.59939009e-01 -2.97312170e-01 -2.88323145e-02
9.94281948e-01 2.47773603e-01 -2.45781332e-01 -3.75520527e-01
-3.03405393e-02 6.54263198e-01 -1.28518268e-01 1.39588103e-01
7.90564954e-01 -6.73311055e-02 -3.23240936e-01 3.93200636e-01
6.92599952e-01 1.31314054e-01 -1.62771809e+00 -3.84230673e-01
-4.12940174e-01 -4.24840152e-01 4.69970852e-02 -9.29178119e-01
-1.18810475e+00 1.21258724e+00 4.68358666e-01 4.01064426e-01
1.41161144e+00 -3.94334465e-01 4.74048346e-01 2.34962642e-01
-1.36185233e-02 -6.06766820e-01 4.30886865e-01 5.81058621e-01
1.13665318e+00 -7.82848239e-01 -4.13150519e-01 -2.46967617e-02
-6.14973307e-01 1.15726507e+00 6.42362237e-01 -5.97698808e-01
7.07476676e-01 5.19507416e-02 4.76731688e-01 -1.57805964e-01
-1.02362502e+00 -4.14794564e-01 3.82564008e-01 8.34667981e-01
4.20665517e-02 -3.93269151e-01 4.43344712e-01 2.22561732e-01
-6.20476641e-02 -6.60293996e-02 6.58446074e-01 3.76015306e-01
-5.44724166e-01 -5.43375731e-01 -3.77946794e-01 1.87401772e-01
-5.95523655e-01 -2.69411862e-01 -6.22475266e-01 4.51110482e-01
1.34879261e-01 9.52398419e-01 2.72371858e-01 -3.58804137e-01
4.51785862e-01 5.19411452e-02 3.98437619e-01 -7.07869530e-01
-8.25453103e-01 -1.30104855e-01 -4.49102223e-01 -8.83067548e-01
-2.14868933e-01 -3.42134118e-01 -8.99964452e-01 -2.24130750e-01
-4.80354466e-02 -5.90275168e-01 4.91320580e-01 6.92120790e-01
7.05110848e-01 8.10826361e-01 6.96526170e-01 -9.57520843e-01
-5.34623623e-01 -8.42799008e-01 -3.46360326e-01 1.50943816e-01
7.99402118e-01 -1.66077048e-01 -6.71939909e-01 1.10970117e-01] | [10.094368934631348, 2.3083271980285645] |
8feebf71-cde2-40f4-9607-e5847d98f65b | learning-attention-propagation-for | 2210.11557 | null | https://arxiv.org/abs/2210.11557v1 | https://arxiv.org/pdf/2210.11557v1.pdf | Learning Attention Propagation for Compositional Zero-Shot Learning | Compositional zero-shot learning aims to recognize unseen compositions of seen visual primitives of object classes and their states. While all primitives (states and objects) are observable during training in some combination, their complex interaction makes this task especially hard. For example, wet changes the visual appearance of a dog very differently from a bicycle. Furthermore, we argue that relationships between compositions go beyond shared states or objects. A cluttered office can contain a busy table; even though these compositions don't share a state or object, the presence of a busy table can guide the presence of a cluttered office. We propose a novel method called Compositional Attention Propagated Embedding (CAPE) as a solution. The key intuition to our method is that a rich dependency structure exists between compositions arising from complex interactions of primitives in addition to other dependencies between compositions. CAPE learns to identify this structure and propagates knowledge between them to learn class embedding for all seen and unseen compositions. In the challenging generalized compositional zero-shot setting, we show that our method outperforms previous baselines to set a new state-of-the-art on three publicly available benchmarks. | ['Muhammad Zeshan Afzal', 'Didier Stricker', 'Alain Pagani', 'Luc van Gool', 'Muhammad Ferjad Naeem', 'Muhammad Gul Zain Ali Khan'] | 2022-10-20 | null | null | null | null | ['compositional-zero-shot-learning'] | ['computer-vision'] | [ 3.90477061e-01 1.88477989e-02 -8.38100389e-02 -2.76750982e-01
-1.98059008e-01 -6.62923634e-01 1.08691728e+00 2.15629399e-01
-2.06160042e-02 2.47126877e-01 6.20824993e-01 -6.84162974e-02
1.21551275e-01 -8.00773561e-01 -1.19656610e+00 -8.66682589e-01
2.80235875e-02 7.99641967e-01 6.52231634e-01 -2.84420371e-01
1.91990994e-02 5.25688112e-01 -1.90354788e+00 7.65362561e-01
7.07478151e-02 6.85843945e-01 2.39736527e-01 6.61438286e-01
-1.67710021e-01 1.35555136e+00 -4.77768630e-01 -2.62943298e-01
2.66558498e-01 -2.75385380e-01 -8.90425086e-01 5.31294405e-01
9.58994210e-01 -6.17623746e-01 -4.25961047e-01 9.23663914e-01
9.79187563e-02 3.89925718e-01 9.81762886e-01 -1.56170213e+00
-9.62068796e-01 6.88314497e-01 -2.50165612e-01 1.76460683e-01
3.11731815e-01 6.07475877e-01 1.54221821e+00 -6.15231395e-01
9.60254490e-01 1.43058980e+00 1.54556006e-01 7.08635747e-01
-1.56203675e+00 -4.10985798e-01 9.53315258e-01 3.73244822e-01
-9.43551660e-01 -4.59760249e-01 7.74621189e-01 -7.59178460e-01
1.13630235e+00 3.36794615e-01 7.87876606e-01 1.37272120e+00
-6.28390461e-02 1.08970821e+00 7.32969820e-01 -2.63630420e-01
2.99069047e-01 4.77064028e-02 5.64976513e-01 1.00842428e+00
2.42756471e-01 -8.98767933e-02 -6.52085900e-01 -2.59826154e-01
4.32499111e-01 5.28748512e-01 -1.52974218e-01 -9.02577221e-01
-1.37582481e+00 5.37875056e-01 5.82557201e-01 1.03603706e-01
9.36426967e-03 5.35224557e-01 6.95279986e-02 2.26654828e-01
-9.76065323e-02 6.09453201e-01 -3.69475365e-01 2.22726643e-01
-1.44046307e-01 2.96850950e-01 9.38815892e-01 1.05466497e+00
9.21419203e-01 -1.84298113e-01 -3.73736829e-01 3.09940159e-01
1.15421243e-01 1.97238341e-01 6.14603758e-02 -8.40528071e-01
3.10618162e-01 9.35872912e-01 1.96382299e-01 -5.25621712e-01
-1.75846696e-01 6.12810366e-02 -4.71230924e-01 4.03268635e-01
5.48714459e-01 3.23012084e-01 -1.25649226e+00 1.87045813e+00
3.87589455e-01 3.70117337e-01 -1.68671295e-01 8.41478050e-01
8.84107471e-01 7.44703054e-01 4.24430184e-02 3.20852727e-01
1.68478918e+00 -1.11372101e+00 -4.55003142e-01 -3.82862449e-01
2.36174390e-01 -4.84392822e-01 1.18544197e+00 1.06324829e-01
-8.20634246e-01 -4.38580334e-01 -9.87412453e-01 -3.92860144e-01
-5.64524233e-01 -4.55650598e-01 7.57550061e-01 9.89584103e-02
-6.78230345e-01 6.08666182e-01 -1.14320993e+00 -3.94477278e-01
3.56358051e-01 2.81313121e-01 -4.00919080e-01 -3.34414840e-01
-6.78566575e-01 8.57081711e-01 7.36099482e-02 -2.80753076e-01
-1.48911417e+00 -8.39187622e-01 -1.20819366e+00 3.56424242e-01
7.78767884e-01 -9.13242817e-01 1.41519272e+00 -9.48316455e-01
-1.12275767e+00 8.35857272e-01 -3.97308946e-01 -1.73454791e-01
1.98655352e-01 -1.76675946e-01 -1.27392383e-02 3.08055617e-02
7.44811520e-02 5.69755256e-01 9.91280615e-01 -1.64915800e+00
-7.08377540e-01 -4.89814699e-01 4.92723376e-01 2.28102714e-01
6.38744906e-02 6.07853569e-02 -4.78026897e-01 -2.33308792e-01
1.92964256e-01 -7.68111706e-01 -9.56435129e-03 4.19027954e-01
-5.38953900e-01 -3.47004503e-01 9.61486459e-01 -6.09877184e-02
8.18569362e-01 -2.17341971e+00 2.79452741e-01 -1.23310983e-01
5.21116972e-01 -2.09477961e-01 -1.64918065e-01 6.48996115e-01
-9.69685763e-02 7.08667096e-04 -9.29874629e-02 -3.96355003e-01
4.37789112e-01 6.45328462e-01 -4.78232533e-01 4.70302492e-01
4.03547823e-01 9.13851440e-01 -1.17502749e+00 -2.09093377e-01
3.63159776e-01 5.19310653e-01 -5.09750366e-01 2.48507157e-01
-6.28087461e-01 1.75319746e-01 -2.02632830e-01 8.03878307e-01
2.27866754e-01 -5.66396296e-01 1.31908685e-01 -2.61200815e-01
-1.65877715e-02 5.23267627e-01 -1.28969419e+00 1.39078975e+00
-2.28836074e-01 5.03593504e-01 -7.17630908e-02 -8.04680288e-01
2.22950831e-01 2.24414319e-01 2.27621198e-01 -3.80941689e-01
-2.87871100e-02 -4.22816753e-01 2.19514206e-01 -6.32528007e-01
3.40207428e-01 -1.29710153e-01 -7.12159052e-02 7.94186354e-01
1.81891561e-01 -2.55016804e-01 1.51035383e-01 5.24874210e-01
1.51174521e+00 1.38571650e-01 4.59643066e-01 -8.77175778e-02
-3.81003357e-02 -2.57998616e-01 6.47250235e-01 9.33205545e-01
-2.07940832e-01 5.76075613e-01 6.93736315e-01 -7.32639968e-01
-1.06037879e+00 -1.49649966e+00 3.56695533e-01 1.47561109e+00
4.93735820e-01 -6.43256664e-01 -1.12187594e-01 -7.19820559e-01
1.37734815e-01 8.18342507e-01 -9.76773620e-01 -2.35536397e-01
-6.15266800e-01 -1.48241177e-01 -5.91373444e-02 6.49074614e-01
-7.35608414e-02 -1.09194279e+00 -8.94265652e-01 4.48251627e-02
-4.06163223e-02 -1.23888767e+00 -7.22334862e-01 4.28529918e-01
-5.18510461e-01 -1.32082188e+00 -9.57658514e-02 -9.17547226e-01
8.52722108e-01 3.41655493e-01 1.45293725e+00 1.63345844e-01
-4.65458572e-01 6.84307396e-01 -2.20097497e-01 -4.21774536e-01
-4.08622026e-01 -3.57947767e-01 -1.43424749e-01 2.87016213e-01
2.94600070e-01 -6.37878835e-01 -5.14138103e-01 1.64042711e-01
-8.03430200e-01 1.97950825e-01 3.72497618e-01 6.97404981e-01
3.46634328e-01 1.16500128e-02 -3.41171265e-01 -9.67002392e-01
1.12219177e-01 -3.83417994e-01 -4.86760169e-01 4.60497797e-01
-1.37406170e-01 4.06896174e-01 5.64679384e-01 -7.35865474e-01
-1.03892124e+00 3.06329012e-01 5.72101355e-01 -7.20553517e-01
-4.27677959e-01 -1.71653613e-01 -3.62137556e-01 4.74861950e-01
5.71230948e-01 9.90996659e-02 -2.96558827e-01 -4.42235231e-01
5.96264005e-01 2.35903319e-02 5.84824741e-01 -1.00147021e+00
9.20992196e-01 1.05858219e+00 -3.49797346e-02 -7.81045139e-01
-7.62626469e-01 -6.53783739e-01 -6.67381287e-01 9.04979482e-02
9.77741480e-01 -8.66174102e-01 -8.65987241e-01 2.74362653e-01
-1.33573842e+00 -6.80640519e-01 -4.94682461e-01 1.60106644e-02
-4.26750958e-01 2.06573918e-01 -8.56693625e-01 -6.54981375e-01
1.35601789e-01 -1.24050844e+00 1.33737755e+00 -1.21954501e-01
-4.95262742e-01 -7.06270218e-01 -8.02168325e-02 9.29221734e-02
-1.12715088e-01 2.24625498e-01 1.34155190e+00 -4.37595934e-01
-1.19092369e+00 1.05306610e-01 -6.71191886e-03 6.51364820e-03
3.95673960e-01 4.29108143e-01 -9.08453345e-01 -1.46398187e-01
-1.42488301e-01 -2.50766426e-01 8.95715117e-01 -3.97927240e-02
9.99483407e-01 -4.42009628e-01 -3.84479254e-01 3.57192010e-01
1.31322324e+00 2.33921155e-01 4.13511485e-01 6.01229109e-02
9.22529399e-01 6.81296885e-01 1.27022430e-01 3.51668537e-01
4.38072443e-01 6.44755900e-01 6.46416306e-01 1.42538533e-01
-3.70602876e-01 -3.46442074e-01 4.27593172e-01 2.80335605e-01
-3.92498858e-02 -1.33723080e-01 -9.12855923e-01 5.93297958e-01
-1.88694644e+00 -1.27052855e+00 2.66669933e-02 2.13023734e+00
7.74048805e-01 -3.41167441e-03 -1.04485884e-01 -1.37339979e-01
4.24993962e-01 2.95319259e-01 -5.56949735e-01 -1.02592297e-01
9.82569382e-02 2.04399869e-01 1.80086091e-01 5.39360225e-01
-1.01139820e+00 1.04106808e+00 6.02984905e+00 2.91612327e-01
-7.60279179e-01 1.37856811e-01 1.70957342e-01 -2.80947447e-01
-5.77614665e-01 2.22562134e-01 -1.04405200e+00 3.16993505e-01
2.00468406e-01 1.23713002e-01 7.20562041e-01 7.09192455e-01
-3.71272594e-01 -1.15205728e-01 -1.96947396e+00 7.50164986e-01
2.94583201e-01 -1.30353642e+00 3.29220593e-01 -8.53669718e-02
8.56155753e-01 -7.52037019e-02 7.86273777e-02 4.87756521e-01
8.58802259e-01 -7.98791587e-01 1.04588556e+00 2.66105056e-01
3.35888892e-01 -2.95500040e-01 1.15414500e-01 3.37105125e-01
-1.40817094e+00 -2.93479115e-01 -1.48581654e-01 -3.85826319e-01
5.67408018e-02 3.45561244e-02 -6.28060818e-01 2.12411527e-02
6.56549215e-01 5.62478721e-01 -4.34887290e-01 6.47503853e-01
-7.03763425e-01 2.55426109e-01 -1.42593518e-01 -7.96508715e-02
2.61920244e-01 6.05127425e-04 6.22244060e-01 6.85088277e-01
-2.57610887e-01 2.46834651e-01 1.38270348e-01 1.30956900e+00
-5.27995415e-02 -5.58251023e-01 -8.45283628e-01 -2.13339180e-01
3.07611376e-01 1.06037641e+00 -6.61295056e-01 -7.61237919e-01
-6.87095463e-01 1.09085751e+00 4.58595961e-01 5.28049111e-01
-9.21689987e-01 1.69919923e-01 1.36703515e+00 1.09600566e-01
6.73112154e-01 -1.25326455e-01 -3.44683118e-02 -1.46177089e+00
5.81555255e-02 -1.07640254e+00 5.05382061e-01 -8.07650149e-01
-1.44130123e+00 1.89315423e-01 5.41054308e-02 -1.08529270e+00
6.64584339e-02 -6.08202934e-01 -8.02752733e-01 3.36791009e-01
-1.18657875e+00 -1.42590773e+00 -3.79939854e-01 7.94287145e-01
7.91842520e-01 9.74605158e-02 9.26023841e-01 -6.58206940e-02
-5.61997056e-01 2.01881185e-01 -3.95781815e-01 2.06701815e-01
5.28285086e-01 -1.43300080e+00 6.31174266e-01 1.00718367e+00
6.57568514e-01 9.00583386e-01 7.42277443e-01 -5.75244188e-01
-1.62905276e+00 -9.18904126e-01 7.07955539e-01 -9.02259886e-01
8.31836939e-01 -1.03313553e+00 -8.73223662e-01 1.35077763e+00
2.38864422e-01 3.13758761e-01 5.18187344e-01 3.01290333e-01
-1.09855866e+00 5.73918149e-02 -4.84387040e-01 1.04054570e+00
1.39034522e+00 -7.85483837e-01 -9.62680578e-01 4.10501331e-01
1.02451670e+00 -2.40168720e-01 -2.47211188e-01 2.91216612e-01
5.98478734e-01 -1.05153191e+00 1.07905960e+00 -1.04749775e+00
4.79810238e-01 -3.89890641e-01 -3.61727118e-01 -1.25816262e+00
-5.05451560e-01 -5.19786775e-01 -7.36659765e-01 8.27958226e-01
1.80215448e-01 -5.88647068e-01 7.46799290e-01 1.05583894e+00
-2.54385114e-01 -5.65995097e-01 -5.83490968e-01 -9.76169169e-01
-2.94179052e-01 -2.85265297e-01 7.47906327e-01 9.86110985e-01
-4.91581634e-02 6.32560551e-01 -1.59417227e-01 4.55812007e-01
6.29386008e-01 4.66715336e-01 1.07816291e+00 -1.10980701e+00
-7.46038258e-01 -5.49268663e-01 -5.88094532e-01 -1.05017924e+00
2.47513786e-01 -7.65444458e-01 2.37025037e-01 -1.77310526e+00
5.58278084e-01 -2.69686133e-01 -3.24330628e-01 8.60673726e-01
-2.08306879e-01 -2.04647943e-01 5.00330925e-01 -3.61562856e-02
-6.67851806e-01 5.22883952e-01 1.29465652e+00 -7.67437756e-01
-5.30018881e-02 -1.53990194e-01 -5.55472970e-01 7.73439467e-01
2.71118969e-01 -4.59347129e-01 -6.32055759e-01 -6.44910574e-01
1.72086567e-01 -2.58874536e-01 7.10909903e-01 -7.81585395e-01
2.65853196e-01 -1.22348495e-01 1.72924593e-01 -4.39041406e-01
8.35155904e-01 -1.14794815e+00 1.64929211e-01 5.31262279e-01
-2.18601257e-01 -1.78060353e-01 -4.91744727e-02 9.21414077e-01
8.08393732e-02 7.14466944e-02 5.29164076e-01 -4.46146995e-01
-9.62367594e-01 4.94258791e-01 -3.30339074e-01 -6.86115250e-02
1.19122684e+00 -2.55641758e-01 -6.98050499e-01 -4.19796795e-01
-8.40473950e-01 1.99803367e-01 6.86130762e-01 6.12314403e-01
4.62726295e-01 -1.10982692e+00 -1.65120423e-01 3.77460629e-01
4.76526499e-01 2.50652522e-01 2.35596001e-01 6.90827072e-01
-2.48396963e-01 -7.57442340e-02 -2.32647926e-01 -5.22943974e-01
-1.54141569e+00 9.47041810e-01 2.99886048e-01 9.28678317e-04
-9.69202340e-01 1.11342525e+00 9.37629998e-01 -3.93065453e-01
5.87449729e-01 -8.55013847e-01 1.35657102e-01 2.39515752e-02
6.69813752e-01 1.10673413e-01 -3.16481769e-01 -4.97621566e-01
-3.02025616e-01 4.09818470e-01 -1.57719985e-01 -3.26595567e-02
1.29274917e+00 2.16795251e-01 -2.35792786e-01 9.73690212e-01
1.08688116e+00 -1.26317754e-01 -1.49285924e+00 -4.79198426e-01
-2.70368010e-01 -5.57098746e-01 -5.07141292e-01 -5.63100159e-01
-7.65088260e-01 1.04530215e+00 1.79727063e-01 5.98674268e-02
6.15673244e-01 5.15649378e-01 5.63642204e-01 7.11808264e-01
4.11401927e-01 -7.39976048e-01 5.31472981e-01 5.34059107e-01
8.71492207e-01 -1.40902948e+00 2.04639230e-02 -4.00100857e-01
-6.96197391e-01 9.18074429e-01 8.51237476e-01 -1.25051573e-01
5.13602376e-01 3.66084397e-01 -2.26242796e-01 -5.23120761e-01
-1.23861861e+00 -5.86825490e-01 2.58636773e-01 5.36352932e-01
-1.74979176e-02 1.78473994e-01 5.19126117e-01 2.86622971e-01
9.44033936e-02 -6.43586576e-01 3.15235108e-01 1.13368130e+00
-2.93511808e-01 -9.39864576e-01 -2.89507359e-01 4.00902480e-01
-3.96650098e-02 -4.67738584e-02 -6.64672852e-01 9.52819467e-01
3.70683789e-01 7.25422204e-01 4.39757556e-01 -3.04106712e-01
2.04846680e-01 2.99462974e-01 7.95959473e-01 -1.32159543e+00
-4.77602929e-01 -8.03214461e-02 -6.32515848e-02 -8.18221867e-01
-2.84079909e-01 -7.99837768e-01 -1.20885217e+00 -9.89817530e-02
-3.68673280e-02 -4.53496009e-01 3.88676763e-01 9.63530004e-01
1.05865225e-01 7.32394218e-01 2.32981980e-01 -8.79294157e-01
-3.95541370e-01 -7.14804888e-01 -5.08898854e-01 9.65489924e-01
6.61916554e-01 -8.33193898e-01 -5.24903059e-01 2.41344079e-01] | [10.267924308776855, 2.1391336917877197] |
bdef150e-0853-4849-aaf3-02400acac815 | modanet-a-large-scale-street-fashion-dataset | 1807.01394 | null | http://arxiv.org/abs/1807.01394v4 | http://arxiv.org/pdf/1807.01394v4.pdf | ModaNet: A Large-Scale Street Fashion Dataset with Polygon Annotations | Understanding clothes from a single image has strong commercial and cultural
impacts on modern societies. However, this task remains a challenging computer
vision problem due to wide variations in the appearance, style, brand and
layering of clothing items. We present a new database called ModaNet, a
large-scale collection of images based on Paperdoll dataset. Our dataset
provides 55,176 street images, fully annotated with polygons on top of the 1
million weakly annotated street images in Paperdoll. ModaNet aims to provide a
technical benchmark to fairly evaluate the progress of applying the latest
computer vision techniques that rely on large data for fashion understanding.
The rich annotation of the dataset allows to measure the performance of
state-of-the-art algorithms for object detection, semantic segmentation and
polygon prediction on street fashion images in detail. The polygon-based
annotation dataset has been released https://github.com/eBay/modanet, we also
host the leaderboard at EvalAI:
https://evalai.cloudcv.org/featured-challenges/136/overview. | ['Shuai Zheng', 'M. Hadi Kiapour', 'Fan Yang', 'Robinson Piramuthu'] | 2018-07-03 | null | null | null | null | ['fashion-understanding'] | ['computer-vision'] | [-5.79559878e-02 -4.55393374e-01 -5.73364571e-02 -3.80797923e-01
-5.16676307e-01 -9.08996999e-01 2.93130130e-01 1.94104090e-01
-2.16405503e-02 3.50498021e-01 -1.35901853e-01 2.75507480e-01
3.23119849e-01 -7.99440503e-01 -1.16311538e+00 -4.06414241e-01
1.44984528e-01 6.66001379e-01 3.01241308e-01 -5.00017166e-01
-8.72295722e-02 1.77866802e-01 -1.52844405e+00 5.55088639e-01
5.65183818e-01 1.21703732e+00 3.83301198e-01 7.67988503e-01
4.78500836e-02 1.41949490e-01 -2.08435088e-01 -9.07898366e-01
5.73857665e-01 5.28319739e-02 -4.78455484e-01 4.70315874e-01
1.12529969e+00 -4.62965854e-02 -1.53148860e-01 1.15878391e+00
2.24821657e-01 -2.79857874e-01 2.82980502e-01 -1.15995848e+00
-9.65785503e-01 5.62078059e-01 -7.49947667e-01 -9.10436735e-02
2.60409743e-01 2.46018931e-01 9.55592752e-01 -8.39395642e-01
1.06623578e+00 1.06781888e+00 9.99561012e-01 1.51217088e-01
-1.01281524e+00 -4.22341287e-01 2.38672376e-01 1.50933579e-01
-1.50654852e+00 -2.29236975e-01 1.05556357e+00 -4.08560663e-01
3.53819162e-01 5.09475887e-01 1.21012175e+00 1.05202568e+00
-7.32365698e-02 1.15896249e+00 1.42174232e+00 -2.91422546e-01
-9.23959762e-02 2.71714449e-01 3.70373696e-01 1.09064436e+00
2.34126106e-01 -3.31706613e-01 -3.60363662e-01 2.34318823e-01
9.99708295e-01 -1.23670831e-01 1.21264085e-01 -4.12586659e-01
-1.27316439e+00 5.56192815e-01 4.87029791e-01 -5.85292988e-02
-3.16521496e-01 6.95444271e-02 5.37159026e-01 3.34500372e-01
7.56576359e-01 2.73929179e-01 -7.75449634e-01 1.96123850e-02
-7.52505004e-01 6.36870265e-01 8.25880468e-01 1.24738908e+00
5.18841147e-01 -2.59913623e-01 1.55774519e-01 1.15281427e+00
1.13271371e-01 8.03277314e-01 -3.88269275e-01 -9.31569636e-01
4.24519539e-01 5.60975254e-01 2.30718721e-02 -1.16930258e+00
-3.72757912e-01 -2.12659478e-01 -5.03219783e-01 -1.24580031e-02
7.58950233e-01 2.42344141e-02 -9.03171241e-01 1.04811502e+00
4.79341805e-01 -2.25098506e-01 -5.40016472e-01 1.23009109e+00
1.33437598e+00 4.36999857e-01 7.96854049e-02 3.05145532e-01
1.63188660e+00 -1.44867694e+00 -4.26127404e-01 -1.94656998e-01
8.81200209e-02 -1.22704637e+00 1.18855309e+00 7.27315426e-01
-1.08120120e+00 -8.52665305e-01 -8.74880791e-01 -1.38596281e-01
-6.81844056e-01 5.53714395e-01 9.02268052e-01 5.96839249e-01
-6.17465615e-01 3.41976255e-01 -4.27221298e-01 -6.64992392e-01
7.73122966e-01 5.60447164e-02 -3.69907826e-01 -1.99044973e-01
-7.03855038e-01 5.37286460e-01 1.87172562e-01 2.93542117e-01
-6.52795672e-01 -7.91698635e-01 -7.00692236e-01 -6.23048723e-01
5.57448030e-01 -4.75117981e-01 9.03197885e-01 -1.20420706e+00
-1.13817704e+00 1.49366856e+00 2.75438577e-01 -2.67967641e-01
8.42233539e-01 -3.29307169e-01 -5.79607546e-01 -4.20723017e-03
1.38296857e-01 9.12752450e-01 7.65420914e-01 -1.50204659e+00
-5.27132034e-01 -4.87020880e-01 -3.39246243e-02 -5.14026955e-02
8.30129534e-02 2.71107201e-02 -1.11858261e+00 -9.07419026e-01
-2.33846739e-01 -1.38250625e+00 -1.39541462e-01 2.72958815e-01
-6.31312490e-01 -1.35624379e-01 7.62845337e-01 -1.02774966e+00
6.88017845e-01 -2.03911710e+00 -4.21156213e-02 6.84184581e-02
7.33821243e-02 2.22597197e-01 -1.74173310e-01 2.51848012e-01
2.82835662e-01 -1.43492594e-01 -1.79959998e-01 -3.33547533e-01
2.08991002e-02 9.80528072e-02 -2.99090315e-02 4.91803378e-01
3.99447009e-02 1.07863545e+00 -5.94399154e-01 -5.99381745e-01
5.56757510e-01 3.54830205e-01 -2.69724339e-01 -1.55880883e-01
-6.35007322e-01 4.60006773e-01 -2.88312942e-01 1.34379387e+00
1.04378474e+00 -1.88849553e-01 2.78348237e-01 -6.59125388e-01
-1.80241600e-01 -4.63039070e-01 -1.23706615e+00 2.06010222e+00
-1.13970578e-01 7.23799765e-01 3.45418900e-01 -5.70981443e-01
9.23033178e-01 -3.29756200e-01 4.15044039e-01 -6.86331272e-01
1.71898767e-01 -3.50505002e-02 -5.02161801e-01 -4.82611090e-01
6.94387138e-01 5.33217430e-01 -4.51487273e-01 -2.41855830e-01
-7.78573472e-03 -3.14108163e-01 4.75185186e-01 4.00074896e-05
5.85211337e-01 6.21083617e-01 -1.50102451e-01 -4.36453402e-01
5.09113558e-02 7.17601836e-01 4.26381439e-01 5.21233916e-01
-1.92261592e-01 9.19634700e-01 2.01463953e-01 -1.04505718e+00
-1.50667369e+00 -1.25720131e+00 -3.35128993e-01 1.14963245e+00
4.05617952e-01 -4.69337791e-01 -9.49795604e-01 -5.39677680e-01
3.04900199e-01 1.85656399e-01 -7.32802272e-01 6.58100963e-01
-5.23206830e-01 -7.97200799e-01 2.80792981e-01 5.19329667e-01
9.23256338e-01 -9.17288363e-01 -1.46457642e-01 -1.60227627e-01
3.61206010e-02 -1.49230373e+00 -5.07188082e-01 -2.58718044e-01
-3.60448301e-01 -1.24439061e+00 -5.87114990e-01 -9.39828336e-01
5.55435538e-01 1.88007116e-01 1.59065080e+00 -2.36782487e-02
-9.29097950e-01 4.50877428e-01 -3.23169529e-01 -6.77932858e-01
-1.16043821e-01 8.25426057e-02 -2.11177945e-01 -8.74208286e-03
3.05129051e-01 -1.59417436e-01 -7.17673242e-01 5.44110119e-01
-5.07726490e-01 4.44991797e-01 2.18511596e-01 2.76056081e-01
1.04822910e+00 -1.04528151e-01 9.92952734e-02 -9.96728599e-01
2.22626656e-01 -2.51562297e-01 -8.41883779e-01 3.73581111e-01
-1.55600682e-01 -6.07447147e-01 3.90603215e-01 -1.69048309e-01
-9.06943738e-01 3.63271683e-01 -3.24094087e-01 -2.88786888e-01
-5.95304966e-01 -1.62542894e-01 -7.26928338e-02 -3.41842957e-02
2.70264149e-01 -7.45854005e-02 -1.30717233e-01 -8.05068612e-01
7.60848165e-01 5.29931128e-01 5.21147311e-01 -6.05962873e-01
5.21090925e-01 6.32191956e-01 -3.58896077e-01 -1.18068731e+00
-1.01896596e+00 -4.50779438e-01 -1.00284755e+00 -6.17132723e-01
1.05787742e+00 -1.07283366e+00 -6.33087158e-01 8.13823402e-01
-8.60515177e-01 -5.59967220e-01 -3.10497910e-01 -8.16405863e-02
-4.25944537e-01 1.18709378e-01 -7.13738441e-01 -4.71644640e-01
-4.31391925e-01 -1.01064861e+00 1.28410900e+00 2.23856106e-01
-1.90515175e-01 -8.77939045e-01 -4.82355282e-02 8.90546799e-01
-1.18961828e-02 7.38107085e-01 4.15576965e-01 -4.11106087e-02
-4.92950767e-01 -1.79005042e-01 -3.64338130e-01 4.40001011e-01
-1.96616486e-01 3.18383366e-01 -9.16812062e-01 7.98932463e-02
-7.82771289e-01 -3.11742246e-01 1.05116999e+00 4.53030109e-01
1.43794453e+00 6.76728040e-02 -1.78315654e-01 7.07293093e-01
1.73985732e+00 -3.26782644e-01 3.69444609e-01 3.55127096e-01
1.14956248e+00 5.65575182e-01 8.10350060e-01 3.55478138e-01
6.44030452e-01 7.25394726e-01 4.48802531e-01 -4.38441813e-01
-6.35651410e-01 -2.46147200e-01 -1.05068553e-02 9.49249685e-01
-6.40488505e-01 2.14784995e-01 -7.99303174e-01 6.03440821e-01
-1.90525353e+00 -6.55349314e-01 -7.58339584e-01 1.77885425e+00
6.52858853e-01 1.43122226e-01 6.76954806e-01 -2.87901849e-01
7.60021389e-01 1.44558936e-01 -3.95608634e-01 -3.38528693e-01
-4.72126484e-01 -8.59863088e-02 9.13467288e-01 -2.05665883e-02
-1.73930907e+00 1.14988029e+00 6.20212793e+00 1.03585172e+00
-7.83102870e-01 2.47605845e-01 7.92223513e-01 -3.57264318e-02
7.87351802e-02 -5.11164725e-01 -9.94371474e-01 6.58589482e-01
4.81475890e-01 5.53260982e-01 5.70035338e-01 1.15392780e+00
4.26505469e-02 -6.96818307e-02 -5.64423501e-01 1.09600699e+00
2.49008864e-01 -1.60452437e+00 -2.97300190e-01 -4.90311161e-02
1.05274940e+00 2.76078105e-01 1.16592526e-01 3.57481502e-02
3.08358043e-01 -6.59963310e-01 1.20748174e+00 5.33415616e-01
9.06948388e-01 -6.56376541e-01 5.28884292e-01 -2.17363462e-01
-1.43018270e+00 9.87455696e-02 -5.09315789e-01 2.20546294e-02
-2.37219594e-02 6.03376091e-01 -3.87644112e-01 3.39587748e-01
1.15339255e+00 9.32633221e-01 -9.52740073e-01 1.00134826e+00
-3.61810513e-02 5.78350842e-01 -4.60942209e-01 1.02247998e-01
8.36384445e-02 -5.50414264e-01 2.89510578e-01 1.43305886e+00
-1.53555274e-01 -3.02867740e-01 6.07939303e-01 6.20304108e-01
-3.02761108e-01 1.19018912e-01 -5.17785192e-01 8.12605992e-02
1.36972442e-01 1.67368639e+00 -1.13928986e+00 -3.08421940e-01
-4.93124008e-01 1.15210664e+00 1.63297921e-01 -7.93713145e-03
-1.17085063e+00 1.53932557e-01 8.12500656e-01 3.94918352e-01
6.02185845e-01 -4.41928804e-01 -4.07080591e-01 -1.00985885e+00
1.45721421e-01 -7.81203806e-01 1.95968360e-01 -8.74113917e-01
-1.69506812e+00 3.94176662e-01 -1.29966035e-01 -9.82456863e-01
9.12554324e-01 -9.35078621e-01 -2.68697470e-01 1.96020365e-01
-1.23110831e+00 -2.05087876e+00 -7.34467506e-01 5.85165679e-01
1.14465773e+00 -3.65119986e-02 7.13401794e-01 4.81557578e-01
-6.68861747e-01 4.03711945e-01 2.27362037e-01 5.01583576e-01
6.81410134e-01 -1.30455089e+00 6.63627625e-01 2.85311788e-01
2.47824937e-01 -3.83625999e-02 6.67265058e-01 -7.43503869e-01
-1.82045424e+00 -1.43086231e+00 2.01244563e-01 -9.71641243e-01
7.52326667e-01 -7.51232147e-01 -3.60566616e-01 7.61812627e-01
3.04914802e-01 1.80261955e-01 5.06810308e-01 1.93173632e-01
-3.41424793e-01 -3.97030085e-01 -1.07334614e+00 4.28875625e-01
1.30921710e+00 -8.70999321e-02 -2.59233981e-01 7.93442786e-01
5.59685051e-01 -5.47147036e-01 -1.27532637e+00 1.86581641e-01
9.04147148e-01 -7.04345345e-01 1.13111901e+00 -1.76292181e-01
6.80795133e-01 -2.61207730e-01 -2.72904038e-01 -9.91135299e-01
-2.45001242e-01 -2.60518700e-01 1.03814535e-01 1.35639024e+00
4.26999509e-01 -8.40617791e-02 8.21913898e-01 3.46323222e-01
-7.10290745e-02 -7.31872261e-01 -5.22098839e-01 -6.68199599e-01
-1.45091608e-01 -5.64397037e-01 5.57919025e-01 9.21102822e-01
-7.18697965e-01 4.95810248e-02 -5.69147110e-01 -4.69999313e-02
1.07711470e+00 6.08852744e-01 1.02225900e+00 -1.27572095e+00
-2.27994341e-02 -2.00639695e-01 -4.09328014e-01 -7.16551781e-01
-2.74240315e-01 -9.81857836e-01 -1.42383829e-01 -1.65422881e+00
3.04192066e-01 -6.48474216e-01 6.95316643e-02 3.94211084e-01
5.51890396e-02 1.10933578e+00 5.65083206e-01 8.37137699e-02
-1.19758523e+00 2.71962341e-02 1.47407126e+00 -5.35460949e-01
1.09488085e-01 -1.46184281e-01 -4.07855898e-01 1.09515858e+00
1.01567304e+00 -1.00149117e-01 2.31166914e-01 -6.49190485e-01
1.47411823e-01 -4.48676914e-01 6.80787861e-01 -1.03593647e+00
-1.76940203e-01 -2.55015828e-02 8.20349813e-01 -8.70562851e-01
5.89703679e-01 -9.26763833e-01 3.10372412e-01 2.99092829e-01
-2.52911281e-02 1.29625514e-01 2.30404899e-01 5.83869636e-01
1.93216130e-01 1.96186408e-01 7.09968865e-01 -6.46943092e-01
-1.43565905e+00 3.88380527e-01 2.61352453e-02 -1.12445734e-01
1.27918458e+00 -2.43215322e-01 -3.34450155e-01 3.05952817e-01
-1.01713037e+00 1.08156793e-01 8.26543331e-01 7.98088372e-01
4.68490869e-01 -1.33416724e+00 -7.99084663e-01 1.32930994e-01
2.33186021e-01 -2.55585849e-01 4.65215623e-01 6.07769668e-01
-1.01919687e+00 1.81993857e-01 -5.36210835e-01 -7.06075191e-01
-1.69846714e+00 5.76996922e-01 -1.03738718e-01 -3.79539467e-02
-8.30464840e-01 9.06641304e-01 -2.70982180e-02 -6.29092157e-01
-6.89850450e-02 -3.40614349e-01 -1.84881091e-02 5.05912416e-02
2.37640783e-01 4.53087181e-01 8.66727009e-02 -6.78723752e-01
-2.47493058e-01 9.10419166e-01 1.03628837e-01 5.05534053e-01
1.52463639e+00 -1.50727674e-01 -2.74390489e-01 4.50943053e-01
8.39709640e-01 5.59880994e-02 -1.31382835e+00 -8.88018459e-02
-2.12351650e-01 -7.24220574e-01 -2.94949152e-02 -1.03495967e+00
-1.30310094e+00 5.18185556e-01 8.65788758e-01 1.99318647e-01
9.54887271e-01 2.75624722e-01 1.04460454e+00 7.89946094e-02
6.36953652e-01 -1.53559864e+00 -2.30274200e-01 1.44841090e-01
1.08363116e+00 -1.62789929e+00 2.39442512e-01 -1.01715922e+00
-8.77702296e-01 1.03820610e+00 6.04863763e-01 -5.28240979e-01
6.46794975e-01 3.37043732e-01 7.16266483e-02 -4.15487140e-01
-2.09823817e-01 -3.22510719e-01 3.23512942e-01 5.26120484e-01
2.56447583e-01 7.08471715e-01 -1.72037572e-01 5.60447633e-01
-3.07756573e-01 -2.56173369e-02 1.15259543e-01 6.52165413e-01
-1.24535024e-01 -9.73553658e-01 -5.33228338e-01 5.38100064e-01
-2.46156678e-01 -2.73713712e-02 -4.89983082e-01 9.32701409e-01
5.78106880e-01 7.66710818e-01 1.80903167e-01 -1.58857867e-01
5.42678654e-01 -2.82036811e-01 9.03967798e-01 -3.02594423e-01
-7.90067971e-01 2.19615251e-01 5.42077363e-01 -5.08525968e-01
-3.78494531e-01 -7.53699958e-01 -6.65502667e-01 -6.20610178e-01
3.75056677e-02 -2.22602159e-01 1.04852378e+00 4.39419836e-01
2.82604337e-01 4.98188496e-01 1.52868003e-01 -1.06255937e+00
2.03227594e-01 -5.55652201e-01 -7.99978912e-01 6.20461166e-01
-2.19072104e-01 -5.31886935e-01 5.01604617e-01 5.28386295e-01] | [9.961243629455566, 0.3327944576740265] |
397ba321-677a-4b00-b25e-6766b2538038 | wavelet-channel-attention-module-with-a | 2007.09163 | null | https://arxiv.org/abs/2007.09163v1 | https://arxiv.org/pdf/2007.09163v1.pdf | Wavelet Channel Attention Module with a Fusion Network for Single Image Deraining | Single image deraining is a crucial problem because rain severely degenerates the visibility of images and affects the performance of computer vision tasks like outdoor surveillance systems and intelligent vehicles. In this paper, we propose the new convolutional neural network (CNN) called the wavelet channel attention module with a fusion network. Wavelet transform and the inverse wavelet transform are substituted for down-sampling and up-sampling so feature maps from the wavelet transform and convolutions contain different frequencies and scales. Furthermore, feature maps are integrated by channel attention. Our proposed network learns confidence maps of four sub-band images derived from the wavelet transform of the original images. Finally, the clear image can be well restored via the wavelet reconstruction and fusion of the low-frequency part and high-frequency parts. Several experimental results on synthetic and real images present that the proposed algorithm outperforms state-of-the-art methods. | ['Yu-Chiang Frank Wang', 'Chao-Han Huck Yang', 'Hao-Hsiang Yang'] | 2020-07-17 | null | null | null | null | ['single-image-deraining'] | ['computer-vision'] | [ 1.79767281e-01 -4.18914676e-01 2.49191806e-01 -5.75337529e-01
-4.50675845e-01 -6.72483072e-02 2.00209692e-01 -5.47529817e-01
-5.36741972e-01 8.08877110e-01 6.00727350e-02 -4.80305962e-02
-9.20047238e-02 -1.00731039e+00 -7.17775464e-01 -1.12062383e+00
-8.76932889e-02 -6.02447748e-01 3.27306777e-01 -2.25994959e-01
-2.04347391e-02 3.01787227e-01 -1.99342632e+00 3.88067186e-01
1.14263356e+00 1.30234718e+00 4.37752157e-01 6.81468189e-01
-8.65314826e-02 3.36780399e-01 -5.75520813e-01 -1.12602971e-01
2.47109935e-01 -3.42867017e-01 1.73737574e-02 1.22531749e-01
4.91441131e-01 -5.39685667e-01 -4.37158644e-01 1.70746899e+00
3.89558464e-01 8.84893462e-02 2.83635825e-01 -9.40293729e-01
-1.09255195e+00 1.81946427e-01 -6.48407757e-01 5.70824921e-01
-1.40448466e-01 1.60031870e-01 5.37902117e-01 -1.09724522e+00
7.31005520e-02 1.33852482e+00 8.88917565e-01 2.60053035e-02
-8.35428894e-01 -7.56794930e-01 9.40652043e-02 7.87571132e-01
-1.09320700e+00 -4.98986989e-02 7.58761048e-01 -1.59790173e-01
6.62982166e-01 1.34954244e-01 7.08000541e-01 7.46670961e-01
5.26974082e-01 3.88178974e-01 1.35654402e+00 -1.33532748e-01
-2.13690311e-01 -8.77006128e-02 1.86862662e-01 7.66521156e-01
5.68897367e-01 3.23466271e-01 -1.91194415e-01 1.77567273e-01
6.18188560e-01 4.40143168e-01 -1.03578317e+00 3.34152639e-01
-1.04829061e+00 9.19138730e-01 7.84822822e-01 2.97922879e-01
-6.72874749e-01 -2.86047813e-04 -1.30122647e-01 5.90239823e-01
5.28820753e-01 -1.53775766e-01 -4.99279350e-01 5.98578572e-01
-7.84646511e-01 1.76604271e-01 3.27529430e-01 4.26534504e-01
9.24983025e-01 4.37860936e-01 -2.38130406e-01 5.30663610e-01
2.87487000e-01 9.40652490e-01 4.08872157e-01 -9.35839474e-01
1.43987566e-01 -8.84413440e-03 2.59478360e-01 -1.01224291e+00
-3.70466441e-01 -5.93631446e-01 -1.29779518e+00 6.02972269e-01
1.03001297e-01 -1.44668430e-01 -1.13322854e+00 1.32667983e+00
4.38072048e-02 7.01848030e-01 3.55981231e-01 1.16920972e+00
1.08899677e+00 1.02591026e+00 -1.73191249e-01 -5.92200339e-01
1.48908269e+00 -7.93737471e-01 -1.19948900e+00 -1.04512669e-01
-4.75675225e-01 -8.65050375e-01 6.58309102e-01 5.47633052e-01
-9.62383568e-01 -1.30710828e+00 -1.34074116e+00 5.09496965e-02
-2.95085549e-01 2.44830981e-01 5.76626420e-01 4.96199995e-01
-6.96725309e-01 7.45576143e-01 -5.30328631e-01 6.79578409e-02
4.66588825e-01 -8.63778368e-02 -4.41318721e-01 -5.09707451e-01
-1.31086516e+00 9.51846063e-01 2.50042230e-01 6.51031315e-01
-6.63155019e-01 -4.82092977e-01 -8.58681321e-01 3.13661546e-01
-1.37775525e-01 -5.24309516e-01 8.23318720e-01 -1.06725192e+00
-1.11043096e+00 2.86989242e-01 -1.06309488e-01 -5.38846791e-01
5.40392809e-02 -3.79859835e-01 -8.00574064e-01 3.59658211e-01
1.10269807e-01 2.32374832e-01 1.61030400e+00 -9.83396351e-01
-9.39531624e-01 -1.59394458e-01 -2.47229069e-01 4.51455712e-02
-1.16485484e-01 -3.66638571e-01 -1.31111845e-01 -7.87035108e-01
3.65983099e-01 -1.81478620e-01 8.59034993e-03 8.94123614e-02
7.25028142e-02 5.74052148e-02 1.22855771e+00 -1.09356475e+00
7.89458156e-01 -2.54025507e+00 -6.69934377e-02 -1.84623063e-01
1.44466415e-01 3.36760044e-01 -2.49983653e-01 -1.58144504e-01
-3.06002021e-01 -1.96994737e-01 -3.47060949e-01 8.46796781e-02
-4.48872685e-01 2.99657524e-01 -3.50601166e-01 7.57799447e-01
4.51851189e-01 3.71367246e-01 -6.33013248e-01 -1.97596192e-01
4.66159582e-01 9.54779923e-01 -2.42390871e-01 2.31177628e-01
1.84027866e-01 5.14624774e-01 -1.84990078e-01 6.50095224e-01
1.29254699e+00 1.46008804e-01 -6.10132098e-01 -7.32281923e-01
-2.66416073e-01 -2.57538676e-01 -8.67139101e-01 1.27346814e+00
-2.66727030e-01 9.07905161e-01 2.90655136e-01 -1.11598277e+00
1.02376270e+00 2.63734251e-01 1.45991534e-01 -7.62865841e-01
2.88174957e-01 2.09332779e-02 6.70384243e-02 -1.16552889e+00
2.11922586e-01 -2.00457245e-01 4.61805671e-01 -1.77347407e-01
2.37282142e-01 -1.80593848e-01 -2.17401296e-01 -3.46926957e-01
5.02418399e-01 -8.66476521e-02 3.26171592e-02 -5.04521057e-02
6.27271712e-01 -3.81405115e-01 7.97504961e-01 5.88688433e-01
-1.35947719e-01 6.69730842e-01 -5.52579127e-02 -7.10583568e-01
-1.02478302e+00 -1.01591182e+00 -3.71063143e-01 7.43217587e-01
2.98722982e-01 2.68596709e-01 -5.68808794e-01 -3.11816454e-01
6.87399432e-02 5.47577143e-01 -7.17142224e-01 -2.61434138e-01
-4.46485281e-01 -8.98605704e-01 1.85023427e-01 3.87246996e-01
1.12553358e+00 -1.10079348e+00 -7.72513866e-01 1.37644887e-01
-3.77288371e-01 -1.17785430e+00 -1.51521981e-01 2.47767687e-01
-6.63324833e-01 -1.21034861e+00 -6.98340237e-01 -9.53653932e-01
4.73939300e-01 8.47445428e-01 7.63376892e-01 2.33685255e-01
-4.52653736e-01 -1.84124008e-01 -5.08951783e-01 -6.55443788e-01
1.76854059e-01 -7.14122057e-01 -2.69584090e-01 4.35184330e-01
1.54571176e-01 -7.57292926e-01 -7.13427961e-01 -2.25288779e-01
-8.61974239e-01 -1.48916736e-01 9.35503781e-01 1.24490237e+00
4.09463555e-01 8.01399887e-01 3.84023130e-01 -4.09828931e-01
5.95369399e-01 -2.10586116e-01 -9.38818991e-01 -2.15173494e-02
-2.83057958e-01 -8.98565203e-02 5.73637366e-01 -3.27135831e-01
-1.32997525e+00 -2.12557837e-01 -5.32388128e-02 -6.36687100e-01
-8.97452533e-02 4.06161010e-01 -5.67103252e-02 -2.25683704e-01
4.08950537e-01 4.57430512e-01 -8.60045943e-03 -6.44821703e-01
2.11187333e-01 6.47514999e-01 1.20558560e+00 2.82753855e-02
1.07741964e+00 7.22509563e-01 -3.09320211e-01 -9.87790465e-01
-8.90046358e-01 -1.87474340e-01 -1.08238287e-01 -3.72877419e-01
1.11827898e+00 -1.14331937e+00 -7.21760929e-01 6.58393800e-01
-1.42178500e+00 3.63803566e-01 -3.73151340e-02 8.13480735e-01
-1.08796610e-02 3.65614235e-01 -5.95012069e-01 -8.29120576e-01
-4.39830184e-01 -1.17462337e+00 8.81084323e-01 8.82335126e-01
7.33688712e-01 -4.62494850e-01 -3.77590865e-01 2.43257359e-02
7.53978610e-01 1.45579547e-01 9.32481885e-01 9.12796259e-02
-5.10551512e-01 -6.44988492e-02 -5.05756080e-01 7.15874910e-01
1.13624550e-01 1.75730698e-02 -1.24519742e+00 -3.50664556e-01
5.67990661e-01 -1.44465473e-02 1.50572121e+00 8.27167332e-01
1.30331111e+00 -1.84159234e-01 7.53075853e-02 1.09465492e+00
1.60361123e+00 4.13164422e-02 8.51707697e-01 1.84762686e-01
1.78153098e-01 5.10380149e-01 5.52533031e-01 2.82831252e-01
2.11499892e-02 6.16189949e-02 7.27390289e-01 -3.60347629e-01
-2.15849176e-01 3.14492136e-01 2.70729631e-01 5.89233398e-01
-5.48055649e-01 6.57371944e-03 -2.40882695e-01 5.81274629e-01
-1.67371511e+00 -1.30709875e+00 -3.42927009e-01 2.03340721e+00
4.21322554e-01 9.53884348e-02 -8.12222540e-01 2.52770036e-01
8.65928233e-01 2.47394547e-01 -3.81897837e-01 5.19607887e-02
-4.59699214e-01 7.14593172e-01 6.67010069e-01 6.12514913e-01
-1.30283034e+00 6.12700582e-01 5.99580908e+00 5.67016602e-01
-1.27206767e+00 1.87960774e-01 4.26624835e-01 3.47740054e-01
-1.26952723e-01 -2.08488598e-01 -2.95873553e-01 5.54214597e-01
3.49559814e-01 9.15826336e-02 5.38998604e-01 5.96439362e-01
1.96881369e-01 -2.40676790e-01 -3.32021594e-01 1.09963024e+00
-2.03060061e-02 -1.10405886e+00 -1.61889240e-01 -3.72516274e-01
5.17404258e-01 1.93824664e-01 2.30840847e-01 3.30616415e-01
1.35743037e-01 -1.08391464e+00 5.68526447e-01 8.71111095e-01
7.25175560e-01 -8.87118697e-01 1.27085161e+00 1.74145624e-01
-1.25027645e+00 -2.33500883e-01 -8.03364515e-01 -4.56913300e-02
-7.04087391e-02 9.42351103e-01 -1.71588659e-01 6.98876321e-01
1.50491142e+00 6.22416675e-01 -2.10772514e-01 1.04313529e+00
-4.64339048e-01 5.43164670e-01 -1.68751597e-01 5.17193675e-01
1.92041740e-01 -3.57048601e-01 5.17918110e-01 1.10548306e+00
5.56434929e-01 4.62268293e-01 -1.36915892e-02 6.85204327e-01
4.56482992e-02 -6.43685400e-01 -4.49292094e-01 3.55610490e-01
3.09494138e-01 1.29803419e+00 -4.35882688e-01 -4.78569597e-01
-6.55345738e-01 8.81255388e-01 -2.94326305e-01 8.11610281e-01
-9.80784714e-01 -8.06620598e-01 8.97461712e-01 -5.00353158e-01
9.30600226e-01 -6.72153383e-02 -3.65069252e-04 -1.05833447e+00
1.30995885e-01 -5.77540815e-01 1.84328794e-01 -1.14162087e+00
-1.40476143e+00 7.57139742e-01 -2.25557432e-01 -1.30118740e+00
4.30704385e-01 -5.12021065e-01 -8.48730206e-01 1.12367868e+00
-2.47472811e+00 -7.58167267e-01 -9.56484377e-01 8.93464983e-01
6.45768166e-01 -1.62620634e-01 6.90015495e-01 2.67752349e-01
-2.59248346e-01 1.20480284e-01 3.11268002e-01 6.79815114e-02
6.35526657e-01 -7.86890328e-01 1.93157211e-01 1.13515699e+00
-1.83879092e-01 1.35843456e-01 8.67541671e-01 -4.98204708e-01
-1.03777611e+00 -1.19793355e+00 5.72760344e-01 5.59528887e-01
1.22809075e-01 1.43909141e-01 -1.19714081e+00 3.61970544e-01
5.00810742e-01 6.18725657e-01 9.00864452e-02 -5.07852376e-01
-5.49790621e-01 -3.89143020e-01 -1.24542892e+00 5.93173802e-02
3.85604650e-01 -3.55876058e-01 -6.81875408e-01 2.96852797e-01
9.58148539e-01 -2.22940981e-01 -5.15429080e-01 6.25745237e-01
4.41755623e-01 -1.44393051e+00 1.13988268e+00 -3.71915966e-01
2.77388006e-01 -6.08972371e-01 -4.06735390e-01 -1.46602499e+00
-6.64752543e-01 -2.09834069e-01 1.30526841e-01 7.06847787e-01
-8.67178291e-03 -7.40656376e-01 1.54901952e-01 -4.36793774e-01
-2.62082785e-01 -4.09357250e-01 -9.65460360e-01 -4.43597406e-01
-4.20577168e-01 -2.72067010e-01 4.25731122e-01 9.54898179e-01
-6.75380170e-01 1.40987247e-01 -5.48766971e-01 9.61188257e-01
1.04316998e+00 3.82997423e-01 1.75995782e-01 -1.45774162e+00
-1.57249108e-01 -2.77413249e-01 -3.69533747e-01 -6.74486458e-01
-1.48945406e-01 -2.55844235e-01 2.21897349e-01 -1.49490631e+00
5.62813990e-02 3.12252909e-01 -3.16313297e-01 4.23532814e-01
-2.69357562e-01 4.71094698e-01 1.41954735e-01 1.40846163e-01
-1.38576522e-01 9.13627326e-01 1.42113328e+00 -4.40372825e-01
2.74068475e-01 -3.26956809e-02 -4.27442789e-01 1.05107832e+00
8.36095691e-01 -4.38834608e-01 -3.65147032e-02 -7.10786462e-01
-2.49979749e-01 2.17101946e-01 5.79076827e-01 -1.26591361e+00
2.70583838e-01 1.74745098e-02 9.56406534e-01 -6.73934758e-01
4.87068743e-01 -1.11971247e+00 -1.04475901e-01 6.23779774e-01
2.09314331e-01 -8.50866288e-02 7.79376701e-02 7.24543035e-01
-5.71168184e-01 9.48890671e-02 1.20368564e+00 -2.90158480e-01
-7.30420947e-01 1.27331331e-01 -4.86600429e-01 -5.67022800e-01
8.21773827e-01 -2.76831180e-01 -5.17220736e-01 -4.25957024e-01
-7.43519723e-01 1.70263704e-02 -1.78658292e-01 3.33751470e-01
1.15380466e+00 -1.17337644e+00 -1.07267356e+00 7.54780293e-01
-2.91867018e-01 3.90949994e-02 5.87991238e-01 8.16634357e-01
-4.93496448e-01 1.97983757e-02 -6.98661566e-01 -4.94521618e-01
-1.16742253e+00 3.89286488e-01 4.33950335e-01 2.16666996e-01
-8.17589462e-01 8.43126833e-01 1.25783056e-01 1.26370475e-01
2.60223389e-01 -6.66701674e-01 -4.96302634e-01 -1.41134322e-01
9.53332245e-01 1.69269338e-01 6.88205194e-03 -5.22662938e-01
-6.50112852e-02 7.63892651e-01 5.76618090e-02 1.23091593e-01
1.78044248e+00 -6.00517653e-02 -3.42050642e-01 3.43692154e-02
1.12049556e+00 -1.53717041e-01 -1.34229016e+00 -4.32017386e-01
-6.89780354e-01 -6.63617134e-01 6.23386204e-01 -5.95640957e-01
-1.54668140e+00 1.03669977e+00 1.03289068e+00 4.43296671e-01
1.65493512e+00 -5.08269846e-01 8.28264177e-01 4.85780507e-01
4.86424975e-02 -7.91899443e-01 -6.87038004e-02 4.42368597e-01
8.46772373e-01 -1.33816898e+00 8.57974663e-02 -2.77820915e-01
-1.67722553e-01 1.36181295e+00 5.39074659e-01 -4.11966294e-01
8.71054590e-01 3.08686674e-01 1.66126341e-01 -1.77194238e-01
-4.51338023e-01 -4.64686483e-01 9.47656482e-02 7.64821589e-01
2.90739052e-02 -1.15230814e-01 -1.30050525e-01 8.25346231e-01
-5.67809418e-02 -2.08366722e-01 4.44157273e-01 6.07000947e-01
-1.16771591e+00 -3.96565735e-01 -1.03808403e+00 2.89606273e-01
-5.27910531e-01 -2.46613070e-01 2.26479322e-01 5.53614616e-01
6.56182408e-01 1.13400090e+00 2.40270689e-01 -4.51334149e-01
1.27851278e-01 -9.46688429e-02 3.94200355e-01 1.34420153e-02
-2.18793094e-01 2.55069464e-01 -3.70444298e-01 -5.47232509e-01
-6.38095319e-01 -2.16073439e-01 -1.07402861e+00 -2.34187409e-01
-4.64056253e-01 2.21089348e-01 6.99182510e-01 7.55204380e-01
5.42823970e-02 9.67092633e-01 6.84512794e-01 -1.13126695e+00
-2.89929360e-01 -1.25850976e+00 -7.46332109e-01 2.20939636e-01
1.19749761e+00 -5.97537577e-01 -5.87227345e-01 3.45042855e-01] | [10.909921646118164, -2.9692516326904297] |
917c12bf-fdd8-4bbb-b47c-84e0da99c574 | term-sets-can-be-strong-document-identifiers | 2305.13859 | null | https://arxiv.org/abs/2305.13859v2 | https://arxiv.org/pdf/2305.13859v2.pdf | Term-Sets Can Be Strong Document Identifiers For Auto-Regressive Search Engines | Auto-regressive search engines emerge as a promising paradigm for next-gen information retrieval systems. These methods work with Seq2Seq models, where each query can be directly mapped to the identifier of its relevant document. As such, they are praised for merits like being end-to-end differentiable. However, auto-regressive search engines also confront challenges in retrieval quality, given the requirement for the exact generation of the document identifier. That's to say, the targeted document will be missed from the retrieval result if a false prediction about its identifier is made in any step of the generation process. In this work, we propose a novel framework, namely AutoTSG (Auto-regressive Search Engine with Term-Set Generation), which is featured by 1) the unordered term-based document identifier and 2) the set-oriented generation pipeline. With AutoTSG, any permutation of the term-set identifier will lead to the retrieval of the corresponding document, thus largely relaxing the requirement of exact generation. Besides, the Seq2Seq model is enabled to flexibly explore the optimal permutation of the document identifier for the presented query, which may further contribute to the retrieval quality. AutoTSG is empirically evaluated with Natural Questions and MS MARCO, where notable improvements can be achieved against the existing auto-regressive search engines. | ['Zhao Cao', 'Zhicheng Dou', 'Yujia Zhou', 'Zheng Liu', 'Peitian Zhang'] | 2023-05-23 | null | null | null | null | ['natural-questions'] | ['miscellaneous'] | [ 3.20951879e-01 -1.23678610e-01 -2.41394266e-01 -9.78754908e-02
-1.03380477e+00 -7.62195706e-01 8.10744703e-01 3.60213369e-02
-3.97862107e-01 5.72477460e-01 2.61455774e-01 -3.97565395e-01
-5.71759582e-01 -8.09528768e-01 -4.45853055e-01 -6.26677334e-01
1.11818075e-01 7.79877961e-01 8.59047696e-02 -6.55479491e-01
4.01724368e-01 2.62592673e-01 -1.78302765e+00 2.41627663e-01
9.83787358e-01 9.33010459e-01 2.65976161e-01 6.81866169e-01
-3.75130385e-01 2.63528824e-01 -7.81643152e-01 -3.11301112e-01
1.65351033e-01 -4.12457883e-01 -7.03361094e-01 -7.10163713e-01
3.88469845e-01 -3.52343708e-01 -3.59202027e-01 7.42889822e-01
5.43555081e-01 2.42742058e-02 5.48438728e-01 -9.65401113e-01
-4.19280887e-01 5.38384438e-01 -1.38473824e-01 1.98652461e-01
6.09759510e-01 1.16847724e-01 1.46743882e+00 -9.00025845e-01
7.00580776e-01 1.16011941e+00 1.79053009e-01 4.39020425e-01
-8.33993554e-01 -4.47381616e-01 1.56175867e-01 3.22823912e-01
-1.46199858e+00 -3.02283078e-01 5.02936900e-01 -1.37793973e-01
1.04704082e+00 8.04444134e-01 5.34106851e-01 7.95926273e-01
3.27418745e-01 7.18883753e-01 5.11260033e-01 -5.54606020e-01
2.16501921e-01 -1.55602023e-02 -5.11889569e-02 4.11574095e-01
1.47982808e-02 3.37688178e-01 -6.22317255e-01 -2.53525525e-01
2.74131566e-01 -7.79045746e-04 -1.56450838e-01 -3.94290872e-02
-9.40721393e-01 6.66544318e-01 4.41347897e-01 3.70719403e-01
-5.91451228e-01 7.02348128e-02 2.26138115e-01 3.22332114e-01
1.99010000e-01 7.69963622e-01 -3.86536658e-01 -1.66499004e-01
-1.21285808e+00 5.98478794e-01 6.06006682e-01 9.45120037e-01
5.67582488e-01 -3.21922362e-01 -1.03251302e+00 6.91017687e-01
2.72913009e-01 5.72561502e-01 6.84844434e-01 -4.53497171e-01
3.41561586e-01 8.59049439e-01 8.25544074e-02 -7.60960579e-01
-2.06943050e-01 -7.33932078e-01 -5.40211141e-01 -3.42061967e-01
1.03348643e-01 -6.54012039e-02 -9.49039340e-01 1.60837364e+00
2.44491294e-01 -7.44834393e-02 5.82360998e-02 8.49087656e-01
6.65108979e-01 7.89740801e-01 -1.04203470e-01 -2.43597046e-01
1.47556925e+00 -7.86300778e-01 -6.62929893e-01 -1.94017023e-01
6.88274920e-01 -6.73824906e-01 9.27064240e-01 2.43571654e-01
-1.02949154e+00 -4.37841266e-01 -9.04667854e-01 -4.94113490e-02
-3.31134915e-01 -1.73893288e-01 4.65017855e-01 4.69447106e-01
-1.17124081e+00 3.13149840e-01 -3.97912472e-01 -6.30947053e-02
-1.53451487e-01 3.47278357e-01 8.51064101e-02 -2.28376210e-01
-1.60583651e+00 7.56807864e-01 5.44863820e-01 1.43897876e-01
-5.94420254e-01 -8.70933056e-01 -4.45793152e-01 5.39802730e-01
5.40770590e-01 -7.94761360e-01 1.25263453e+00 -4.99339759e-01
-1.17733848e+00 3.35535586e-01 -3.92978966e-01 -3.79302144e-01
3.89061421e-01 -2.28608325e-01 -5.47917962e-01 1.77777201e-01
5.84788062e-02 6.14129961e-01 9.30334508e-01 -7.66097486e-01
-8.11462998e-01 -3.13200891e-01 -2.43874895e-03 3.87462467e-01
-5.30043244e-01 8.10886454e-03 -7.27589965e-01 -5.46687782e-01
-7.23595694e-02 -9.13157165e-01 -1.11632161e-01 -5.02932549e-01
-4.28934693e-01 -6.66715384e-01 4.74205643e-01 -4.65104222e-01
1.86269164e+00 -2.07368779e+00 7.62511184e-03 3.24063331e-01
7.61836115e-03 5.97036541e-01 -3.86270434e-01 8.09058666e-01
5.21319099e-02 2.90448636e-01 2.69953460e-01 1.33748397e-01
-6.15539774e-03 -1.86311454e-01 -5.31122804e-01 -1.40789539e-01
3.95005643e-01 1.25255764e+00 -1.11958122e+00 -4.86255229e-01
-1.77510917e-01 2.35138968e-01 -6.75362706e-01 1.74640492e-01
-5.98627269e-01 -7.05574378e-02 -6.72680438e-01 4.43781525e-01
4.57056671e-01 -2.43521035e-01 1.05524696e-01 1.70870069e-02
-1.78394839e-01 5.42135656e-01 -8.62185299e-01 1.51270103e+00
-4.32660848e-01 3.09646398e-01 -3.96907300e-01 -4.74440634e-01
9.84613359e-01 3.03662956e-01 4.06067342e-01 -1.23347151e+00
-4.26207751e-01 3.48073363e-01 9.13194790e-02 -3.76578957e-01
9.96395350e-01 1.38504133e-01 -5.96332476e-02 3.49037737e-01
-2.47746885e-01 -1.12398811e-01 5.01701057e-01 3.63953441e-01
1.19984293e+00 4.91406918e-02 -1.22366689e-01 -8.71425569e-02
6.63064837e-01 1.25326231e-01 3.97559196e-01 1.10091329e+00
4.06801641e-01 5.53344369e-01 3.86517763e-01 -2.68615156e-01
-9.70123291e-01 -7.83922851e-01 -3.80563587e-02 1.21832287e+00
-4.04046103e-02 -6.23901188e-01 -5.35933673e-01 -7.51404822e-01
7.60083646e-03 9.07189667e-01 -3.23546082e-01 -4.30641919e-01
-7.39717484e-01 -3.28915954e-01 6.46839023e-01 9.27715376e-02
1.15255073e-01 -1.21062970e+00 -5.74413657e-01 3.43859881e-01
-2.35797212e-01 -5.35194874e-01 -8.89323890e-01 1.16350658e-01
-6.24076068e-01 -8.94141912e-01 -8.11726391e-01 -5.45688450e-01
4.88236308e-01 2.17296600e-01 1.01762581e+00 2.48157024e-01
-3.14426750e-01 3.10183167e-01 -5.72317779e-01 -2.53451198e-01
-4.46369678e-01 5.47954023e-01 -1.01700723e-01 -1.50184169e-01
3.37312281e-01 -8.63683298e-02 -9.62771475e-01 3.85207027e-01
-1.28194237e+00 -2.23137870e-01 6.96207702e-01 1.01061821e+00
4.59667534e-01 1.01719484e-01 8.71737599e-01 -6.24870241e-01
9.42865610e-01 -5.89899480e-01 -7.46243119e-01 6.76876903e-01
-1.12978828e+00 4.67217982e-01 7.43094981e-01 -3.78355354e-01
-8.75356793e-01 -2.81180084e-01 -1.52586907e-01 -2.10913762e-01
2.64858156e-01 9.20798004e-01 1.91536143e-01 3.78247142e-01
6.56998575e-01 6.46379888e-01 9.51410383e-02 -4.39264596e-01
3.41578007e-01 7.64072955e-01 2.70149350e-01 -2.53837526e-01
8.35351288e-01 -4.86486107e-02 -2.51053758e-02 -5.10325074e-01
-6.21784508e-01 -7.11930752e-01 -2.34592304e-01 -1.00761488e-01
2.78741390e-01 -6.39504313e-01 -7.86542892e-01 1.21545844e-01
-1.09144819e+00 1.62122458e-01 -2.43338645e-01 1.90260157e-01
-1.49377272e-01 3.27325404e-01 -2.27316648e-01 -8.73664141e-01
-7.75679171e-01 -1.12011468e+00 1.12595189e+00 3.75737429e-01
-4.14340764e-01 -6.94774687e-01 -8.07272084e-03 2.60008186e-01
6.96655691e-01 -4.62135762e-01 1.32465255e+00 -1.06102312e+00
-6.73162997e-01 -5.53905487e-01 5.07030962e-03 1.03519700e-01
5.98449148e-02 -1.32782429e-01 -6.34175479e-01 -4.96640503e-01
-2.17884183e-01 -9.36767459e-02 6.47986889e-01 9.72551256e-02
8.96950483e-01 -5.61962962e-01 -3.96091670e-01 2.81199783e-01
1.20747280e+00 4.39187676e-01 6.60882115e-01 4.40575361e-01
1.37527406e-01 6.06222093e-01 9.80458498e-01 4.20015335e-01
1.19648986e-01 1.06119049e+00 4.12214667e-01 1.72438189e-01
-5.06175272e-02 -5.74423373e-01 2.40032420e-01 6.32667363e-01
3.73803645e-01 -6.50490999e-01 -8.18130910e-01 4.11903292e-01
-1.62795734e+00 -8.77134025e-01 6.44739866e-02 2.44352937e+00
9.23269331e-01 -6.64842222e-03 -9.62524563e-02 3.32587697e-02
5.03660858e-01 1.26047924e-01 -7.38701880e-01 -3.48218143e-01
-7.37716556e-02 2.81633079e-01 1.48709089e-01 4.42211419e-01
-5.40222168e-01 1.04873669e+00 6.51387310e+00 1.04349530e+00
-1.10801041e+00 -6.02732062e-01 2.95631826e-01 -3.80459428e-01
-7.26540864e-01 1.95215032e-01 -1.30741930e+00 6.15504920e-01
9.31808114e-01 -5.55031538e-01 4.46964025e-01 5.38323939e-01
1.40181467e-01 8.17308947e-03 -1.13113534e+00 5.77147901e-01
-1.91201009e-02 -1.22292948e+00 5.78588545e-01 1.90971911e-01
3.66592705e-01 -2.78254837e-01 2.36717924e-01 6.00337267e-01
-1.07334331e-01 -1.05344760e+00 4.93490338e-01 7.44187713e-01
8.60507548e-01 -8.48871589e-01 7.13293433e-01 6.45355999e-01
-8.63113165e-01 -2.62375355e-01 -1.87506884e-01 3.63578528e-01
9.08358842e-02 4.96840864e-01 -1.37228131e+00 8.30728710e-01
4.96360511e-01 2.63942540e-01 -5.85924149e-01 1.08825111e+00
-7.86263123e-02 4.71839190e-01 -3.53439718e-01 -4.99722213e-01
3.67339283e-01 -1.10272719e-02 7.67051399e-01 1.09570777e+00
5.74656367e-01 -1.95692256e-02 -7.01575875e-02 7.86757171e-01
3.33014391e-02 1.95911631e-01 -3.52579802e-01 -2.33162880e-01
8.17522585e-01 1.12758958e+00 -1.75264612e-01 -1.48501188e-01
-3.23049314e-02 7.21650720e-01 4.25304733e-02 4.17441994e-01
-4.47912127e-01 -6.17863476e-01 4.47966248e-01 1.42102301e-01
4.16358888e-01 3.09464782e-01 2.18176797e-01 -6.92259192e-01
2.03693077e-01 -1.27968884e+00 5.59874475e-01 -7.31550634e-01
-9.88965750e-01 7.17810333e-01 -7.59242848e-02 -1.20975399e+00
-9.21307445e-01 -1.06938265e-01 -1.92433804e-01 1.15682387e+00
-1.49897528e+00 -7.54400134e-01 2.83593219e-02 2.73688644e-01
4.97981161e-01 -2.33867839e-01 8.98971021e-01 2.23186165e-01
-2.52983004e-01 9.97953176e-01 2.33138919e-01 -2.54294455e-01
7.51462519e-01 -9.59672809e-01 3.76952261e-01 6.16937995e-01
1.42260149e-01 9.99754310e-01 6.53084517e-01 -8.13945711e-01
-1.74518025e+00 -8.92115355e-01 1.24077439e+00 -3.77954960e-01
4.26769197e-01 -2.05615327e-01 -9.32194769e-01 2.70804137e-01
-4.01409306e-02 -3.98804128e-01 4.84313309e-01 2.54042029e-01
-1.03431307e-01 -3.24869245e-01 -8.05258334e-01 8.24810266e-01
7.76637852e-01 -3.52839470e-01 -4.81601685e-01 3.36065203e-01
9.79835033e-01 -4.13378209e-01 -7.35082686e-01 5.12832642e-01
7.21242189e-01 -6.99840546e-01 1.01099145e+00 -6.38391316e-01
4.81749475e-01 -2.89672971e-01 1.99111938e-01 -1.21664739e+00
-4.20107991e-01 -7.63463497e-01 -3.59156311e-01 1.20304012e+00
6.29809737e-01 -5.43975353e-01 7.59736240e-01 5.63201785e-01
-1.50148454e-03 -1.03685689e+00 -1.03909218e+00 -8.69304776e-01
-8.80191103e-03 -1.50027260e-01 9.79809403e-01 3.02374631e-01
1.04882687e-01 4.48006958e-01 -3.38789940e-01 -9.43577588e-02
7.12610483e-02 5.59008941e-02 5.49046218e-01 -1.18291032e+00
-3.74966532e-01 -6.16189599e-01 -8.05525854e-03 -1.46697378e+00
-1.43906459e-01 -9.52283323e-01 2.55135775e-01 -1.31011200e+00
2.06107765e-01 -5.29550076e-01 -4.33614343e-01 3.07794988e-01
-4.74935055e-01 -3.16303670e-01 1.25041187e-01 2.62237400e-01
-6.41708732e-01 6.24268830e-01 1.20610237e+00 -2.19922125e-01
-4.22528446e-01 2.98409343e-01 -7.30338395e-01 -3.88254672e-02
4.80242640e-01 -5.02990425e-01 -5.25145769e-01 -3.51538986e-01
7.34562755e-01 4.86897141e-01 8.78396258e-02 -3.91717523e-01
6.53332591e-01 -5.78656420e-02 6.51135156e-03 -9.30081189e-01
2.93845981e-01 -6.89760506e-01 1.36879727e-01 5.74874461e-01
-6.76578581e-01 3.32742095e-01 1.18165664e-01 5.13958037e-01
-2.95308024e-01 -5.36922455e-01 6.81007132e-02 8.69645029e-02
-4.04703081e-01 2.18028173e-01 -3.28548849e-01 7.79228881e-02
5.92977643e-01 -1.85864002e-01 -3.31109047e-01 -3.78362149e-01
-5.09464256e-02 5.77132046e-01 7.45880976e-02 7.54630506e-01
6.91234708e-01 -1.11040676e+00 -9.08321023e-01 3.08182865e-01
3.26671988e-01 2.55897660e-02 1.82734892e-01 4.87780839e-01
-1.06491201e-01 1.19379938e+00 3.46880108e-01 -5.02254069e-01
-1.29627442e+00 4.78925288e-01 5.17600216e-02 -7.64989316e-01
-2.13433459e-01 6.36671722e-01 -9.88650769e-02 -3.48687857e-01
2.26497754e-01 2.44026612e-02 -3.37546676e-01 9.95569602e-02
7.06693947e-01 2.81271249e-01 3.26242924e-01 -1.30264297e-01
-3.21806252e-01 1.66333288e-01 -6.14634037e-01 -2.47232035e-01
1.11555052e+00 -4.12134044e-02 -2.47978121e-01 4.11114022e-02
1.02845716e+00 3.91219631e-02 -7.58759022e-01 -4.30303305e-01
3.58266354e-01 -3.06372583e-01 6.31189123e-02 -1.19686604e+00
-7.57415652e-01 4.72406983e-01 4.66919780e-01 2.22010598e-01
1.19721866e+00 -1.17156982e-01 8.42255771e-01 5.16003489e-01
3.19239587e-01 -9.58865643e-01 1.07804351e-01 8.89868855e-01
8.55434060e-01 -9.01370943e-01 -7.97538012e-02 -3.44724245e-02
-3.14682424e-01 1.08436787e+00 4.62431073e-01 3.69431108e-01
2.11300716e-01 -1.12337574e-01 -7.37178028e-02 -2.16050461e-01
-1.12797225e+00 -6.24523573e-02 7.69019663e-01 2.02122197e-01
4.87968117e-01 -2.74725586e-01 -5.28131664e-01 3.32983822e-01
-4.58601356e-01 -2.35021766e-02 -5.29955439e-02 7.43727744e-01
-3.74741703e-01 -1.36192143e+00 -3.26198101e-01 6.03355289e-01
-3.69852990e-01 -6.59669399e-01 -3.10813099e-01 4.33017701e-01
-3.38304788e-01 1.01326931e+00 2.80602649e-02 -3.38179916e-01
4.24256444e-01 3.00303698e-01 5.68883158e-02 -6.70215011e-01
-7.65081346e-01 1.00170210e-01 1.68152042e-02 -3.62261504e-01
3.79348069e-01 -5.18414438e-01 -1.03473628e+00 5.50968526e-03
-5.38499117e-01 6.14257395e-01 6.73175454e-01 9.70454752e-01
7.77311146e-01 5.46998799e-01 8.17165792e-01 -1.89435005e-01
-9.54002202e-01 -1.06094229e+00 -1.67789772e-01 1.57052845e-01
4.00633663e-01 -1.48652211e-01 -2.94283926e-01 -3.47901672e-01] | [11.529601097106934, 7.595560073852539] |
20d47ebd-d55b-4ba0-baf4-57669ee12a05 | clifford-neural-layers-for-pde-modeling | 2209.04934 | null | https://arxiv.org/abs/2209.04934v2 | https://arxiv.org/pdf/2209.04934v2.pdf | Clifford Neural Layers for PDE Modeling | Partial differential equations (PDEs) see widespread use in sciences and engineering to describe simulation of physical processes as scalar and vector fields interacting and coevolving over time. Due to the computationally expensive nature of their standard solution methods, neural PDE surrogates have become an active research topic to accelerate these simulations. However, current methods do not explicitly take into account the relationship between different fields and their internal components, which are often correlated. Viewing the time evolution of such correlated fields through the lens of multivector fields allows us to overcome these limitations. Multivector fields consist of scalar, vector, as well as higher-order components, such as bivectors and trivectors. Their algebraic properties, such as multiplication, addition and other arithmetic operations can be described by Clifford algebras. To our knowledge, this paper presents the first usage of such multivector representations together with Clifford convolutions and Clifford Fourier transforms in the context of deep learning. The resulting Clifford neural layers are universally applicable and will find direct use in the areas of fluid dynamics, weather forecasting, and the modeling of physical systems in general. We empirically evaluate the benefit of Clifford neural layers by replacing convolution and Fourier operations in common neural PDE surrogates by their Clifford counterparts on 2D Navier-Stokes and weather modeling tasks, as well as 3D Maxwell equations. For similar parameter count, Clifford neural layers consistently improve generalization capabilities of the tested neural PDE surrogates. Source code for our PyTorch implementation is available at https://microsoft.github.io/cliffordlayers/. | ['Jayesh K. Gupta', 'Max Welling', 'Rianne van den Berg', 'Johannes Brandstetter'] | 2022-09-08 | null | null | null | null | ['weather-forecasting'] | ['miscellaneous'] | [-5.84137082e-01 -8.70588422e-01 6.10637009e-01 -6.23798557e-02
2.40875334e-01 -6.41897440e-01 9.19652462e-01 -6.77052001e-03
-5.86384237e-01 8.61819029e-01 -2.15556324e-01 -4.81379330e-01
-1.44867614e-01 -9.33336496e-01 -6.96295142e-01 -8.07422280e-01
-5.31343699e-01 3.16225380e-01 -1.36491045e-01 -5.88920295e-01
-2.87054852e-02 8.34209979e-01 -1.51766968e+00 -4.24613096e-02
7.81800330e-01 1.04837418e+00 2.93350983e-02 8.67675304e-01
-1.53866131e-03 6.09640181e-01 -4.41874534e-01 -1.23359904e-01
2.80574352e-01 5.40019609e-02 -5.05497456e-01 -4.91547942e-01
3.06595474e-01 -2.83608824e-01 -4.93552357e-01 8.22658300e-01
5.25976777e-01 5.99846840e-01 9.74249244e-01 -1.24844813e+00
-8.25643897e-01 -1.01008974e-02 -3.54223609e-01 6.02653205e-01
-9.35426727e-02 3.22492003e-01 7.62084007e-01 -1.13883758e+00
3.07052761e-01 1.29955888e+00 1.09886241e+00 1.73989713e-01
-1.35322368e+00 -5.30153930e-01 -1.04838960e-01 3.03175896e-02
-1.44912136e+00 4.71013263e-02 6.53930366e-01 -6.55665517e-01
1.29397523e+00 3.70352685e-01 7.91407585e-01 8.76035631e-01
7.06620216e-01 4.31987554e-01 1.04765773e+00 2.67196279e-02
2.79693782e-01 1.60972387e-01 4.21181530e-01 5.02105474e-01
2.10411698e-01 3.71547163e-01 -1.35726646e-01 -4.53792602e-01
9.67876971e-01 7.66822174e-02 -3.88676882e-01 -2.21950740e-01
-1.16677225e+00 1.00387096e+00 2.78297275e-01 3.02422047e-01
-5.41612506e-01 4.91806597e-01 4.20811296e-01 1.71726152e-01
6.55190051e-01 5.56639194e-01 -4.40576613e-01 -7.32569918e-02
-6.39096797e-01 8.71604621e-01 1.06483567e+00 7.17323482e-01
7.09762454e-01 4.25423086e-01 1.12278517e-02 6.95510983e-01
-9.31977388e-03 7.45619059e-01 1.60701245e-01 -1.08707869e+00
-9.26266909e-02 1.80225179e-01 1.57038450e-01 -9.82826293e-01
-5.23413539e-01 -5.53215563e-01 -1.52013993e+00 4.03858036e-01
2.13946998e-01 -4.64493841e-01 -5.66813767e-01 1.59455109e+00
2.24624440e-01 5.11955202e-01 3.26189250e-02 1.03904915e+00
7.85783529e-01 1.02851295e+00 1.95178809e-03 1.58896595e-01
1.22353578e+00 -4.60748643e-01 -6.22587025e-01 2.60051042e-01
7.04199791e-01 -8.25456858e-01 7.18261898e-01 1.08829707e-01
-1.26693094e+00 -5.03517866e-01 -9.03952003e-01 -1.39053375e-01
-6.91004932e-01 -1.02869019e-01 1.10120320e+00 1.11035913e-01
-1.27797151e+00 1.20785236e+00 -1.05333960e+00 1.02587424e-01
1.98966742e-01 3.35635990e-01 -2.56728590e-01 4.81498003e-01
-1.53769779e+00 8.62120271e-01 -1.62713662e-01 3.54243308e-01
-5.72091579e-01 -1.44078672e+00 -8.94810379e-01 1.39047027e-01
-3.42732579e-01 -8.44628453e-01 1.19890022e+00 -3.64886642e-01
-1.21325064e+00 4.35256958e-01 -2.42485963e-02 -4.85415488e-01
4.87064749e-01 -1.86844870e-01 -3.18922698e-01 -2.79228777e-01
-1.48622751e-01 2.60327607e-01 5.00292838e-01 -7.32303441e-01
-2.01208502e-01 1.67153794e-02 2.49235988e-01 1.73406526e-01
-2.16773957e-01 -1.98938832e-01 2.11870596e-01 -6.99368536e-01
-2.72686362e-01 -9.35761273e-01 -3.68623078e-01 2.62919754e-01
-2.84749474e-02 -2.99606264e-01 1.02060866e+00 -4.27551091e-01
8.32009256e-01 -2.05369687e+00 1.90463334e-01 -9.46459845e-02
4.16918516e-01 3.92198980e-01 4.33816807e-03 5.13460040e-01
-3.31091255e-01 1.42005635e-02 -3.75154495e-01 -3.04927379e-01
1.11978978e-01 1.78608552e-01 -6.00502431e-01 8.31640720e-01
4.55530465e-01 7.84272432e-01 -8.80225241e-01 -3.93227637e-02
4.04922158e-01 1.09583962e+00 -6.29011095e-01 9.14655253e-02
-6.85460120e-02 4.74794209e-01 -3.42899203e-01 7.89162591e-02
1.18299460e+00 -3.28614414e-01 -4.54231977e-01 -3.20544243e-01
-5.37374794e-01 2.05908224e-01 -1.40111434e+00 1.05839694e+00
-7.65333235e-01 8.14594865e-01 4.32684928e-01 -1.18931866e+00
6.52638197e-01 3.88915122e-01 6.36598349e-01 -5.02112806e-01
9.24467742e-02 3.23809475e-01 1.13403387e-01 -2.47767657e-01
5.57189763e-01 -3.20591360e-01 1.80831701e-01 4.73866761e-01
9.75060984e-02 -4.94891822e-01 3.88949722e-01 1.96315005e-01
8.26468706e-01 2.84320209e-02 1.83505446e-01 -7.76558876e-01
7.50097513e-01 -5.49799986e-02 1.80286363e-01 4.82210726e-01
1.62834585e-01 3.09371471e-01 3.22643489e-01 -7.63173282e-01
-1.06230068e+00 -1.15969396e+00 -5.81030428e-01 6.59867883e-01
-3.49689997e-03 -2.47883037e-01 -2.90235639e-01 2.79231548e-01
3.31748366e-01 6.01483405e-01 -5.93807459e-01 -2.15949550e-01
-8.04108441e-01 -9.98635530e-01 7.18939006e-01 6.76313818e-01
5.68439484e-01 -9.51785326e-01 -5.88101089e-01 3.02852690e-01
3.39816123e-01 -9.46602762e-01 -2.59652227e-01 3.74163568e-01
-7.75970161e-01 -6.68143332e-01 -1.01841021e+00 -4.35197741e-01
2.27850512e-01 1.12563677e-01 1.11359847e+00 -1.13972999e-01
-4.53833729e-01 3.48342836e-01 3.93578082e-01 -2.79014528e-01
-2.47257769e-01 -3.86472642e-01 3.07376087e-01 -9.32495967e-02
5.68902083e-02 -1.05679429e+00 -6.04556978e-01 1.51523665e-01
-9.85427856e-01 -4.98001724e-02 4.94430736e-02 7.26617932e-01
2.59115249e-01 -2.39911810e-01 6.19432516e-02 -5.01058221e-01
9.69094515e-01 -5.16649604e-01 -9.34188366e-01 -2.61017740e-01
-5.27068302e-02 2.12366492e-01 1.11704695e+00 -6.87556267e-01
-1.03052580e+00 -3.84088337e-01 -1.49176419e-01 -7.27731407e-01
-4.55791429e-02 5.95513463e-01 5.08902431e-01 -3.06238472e-01
5.94532311e-01 1.98629841e-01 -9.31507424e-02 -4.45967704e-01
1.47898287e-01 2.23841146e-01 5.47935486e-01 -7.61212468e-01
5.66473126e-01 6.72551751e-01 5.30624628e-01 -1.14347196e+00
-2.98463225e-01 -4.76096511e-01 -3.50052208e-01 -1.30088225e-01
7.58430660e-01 -8.17645311e-01 -1.07841647e+00 9.77782428e-01
-1.72997940e+00 -3.70305061e-01 -2.45715812e-01 7.34349489e-01
-3.57579857e-01 2.20705703e-01 -9.10870969e-01 -7.76484370e-01
-2.18248144e-01 -1.17726290e+00 9.17225301e-01 9.34172198e-02
-2.61384845e-01 -1.58077061e+00 2.36074671e-01 -4.43573475e-01
8.34663630e-01 2.50577629e-01 1.00582206e+00 -1.83686256e-01
-3.70802701e-01 4.59447764e-02 -1.56078801e-01 5.64664245e-01
-2.66296953e-01 1.60252154e-01 -9.33159411e-01 -1.53119043e-01
3.11573505e-01 8.23230073e-02 8.97641599e-01 5.34612358e-01
1.04877031e+00 -1.53136253e-01 -3.05883020e-01 1.16441369e+00
1.33295441e+00 1.00826986e-01 2.25207180e-01 -2.25980327e-01
8.27105165e-01 4.22596663e-01 -1.49124444e-01 7.30014265e-01
2.21667841e-01 4.14686233e-01 2.43622392e-01 -3.58090371e-01
2.11799011e-01 4.95287597e-01 2.20119730e-01 7.76200354e-01
-5.45882702e-01 -1.65229887e-01 -1.01370156e+00 3.15855205e-01
-1.45787466e+00 -9.44102347e-01 -7.51358867e-01 1.98499739e+00
6.03198946e-01 -2.25671545e-01 -3.23417991e-01 -2.00951993e-02
5.50892234e-01 1.52926341e-01 -5.42544425e-01 -7.34713435e-01
-2.09867254e-01 5.86016715e-01 6.21164143e-01 6.71284258e-01
-1.26035166e+00 5.27293205e-01 5.67974091e+00 4.98188138e-01
-1.64946961e+00 1.24961756e-01 3.18712115e-01 1.42176105e-02
-1.63510308e-01 -1.83198243e-01 -6.69455111e-01 2.80535638e-01
8.95875812e-01 -2.16360897e-01 5.35353065e-01 4.69666094e-01
3.38082939e-01 1.90952763e-01 -1.21470165e+00 9.16065633e-01
-5.52617371e-01 -1.60720325e+00 -4.09345999e-02 1.78273752e-01
7.50037909e-01 3.15888703e-01 1.95933238e-01 3.16864341e-01
4.37586039e-01 -1.23209691e+00 5.71056902e-01 5.84129274e-01
6.24388874e-01 -4.79491711e-01 6.67975962e-01 2.05998644e-01
-1.14994633e+00 3.51752639e-01 -4.56684798e-01 -6.09732509e-01
3.89434338e-01 8.89423370e-01 -2.80191213e-01 2.83938795e-01
7.85665691e-01 9.57847178e-01 5.04674986e-02 7.84100652e-01
4.59973723e-01 4.52225864e-01 -5.51404417e-01 -1.48585707e-01
4.50839847e-01 -7.13510215e-01 7.78232157e-01 1.28944635e+00
5.14627159e-01 1.93977922e-01 -2.28320494e-01 1.40090287e+00
2.55585223e-01 -1.18025951e-01 -7.16820359e-01 -1.69446573e-01
4.67705773e-03 1.27570570e+00 -5.22936702e-01 -4.18816119e-01
-4.83763665e-01 6.07015789e-01 1.30427688e-01 7.67333388e-01
-1.04194200e+00 -3.97654682e-01 1.45806718e+00 1.35982215e-01
3.40136886e-01 -7.49091387e-01 -2.81918377e-01 -1.23392320e+00
-5.82571924e-02 -3.63576263e-01 -2.58925389e-02 -7.19160140e-01
-1.52139449e+00 6.86233401e-01 2.18234286e-01 -1.05743253e+00
-2.49914572e-01 -1.02191722e+00 -8.88408661e-01 1.37276340e+00
-1.42572737e+00 -5.86273611e-01 -3.53989661e-01 6.79401398e-01
-1.88468561e-01 7.27726519e-02 9.31401610e-01 4.61559087e-01
-2.94075996e-01 -2.20739674e-02 4.61952537e-01 1.23449014e-02
2.88521141e-01 -1.21851397e+00 5.93119621e-01 4.70982462e-01
-4.10045177e-01 8.32443357e-01 9.06554222e-01 -3.13178807e-01
-1.45828354e+00 -1.01769030e+00 6.89045250e-01 -1.10251114e-01
9.50780869e-01 -5.62735796e-01 -1.21032572e+00 4.69469875e-01
1.81565389e-01 5.03117263e-01 1.85515970e-01 -2.58242786e-01
-1.53599024e-01 -6.34733513e-02 -6.70852959e-01 6.85714006e-01
8.93137336e-01 -3.58648747e-01 -1.07358709e-01 5.64516664e-01
3.97096783e-01 -4.66938496e-01 -9.02047932e-01 5.89913011e-01
3.63466471e-01 -1.11307549e+00 1.34624910e+00 -5.03344893e-01
5.57201624e-01 -1.52422234e-01 3.50460410e-02 -1.49023998e+00
-1.43781960e-01 -6.90918148e-01 -1.50693625e-01 6.79400742e-01
2.15516984e-02 -1.11097956e+00 1.64718211e-01 3.30569863e-01
-3.20698172e-01 -1.02292097e+00 -8.28057051e-01 -7.98905432e-01
6.74069643e-01 -4.69704926e-01 4.68640566e-01 1.02686262e+00
-3.16268593e-01 1.56090558e-01 -1.76380783e-01 2.34688327e-01
4.10877913e-01 1.45016715e-01 4.48063016e-01 -1.42486715e+00
-3.33626807e-01 -7.90895641e-01 -3.82181168e-01 -9.24008906e-01
4.01109695e-01 -1.12022710e+00 -4.68382016e-02 -1.30895042e+00
-3.80805492e-01 -5.88831604e-01 -2.98963650e-03 -6.13995343e-02
1.41762465e-01 1.79026008e-01 2.88265459e-02 -6.56407923e-02
1.73551992e-01 6.42239273e-01 1.36225212e+00 8.39531329e-03
-1.31302580e-01 -2.53093001e-02 -1.64157972e-01 7.28008687e-01
9.11837876e-01 -6.93831965e-02 -2.91047275e-01 -5.45397937e-01
8.03836137e-02 1.83169946e-01 8.67882490e-01 -1.16160917e+00
2.77888000e-01 -1.19109698e-01 2.87066162e-01 -3.44009787e-01
6.86791718e-01 -4.03514236e-01 3.32502574e-01 4.97372031e-01
-3.20011750e-02 6.46811843e-01 5.17963827e-01 6.85733110e-02
-2.99162567e-01 5.51522151e-02 7.85887182e-01 -2.09028378e-01
-5.18372118e-01 4.99800891e-01 -8.23399484e-01 1.97465539e-01
8.13328683e-01 5.20926639e-02 -1.67187691e-01 -5.04033089e-01
-4.19306457e-01 1.37339374e-02 1.73434705e-01 9.41479057e-02
4.46112007e-01 -1.15355921e+00 -7.35091031e-01 4.54774261e-01
-5.95959544e-01 1.57806531e-01 3.56063247e-01 1.03767717e+00
-8.77582788e-01 5.00555396e-01 -4.25673634e-01 -7.33713269e-01
-8.41930747e-01 3.32062185e-01 8.02642643e-01 -3.07951957e-01
-5.53805947e-01 7.94280648e-01 8.02161634e-01 -5.66181302e-01
-5.98786436e-02 -6.08581901e-01 -7.34810010e-02 -4.33119573e-02
2.41460979e-01 6.19375229e-01 1.32389277e-01 -6.71139896e-01
-5.00427306e-01 6.40227497e-01 4.10085291e-01 -1.01370968e-01
1.46747541e+00 3.47118527e-01 -4.79176581e-01 4.52847540e-01
1.60881162e+00 -2.17641398e-01 -1.08352482e+00 3.42381895e-02
-5.39267778e-01 1.17437050e-01 2.54244626e-01 -2.40348905e-01
-1.15628874e+00 1.32562673e+00 2.13824168e-01 2.35952839e-01
7.02234626e-01 -2.53088027e-01 8.58045638e-01 4.14097786e-01
-5.02853617e-02 -4.66564536e-01 -3.49536449e-01 1.14371252e+00
9.81435776e-01 -8.41432095e-01 -1.17554508e-01 -4.18582886e-01
-1.15295902e-01 1.23357856e+00 5.17054677e-01 -7.51520753e-01
1.22661006e+00 8.80404413e-01 2.04999354e-02 -1.11885443e-01
-7.70079255e-01 1.68216944e-01 3.11032265e-01 2.35974506e-01
7.98908234e-01 9.53505486e-02 -4.25448745e-01 3.92023981e-01
-2.73477316e-01 -1.41621763e-02 6.93391144e-01 8.56975138e-01
2.50503033e-01 -8.17795277e-01 -3.77064973e-01 3.69603008e-01
-3.18211466e-01 -3.05654556e-01 2.47400463e-01 7.88525045e-01
1.62699103e-01 3.84773850e-01 5.97682297e-01 -6.07234612e-02
2.99428701e-01 -1.03271648e-01 3.61801565e-01 -1.88941285e-01
-6.47729099e-01 -2.64862210e-01 -3.18429887e-01 -3.80460292e-01
-3.68727833e-01 -6.91684783e-01 -1.47719872e+00 -6.93936706e-01
1.26653075e-01 2.42531881e-01 6.24530673e-01 8.07525337e-01
3.14946234e-01 6.75905526e-01 1.57976449e-01 -1.34243846e+00
-8.03580999e-01 -9.51083958e-01 -6.61733687e-01 4.38879758e-01
7.32817590e-01 -8.90423000e-01 -5.33639193e-01 -1.27308786e-01] | [6.522068023681641, 3.3594846725463867] |
1c4d7019-60c1-46cf-8bf5-60862580e8c9 | visual-transformers-with-primal-object | 2112.05485 | null | https://arxiv.org/abs/2112.05485v2 | https://arxiv.org/pdf/2112.05485v2.pdf | Visual Transformers with Primal Object Queries for Multi-Label Image Classification | Multi-label image classification is about predicting a set of class labels that can be considered as orderless sequential data. Transformers process the sequential data as a whole, therefore they are inherently good at set prediction. The first vision-based transformer model, which was proposed for the object detection task introduced the concept of object queries. Object queries are learnable positional encodings that are used by attention modules in decoder layers to decode the object classes or bounding boxes using the region of interests in an image. However, inputting the same set of object queries to different decoder layers hinders the training: it results in lower performance and delays convergence. In this paper, we propose the usage of primal object queries that are only provided at the start of the transformer decoder stack. In addition, we improve the mixup technique proposed for multi-label classification. The proposed transformer model with primal object queries improves the state-of-the-art class wise F1 metric by 2.1% and 1.8%; and speeds up the convergence by 79.0% and 38.6% on MS-COCO and NUS-WIDE datasets respectively. | ['LongLong Yu', 'Joost Van de Weijer', 'Vacit Oguz Yazici'] | 2021-12-10 | null | null | null | null | ['multi-label-image-classification'] | ['computer-vision'] | [ 0.59241545 -0.06540931 -0.22908238 -0.5646367 -1.0390016 -0.628068
0.5434346 0.45356372 -0.46180272 0.294161 -0.18605775 -0.28565636
0.13675247 -0.66526634 -1.0219728 -0.62266684 0.45923638 0.73499537
0.57515526 0.15163817 0.37443495 0.08282934 -1.9029074 0.98847884
0.52509874 1.5366338 0.6051833 0.6996431 -0.23274072 1.1625754
-0.29409006 -0.53182703 0.04411706 -0.1647634 -0.83205014 -0.00794845
0.7436164 -0.47511172 -0.07128289 0.98295313 0.46523145 -0.24602877
0.77863187 -1.2234259 -0.7379933 0.57985705 -0.7629924 0.17712332
-0.08110293 -0.06976765 1.3116617 -1.1365013 0.3290664 1.215735
0.4939935 0.6729979 -0.80557686 -0.60538465 0.24156815 0.55633855
-1.318081 -0.56240004 0.42137054 -0.34151047 1.0787807 0.5799817
0.28192288 0.7295433 0.17520109 1.2382259 0.94752026 -0.43480468
-0.0476182 0.11661714 0.35140482 0.9958638 -0.13114133 -0.20559172
-0.6258522 0.02479758 0.41203395 0.23099464 0.06937653 -0.11433946
-1.2053132 0.8184504 0.4832213 0.1308705 -0.24618948 0.4566258
0.58565676 0.16585125 0.57478255 -0.1512195 -0.47188154 0.1871239
-0.82025474 0.04376677 0.44609708 1.2155962 0.5216796 -0.4244912
-0.61863786 1.011777 0.7205574 0.4250087 0.42759907 -0.5909165
0.7575172 0.73746294 -0.04500975 -0.67509365 -0.35811496 -0.65987
-0.62043774 -0.29149368 0.3485169 0.30925956 -1.1430207 1.6119897
0.43713537 0.2868105 0.02277454 0.8015053 0.9979098 0.7811119
0.19404772 -0.16963904 1.5605446 -1.403102 -0.61321247 -0.37308583
0.82884455 -0.8563859 0.8696361 0.27722043 -1.023802 -0.6150978
-0.8650366 -0.36794716 -0.29904604 0.43594706 0.3945783 0.38708484
-1.0512265 0.35201874 -0.6290268 -0.08467545 0.86141545 0.56073797
0.08035655 -0.29581216 -0.78622967 0.7527892 0.33248308 0.08770356
-1.367037 -0.6529756 -0.58814627 0.14442284 0.39773712 -0.3586319
1.520963 -1.0070499 -1.0297375 1.111568 -0.31903756 -0.4731574
0.56370544 -0.2104098 -0.18211511 0.12274452 0.27750126 1.0569447
0.84177685 -1.2161446 -1.2598258 -0.5409067 0.10850734 0.3705254
-0.45440197 0.03108183 -0.652426 -0.39999822 0.3402275 -1.0094414
0.07296106 0.11052195 -0.48942986 -0.42372915 0.834493 -0.59212255
1.1788384 -2.0527165 0.04854523 -0.24080117 0.32111192 0.07695953
-0.16171351 0.11895776 0.03355694 0.11007778 0.05645892 -0.6504996
-0.11822558 0.10796653 -0.1681099 0.52584887 0.06069848 0.9699355
-0.78869855 -0.6900424 0.13381407 0.2438688 -0.507836 0.08763064
-0.5712041 0.1539284 -0.35803634 0.7992331 0.62373686 -0.5236109
0.05481358 -0.5058747 -0.02099149 0.22816165 -0.9289896 1.7623324
-0.44330364 0.48896497 -0.4063937 -1.0818831 0.6011495 0.5079752
0.5746589 -0.9768864 0.11870577 0.26668978 -0.22130728 -0.40484977
0.3882714 -0.0313665 -0.15932934 0.33957586 0.10601117 0.4128854
0.11818974 0.04045766 0.9025799 0.04961271 0.15445828 -0.20013046
0.57359016 -0.05749515 0.54268974 0.7275035 -0.01220355 0.57411563
0.37775367 -0.55384266 -1.1424013 -0.7698119 -0.09377838 1.594324
0.20232779 -0.21119694 -0.49282053 -1.0160183 -0.08168983 0.6544025
-0.6626428 -0.11355489 -0.46139416 -0.84045595 0.5058471 0.38558993
0.61154515 -0.7610605 -0.7640368 0.32533994 -0.32710907 -1.1332939
-0.6504958 0.48598462 -0.8898617 -1.1872606 -0.6833968 -1.0577163
0.66153973 0.24697867 1.0491241 0.12988202 -0.04136962 0.00740068
-0.4898125 -0.51042867 -0.47308233 0.12955114 -0.2822911 0.40673634
0.22692329 -0.12459654 -0.5364781 0.4633809 -0.79263824 0.645377
0.53781813 0.94316554 0.82437617 -0.22636947 0.5688459 -0.9902766
0.10473999 -0.51271784 -0.602551 0.6953389 -0.66533077 0.20823562
0.5824467 -0.38082647 -0.7587326 0.38384762 -0.12725405 -0.45837176
0.08116675 0.12916356 0.07513645 -0.00867742 0.37959617 0.45426118
-0.23727715 -0.6674158 0.26225552 0.9728959 0.0958699 -0.16888307
0.05580428 0.32071498 -0.03168759 -0.21073635 -1.214547 -0.6834246
-0.48946288 -0.42129692 0.9349218 -0.90565085 -0.9540363 0.67815405
-1.2962713 -0.16283642 0.06508739 -0.03159558 -0.45783862 0.13198365
-0.5996704 -0.6015968 -0.6640314 -1.4820313 1.4020444 0.11577423
0.30161044 -0.5497719 -0.4635109 0.5702115 0.19586155 -0.22944435
1.0730093 -0.8815026 -0.7178726 0.05100081 -0.46380302 0.21078746
-0.12887506 -0.43209475 -1.2498063 -0.41442043 -0.05474404 -0.5155891
0.9503054 0.1178821 1.4788007 -0.45370513 -0.61173946 0.27531672
1.8404024 0.35372737 0.34646508 0.20858775 0.88849247 0.4361076
0.78791714 0.49098277 0.6390134 0.8700881 1.0437562 0.06382861
-0.26238924 -0.05045055 0.15141486 0.7095626 0.51573825 -0.6379743
-0.94977736 0.6570635 -2.0249765 -0.6563546 -0.41064578 2.043995
0.8129925 0.15195239 -0.08224038 0.2048782 0.73665094 -0.02765556
-0.729368 -0.22651991 0.16975994 -0.04141792 0.7922244 0.4293946
-1.2108405 0.97706413 5.564152 0.9975985 -1.1218767 0.5000034
0.9316694 -0.2944173 -0.16080494 -0.29298866 -1.237931 0.5059461
1.0094961 0.35578004 0.1681632 0.7325996 -0.11181447 -0.17340505
-1.5253463 1.1879535 0.16653821 -1.3533659 0.21888444 -0.12322348
0.511714 0.06080145 0.28369635 0.25503588 0.03179345 -0.84333044
1.2385571 0.4093302 0.9482789 -0.6463537 0.63442314 0.41617504
-1.2411735 -0.44691992 -0.3450108 0.19640821 -0.04626996 0.47395822
-1.1220175 0.29154092 0.56522363 0.73615736 -0.7532005 1.0783641
0.22731178 0.67865896 -0.3622491 -0.17221136 0.4903383 0.36636803
0.02348413 1.0890586 0.19420332 -0.0722736 0.20262389 0.3358098
-0.25107634 0.09288787 -0.29534733 0.319169 0.36947995 1.0702343
-0.70363784 -0.49038377 -0.54990196 1.1675938 0.43395358 -0.02924122
-1.2846717 0.09649491 0.38182956 -0.07616226 0.53417534 0.30192983
-0.47901186 -0.74136144 -0.01942683 -0.8686217 0.506633 -0.7669875
-0.8158212 0.4025624 -0.33946767 -1.2124227 -0.05269618 -0.61863357
0.05513474 0.619065 -1.5792044 -1.4868039 -0.2761117 0.3937557
0.9459937 -0.04982553 0.865124 0.81138927 -0.519992 0.89379513
0.39872882 -0.04540475 0.50079817 -1.0901957 0.24395938 0.532315
0.4000268 -0.04436554 0.4118542 -0.3240787 -1.4034045 -1.3148082
1.1212268 -0.4417077 0.16170646 -0.4636442 -0.594735 0.6598046
-0.02300377 0.05434491 0.6139196 -0.27683195 -0.3459133 -0.417664
-1.2338445 0.34093928 0.98308074 -0.47877768 -0.15436263 0.657492
0.87286854 -0.49974054 -0.7044982 0.23860098 0.6205571 -0.6865845
0.99542224 -0.45181346 0.47203305 -0.2359836 -0.48799157 -0.7040657
-0.37633044 0.15859447 -0.19880393 1.148554 0.6020944 -0.25712445
0.75013083 0.14450157 -0.22830957 -1.0675344 -1.0413134 -0.2586861
-0.3330396 -0.5424867 0.44300687 0.6673977 -0.5271085 0.62210155
-0.2961807 0.2877406 0.7130712 0.13076352 0.12491468 -1.0674565
-0.09661288 -0.21035418 -0.19176473 -1.3676957 -0.08451895 -1.2200224
0.12723511 -1.5425205 0.44070974 -0.86269623 -0.5189199 0.8048469
-0.05091312 0.40793136 0.40418357 0.42642477 -1.0192987 0.27532187
1.1030185 -0.4540402 0.24392946 0.12628852 -0.6406534 0.42982867
0.66364264 -0.8292616 -0.52172124 -1.0064305 0.2189619 0.18713397
0.3586699 -1.0250843 0.451788 0.11262757 0.23576793 -1.044618
0.37405288 -0.9742633 0.06159892 0.5660266 -0.5987253 0.03714819
0.19340453 0.6316092 0.02883752 -0.5163804 0.87377304 -0.2461373
-0.9792207 0.31718028 -0.174388 -0.07473844 1.1922599 -0.15604387
-0.12184609 0.05224465 -0.5901393 0.42225376 0.09821142 0.48116866
0.6482009 -1.3269901 -0.6872471 0.00974588 0.22457056 0.15853359
0.19795406 0.6155025 -0.40293187 0.6398285 -0.01423405 -0.9797361
-1.5989418 0.61361337 0.47882894 -0.43689173 -0.30039343 1.2254227
0.33066377 -0.2590377 0.40068445 -0.41119343 -0.41935048 0.25300294
0.4938875 0.29604366 0.17391995 -0.656214 -0.5048298 0.65625215
-0.38917816 -0.01567517 1.2949303 -0.18086937 -0.09139015 0.5958039
1.5522057 -0.6321551 -1.1281924 -0.40012294 0.10683265 -0.27284116
0.21097967 -1.0636203 -1.109588 0.926232 1.091332 0.06219621
1.2062953 0.10896109 0.69465524 0.15039638 0.36832777 -0.9958543
0.26032716 0.609484 0.66653895 -1.3919338 -0.33994374 -0.4723289
-0.45625246 0.87734586 0.6981015 0.42022753 0.464688 0.16612904
-0.21389653 -0.21168663 -1.0335236 -0.05497519 0.2698432 0.22138114
0.41557634 -0.08540051 -0.2948083 0.32856983 0.36404622 0.08037347
0.09831642 0.8067371 -0.7338206 -0.9734219 -0.19457427 0.8494608
-0.70458585 -0.38422436 0.16064881 0.08524093 0.39417532 0.99735516
0.11454567 -0.42569143 0.19072396 0.28473166 0.4548198 -0.7483628
-0.6577564 -0.16691782 0.12060159 -0.32390144 -0.1678953 -0.552277
-1.3336643 -0.10807614 -0.5072971 -0.10976295 0.7864402 1.024293
0.36823645 0.6690282 0.7176266 -0.33030215 -0.74235547 -1.0442302
-0.19013393 0.18520604 0.39231813 -0.47676226 0.14822991 0.20837475] | [9.79572868347168, 3.9635047912597656] |
dd5006ef-4060-463f-8890-0ea37480eef0 | multi-scale-temporal-network-for-continuous | 2204.03864 | null | https://arxiv.org/abs/2204.03864v2 | https://arxiv.org/pdf/2204.03864v2.pdf | Multi-scale temporal network for continuous sign language recognition | Continuous Sign Language Recognition (CSLR) is a challenging research task due to the lack of accurate annotation on the temporal sequence of sign language data. The recent popular usage is a hybrid model based on "CNN + RNN" for CSLR. However, when extracting temporal features in these works, most of the methods using a fixed temporal receptive field and cannot extract the temporal features well for each sign language word. In order to obtain more accurate temporal features, this paper proposes a multi-scale temporal network (MSTNet). The network mainly consists of three parts. The Resnet and two fully connected (FC) layers constitute the frame-wise feature extraction part. The time-wise feature extraction part performs temporal feature learning by first extracting temporal receptive field features of different scales using the proposed multi-scale temporal block (MST-block) to improve the temporal modeling capability, and then further encoding the temporal features of different scales by the transformers module to obtain more accurate temporal features. Finally, the proposed multi-level Connectionist Temporal Classification (CTC) loss part is used for training to obtain recognition results. The multi-level CTC loss enables better learning and updating of the shallow network parameters in CNN, and the method has no parameter increase and can be flexibly embedded in other models. Experimental results on two publicly available datasets demonstrate that our method can effectively extract sign language features in an end-to-end manner without any prior knowledge, improving the accuracy of CSLR and achieving competitive results. | ['Quan Gan', 'Fei Yuan', 'Jing Li', 'Qidan Zhu'] | 2022-04-08 | null | null | null | null | ['sign-language-recognition'] | ['computer-vision'] | [-2.40215380e-02 -7.89813995e-01 -3.52403760e-01 -3.55837464e-01
-6.68998837e-01 -1.69123396e-01 4.27334189e-01 -7.66238809e-01
-8.57994378e-01 3.84329051e-01 2.79765666e-01 -4.68100794e-02
-1.31730050e-01 -3.56667548e-01 -3.57986838e-01 -9.48656976e-01
-5.18494770e-02 -2.52790660e-01 8.67815912e-01 -1.16022579e-01
6.27938062e-02 6.19683146e-01 -1.45570517e+00 4.65693355e-01
7.51672804e-01 1.21545660e+00 2.88151413e-01 4.78633106e-01
-2.81853706e-01 1.04800653e+00 -2.01983407e-01 1.48578301e-01
2.41598219e-01 -4.30952489e-01 -4.68322128e-01 -1.88945845e-01
1.75718427e-01 -5.20010412e-01 -9.60530639e-01 6.21278882e-01
7.70412862e-01 3.11135292e-01 2.17677757e-01 -9.75158572e-01
-3.10766608e-01 1.48438230e-01 -3.51887375e-01 2.64164537e-01
-1.48595184e-01 6.26275539e-01 7.09759951e-01 -6.62409067e-01
5.84314585e-01 1.19110668e+00 5.68223596e-01 7.21350610e-01
-5.88183939e-01 -8.76305819e-01 5.45270801e-01 8.03695202e-01
-1.27455759e+00 -1.49448529e-01 8.64472687e-01 -2.05730930e-01
1.02516389e+00 8.21510237e-03 1.11310184e+00 1.14293969e+00
1.72214895e-01 1.17017889e+00 1.26541412e+00 -1.88709646e-01
-1.01668045e-01 -5.50791681e-01 2.70506322e-01 6.77129090e-01
-4.56659824e-01 4.10492301e-01 -5.55438519e-01 3.59753966e-01
8.88883173e-01 3.66682589e-01 -2.58157164e-01 -2.94109527e-02
-1.07224953e+00 4.85008508e-01 8.07294965e-01 7.72659838e-01
-4.28670615e-01 4.66137022e-01 6.55208826e-01 2.83044785e-01
1.26110226e-01 -5.07257938e-01 -5.73164880e-01 -3.48390549e-01
-8.94246101e-01 -1.18804194e-01 1.80052295e-01 6.73288405e-01
2.22898960e-01 1.08858474e-01 -4.13714051e-01 9.60950136e-01
2.60627925e-01 4.12019312e-01 1.10365236e+00 -5.66146791e-01
3.89668196e-01 6.12653971e-01 -2.84542829e-01 -4.43437368e-01
-4.91072327e-01 -4.66930896e-01 -8.12853694e-01 3.90693218e-01
4.69067514e-01 1.50207624e-01 -1.52752805e+00 1.65389597e+00
7.33323470e-02 5.20719886e-01 -1.09662838e-01 1.17127240e+00
7.19089448e-01 5.87050796e-01 3.07078302e-01 -1.40700281e-01
1.38395476e+00 -1.04657841e+00 -7.04405189e-01 3.43986340e-02
6.65904939e-01 -5.62339425e-01 9.72393513e-01 3.22825700e-01
-6.86882675e-01 -5.63150644e-01 -7.91940928e-01 -4.03011441e-01
-4.61348623e-01 6.30697370e-01 6.92709446e-01 8.11919272e-02
-8.26841950e-01 3.80749553e-01 -1.30632782e+00 -3.32114518e-01
3.32110912e-01 3.96573752e-01 -2.77090311e-01 -1.83454692e-01
-1.18266058e+00 8.75738025e-01 2.78898239e-01 6.74105227e-01
-6.02039933e-01 -3.99630427e-01 -8.71348023e-01 -2.33492285e-01
6.46410137e-02 -4.00219291e-01 1.05652535e+00 -1.00284660e+00
-1.79783928e+00 5.04742801e-01 -4.72162724e-01 -4.95533466e-01
8.70902359e-01 -1.03545070e-01 -4.88838226e-01 2.62870789e-01
-3.84374619e-01 6.95693433e-01 9.63022828e-01 -6.34565413e-01
-9.36627865e-01 -2.91195631e-01 -1.19463079e-01 7.19682351e-02
-2.93817788e-01 2.03036025e-01 -6.53894544e-01 -7.89305627e-01
3.74284655e-01 -9.43850338e-01 -1.72971189e-01 2.01533511e-01
2.08508909e-01 -4.53292400e-01 1.01446474e+00 -1.08267117e+00
1.35038757e+00 -2.21242309e+00 6.29564002e-02 7.84035176e-02
-7.86017478e-02 6.15764320e-01 -2.89601356e-01 1.95942428e-02
-5.03269881e-02 -2.78603792e-01 -2.61927068e-01 -2.31969878e-01
-2.80505419e-01 2.69486278e-01 -2.02170387e-01 4.00251806e-01
1.62371650e-01 1.03272605e+00 -7.21992016e-01 -5.76106429e-01
6.27865314e-01 8.59954953e-01 -3.06195855e-01 -1.69953510e-01
-8.32796916e-02 6.91891253e-01 -5.23349345e-01 6.37709320e-01
4.43276048e-01 7.76195992e-03 -1.18996866e-01 -3.73681366e-01
-4.26771343e-01 3.39204818e-01 -9.12213564e-01 1.84925210e+00
-7.19259799e-01 6.97659373e-01 -3.12531203e-01 -7.98057437e-01
7.71828115e-01 4.72576946e-01 7.18748391e-01 -1.16906953e+00
3.79754990e-01 3.95517260e-01 1.40289828e-01 -7.22880602e-01
-1.83622483e-02 -8.50869417e-02 1.11115396e-01 1.28194585e-01
2.42448244e-02 5.24826884e-01 1.53815851e-01 -2.91458815e-01
1.18693590e+00 4.32400107e-01 -1.63937494e-01 4.37662989e-01
7.48776555e-01 -1.62853628e-01 8.94900143e-01 1.73292384e-01
-4.40801501e-01 4.48238045e-01 2.05192734e-02 -5.75246930e-01
-8.05476367e-01 -7.59430051e-01 -1.31531181e-02 8.53054464e-01
9.67935771e-02 -2.76563391e-02 -1.48070753e-01 -7.63942361e-01
-1.15099952e-01 2.81159788e-01 -5.87662637e-01 -2.48132452e-01
-1.13555372e+00 -4.68350768e-01 5.80491185e-01 8.80923092e-01
1.01004362e+00 -1.42788815e+00 -8.04067552e-01 2.75538415e-01
-3.27835649e-01 -1.16442668e+00 -7.70365834e-01 -1.19830370e-02
-1.03832579e+00 -8.93718779e-01 -9.81781542e-01 -1.11670458e+00
4.95701194e-01 1.16651334e-01 -2.96743587e-02 1.08893774e-02
-4.79067028e-01 8.54651406e-02 -6.46801293e-01 -8.96189511e-02
1.92874417e-01 -3.14809047e-02 -3.21639389e-01 3.88086528e-01
4.05259848e-01 -8.25630069e-01 -7.35256970e-01 3.05140406e-01
-8.31999481e-01 3.98860015e-02 9.35438871e-01 1.07131684e+00
4.78831530e-01 -1.33584365e-01 3.69659334e-01 6.44255280e-02
1.38128817e-01 1.83055669e-01 -6.26467705e-01 3.43019098e-01
-3.08877438e-01 1.81550652e-01 7.02700973e-01 -9.37443018e-01
-1.14831591e+00 1.14691742e-01 -2.52852261e-01 -6.79629624e-01
9.65236649e-02 4.06593770e-01 1.65245548e-01 -2.71541059e-01
1.15315787e-01 9.43918705e-01 1.28777802e-01 -6.60623729e-01
2.36565068e-01 5.02000272e-01 4.61565137e-01 -1.75945237e-01
6.23625576e-01 7.12498486e-01 -1.24924317e-01 -6.16705000e-01
-6.06012225e-01 -6.80449963e-01 -8.33233058e-01 -4.03273970e-01
8.59134316e-01 -8.07831347e-01 -8.95798922e-01 9.63922799e-01
-9.91998553e-01 -4.85408306e-01 -2.96517581e-01 1.11426461e+00
-5.20461679e-01 3.93727392e-01 -7.54273057e-01 -7.99784303e-01
-3.41427118e-01 -1.10032940e+00 8.91703248e-01 3.10650140e-01
3.67874503e-01 -7.05359519e-01 -9.24521387e-02 2.37911358e-01
5.52979946e-01 6.93807825e-02 5.86702168e-01 -9.13271904e-02
-7.90827215e-01 -3.30507100e-01 -3.98125142e-01 5.75335741e-01
-2.64321510e-02 -1.59741908e-01 -8.22510302e-01 -3.08194011e-01
-1.70888633e-01 -3.22428197e-01 1.41320777e+00 6.32027090e-01
1.17419088e+00 -1.80627868e-01 -1.61391720e-01 7.77747750e-01
1.40405178e+00 4.35025513e-01 8.55243385e-01 2.85718322e-01
6.41186714e-01 3.87027651e-01 7.57465780e-01 2.38080099e-01
4.35518563e-01 7.34177589e-01 2.87329815e-02 -6.71156496e-02
-5.92939317e-01 -1.68427199e-01 7.04995573e-01 1.04258800e+00
-3.71330500e-01 2.32431978e-01 -6.85462058e-01 7.03032136e-01
-2.09245729e+00 -1.20462286e+00 2.26047933e-01 1.98676491e+00
8.15296412e-01 -4.62937355e-02 1.35436624e-01 3.21112156e-01
4.65632617e-01 2.25868300e-01 -7.40303278e-01 8.45136773e-03
-2.69523352e-01 3.64044994e-01 5.29152155e-01 3.72637242e-01
-1.00964725e+00 1.20403540e+00 4.91545486e+00 9.60430026e-01
-1.75225747e+00 2.46566445e-01 9.43195261e-03 -3.49551290e-01
2.73643255e-01 6.04124852e-02 -7.67698944e-01 3.85719389e-01
6.02130830e-01 2.02229604e-01 3.74148071e-01 5.24832308e-01
6.58251882e-01 1.03838123e-01 -6.67748153e-01 1.17485332e+00
-1.20839432e-01 -8.73476148e-01 5.48293851e-02 -6.68339571e-03
3.04335445e-01 2.39455596e-01 -8.64824876e-02 5.52813709e-01
-4.20624167e-02 -7.96603084e-01 7.34735727e-01 8.92490566e-01
1.02088583e+00 -5.70406556e-01 6.00728095e-01 4.00905788e-01
-1.81034422e+00 -4.55353081e-01 -1.33949786e-01 1.21548198e-01
2.88706899e-01 1.38395429e-01 -2.93305248e-01 4.09629166e-01
7.82863379e-01 1.03644800e+00 -4.41179663e-01 1.42083025e+00
-2.99523413e-01 6.75763309e-01 -5.12016356e-01 -1.73146740e-01
5.09201586e-01 -5.57566993e-04 4.48521435e-01 1.22046006e+00
2.82610536e-01 2.52221704e-01 1.12147942e-01 4.16568071e-01
2.96114475e-01 7.77172074e-02 -2.38714397e-01 4.70723845e-02
1.11751016e-02 1.01462436e+00 -4.64843959e-01 -1.64351612e-01
-4.27342236e-01 1.15460300e+00 1.51576206e-01 5.70537925e-01
-8.10706198e-01 -4.97183293e-01 3.74704927e-01 -2.25332052e-01
6.15341127e-01 -5.49754322e-01 -9.79009718e-02 -1.32652330e+00
5.36651492e-01 -5.23998678e-01 5.22233844e-01 -6.15072966e-01
-1.12311435e+00 4.80150700e-01 -2.42323011e-01 -1.78434300e+00
-1.59783587e-01 -6.28400028e-01 -5.18912792e-01 9.64352012e-01
-1.97557163e+00 -1.76918733e+00 -4.36666012e-01 1.04448009e+00
6.62584603e-01 1.20961769e-02 3.60480726e-01 3.66821915e-01
-4.68400568e-01 7.85400629e-01 -1.19875908e-01 4.69104886e-01
5.63563526e-01 -7.73737431e-01 -8.10250789e-02 9.09216762e-01
-2.49020904e-01 4.12882417e-01 -6.36331365e-03 -5.64528883e-01
-1.06945038e+00 -1.20288193e+00 1.13267779e+00 5.63470982e-02
5.94637394e-01 -1.04658172e-01 -7.18324780e-01 4.72972661e-01
-3.44882071e-01 4.35638130e-01 1.38112664e-01 -3.78897995e-01
-5.38932443e-01 -4.62236702e-01 -9.12752986e-01 6.13164783e-01
1.39907229e+00 -7.39051938e-01 -6.19221330e-01 1.49652898e-01
6.61761105e-01 -2.65576065e-01 -7.30724394e-01 6.82465911e-01
1.16325462e+00 -5.44988453e-01 7.76849270e-01 -4.88475978e-01
1.71220422e-01 -5.08705258e-01 8.15979838e-02 -7.26240814e-01
-1.88874036e-01 -2.23843887e-01 -4.49542589e-02 9.56853271e-01
9.57420990e-02 -8.52597237e-01 5.41409016e-01 2.93011010e-01
-2.12148666e-01 -8.89047682e-01 -1.35366118e+00 -1.13348353e+00
-2.03253761e-01 -7.54843235e-01 2.35984623e-01 5.55809557e-01
-2.25808367e-01 -5.63796684e-02 -4.43967849e-01 2.43756976e-02
5.10032833e-01 1.70737222e-01 3.16286147e-01 -8.93553257e-01
-7.09340498e-02 -6.80169404e-01 -6.28875256e-01 -1.29004633e+00
1.15908772e-01 -9.51111674e-01 5.52267171e-02 -1.56722271e+00
-6.68959022e-02 -3.66662949e-01 -6.40656829e-01 7.87464976e-01
1.63546667e-01 1.81039408e-01 2.53757328e-01 3.29154998e-01
-4.13682371e-01 8.74329329e-01 1.48553503e+00 -2.33127713e-01
-4.29036528e-01 7.13614300e-02 2.78494895e-01 7.24967718e-01
5.90610862e-01 -2.67260939e-01 -2.29381755e-01 -3.22486579e-01
-5.33557177e-01 2.59432904e-02 7.04231799e-01 -1.13551855e+00
5.84443450e-01 -3.74982841e-02 3.49063218e-01 -9.24643934e-01
4.91329700e-01 -9.21327233e-01 -2.86184758e-01 7.35591769e-01
-3.12747419e-01 -2.85488635e-01 2.50746757e-01 4.60145980e-01
-5.17342567e-01 2.61710167e-01 1.03625906e+00 -5.75432740e-03
-1.11710584e+00 7.50591040e-01 -1.40434399e-01 -3.41211975e-01
8.71884346e-01 -3.44401658e-01 -3.20000537e-02 -1.27641872e-01
-9.18748736e-01 3.76106590e-01 -5.36213815e-02 6.73695028e-01
9.47156310e-01 -1.46412230e+00 -5.61120570e-01 2.03685984e-01
1.44427314e-01 -2.63530910e-01 6.17207110e-01 1.29041076e+00
-4.16501969e-01 6.43479466e-01 -2.34336600e-01 -7.01843798e-01
-1.30743957e+00 1.73091233e-01 7.19937861e-01 -5.29271126e-01
-1.12762666e+00 7.19844460e-01 6.86454447e-03 -1.42268479e-01
5.02445936e-01 -6.08446479e-01 -4.12053883e-01 -2.52952576e-02
5.61306655e-01 1.69735372e-01 -8.44614357e-02 -7.64824092e-01
-4.22375619e-01 1.02482128e+00 -9.31460559e-02 -4.22995478e-01
1.36428475e+00 3.40125933e-02 4.89708334e-02 5.31310081e-01
1.35680175e+00 -5.03459513e-01 -1.49593925e+00 -6.50877774e-01
-1.67227879e-01 -3.16555798e-01 1.57465369e-01 -9.73990262e-01
-1.25718701e+00 1.17024624e+00 1.07584178e+00 -6.89044774e-01
1.58211172e+00 -3.24051112e-01 1.32300663e+00 2.47251764e-01
4.82743800e-01 -1.09661722e+00 -1.65756661e-02 7.85430908e-01
1.00458765e+00 -9.70589042e-01 -4.58659410e-01 1.52396262e-02
-4.74341661e-01 1.28175640e+00 4.72693563e-01 -2.78618932e-01
9.39404130e-01 2.13350970e-02 1.82304338e-01 7.61288255e-02
-7.29644299e-01 -5.67767859e-01 5.87800682e-01 2.08800867e-01
1.57112539e-01 -2.09653407e-01 -8.16110492e-01 5.87581992e-01
1.52669959e-02 4.99726772e-01 -1.89869449e-01 9.93390024e-01
-2.19457626e-01 -1.09071636e+00 7.05250800e-02 4.28684592e-01
-1.56347722e-01 -3.17413472e-02 1.15940489e-01 7.98585832e-01
4.05257553e-01 6.19906545e-01 -1.51099876e-01 -5.92170358e-01
4.88973051e-01 2.44707420e-01 6.00538731e-01 -1.91989786e-03
-6.90131068e-01 3.29601109e-01 -1.45617127e-01 -8.52929950e-01
-5.82629383e-01 -7.11676538e-01 -1.71893477e+00 6.88095093e-02
-2.50296086e-01 -2.89149702e-01 6.82696819e-01 1.28720677e+00
9.71882865e-02 6.18713021e-01 5.94767392e-01 -6.63713455e-01
-5.09577632e-01 -1.09429812e+00 -3.84913743e-01 3.70316565e-01
5.12671113e-01 -6.99581563e-01 -2.36380890e-01 7.52187669e-02] | [9.20715618133545, -6.473455429077148] |
7115507e-e403-4a3a-9efe-53def9e1d161 | funcgnn-a-graph-neural-network-approach-to | 2007.13239 | null | https://arxiv.org/abs/2007.13239v3 | https://arxiv.org/pdf/2007.13239v3.pdf | funcGNN: A Graph Neural Network Approach to Program Similarity | Program similarity is a fundamental concept, central to the solution of software engineering tasks such as software plagiarism, clone identification, code refactoring and code search. Accurate similarity estimation between programs requires an in-depth understanding of their structure, semantics and flow. A control flow graph (CFG), is a graphical representation of a program which captures its logical control flow and hence its semantics. A common approach is to estimate program similarity by analysing CFGs using graph similarity measures, e.g. graph edit distance (GED). However, graph edit distance is an NP-hard problem and computationally expensive, making the application of graph similarity techniques to complex software programs impractical. This study intends to examine the effectiveness of graph neural networks to estimate program similarity, by analysing the associated control flow graphs. We introduce funcGNN, which is a graph neural network trained on labeled CFG pairs to predict the GED between unseen program pairs by utilizing an effective embedding vector. To our knowledge, this is the first time graph neural networks have been applied on labeled CFGs for estimating the similarity between high-level language programs. Results: We demonstrate the effectiveness of funcGNN to estimate the GED between programs and our experimental analysis demonstrates how it achieves a lower error rate (0.00194), with faster (23 times faster than the quickest traditional GED approximation method) and better scalability compared with the state of the art methods. funcGNN posses the inductive learning ability to infer program structure and generalise to unseen programs. The graph embedding of a program proposed by our methodology could be applied to several related software engineering problems (such as code plagiarism and clone identification) thus opening multiple research directions. | ['Avijit Roy', 'Karl Meinke', 'Aravind Nair'] | 2020-07-26 | null | null | null | null | ['code-search', 'code-search', 'graph-similarity'] | ['computer-code', 'computer-vision', 'graphs'] | [ 3.05481344e-01 1.56121895e-01 -2.09502473e-01 -1.24156721e-01
-1.13515005e-01 -6.25800252e-01 3.18476975e-01 9.08951104e-01
-2.45362595e-02 -9.80363563e-02 -4.74726669e-02 -8.84120643e-01
-1.14527173e-01 -1.01937151e+00 -8.02091479e-01 -7.44994730e-03
-4.20038581e-01 3.26665230e-02 2.75442690e-01 -2.07797468e-01
7.45236099e-01 3.20524156e-01 -1.80096209e+00 1.15951449e-01
8.88411641e-01 4.41909134e-01 3.08133692e-01 9.25888538e-01
-4.53798234e-01 9.21720028e-01 -5.25404274e-01 -5.34025788e-01
-1.58819824e-01 -4.67683911e-01 -1.07607520e+00 -3.92481685e-01
6.71208560e-01 -2.07367353e-02 -2.69793212e-01 1.56534946e+00
9.41846892e-02 -6.79312050e-02 4.73145992e-01 -1.53658998e+00
-9.99551892e-01 8.66631329e-01 -3.76788974e-01 3.80722016e-01
7.80295610e-01 -1.18681014e-01 1.14969242e+00 -5.30282795e-01
7.09158659e-01 9.95608449e-01 1.02102864e+00 3.72130543e-01
-1.36056662e+00 -2.92476952e-01 -3.82864207e-01 3.05122763e-01
-1.00266886e+00 1.47300273e-01 7.61188269e-01 -8.01973104e-01
1.27280021e+00 3.12022775e-01 4.34187651e-01 5.67775428e-01
4.28684115e-01 5.25834203e-01 7.53118098e-01 -7.05900311e-01
2.40677148e-01 1.79244906e-01 6.72601461e-01 1.22764480e+00
2.46975154e-01 6.39247373e-02 1.80547610e-02 -5.73301494e-01
2.56252229e-01 8.51282254e-02 -5.98161399e-01 -6.38092995e-01
-1.04410064e+00 1.06397974e+00 5.78309715e-01 8.05632055e-01
2.27903247e-01 4.45220113e-01 9.20863807e-01 9.10321236e-01
1.84343919e-01 8.28880370e-01 -9.60522965e-02 -1.72334194e-01
-8.39086354e-01 7.70807937e-02 1.21539450e+00 1.10276604e+00
1.06351364e+00 3.83681022e-02 2.20349222e-01 5.10380745e-01
3.29748839e-01 1.10604495e-01 7.29422092e-01 -4.99814391e-01
4.80176419e-01 1.32613277e+00 -7.19415128e-01 -1.66217268e+00
-1.61790177e-01 3.39866616e-02 -5.43549359e-01 3.82484317e-01
3.27376932e-01 4.68325227e-01 -2.81610966e-01 1.52339864e+00
6.35499507e-02 1.31657720e-01 -8.15200582e-02 2.60377496e-01
8.05869341e-01 4.36464071e-01 -3.30955267e-01 2.33990297e-01
1.26801527e+00 -7.64654756e-01 -3.75427425e-01 -2.05815926e-01
1.37494922e+00 -6.90293670e-01 1.09077954e+00 1.44564593e-02
-5.96103549e-01 -4.52323705e-01 -1.20795643e+00 1.66923751e-03
-7.03961611e-01 2.01248983e-03 6.75916314e-01 9.63912070e-01
-1.32039785e+00 1.05175805e+00 -7.52400100e-01 -6.63623154e-01
1.99011236e-01 4.66887891e-01 -6.76972151e-01 -8.91215503e-02
-7.38426685e-01 7.78053999e-01 7.06394732e-01 -2.25122854e-01
-5.26127636e-01 -7.80377805e-01 -1.36409616e+00 3.51298749e-01
2.74155796e-01 -4.18897569e-01 9.72299159e-01 -1.18243051e+00
-8.79822671e-01 9.53396380e-01 4.74414118e-02 -4.60609406e-01
-1.76350288e-02 3.20060432e-01 -4.04385626e-01 -2.30082184e-01
-5.01342081e-02 -3.63374054e-02 6.36157095e-01 -8.92599940e-01
-1.88338950e-01 -3.61306816e-01 3.60907942e-01 -5.59908032e-01
-5.12843132e-01 2.22447202e-01 1.17996626e-01 -5.28245091e-01
-1.72806934e-01 -9.32220399e-01 -2.68609240e-03 7.61625767e-02
-1.57507107e-01 -3.53057057e-01 8.14081192e-01 -8.91490757e-01
1.62477314e+00 -2.17153621e+00 2.45476410e-01 2.49010339e-01
7.41222739e-01 3.85053664e-01 -2.45892853e-01 8.28993201e-01
-4.37224716e-01 2.38110542e-01 -7.55330563e-01 3.27702165e-01
2.34201878e-01 -6.10703230e-03 -1.32634059e-01 6.30594194e-01
1.94470167e-01 9.91280437e-01 -1.12980258e+00 -2.62323350e-01
1.09581061e-01 1.66700050e-01 -7.92874813e-01 3.23207676e-01
-1.63390428e-01 -4.05485749e-01 -2.68853396e-01 4.67017889e-01
5.42714834e-01 -2.27214366e-01 3.52850676e-01 -9.79590267e-02
-3.42005640e-02 -4.08440502e-03 -9.82108712e-01 1.52933288e+00
-8.20325136e-01 1.16560292e+00 -3.88179690e-01 -1.39596450e+00
1.26498091e+00 5.99061809e-02 -8.57384726e-02 -4.11710709e-01
7.36270249e-02 2.69655943e-01 9.01382118e-02 -8.21728647e-01
5.67639947e-01 3.18397135e-01 -2.64384747e-01 7.92393208e-01
1.60527870e-01 9.78857465e-03 3.20400894e-01 3.43227237e-01
1.63750005e+00 -3.70356329e-02 7.51052618e-01 -5.60326397e-01
1.02184963e+00 -1.00477085e-01 -2.42636427e-02 6.98911905e-01
-3.52473222e-02 1.83666751e-01 1.03537786e+00 -6.02319002e-01
-1.03279614e+00 -7.28727579e-01 3.74565333e-01 9.18870032e-01
-3.75142507e-02 -8.43967855e-01 -1.05386436e+00 -9.20215607e-01
1.88530251e-01 8.16066444e-01 -6.78856313e-01 -6.64069653e-01
-8.29897761e-01 -2.68381655e-01 8.36217165e-01 6.15329027e-01
2.51900181e-02 -1.05334461e+00 -5.03212869e-01 1.85013756e-01
1.72257543e-01 -6.74145937e-01 -6.91825032e-01 1.03397723e-02
-9.29626644e-01 -1.49933934e+00 -1.22312471e-01 -1.09237874e+00
8.67684424e-01 -1.28867512e-03 1.51806235e+00 6.94198012e-01
-4.39554870e-01 5.47596633e-01 -3.35639954e-01 9.81444865e-02
-1.37011325e+00 7.93370605e-02 -3.49399894e-01 -4.05260205e-01
6.03403568e-01 -6.34980917e-01 -6.31126463e-02 1.63465932e-01
-1.25846219e+00 -4.37286586e-01 3.25099587e-01 8.68324280e-01
-1.56566277e-02 6.87313452e-02 1.23203993e-01 -1.19628787e+00
9.08463955e-01 -5.54650366e-01 -9.35225010e-01 4.16000664e-01
-7.88760245e-01 5.36704659e-01 9.81550217e-01 -3.45796317e-01
-5.89773059e-01 -1.23833582e-01 1.56462640e-02 -2.59080619e-01
-1.21566497e-01 7.48125136e-01 7.21664121e-03 -6.36146367e-01
8.99019718e-01 2.62117654e-01 1.51988566e-01 -2.68432170e-01
4.21209067e-01 6.97494268e-01 5.34679055e-01 -4.96034831e-01
8.62966776e-01 -3.96798775e-02 3.58142853e-02 -6.83582604e-01
6.69188127e-02 -5.46963930e-01 -7.09358454e-01 1.79839637e-02
7.06012547e-01 -2.61325002e-01 -7.86056280e-01 2.28354543e-01
-1.40052283e+00 4.86105010e-02 -1.80267971e-02 2.42211118e-01
-5.59132159e-01 1.08556342e+00 -5.94308794e-01 -4.41925347e-01
-3.49925846e-01 -1.27781153e+00 9.32527244e-01 -1.93856448e-01
-4.83923733e-01 -1.37679231e+00 4.72854912e-01 1.96520016e-01
3.61975044e-01 4.58858967e-01 1.58749747e+00 -7.48858035e-01
-6.58875644e-01 -3.77210557e-01 -1.51789770e-01 2.74814606e-01
1.84848309e-01 2.23885193e-01 -6.79544985e-01 -5.44161975e-01
2.74755266e-02 4.96614315e-02 4.66329396e-01 -1.86910123e-01
9.19726610e-01 -3.35665554e-01 -3.55552673e-01 4.66240168e-01
1.90791833e+00 1.58805922e-01 6.09234095e-01 3.62678319e-01
8.82644534e-01 6.02101088e-01 2.08278582e-01 1.97637990e-01
7.56118968e-02 4.71168458e-01 6.65300846e-01 3.34298700e-01
-6.63190559e-02 -2.28690267e-01 5.56123972e-01 1.18234265e+00
2.13343441e-01 -1.55683875e-01 -1.20074511e+00 5.73024273e-01
-1.62826514e+00 -8.55330169e-01 -4.84160841e-01 2.35229492e+00
4.16890442e-01 -1.16562530e-01 -1.54975817e-01 3.26496363e-01
9.94879365e-01 8.22504982e-02 -1.12881996e-01 -1.09274888e+00
4.82675880e-01 3.12008888e-01 3.41475517e-01 4.21103090e-01
-7.40841925e-01 5.73539793e-01 5.11983442e+00 5.59993446e-01
-7.85510123e-01 2.14684270e-02 -1.32044584e-01 7.61034846e-01
-4.30066168e-01 3.63291144e-01 -3.04481089e-01 4.36110616e-01
1.18520021e+00 -6.17435932e-01 7.71242499e-01 1.14337647e+00
-4.12956774e-01 1.47087827e-01 -1.61249888e+00 7.24336565e-01
4.81579483e-01 -1.29047716e+00 -2.34391138e-01 -1.15767248e-01
3.97499204e-01 -5.46955736e-03 -4.05024409e-01 5.35791159e-01
1.42973661e-01 -9.23449755e-01 3.07880968e-01 3.24901879e-01
4.62362587e-01 -7.98446417e-01 8.97108614e-01 1.90948293e-01
-1.49768031e+00 -1.52190685e-01 -5.12285292e-01 4.46773283e-02
-3.43104869e-01 3.45970541e-01 -9.82124805e-01 6.41072929e-01
5.07598460e-01 8.32056046e-01 -1.06622279e+00 9.48768973e-01
3.46634947e-02 2.69129664e-01 2.24369109e-01 -4.12134439e-01
1.31910667e-01 -2.17460528e-01 5.44089854e-01 1.28824246e+00
5.21818817e-01 -6.48493469e-01 -2.99731735e-02 1.54528093e+00
-9.56641212e-02 1.63800105e-01 -1.24636519e+00 -4.54761595e-01
2.07464501e-01 1.04596770e+00 -7.77079582e-01 -2.08895266e-01
-6.84533477e-01 9.04659390e-01 6.04524970e-01 -7.20953271e-02
-7.45507121e-01 -9.88320291e-01 4.97807831e-01 -1.87216457e-02
3.19897652e-01 -1.42305139e-02 1.83883995e-01 -9.92445648e-01
2.82460511e-01 -1.06277585e+00 3.67892116e-01 -5.94339430e-01
-1.08418489e+00 8.32288563e-01 -6.21698126e-02 -1.20638061e+00
-4.71426129e-01 -8.30017447e-01 -8.81844044e-01 7.52050221e-01
-1.07577550e+00 -7.36218333e-01 -4.43371058e-01 3.09564292e-01
2.37270519e-01 -3.66510034e-01 9.03423011e-01 3.33206624e-01
-3.27926278e-01 7.11419463e-01 1.28604949e-01 2.66763598e-01
2.38393068e-01 -1.57062387e+00 9.31284249e-01 9.42143977e-01
1.04502246e-01 9.33812559e-01 7.14581370e-01 -5.44730783e-01
-1.85376823e+00 -1.12172377e+00 1.05887628e+00 -5.21909416e-01
1.12019122e+00 -3.54454935e-01 -1.33406436e+00 6.97081149e-01
1.52278230e-01 1.73830464e-01 6.25248611e-01 -1.70649767e-01
-7.31091261e-01 1.67939991e-01 -1.00126266e+00 4.14123446e-01
9.86385703e-01 -1.18269408e+00 -7.22714186e-01 2.28860110e-01
7.81738639e-01 -2.16546357e-01 -1.27608466e+00 5.01891077e-02
3.62772733e-01 -1.26387262e+00 7.03715861e-01 -5.92535496e-01
6.49617910e-01 -4.43469346e-01 -1.56809300e-01 -1.21441734e+00
-1.60229176e-01 -4.31426525e-01 -1.99337274e-01 1.30476630e+00
1.56331226e-01 -7.21065819e-01 6.93063438e-01 3.17565829e-01
-3.92457917e-02 -3.81745487e-01 -5.24100065e-01 -9.45493996e-01
-3.37369740e-02 -3.84693950e-01 7.31487989e-01 1.39553022e+00
3.62567693e-01 1.24301568e-01 1.43671498e-01 2.31269240e-01
5.05514562e-01 3.48149925e-01 7.33565867e-01 -1.33357680e+00
-5.88507771e-01 -7.47711420e-01 -1.25137186e+00 -3.92465383e-01
7.90184617e-01 -1.55501997e+00 -2.24162519e-01 -1.08667910e+00
4.89950553e-02 5.55953644e-02 1.13517120e-01 2.39713728e-01
-1.21440686e-01 -3.13527495e-01 1.73617080e-01 -1.55281872e-01
-1.45133585e-01 2.37070173e-01 6.04884624e-01 -6.19307458e-01
2.04602834e-02 -4.03800532e-02 -4.14142013e-01 5.70010841e-01
5.90046644e-01 -8.51252675e-01 -5.66996872e-01 -2.72275031e-01
5.52910864e-01 1.95761651e-01 3.91796142e-01 -1.00216520e+00
3.00744623e-01 2.81042278e-01 -5.08320272e-01 -7.73405358e-02
-5.34915447e-01 -9.37627673e-01 4.41095620e-01 9.96242762e-01
-3.28944892e-01 6.25906467e-01 2.73944676e-01 7.89667845e-01
-4.19723392e-01 -1.01764941e+00 5.07209778e-01 -1.57903627e-01
-9.94895816e-01 4.13147062e-02 -3.93909544e-01 3.78930643e-02
1.02821863e+00 -4.83889371e-01 -3.20676923e-01 1.07319944e-01
-2.74421990e-01 -3.65059048e-01 7.88237929e-01 6.56474173e-01
6.12531662e-01 -1.35957837e+00 -3.56112689e-01 3.16068351e-01
5.72285414e-01 -6.42358601e-01 -6.93748072e-02 6.69864774e-01
-1.00148571e+00 1.75756916e-01 -2.45492846e-01 -5.87722301e-01
-1.74462140e+00 1.00560403e+00 3.26232463e-01 -3.49656045e-01
-5.34730911e-01 3.50993723e-01 -1.02393059e-02 -1.08080196e+00
-7.63054192e-02 -6.61806226e-01 -3.09505224e-01 -2.80879915e-01
5.05050182e-01 6.60269380e-01 2.94934213e-01 -5.08448124e-01
-3.53616863e-01 7.23735869e-01 2.71859095e-02 7.38934159e-01
1.12118411e+00 3.63408476e-01 -9.89318430e-01 2.36980140e-01
1.90169668e+00 -1.24418542e-01 -2.68098325e-01 -6.98838383e-02
6.25333071e-01 -6.51636600e-01 -1.49009004e-03 -1.40419111e-01
-1.00269353e+00 9.17148709e-01 7.19359457e-01 5.94731510e-01
8.93089712e-01 -7.25307688e-02 5.19788265e-01 6.77310407e-01
3.65417480e-01 -4.70674396e-01 -1.45303933e-02 4.96493191e-01
6.47611320e-01 -1.07160783e+00 -2.29446083e-01 -4.28917617e-01
1.86905004e-02 1.69783592e+00 5.15962958e-01 -2.32330263e-01
4.49037492e-01 3.48334908e-01 -3.94199431e-01 -4.85269517e-01
-3.60881448e-01 1.32169634e-01 4.31464285e-01 6.69360042e-01
7.07904279e-01 -1.02959119e-01 -2.74433792e-01 1.37712304e-02
-8.86650831e-02 -7.54133463e-02 9.58908379e-01 1.07249153e+00
-2.95260817e-01 -1.28412819e+00 -2.92709708e-01 6.03817821e-01
-1.48726121e-01 -3.23341697e-01 -5.27165115e-01 9.24591482e-01
-1.47387549e-01 6.92072630e-01 -5.74458577e-02 -4.61767197e-01
3.73409092e-01 -1.28906101e-01 4.20801699e-01 -7.31670618e-01
-9.91560102e-01 -1.06481171e+00 -1.16282284e-01 -7.16912627e-01
-3.61496091e-01 -3.75064671e-01 -1.01135540e+00 -5.36532938e-01
-5.96244991e-01 8.14024881e-02 5.90259135e-01 6.01691425e-01
4.42232341e-01 5.59508741e-01 5.44027448e-01 -4.05333668e-01
-5.93036115e-01 -4.82841462e-01 -4.40189421e-01 6.25785708e-01
3.12282652e-01 -2.85523117e-01 -5.48290551e-01 8.52254927e-02] | [7.254534721374512, 7.803074836730957] |
7ed83ad7-eb24-41e4-9234-ba59b8cf2127 | unsupervised-word-segmentation-from-speech | 1806.06734 | null | http://arxiv.org/abs/1806.06734v1 | http://arxiv.org/pdf/1806.06734v1.pdf | Unsupervised Word Segmentation from Speech with Attention | We present a first attempt to perform attentional word segmentation directly
from the speech signal, with the final goal to automatically identify lexical
units in a low-resource, unwritten language (UL). Our methodology assumes a
pairing between recordings in the UL with translations in a well-resourced
language. It uses Acoustic Unit Discovery (AUD) to convert speech into a
sequence of pseudo-phones that is segmented using neural soft-alignments
produced by a neural machine translation model. Evaluation uses an actual Bantu
UL, Mboshi; comparisons to monolingual and bilingual baselines illustrate the
potential of attentional word segmentation for language documentation. | ['François Yvon', 'Marcely Zanon-Boito', 'Laurent Besacier', 'Alexandre Berard', 'Lucas Ondel', 'Aline Villavicencio', 'Pierre Godard'] | 2018-06-18 | null | null | null | null | ['acoustic-unit-discovery'] | ['speech'] | [ 6.89032376e-01 3.59789342e-01 -1.66356951e-01 -1.91269338e-01
-1.55890739e+00 -7.69840062e-01 3.96766663e-01 -2.85924345e-01
-5.98408639e-01 7.70136893e-01 5.42608738e-01 -8.19939613e-01
5.25796294e-01 -1.70561507e-01 -6.42200649e-01 -2.44622096e-01
3.39108497e-01 8.46047699e-01 -2.69645810e-01 -1.04178652e-01
2.04721376e-01 1.38105273e-01 -8.14488292e-01 4.12025869e-01
7.11031437e-01 1.83285981e-01 8.53982449e-01 7.10360885e-01
-1.81076288e-01 3.16649497e-01 -7.83324003e-01 -9.28302854e-03
3.14147681e-01 -8.41581583e-01 -1.22599792e+00 -2.92542838e-02
6.36214733e-01 9.16544721e-02 1.33171186e-01 1.15251076e+00
3.92955482e-01 2.07235649e-01 6.51539028e-01 -1.70299068e-01
-9.88292694e-01 1.45036554e+00 4.77844663e-03 8.89549971e-01
4.69147921e-01 9.86976922e-02 1.49872339e+00 -1.24346435e+00
7.52037048e-01 1.40917146e+00 6.83212280e-01 4.94461179e-01
-1.68575966e+00 -6.05111957e-01 -8.16430151e-02 -6.30983636e-02
-1.46432710e+00 -8.92682254e-01 4.53037143e-01 -5.47322929e-01
1.64608836e+00 2.14007244e-01 4.81850415e-01 1.34135544e+00
-1.21340342e-02 6.44367874e-01 1.16751778e+00 -1.14757752e+00
4.52125669e-02 -6.94738850e-02 2.11469933e-01 2.38962471e-01
-1.56237826e-01 5.75869391e-03 -8.95588100e-01 1.18583530e-01
7.64303684e-01 -1.00294542e+00 -2.58495450e-01 5.06138325e-01
-1.59396780e+00 8.85451794e-01 -2.09270179e-01 7.86586940e-01
-6.85434699e-01 -1.76363319e-01 2.82879412e-01 4.56298113e-01
4.42790955e-01 9.75467026e-01 -5.93066871e-01 -3.54321212e-01
-1.26716423e+00 -5.23629069e-01 6.55355275e-01 1.09647942e+00
5.57651043e-01 7.44200945e-01 -1.52465194e-01 1.06577873e+00
4.02321011e-01 6.62961304e-01 1.18495381e+00 -7.10197568e-01
5.62429786e-01 -1.01761729e-01 -3.40490311e-01 -1.14325948e-01
9.63859409e-02 -9.06962380e-02 1.39423415e-01 -3.34109426e-01
2.08861634e-01 -2.60096252e-01 -1.18875313e+00 1.79464900e+00
-1.93273962e-01 -2.59878337e-01 2.87896723e-01 5.88936269e-01
4.72461671e-01 1.01273263e+00 1.82731301e-01 -6.38965368e-01
1.44364762e+00 -9.30461466e-01 -8.61359119e-01 -8.33850861e-01
4.15364176e-01 -1.16404819e+00 1.71206832e+00 3.19644630e-01
-1.35960281e+00 -6.27990842e-01 -1.08952498e+00 -2.55129695e-01
-3.24681431e-01 1.84737310e-01 7.30966404e-02 7.68159270e-01
-1.02927458e+00 1.18518524e-01 -7.12817967e-01 -5.04581630e-01
-9.59220678e-02 3.71595383e-01 -1.62103832e-01 3.36677223e-01
-1.40913904e+00 1.05429816e+00 6.71294928e-01 -9.99542624e-02
-1.03940868e+00 -3.49485368e-01 -8.86204779e-01 -1.92284331e-01
7.73868635e-02 -9.91375744e-02 1.55041385e+00 -1.36990535e+00
-1.50562847e+00 1.21354198e+00 -1.33128196e-01 -6.04571283e-01
-1.05609015e-01 -3.61230969e-01 -5.24515450e-01 6.95451116e-03
4.95184660e-01 7.18641639e-01 8.56832325e-01 -9.49264884e-01
-6.38059735e-01 -1.26292661e-01 -6.10199690e-01 3.97021890e-01
-1.44471496e-01 5.89795411e-01 -2.55204350e-01 -1.18022490e+00
-8.57102945e-02 -1.00972521e+00 1.84433639e-01 -1.26114750e+00
-4.45311397e-01 -2.11956665e-01 2.35069633e-01 -1.50082290e+00
1.42837787e+00 -2.10555673e+00 4.25170958e-01 1.53518006e-01
-4.29034799e-01 1.93545148e-01 -2.26006389e-01 3.11794877e-01
-2.45696932e-01 2.25089118e-01 -4.78198320e-01 -4.63541299e-01
-8.57206956e-02 5.21699309e-01 -5.19274354e-01 3.44992936e-01
1.44089684e-01 9.93880153e-01 -7.26170123e-01 -4.47469771e-01
-9.72146690e-02 2.33495116e-01 -2.14272782e-01 6.05161786e-02
-2.34141603e-01 4.73745793e-01 3.17365199e-01 6.99788272e-01
-2.95447558e-01 6.99474514e-01 3.96880537e-01 1.32000878e-01
-6.16733611e-01 1.25411487e+00 -6.05045199e-01 1.92829609e+00
-6.72327578e-01 9.97478604e-01 2.12491587e-01 -8.51447582e-01
7.09991932e-01 6.13773704e-01 -1.97776616e-01 -6.56606257e-01
3.20505530e-01 6.85549974e-01 6.12974882e-01 -2.94356406e-01
6.59250557e-01 -3.83588344e-01 -4.00840700e-01 5.43198466e-01
5.45882225e-01 -2.99459636e-01 4.35897484e-02 -3.22096944e-01
7.94564128e-01 6.86303526e-02 5.93691111e-01 -7.40166068e-01
2.61757612e-01 4.16940004e-01 4.63009447e-01 6.44335151e-01
-7.51550123e-02 6.38706446e-01 -1.40103087e-01 1.49117395e-01
-1.27707541e+00 -1.22724092e+00 -2.04496473e-01 1.50688946e+00
-6.08473659e-01 -1.91561177e-01 -1.09942579e+00 -3.67622614e-01
-5.22710264e-01 1.47100902e+00 -3.29261005e-01 1.20099746e-01
-1.19830239e+00 -3.08782250e-01 8.50714386e-01 4.04734462e-01
-3.52626652e-01 -1.70261610e+00 -1.24191239e-01 4.67620075e-01
-5.75875163e-01 -1.15491641e+00 -1.00279689e+00 6.32978499e-01
-5.16452014e-01 -4.10414964e-01 -6.65443182e-01 -1.40350330e+00
2.37573296e-01 -1.34437382e-01 1.21308184e+00 -3.93366188e-01
-2.04892173e-01 3.22092891e-01 -2.03245118e-01 -4.60819066e-01
-1.07731211e+00 5.30316114e-01 5.19246817e-01 -2.47140810e-01
9.60410416e-01 -2.54099429e-01 1.43576950e-01 -4.44300361e-02
-5.30273080e-01 -2.24028543e-01 8.82516503e-01 7.83094704e-01
8.94253433e-01 -6.78160667e-01 6.65828228e-01 -7.88792491e-01
7.89655030e-01 -4.54411030e-01 -3.60803515e-01 5.67318387e-02
-5.23195565e-01 -8.68851133e-03 4.61291283e-01 -6.43627822e-01
-9.35412526e-01 8.20532292e-02 -2.29418516e-01 -1.68040588e-01
-3.52943510e-01 5.35906613e-01 -4.05549437e-01 3.58296663e-01
7.40231872e-01 3.38137060e-01 -1.93272129e-01 -6.54075384e-01
6.89426601e-01 1.19867814e+00 1.16215122e+00 -6.12030029e-01
5.23329794e-01 -2.52091020e-01 -9.21917677e-01 -1.33194959e+00
-5.46032012e-01 -3.91269147e-01 -1.12542653e+00 -3.84344608e-02
1.11564076e+00 -8.13671768e-01 2.20610395e-01 -1.27123117e-01
-1.50755572e+00 -5.57433963e-01 -5.57273269e-01 9.14777994e-01
-6.65202320e-01 2.62783468e-01 -7.29581058e-01 -6.41245365e-01
-5.21566570e-01 -1.06442428e+00 8.88226748e-01 -2.22764820e-01
-1.02119172e+00 -8.83712888e-01 4.42042917e-01 5.00277996e-01
1.85311377e-01 -5.14966488e-01 1.03733814e+00 -1.24633253e+00
-2.05441758e-01 1.37008011e-01 2.44177654e-01 8.24544609e-01
2.38092020e-01 -2.44088486e-01 -9.31135416e-01 -1.54002726e-01
1.34840295e-01 -1.42262086e-01 4.38741952e-01 4.05196220e-01
1.07675456e-01 -4.27537173e-01 5.62707223e-02 1.89759478e-01
9.38658357e-01 7.11020947e-01 3.07292581e-01 2.06640407e-01
6.31190598e-01 4.79949772e-01 4.53330517e-01 -3.74374628e-01
-4.31370214e-02 4.60759968e-01 -2.42582530e-01 -1.98003929e-02
-6.07610106e-01 -1.58546776e-01 1.02089965e+00 1.93767214e+00
2.15443954e-01 -6.96429610e-02 -1.34255075e+00 1.22514272e+00
-1.31531084e+00 -5.38087726e-01 -5.35887629e-02 2.17946744e+00
1.36578953e+00 3.33379000e-01 4.49429220e-03 -2.69159436e-01
9.11430478e-01 -4.40871604e-02 -2.18150213e-01 -8.68040442e-01
-2.80779094e-01 6.50631726e-01 5.12898505e-01 1.10580778e+00
-8.92675877e-01 1.61514139e+00 7.62787056e+00 1.00407553e+00
-9.22835052e-01 6.20043695e-01 3.73004556e-01 -1.99665874e-01
-2.07546338e-01 1.57580301e-02 -9.18041468e-01 3.41002017e-01
1.82151699e+00 -1.49320692e-01 8.70580912e-01 6.44727588e-01
2.44940102e-01 1.75937161e-01 -1.37109971e+00 6.52335405e-01
4.99470443e-01 -1.02180791e+00 1.48233801e-01 3.12897526e-02
6.26394749e-01 5.85466087e-01 -5.10725677e-02 3.76985013e-01
4.17891860e-01 -1.10929120e+00 1.10731637e+00 1.29837528e-01
1.23213661e+00 -6.80945694e-01 3.95799726e-01 2.26085857e-01
-9.79059517e-01 3.88619125e-01 -2.50565767e-01 4.08906117e-02
3.73193741e-01 -1.68160692e-01 -1.35190642e+00 3.41231041e-02
3.06961894e-01 2.12416857e-01 -3.05943370e-01 4.54975367e-01
-3.97866070e-01 1.38768172e+00 -2.49681354e-01 5.57920672e-02
4.45532858e-01 -2.96035707e-01 1.05513418e+00 1.85360789e+00
2.88123906e-01 -2.28291050e-01 2.09312230e-01 8.15823972e-01
-1.45302534e-01 6.21847391e-01 -6.71999514e-01 -5.53364277e-01
6.16384864e-01 8.00282061e-01 -8.85958135e-01 -3.20265859e-01
-4.39993352e-01 1.13947093e+00 1.93432346e-01 1.85888425e-01
-4.02355880e-01 -4.82163191e-01 6.44897342e-01 -1.64705217e-01
1.56580687e-01 -6.34104490e-01 -5.43824971e-01 -8.12881052e-01
-3.00403237e-01 -1.21444535e+00 7.53784478e-02 -6.76333666e-01
-1.26361585e+00 1.23539317e+00 -4.54512239e-02 -8.09769213e-01
-6.94950640e-01 -6.70233190e-01 -2.47743726e-01 1.50046408e+00
-8.49061370e-01 -1.02913558e+00 5.44716001e-01 1.52429000e-01
1.48576164e+00 -6.10939920e-01 1.12490308e+00 3.60335767e-01
-5.33840716e-01 3.64309311e-01 -1.66852042e-01 3.55153114e-01
6.45337462e-01 -1.53450716e+00 9.58569109e-01 1.30914474e+00
9.36390817e-01 8.79276693e-01 6.93504035e-01 -9.00361538e-01
-8.28313351e-01 -7.53550947e-01 1.25094748e+00 -7.04775274e-01
9.43388939e-01 -7.65111089e-01 -8.47909212e-01 1.30343962e+00
9.42276120e-01 -8.19278836e-01 1.05804551e+00 -5.82733974e-02
3.82126458e-02 5.18876910e-01 -5.40272534e-01 8.74088943e-01
9.33512270e-01 -1.19436598e+00 -1.51620519e+00 3.79286021e-01
1.11516976e+00 -1.74906105e-01 -6.66886330e-01 -1.73062459e-01
3.30653310e-01 -1.87905300e-02 7.71853626e-01 -5.99939525e-01
-2.06268094e-02 -1.71280816e-01 -4.18619066e-01 -1.56103814e+00
-2.80325145e-01 -1.15172625e+00 4.83553439e-01 1.38933432e+00
1.00279832e+00 -3.12905192e-01 4.93177399e-02 1.94905788e-01
-6.37214422e-01 8.46300051e-02 -1.32724869e+00 -8.61892283e-01
2.78419077e-01 -7.49166667e-01 1.39205560e-01 1.07216990e+00
1.55524313e-01 1.17177355e+00 -8.84783641e-02 7.46678859e-02
2.33527899e-01 -2.96655416e-01 -6.14318950e-03 -1.09883404e+00
-3.51606399e-01 -5.07892668e-01 3.07893697e-02 -9.79524672e-01
5.27210534e-01 -1.27319574e+00 7.09774077e-01 -1.28584027e+00
-4.19404209e-01 -1.48704663e-01 -5.36055565e-02 7.39423096e-01
1.35854661e-01 5.55385172e-01 6.20444044e-02 4.96033847e-01
3.35241482e-02 2.56361693e-01 4.36986774e-01 -9.99785364e-02
-4.47511047e-01 -3.31352919e-01 -4.13611084e-01 6.13525569e-01
7.43732870e-01 -7.46103466e-01 -1.10777661e-01 -6.19042993e-01
-3.46741453e-02 -1.45538181e-01 -4.57874537e-01 -7.57130444e-01
-2.85113044e-03 1.16258198e-02 -1.30819678e-01 -5.23919702e-01
1.68477356e-01 -4.81788516e-01 -9.17635262e-02 1.44834071e-01
-3.82424802e-01 4.38456595e-01 4.96081769e-01 1.33869369e-02
-2.77385831e-01 -8.11357319e-01 7.07326114e-01 -4.08278972e-01
-7.02453673e-01 -4.83112514e-01 -1.14478612e+00 2.90402204e-01
7.99291670e-01 -3.04575413e-01 9.15907174e-02 4.95228618e-02
-1.02282059e+00 -3.52035165e-01 2.14659780e-01 6.15997136e-01
1.55748174e-01 -1.28224671e+00 -8.63699496e-01 4.63246495e-01
-2.04968914e-01 -3.94792140e-01 -6.38029635e-01 6.75691187e-01
-5.95200837e-01 8.21201622e-01 -1.83025211e-01 -3.39578182e-01
-1.15776372e+00 3.76095057e-01 1.62573427e-01 1.95732221e-01
-5.23261309e-01 9.85225141e-01 -5.95873334e-02 -4.42153394e-01
1.12003237e-01 -3.98007363e-01 -3.35985683e-02 3.35726410e-01
2.94744402e-01 1.47192791e-01 2.03580678e-01 -1.20720077e+00
-2.67305285e-01 5.60284965e-03 3.44310328e-02 -8.55493724e-01
9.55605865e-01 -4.50595975e-01 -2.10187063e-01 1.19341457e+00
8.00735533e-01 6.97627246e-01 -5.52829146e-01 -2.09271580e-01
4.86472011e-01 1.18668295e-01 1.67490244e-01 -6.47413552e-01
-3.43162477e-01 8.10323238e-01 2.76184618e-01 3.48947316e-01
6.26927078e-01 9.90820825e-02 9.17837739e-01 3.60146105e-01
6.18544370e-02 -1.68406165e+00 -3.87986898e-01 9.50804353e-01
1.01896608e+00 -8.43258321e-01 -5.38076639e-01 -2.23440051e-01
-6.77608788e-01 1.19015551e+00 2.24062726e-01 -7.50962943e-02
3.39690924e-01 3.09190750e-01 4.70342070e-01 -1.18447274e-01
-2.87827522e-01 -3.94689560e-01 3.72619927e-01 5.30166626e-01
6.62943661e-01 2.32623145e-01 -4.90507334e-01 4.90745425e-01
-8.79714251e-01 -7.22993195e-01 5.59863269e-01 6.60441577e-01
-4.52099681e-01 -1.19783080e+00 -6.82133913e-01 1.67531475e-01
-8.31439316e-01 -9.68935370e-01 -7.04304755e-01 7.83304989e-01
6.36488721e-02 8.59184563e-01 4.68351871e-01 -5.54150604e-02
9.63396803e-02 8.88055205e-01 9.61113125e-02 -1.44649470e+00
-8.06836426e-01 7.74384975e-01 4.13310796e-01 -1.13617204e-01
-2.98172593e-01 -1.21290660e+00 -1.13289237e+00 5.37597537e-01
-2.06287354e-01 2.97771215e-01 7.12113261e-01 1.08300376e+00
-6.59453645e-02 6.78812921e-01 9.51106921e-02 -9.44968164e-01
-2.62929529e-01 -1.53186309e+00 -4.00755495e-01 -1.12648390e-01
1.70991629e-01 -1.62503019e-01 -3.84599119e-01 5.62690496e-01] | [14.366455078125, 6.97222375869751] |
2733e118-6fb3-4ea7-a271-3749bedfa77e | provably-efficient-adversarial-imitation | 2306.06563 | null | https://arxiv.org/abs/2306.06563v1 | https://arxiv.org/pdf/2306.06563v1.pdf | Provably Efficient Adversarial Imitation Learning with Unknown Transitions | Imitation learning (IL) has proven to be an effective method for learning good policies from expert demonstrations. Adversarial imitation learning (AIL), a subset of IL methods, is particularly promising, but its theoretical foundation in the presence of unknown transitions has yet to be fully developed. This paper explores the theoretical underpinnings of AIL in this context, where the stochastic and uncertain nature of environment transitions presents a challenge. We examine the expert sample complexity and interaction complexity required to recover good policies. To this end, we establish a framework connecting reward-free exploration and AIL, and propose an algorithm, MB-TAIL, that achieves the minimax optimal expert sample complexity of $\widetilde{O} (H^{3/2} |S|/\varepsilon)$ and interaction complexity of $\widetilde{O} (H^{3} |S|^2 |A|/\varepsilon^2)$. Here, $H$ represents the planning horizon, $|S|$ is the state space size, $|A|$ is the action space size, and $\varepsilon$ is the desired imitation gap. MB-TAIL is the first algorithm to achieve this level of expert sample complexity in the unknown transition setting and improves upon the interaction complexity of the best-known algorithm, OAL, by $O(H)$. Additionally, we demonstrate the generalization ability of MB-TAIL by extending it to the function approximation setting and proving that it can achieve expert sample and interaction complexity independent of $|S|$ | ['Zhi-Quan Luo', 'Yang Yu', 'Ziniu Li', 'Tian Xu'] | 2023-06-11 | null | null | null | null | ['imitation-learning'] | ['methodology'] | [ 1.99688986e-01 5.53105116e-01 -3.80828559e-01 3.95072460e-01
-1.08469963e+00 -8.12022448e-01 3.37906122e-01 4.22321595e-02
-8.35070193e-01 9.45396423e-01 -5.58276415e-01 -7.43885994e-01
-6.12707376e-01 -6.08368039e-01 -1.13309288e+00 -7.59213626e-01
-7.25570560e-01 3.69443208e-01 1.22313559e-01 -1.60632074e-01
2.82973975e-01 3.57685745e-01 -1.19027531e+00 -6.93501532e-01
9.96538222e-01 1.21812689e+00 -5.65887764e-02 8.84267271e-01
2.44335964e-01 7.18650460e-01 -5.34955561e-01 4.50414345e-02
5.90492249e-01 -7.47658074e-01 -6.92195356e-01 2.11556293e-02
1.25262849e-02 -5.16060531e-01 -4.99410957e-01 1.25449669e+00
5.00355244e-01 5.11866808e-01 3.98572505e-01 -1.40754843e+00
4.08604555e-02 5.47462583e-01 -2.06081435e-01 1.11516520e-01
2.72311866e-01 6.21378541e-01 7.33109474e-01 -2.42920533e-01
6.00739479e-01 1.10354674e+00 3.29593599e-01 6.05027676e-01
-1.04520345e+00 -8.23653162e-01 5.18921256e-01 -1.09132513e-01
-1.00444555e+00 -7.96230137e-02 3.45379025e-01 -4.40680891e-01
9.39337015e-01 7.75594711e-02 6.61455035e-01 9.06452656e-01
2.46972755e-01 9.15460825e-01 1.37017739e+00 -3.92775267e-01
6.59304976e-01 -4.63296920e-02 -3.81419450e-01 8.31505299e-01
1.00572892e-01 6.42149150e-01 -2.32947186e-01 -3.99033800e-02
9.02478278e-01 -7.88825080e-02 -2.90694296e-01 -2.63268769e-01
-9.18365479e-01 1.03284919e+00 2.12358236e-01 -2.16337413e-01
-2.33416110e-01 6.62205935e-01 1.24295667e-01 7.97868490e-01
-6.57183155e-02 7.93304145e-01 -2.99839914e-01 -6.53041303e-01
-5.97915530e-01 4.87700731e-01 8.23472261e-01 1.17580509e+00
5.18068075e-01 4.07449901e-01 5.83583536e-03 6.73390850e-02
-6.92560747e-02 1.00340497e+00 -2.30794195e-02 -1.52908254e+00
7.21507192e-01 3.69756907e-01 7.94478714e-01 -2.47268647e-01
-2.17982963e-01 -4.50762480e-01 -3.75237793e-01 7.09386706e-01
6.58653259e-01 -6.81783974e-01 -7.05679476e-01 2.08918643e+00
2.04637021e-01 -1.47987127e-01 8.87271687e-02 5.16900897e-01
-3.10439616e-01 6.05278790e-01 -4.04708594e-01 -6.25810921e-01
7.72643328e-01 -8.59399557e-01 -2.78463930e-01 -4.86788958e-01
3.99794102e-01 -2.02939376e-01 1.29723287e+00 4.62150991e-01
-1.42445695e+00 -5.93066216e-02 -1.05151725e+00 7.13476062e-01
1.44671118e-02 -3.26350480e-01 4.41925406e-01 5.86717904e-01
-8.26556861e-01 8.25900912e-01 -1.02333474e+00 2.57081300e-01
3.79297078e-01 6.09702229e-01 -5.02825640e-02 -1.16876557e-01
-1.05032015e+00 8.86671484e-01 2.04704165e-01 1.10475468e-02
-1.50641143e+00 -4.56192285e-01 -7.84327269e-01 -7.60047361e-02
1.19578207e+00 -2.23654076e-01 1.47926605e+00 -4.57228541e-01
-1.83373725e+00 -5.39653189e-02 1.39871299e-01 -8.71339917e-01
8.56256783e-01 -1.24940194e-01 1.03754312e-01 3.75052661e-01
4.93909083e-02 2.74999708e-01 8.43614697e-01 -8.25445592e-01
-7.35105753e-01 -3.31108898e-01 6.33805811e-01 2.09619567e-01
-8.48489702e-02 -2.95574605e-01 1.65437479e-02 -2.33507484e-01
-2.13518173e-01 -1.35705030e+00 -7.26151884e-01 -9.00799856e-02
-1.45742506e-01 -2.41990089e-02 3.86412144e-01 -3.12320262e-01
1.08290958e+00 -1.98668706e+00 2.07032531e-01 4.89402622e-01
1.39560446e-01 2.65967548e-01 6.87889159e-02 6.63101315e-01
6.29232764e-01 5.81816360e-02 -4.89364356e-01 -8.61094296e-02
4.11983967e-01 2.92139560e-01 -4.00134116e-01 4.88595545e-01
-1.41369775e-01 7.15564847e-01 -1.18708420e+00 -7.82105401e-02
2.00917751e-01 -1.87417062e-03 -5.26461959e-01 1.72320336e-01
-4.60461259e-01 5.95711708e-01 -7.83120036e-01 4.93825167e-01
1.99636087e-01 -1.17284030e-01 1.70996010e-01 6.81380928e-01
-3.00216466e-01 2.00223867e-02 -1.35086632e+00 1.32974231e+00
-5.22772789e-01 4.29767609e-01 4.21811551e-01 -9.76629853e-01
5.97535551e-01 2.27902576e-01 5.23946226e-01 -5.84022999e-01
2.08973765e-01 4.97112930e-01 -5.91705851e-02 -2.67981052e-01
8.99017975e-02 -2.69543856e-01 -4.29203182e-01 5.33615112e-01
-2.67292500e-01 -4.18210834e-01 2.10396364e-01 -1.12251639e-02
1.56098342e+00 2.80677110e-01 3.34005296e-01 -7.01487437e-02
1.87215284e-01 7.49353366e-03 6.23739958e-01 1.21230209e+00
-5.61161220e-01 -3.18124652e-01 9.70033765e-01 -3.12702544e-02
-9.58529770e-01 -1.04807985e+00 1.71810344e-01 7.28912771e-01
5.54480135e-01 -7.81982318e-02 -8.48583698e-01 -8.25083971e-01
6.10964820e-02 7.99475908e-01 -7.19073296e-01 -1.94379836e-01
-6.36957824e-01 -1.07751220e-01 5.56290090e-01 6.38972521e-01
5.26560426e-01 -1.23157704e+00 -1.21906590e+00 3.32280278e-01
3.19783568e-01 -8.87320459e-01 -6.30898237e-01 5.14389932e-01
-6.66886449e-01 -1.15350199e+00 -6.64024293e-01 -4.10296738e-01
8.15775990e-01 -2.11480111e-01 5.53395152e-01 -3.78948539e-01
-1.57287791e-01 6.91089988e-01 -1.63074121e-01 -5.43785214e-01
-4.58149612e-01 -1.60879135e-01 2.25681543e-01 -6.10835969e-01
-3.53275567e-01 -5.16166508e-01 -7.59818971e-01 1.85874969e-01
-8.13591957e-01 -3.47253412e-01 5.94960630e-01 9.68990028e-01
7.04675376e-01 2.54188597e-01 5.37188947e-01 -7.28990436e-01
6.97184026e-01 -1.57432407e-01 -1.28879023e+00 -7.30864480e-02
-8.35654497e-01 3.13683808e-01 1.08036947e+00 -7.71470368e-01
-5.72528005e-01 3.18061821e-02 1.07453652e-01 -5.94953120e-01
1.96833193e-01 3.48006189e-01 1.15144983e-01 -2.96146452e-01
6.23687923e-01 3.76568645e-01 2.41656303e-01 -2.12183282e-01
4.14819896e-01 2.70464450e-01 5.43371737e-01 -7.84406900e-01
8.12439501e-01 3.30415249e-01 3.34241599e-01 -3.48599464e-01
-5.04063129e-01 6.81879520e-02 -6.25855848e-02 -8.30158517e-02
3.92992258e-01 -6.24042034e-01 -1.36233592e+00 3.55877518e-03
-4.17056471e-01 -9.57497299e-01 -7.08815396e-01 5.92074513e-01
-1.23799384e+00 2.64285982e-01 -3.50230813e-01 -1.50340700e+00
-2.69196451e-01 -1.41618693e+00 5.00502825e-01 1.72355413e-01
1.03900991e-01 -5.92474222e-01 -2.30757087e-01 5.64577281e-02
3.80923390e-01 5.47577620e-01 8.35094452e-01 -4.45450187e-01
-8.08449149e-01 -2.67522752e-01 3.88421416e-01 4.98468906e-01
-2.71089405e-01 -5.87664366e-01 -2.42174074e-01 -7.95768738e-01
7.87395760e-02 -4.28834349e-01 5.50066352e-01 2.98518807e-01
8.41938496e-01 -9.20588136e-01 -2.52971143e-01 2.20624089e-01
1.35337210e+00 7.91127264e-01 3.34478378e-01 5.75991452e-01
2.01482683e-01 1.28578767e-01 9.37060118e-01 8.48770022e-01
1.08025365e-01 4.99129266e-01 8.29658031e-01 2.49444455e-01
4.55870360e-01 -2.82102436e-01 7.68802464e-01 1.39544725e-01
1.11751705e-01 -1.33237690e-01 -6.05686367e-01 6.29754066e-01
-1.70740581e+00 -7.35155225e-01 7.51980245e-01 2.78555298e+00
8.62748384e-01 6.65366828e-01 3.52346122e-01 1.33370876e-01
4.16046321e-01 -3.50416228e-02 -1.05153787e+00 -7.00672448e-01
3.56723160e-01 7.16868877e-01 9.29557085e-01 6.40031874e-01
-8.43082726e-01 9.28022623e-01 5.39290857e+00 9.05089855e-01
-8.97682726e-01 -1.84642136e-01 1.94187835e-01 -3.80973607e-01
-4.83324081e-02 3.00495744e-01 -7.18090773e-01 6.97077692e-01
1.13819826e+00 -3.05211008e-01 9.58053470e-01 1.10093117e+00
5.17202429e-02 -3.61415446e-01 -1.24809802e+00 5.69471061e-01
-3.92544597e-01 -8.85461390e-01 -6.28338754e-01 2.59416491e-01
7.99147606e-01 -1.64828882e-01 2.95855463e-01 5.86081803e-01
6.37224555e-01 -9.79560494e-01 6.43471837e-01 8.83707777e-03
9.23032463e-01 -1.11569786e+00 4.51174945e-01 8.33429039e-01
-1.15397108e+00 -7.08024085e-01 -7.44351745e-02 -1.06400140e-01
2.89250851e-01 -5.16919084e-02 -7.58739114e-01 3.53997231e-01
5.27428865e-01 1.26363593e-03 1.57466426e-01 7.84422696e-01
-4.52434123e-01 3.69593263e-01 -6.49403512e-01 -2.11695015e-01
7.91881919e-01 -2.05698922e-01 8.55383694e-01 6.62281692e-01
3.81913304e-01 1.66126505e-01 7.25279391e-01 7.12763906e-01
5.07920422e-02 -3.80172044e-01 -5.67172647e-01 -4.15015101e-01
7.61133909e-01 5.50787687e-01 -4.00178015e-01 -1.22184291e-01
1.08126044e-01 8.22193980e-01 1.57440349e-01 3.17239106e-01
-1.08335686e+00 -8.12500000e-01 5.14127731e-01 -1.15354685e-02
5.26527107e-01 -4.24485803e-01 1.27193227e-01 -6.41119063e-01
2.99958438e-01 -9.74401593e-01 2.47484609e-01 -1.51235655e-01
-5.29869795e-01 4.17619020e-01 1.13949649e-01 -1.39849412e+00
-6.93454206e-01 -3.83695453e-01 -3.09637785e-01 5.22323310e-01
-1.18187010e+00 -4.72442448e-01 1.71086043e-01 5.81900895e-01
5.58534205e-01 -1.94091022e-01 7.06468046e-01 -2.26791471e-01
-5.92293322e-01 7.76563883e-01 4.29516494e-01 -2.90940225e-01
1.06705226e-01 -1.38905251e+00 2.27950498e-01 8.64016473e-01
-4.59795475e-01 4.05830562e-01 9.48839426e-01 -5.02284706e-01
-1.79686570e+00 -9.38912451e-01 1.41946509e-01 -1.90741122e-01
7.32092440e-01 -1.52255714e-01 -3.23945642e-01 7.89052248e-01
-2.49155372e-01 5.83864190e-02 7.62804830e-03 -4.68337089e-01
-7.03428462e-02 4.91800532e-03 -1.39294016e+00 1.00612319e+00
1.05980158e+00 -5.24021387e-01 -2.84446120e-01 1.89666212e-01
7.58476257e-01 -7.79632747e-01 -1.18536055e+00 4.62159008e-01
3.75633985e-01 -6.51748955e-01 7.43265569e-01 -3.86917949e-01
-2.02923082e-02 -9.96389836e-02 6.87236851e-03 -1.17671382e+00
1.81070954e-01 -1.46764016e+00 -5.93107939e-01 4.22862738e-01
5.51667869e-01 -9.76964653e-01 6.99555099e-01 6.02546990e-01
-2.01801673e-01 -1.23719525e+00 -1.38594866e+00 -1.33341599e+00
4.29275751e-01 -2.80076981e-01 1.99378431e-01 2.34549120e-01
3.11857373e-01 -2.40105540e-01 -4.44419026e-01 1.32700354e-01
6.62156999e-01 1.56005248e-01 8.08381438e-01 -5.95166266e-01
-9.11350965e-01 -4.89019126e-01 4.88383062e-02 -1.17646086e+00
1.85896099e-01 -3.72433811e-01 2.96043187e-01 -1.26279914e+00
-1.60380021e-01 -7.03538835e-01 -1.63721636e-01 3.91737580e-01
-3.47096473e-02 -5.17309725e-01 4.71804261e-01 2.12800018e-02
-7.02280641e-01 5.82919955e-01 1.39487827e+00 -6.89271539e-02
-5.53575575e-01 3.63690943e-01 -4.15647238e-01 6.79758668e-01
7.57828474e-01 -3.01945806e-01 -7.47430921e-01 -9.16732624e-02
1.30629629e-01 7.51622915e-01 1.35034233e-01 -9.42955196e-01
1.26097612e-02 -5.61233878e-01 -2.10413471e-01 -1.78796679e-01
5.09703636e-01 -6.52488589e-01 1.72014683e-02 9.90251422e-01
-6.62359953e-01 7.82651082e-02 1.16710082e-01 1.03004813e+00
2.18676731e-01 -2.10857019e-01 7.10393429e-01 -2.81750560e-01
-4.24490035e-01 4.48319823e-01 -4.93108809e-01 2.62987584e-01
1.17530036e+00 -3.20384920e-01 -2.72174120e-01 -5.55411100e-01
-6.46908998e-01 6.01361513e-01 3.49534512e-01 -4.76911515e-02
4.15781170e-01 -8.74755144e-01 -1.37909487e-01 2.61032414e-02
-2.45508432e-01 2.62861997e-01 8.96114707e-02 1.01772153e+00
-1.38575941e-01 3.69181901e-01 9.01702866e-02 -2.14177206e-01
-7.63303161e-01 6.65093303e-01 4.00576144e-01 -5.17138660e-01
-8.11848640e-01 9.07996118e-01 -1.11406259e-01 -4.93103340e-02
5.50563693e-01 -3.88417125e-01 4.30954337e-01 -6.28729582e-01
3.22413921e-01 6.95306480e-01 -3.61702114e-01 6.86663836e-02
-2.06346273e-01 2.33567312e-01 4.92932051e-02 -6.17152989e-01
1.03562427e+00 -7.44106099e-02 4.35142100e-01 2.78689027e-01
8.67079198e-01 -1.54298484e-01 -2.08053732e+00 -6.54033124e-02
-2.37333681e-02 -1.59155965e-01 -2.98677444e-01 -7.02440679e-01
-6.27435684e-01 7.94747889e-01 6.81870461e-01 1.66786462e-01
1.02623212e+00 -1.88939333e-01 8.43219042e-01 6.02593362e-01
8.78279507e-01 -1.42931437e+00 2.74420679e-01 5.98894656e-01
6.78306937e-01 -8.96738470e-01 -1.03263088e-01 6.78826962e-03
-7.15179801e-01 7.40536451e-01 6.42544985e-01 -4.57661420e-01
1.95443079e-01 2.39896908e-01 -4.93109912e-01 1.68329477e-01
-6.70264661e-01 -2.47115850e-01 -1.67525157e-01 3.81295979e-01
-2.84571558e-01 2.70969152e-01 -3.28260422e-01 4.18203294e-01
-1.72115594e-01 -2.55127829e-02 5.55786431e-01 1.43592691e+00
-8.01960289e-01 -1.16053200e+00 -1.74843892e-01 3.05338264e-01
-5.57110548e-01 1.70149669e-01 5.37258200e-02 9.51266408e-01
-1.32179603e-01 1.02090216e+00 -3.05861145e-01 -1.12285778e-01
3.19511503e-01 -3.82791273e-02 7.41776884e-01 -3.08038235e-01
-4.21879917e-01 1.57720536e-01 1.45975083e-01 -8.68081093e-01
3.66256312e-02 -3.43546033e-01 -1.66803670e+00 -2.17485756e-01
-2.78348804e-01 2.44207934e-01 5.14975965e-01 1.09101605e+00
2.85516679e-01 3.51646781e-01 1.07660019e+00 -6.45051360e-01
-1.29305375e+00 -6.53445959e-01 -5.53961039e-01 -8.89576599e-02
5.81357837e-01 -8.37329149e-01 -7.66881347e-01 -5.79314232e-01] | [4.328333854675293, 2.7797763347625732] |
63432856-6852-42fb-b776-22b6c65c973d | hpointloc-point-based-indoor-place | 2212.14649 | null | https://arxiv.org/abs/2212.14649v1 | https://arxiv.org/pdf/2212.14649v1.pdf | HPointLoc: Point-based Indoor Place Recognition using Synthetic RGB-D Images | We present a novel dataset named as HPointLoc, specially designed for exploring capabilities of visual place recognition in indoor environment and loop detection in simultaneous localization and mapping. The loop detection sub-task is especially relevant when a robot with an on-board RGB-D camera can drive past the same place (``Point") at different angles. The dataset is based on the popular Habitat simulator, in which it is possible to generate photorealistic indoor scenes using both own sensor data and open datasets, such as Matterport3D. To study the main stages of solving the place recognition problem on the HPointLoc dataset, we proposed a new modular approach named as PNTR. It first performs an image retrieval with the Patch-NetVLAD method, then extracts keypoints and matches them using R2D2, LoFTR or SuperPoint with SuperGlue, and finally performs a camera pose optimization step with TEASER++. Such a solution to the place recognition problem has not been previously studied in existing publications. The PNTR approach has shown the best quality metrics on the HPointLoc dataset and has a high potential for real use in localization systems for unmanned vehicles. The proposed dataset and framework are publicly available: https://github.com/metra4ok/HPointLoc. | ['Aleksandr I. Panov', 'Aleksei Staroverov', 'Ruslan Musaev', 'Yaroslav Solomentsev', 'Dmitry Yudin'] | 2022-12-30 | null | null | null | null | ['simultaneous-localization-and-mapping', 'visual-place-recognition'] | ['computer-vision', 'computer-vision'] | [-3.23162615e-01 -1.95555165e-01 2.53950149e-01 -2.50914961e-01
-5.52642226e-01 -9.02833402e-01 7.51612306e-01 3.85174602e-02
-6.59264684e-01 6.78448200e-01 -3.18818778e-01 -2.26975143e-01
-1.49575114e-01 -7.09190130e-01 -9.38616097e-01 -6.22135520e-01
-3.44871789e-01 5.86942315e-01 2.66863286e-01 -3.90792161e-01
4.31600451e-01 9.53736246e-01 -1.83851874e+00 -4.52377021e-01
5.83620310e-01 1.01369846e+00 7.10456312e-01 6.37153149e-01
2.89589077e-01 3.54873031e-01 -2.32412443e-01 1.68411404e-01
7.44914711e-01 1.14708863e-01 -3.51529837e-01 -2.80852824e-01
5.49697042e-01 -1.14493422e-01 -2.30731651e-01 8.21328223e-01
6.58762097e-01 3.19889218e-01 3.16764474e-01 -1.56345403e+00
-2.23715510e-02 -1.64789110e-01 -4.97141659e-01 6.52216375e-02
9.88382697e-01 1.83470175e-02 6.37509048e-01 -9.49126065e-01
9.00492489e-01 9.58788574e-01 8.14040184e-01 -1.35108516e-01
-1.04266632e+00 -4.08900052e-01 -2.99433023e-01 3.34011346e-01
-1.93116963e+00 -3.07238549e-01 6.15940094e-01 -4.08375680e-01
9.33461785e-01 3.75876963e-01 7.40690708e-01 8.46875906e-01
2.87637711e-01 4.45082694e-01 1.14320958e+00 -2.76566744e-01
3.83445650e-01 2.42969662e-01 -2.63824195e-01 6.64270222e-01
2.01698840e-01 2.51536280e-01 -5.23523927e-01 4.53186184e-02
8.63295078e-01 2.00087503e-01 -5.21870494e-01 -1.08564687e+00
-1.60349226e+00 5.19789755e-01 9.38032210e-01 1.71787262e-01
-3.42255592e-01 9.57765579e-02 -4.76080030e-02 3.63356590e-01
-6.77238554e-02 5.71665823e-01 -2.32667655e-01 -1.46284088e-01
-6.09947443e-01 4.25992548e-01 8.45227480e-01 1.52346742e+00
1.16376591e+00 -5.22738516e-01 4.49383020e-01 5.03224790e-01
5.33985734e-01 8.44332755e-01 2.74193197e-01 -8.13413084e-01
5.15396357e-01 5.46751201e-01 5.39353371e-01 -1.31416273e+00
-6.51833773e-01 -1.77662700e-01 -4.63928401e-01 2.54834503e-01
1.52776837e-01 -3.99134122e-02 -7.40552068e-01 1.10970330e+00
5.91566205e-01 1.27925038e-01 2.05179065e-01 1.14754808e+00
8.86121631e-01 8.67371202e-01 -5.39171159e-01 4.23923999e-01
1.17352641e+00 -8.67591858e-01 -4.37286466e-01 -5.15339553e-01
7.21355557e-01 -9.43025827e-01 5.25787711e-01 2.89487958e-01
-3.81732374e-01 -3.92038494e-01 -1.26451349e+00 -2.65768498e-01
-8.36841166e-01 4.09582347e-01 4.27786976e-01 3.06634247e-01
-1.38004029e+00 5.75955100e-02 -6.50689065e-01 -1.04328108e+00
-1.18126847e-01 4.06266093e-01 -8.94563556e-01 -1.89662859e-01
-7.95874655e-01 1.14939547e+00 4.21951294e-01 3.72509807e-01
-1.04736984e+00 -3.29385132e-01 -1.14429510e+00 -4.01862890e-01
3.25061262e-01 -3.78345668e-01 8.72881532e-01 -4.26073223e-01
-1.31345153e+00 9.59516883e-01 -1.38329223e-01 -5.21496236e-01
6.49825215e-01 -2.95350522e-01 -3.20714951e-01 -9.03089270e-02
3.94793361e-01 8.42197478e-01 4.12057579e-01 -1.38206613e+00
-7.71392405e-01 -6.02459073e-01 1.23687036e-01 5.63935101e-01
5.51801503e-01 -3.79281759e-01 -5.40598273e-01 -1.63323402e-01
6.57609820e-01 -1.31685650e+00 -3.35281312e-01 1.47891551e-01
-3.67699504e-01 2.30454668e-01 8.19617450e-01 -3.54405344e-01
4.77882683e-01 -2.31684446e+00 1.25116587e-01 4.96602505e-01
-2.75307238e-01 -9.35953781e-02 2.27215365e-02 9.01137233e-01
1.23181328e-01 -3.46414715e-01 -4.94398586e-02 -3.37446421e-01
5.19667901e-02 3.08584273e-01 -1.39203981e-01 1.09874904e+00
-5.22284746e-01 4.24993694e-01 -8.80516768e-01 -2.24744529e-01
6.77974224e-01 5.37234545e-01 -4.07935560e-01 1.41018897e-01
1.95760690e-02 4.90397513e-01 -3.71844560e-01 7.63531864e-01
1.16758168e+00 4.00161058e-01 -1.96168020e-01 -1.23650888e-02
-9.43799138e-01 7.17929564e-03 -1.60715210e+00 2.09517503e+00
-4.82690752e-01 8.93535972e-01 1.81968778e-01 -4.36346620e-01
1.24758112e+00 -8.53340402e-02 3.33412021e-01 -7.83163965e-01
4.50546145e-02 5.08875847e-01 -6.18717909e-01 -3.42193156e-01
9.81012404e-01 4.39246446e-01 -9.43987221e-02 -2.68918842e-01
-2.00986955e-02 -4.38752890e-01 -2.77307760e-02 -9.90366861e-02
1.18544018e+00 4.16594237e-01 4.22474861e-01 -3.59056175e-01
6.32151783e-01 5.99371552e-01 3.03623527e-01 5.71100295e-01
-1.72785580e-01 7.40957439e-01 9.69289988e-02 -3.13427389e-01
-1.03878772e+00 -9.60219741e-01 -2.29864448e-01 4.63149279e-01
7.95228243e-01 -2.29262844e-01 -3.50992024e-01 -1.48201421e-01
2.56925404e-01 2.11249456e-01 -3.88384581e-01 2.81701237e-01
-1.35441139e-01 -2.13444814e-01 4.04317468e-01 -3.85136232e-02
8.24501812e-01 -6.64569497e-01 -1.07764101e+00 3.66732217e-02
-1.39327854e-01 -1.08104956e+00 8.09822679e-02 5.24328947e-01
-4.81392682e-01 -1.11685121e+00 -7.04874873e-01 -9.94332373e-01
6.83169425e-01 7.46241331e-01 6.01142585e-01 -2.63909876e-01
-2.37536371e-01 6.14780128e-01 -6.09854102e-01 -2.05756694e-01
2.27548957e-01 -6.74142269e-04 1.18216373e-01 -4.30078991e-02
2.39923984e-01 -3.73804241e-01 -7.05499828e-01 7.38463163e-01
-4.34796333e-01 -1.94576710e-01 5.66326022e-01 4.39262360e-01
8.25507164e-01 -1.50116205e-01 -2.96794027e-01 -5.90358526e-02
1.93583239e-02 -6.00799084e-01 -1.19326138e+00 -1.02660634e-01
-2.37059802e-01 -2.29562789e-01 2.83485740e-01 1.12562440e-01
-6.04341269e-01 6.41294181e-01 -3.22228111e-02 -3.96162003e-01
-3.94691676e-01 2.91606992e-01 -2.48856246e-01 -7.69414723e-01
6.04475319e-01 2.46594146e-01 -4.01959121e-02 -4.20651853e-01
2.68466085e-01 7.23659456e-01 5.85783124e-01 -5.75489625e-02
1.03102577e+00 8.31772029e-01 2.06462830e-01 -1.16387367e+00
-2.97807734e-02 -1.05217063e+00 -8.50025237e-01 -1.67145774e-01
8.82981777e-01 -1.17208791e+00 -6.97959006e-01 4.84280735e-01
-1.16243601e+00 -2.44377300e-01 7.22771510e-02 6.52743995e-01
-7.37510383e-01 7.97702968e-02 6.75901100e-02 -7.00323522e-01
1.50388837e-01 -1.21438169e+00 1.15745711e+00 4.14881468e-01
2.00678989e-01 -8.40478241e-01 6.32929087e-01 9.04278234e-02
2.04815924e-01 4.81901258e-01 -8.63644406e-02 -3.83658230e-01
-1.15723705e+00 -3.49701673e-01 -2.22217515e-02 -2.36625880e-01
-1.81282371e-01 -3.42881054e-01 -9.38099265e-01 -4.25439090e-01
-2.40823850e-01 4.02131379e-02 4.64736074e-01 3.61932255e-02
3.98360342e-01 6.45416975e-02 -6.13740325e-01 9.65068281e-01
1.89301944e+00 3.07730287e-01 8.28632474e-01 1.08099413e+00
5.93414783e-01 3.56989741e-01 1.12590957e+00 4.83935952e-01
7.51289248e-01 1.08776414e+00 1.00269234e+00 -8.90271217e-02
2.31098637e-01 -5.08825302e-01 2.47031435e-01 1.87982917e-01
3.36026140e-02 -1.58034533e-01 -1.12581956e+00 7.30075657e-01
-1.93831503e+00 -6.12968445e-01 -4.05454934e-01 2.44552398e+00
-2.07960054e-01 -3.12141448e-01 -3.84594202e-01 -2.85781845e-02
5.56927443e-01 1.35266542e-01 -1.28225073e-01 -1.00558773e-01
-8.55005383e-02 -2.91852474e-01 1.28499436e+00 6.74186468e-01
-1.17463362e+00 8.81917059e-01 5.13354683e+00 3.96720052e-01
-1.08673275e+00 5.36260046e-02 -1.89839587e-01 2.02215716e-01
1.10692613e-01 3.93846422e-01 -8.80088389e-01 1.97061479e-01
6.04206085e-01 2.53886729e-01 4.82070625e-01 1.18515074e+00
2.19677299e-01 -8.51640761e-01 -9.82325196e-01 1.35909641e+00
2.97824442e-01 -1.04187787e+00 -5.11995077e-01 3.02612305e-01
5.17203391e-01 5.92230797e-01 -1.10056199e-01 6.48243055e-02
-7.69017115e-02 -7.08182633e-01 9.84148562e-01 2.68528223e-01
3.59022796e-01 -5.95198035e-01 7.79842913e-01 4.37912941e-01
-1.46414971e+00 -1.21565513e-01 -5.64369500e-01 -5.75104542e-02
3.23809609e-02 1.53180316e-01 -1.34947252e+00 7.92010009e-01
8.31438005e-01 8.69476438e-01 -7.52045929e-01 1.79041171e+00
-2.13389769e-01 -3.41400087e-01 -7.79688954e-01 -1.79825649e-01
3.99685264e-01 -3.95827115e-01 8.19354236e-01 8.17999184e-01
8.79286110e-01 -2.33711600e-01 1.06154583e-01 6.57360315e-01
4.76501465e-01 1.13014318e-01 -1.23801327e+00 6.06353104e-01
6.16790771e-01 1.36514604e+00 -7.96806753e-01 1.23150721e-01
-2.61562690e-02 1.06103110e+00 8.01555365e-02 2.03879282e-01
-7.90923774e-01 -5.93947649e-01 6.09775543e-01 1.92727268e-01
3.84517968e-01 -6.57785177e-01 1.94561630e-01 -1.03830683e+00
9.29859132e-02 -4.47723866e-01 -4.60855849e-02 -1.23899865e+00
-3.08099627e-01 5.29799819e-01 -2.57548932e-02 -1.78816795e+00
-2.27474980e-02 -6.86801255e-01 -1.66671321e-01 7.18533158e-01
-1.65097284e+00 -1.07368624e+00 -7.77685702e-01 8.49701643e-01
2.63544351e-01 1.32825285e-01 7.11712420e-01 5.16760647e-01
-2.38282785e-01 -2.69791558e-02 4.65391248e-01 -2.11931840e-01
6.85925305e-01 -1.27116084e+00 2.20548049e-01 9.46606338e-01
8.61954540e-02 5.37660897e-01 9.44258153e-01 -5.60252726e-01
-2.09061790e+00 -8.78846347e-01 7.61750877e-01 -5.85032642e-01
3.68066609e-01 -6.27277970e-01 -2.45351151e-01 8.15701067e-01
3.79403122e-02 3.79026495e-02 8.82790163e-02 -4.61245179e-01
1.12155959e-01 -4.19534832e-01 -1.44190490e+00 4.78639841e-01
1.01656091e+00 -3.79841149e-01 -4.22859222e-01 4.73433614e-01
2.67673999e-01 -9.42537963e-01 -6.81323111e-01 1.06031723e-01
2.78624773e-01 -1.21687555e+00 1.00037575e+00 6.18000567e-01
-2.50403821e-01 -8.59527946e-01 -4.50763375e-01 -1.49719787e+00
7.89855644e-02 -4.69681591e-01 5.25834978e-01 1.06434357e+00
2.64474809e-01 -9.74117339e-01 5.22720516e-01 9.31814387e-02
-3.19330364e-01 -1.60093009e-01 -1.23657668e+00 -8.43599200e-01
-7.87692368e-01 -3.07469428e-01 6.87966406e-01 5.41808486e-01
-2.07468927e-01 -2.50665933e-01 -2.42367774e-01 8.43198657e-01
7.31387794e-01 7.18102604e-02 1.42235005e+00 -1.08729911e+00
2.72119939e-01 9.24350768e-02 -1.16795266e+00 -1.16627383e+00
-1.63377002e-01 -8.41320097e-01 4.14662987e-01 -1.67785645e+00
-5.48982203e-01 -8.55336249e-01 1.86653897e-01 2.06005454e-01
7.93474257e-01 3.44801664e-01 1.30125389e-01 3.54048103e-01
-6.57695830e-01 5.09814382e-01 7.69389570e-01 8.32712799e-02
-2.77496636e-01 -1.29344478e-01 6.91068470e-02 4.53179777e-01
5.95220983e-01 -2.45837986e-01 -2.63424158e-01 -4.58618462e-01
4.02057350e-01 1.54953510e-01 7.84964919e-01 -1.73321915e+00
6.03814721e-01 1.44126281e-01 3.70370239e-01 -1.07987034e+00
7.97483265e-01 -1.44649601e+00 7.30411589e-01 4.73224312e-01
3.48369598e-01 4.65585977e-01 1.53511152e-01 5.66656947e-01
-2.56790906e-01 -1.63363948e-01 4.24833536e-01 -3.27686399e-01
-1.43115127e+00 1.20548323e-01 -2.57005543e-01 -4.63647217e-01
1.51283336e+00 -4.81800169e-01 -2.85977662e-01 -2.16232076e-01
-4.70903337e-01 4.63030815e-01 1.15918648e+00 5.27273953e-01
8.12120318e-01 -1.37616324e+00 -1.17635585e-01 5.06030917e-01
6.18508995e-01 1.90959886e-01 1.01773985e-01 1.01033986e+00
-1.35762393e+00 5.72563708e-01 -5.23053348e-01 -1.00569379e+00
-1.03204250e+00 4.51636791e-01 4.87268388e-01 3.05855751e-01
-6.89375043e-01 6.49702787e-01 -2.31029764e-01 -9.15268898e-01
2.30060264e-01 -3.42470855e-01 -1.96589649e-01 6.22443072e-02
3.21134299e-01 5.30850232e-01 2.76021183e-01 -1.11654472e+00
-8.28505456e-01 9.34079230e-01 5.73470235e-01 -2.83474028e-01
1.30722332e+00 -5.82493186e-01 -2.75614113e-01 3.10930908e-01
1.24117208e+00 3.84359993e-02 -1.13578165e+00 3.18668127e-01
1.38449827e-02 -8.17748904e-01 1.72587819e-02 -4.32688534e-01
-5.74198127e-01 3.60696346e-01 1.16783464e+00 -1.09168030e-02
6.89472616e-01 -1.42144859e-01 2.76449859e-01 6.95406318e-01
1.30478764e+00 -9.02424097e-01 -4.81394023e-01 7.11997092e-01
9.44782197e-01 -1.38226771e+00 -2.81421584e-03 -1.32642269e-01
-4.11373764e-01 8.34072113e-01 4.57741410e-01 -3.15603167e-01
5.14609516e-01 2.14056354e-02 2.31311247e-01 -2.56628662e-01
-5.67728281e-02 -2.65582114e-01 -1.12510607e-01 7.49506056e-01
-1.80401787e-01 1.42044470e-01 4.31076176e-02 -2.63031602e-01
-4.48032260e-01 -9.91042852e-02 5.90247989e-01 1.24309444e+00
-5.79201460e-01 -7.56654441e-01 -9.62235212e-01 -2.54432887e-01
3.21293384e-01 1.87739402e-01 -2.44153693e-01 8.98028672e-01
3.45235676e-01 9.21030641e-01 -7.33230216e-03 -4.32751834e-01
5.64332306e-01 -3.33958566e-01 3.36902231e-01 -2.34892368e-01
-3.69148552e-01 -2.22909555e-01 -1.46860313e-02 -1.03829861e+00
-3.24804753e-01 -7.57583201e-01 -1.12198806e+00 -1.69350669e-01
-4.49858196e-02 2.76880026e-01 1.35938430e+00 5.43336034e-01
5.51488221e-01 -1.49815962e-01 7.07026124e-01 -1.56556511e+00
5.89829348e-02 -6.31981850e-01 -7.38443792e-01 -1.69536293e-01
7.42857993e-01 -9.05822575e-01 -3.86208951e-01 -4.41769481e-01] | [7.358870029449463, -2.054351568222046] |
18cca702-24d0-4ef8-a633-a6bad2a8cc56 | detecting-gender-bias-in-transformer-based | 2110.15733 | null | https://arxiv.org/abs/2110.15733v1 | https://arxiv.org/pdf/2110.15733v1.pdf | Detecting Gender Bias in Transformer-based Models: A Case Study on BERT | In this paper, we propose a novel gender bias detection method by utilizing attention map for transformer-based models. We 1) give an intuitive gender bias judgement method by comparing the different relation degree between the genders and the occupation according to the attention scores, 2) design a gender bias detector by modifying the attention module, 3) insert the gender bias detector into different positions of the model to present the internal gender bias flow, and 4) draw the consistent gender bias conclusion by scanning the entire Wikipedia, a BERT pretraining dataset. We observe that 1) the attention matrices, Wq and Wk introduce much more gender bias than other modules (including the embedding layer) and 2) the bias degree changes periodically inside of the model (attention matrix Q, K, V, and the remaining part of the attention layer (including the fully-connected layer, the residual connection, and the layer normalization module) enhance the gender bias while the averaged attentions reduces the bias). | ['Caiwen Ding', 'Hang Liu', 'Binghui Wang', 'Weiwen Jiang', 'Yueying Liang', 'Lei Yang', 'Junhuan Yang', 'Rajat Sainju', 'Hongwu Peng', 'Bingbing Li'] | 2021-10-15 | null | null | null | null | ['gender-bias-detection', 'gender-bias-detection'] | ['miscellaneous', 'natural-language-processing'] | [-2.79587001e-01 4.09650564e-01 -2.66580969e-01 -4.11534399e-01
3.59666407e-01 -1.85041532e-01 6.20977819e-01 -9.99644049e-04
-6.15646839e-01 4.43180114e-01 3.84237647e-01 -1.27002046e-01
4.74650264e-02 -1.15972841e+00 -5.12180507e-01 -6.22763872e-01
1.90116420e-01 3.81329954e-01 2.15039000e-01 -4.63956863e-01
7.22692072e-01 1.50918802e-02 -1.40449011e+00 -8.50241929e-02
1.11560202e+00 8.62977087e-01 -6.84914785e-03 1.13422684e-01
-3.33421290e-01 5.25421441e-01 -5.87550640e-01 -8.73695076e-01
-1.84640199e-01 -5.55335343e-01 -9.30411339e-01 -4.35724616e-01
4.35323924e-01 -2.79048294e-01 -2.28239521e-01 1.50325537e+00
7.98709929e-01 -1.26218289e-01 1.01695287e+00 -1.38840020e+00
-7.65695989e-01 9.15997028e-01 -8.80755544e-01 4.02648062e-01
-1.63052291e-01 -1.68534875e-01 7.52603829e-01 -9.74974036e-01
4.73399580e-01 1.64798129e+00 5.82383692e-01 6.82066023e-01
-1.01363313e+00 -1.28863609e+00 5.83514690e-01 4.95228171e-01
-1.28015125e+00 -1.09208710e-01 9.97537434e-01 -4.88170266e-01
3.82842720e-01 4.16986823e-01 8.21906984e-01 1.20957589e+00
4.16240603e-01 5.01777709e-01 1.00431097e+00 -2.04973653e-01
-3.24792087e-01 6.09176755e-01 5.15367389e-01 8.11951756e-01
4.65769470e-01 -2.02439204e-01 -3.43875855e-01 8.28135386e-02
7.62957811e-01 -8.75863731e-02 2.34331563e-01 -2.95511812e-01
-1.20864880e+00 7.02636421e-01 7.02195287e-01 3.06023449e-01
-1.15984239e-01 1.99668124e-01 5.62626123e-01 2.64204830e-01
3.45245451e-01 3.18855494e-01 -4.15455401e-01 3.97433370e-01
-7.66625762e-01 2.11464778e-01 2.99911410e-01 7.55811095e-01
1.10209739e+00 -4.34856080e-02 -6.58663929e-01 8.64749193e-01
6.47131145e-01 5.51171541e-01 7.47074187e-01 -5.81507444e-01
8.84848714e-01 9.87972796e-01 -1.94155321e-01 -1.50460410e+00
-4.05984133e-01 -6.41451359e-01 -1.00230503e+00 -3.67351323e-01
3.76135141e-01 -8.69921669e-02 -6.01561546e-01 2.03120446e+00
1.82766870e-01 -4.68288571e-01 -4.84130621e-01 7.49226093e-01
1.23777938e+00 1.82397410e-01 4.80845690e-01 9.97153521e-02
1.86715651e+00 -7.29259908e-01 -1.04207623e+00 -5.39005339e-01
3.91592026e-01 -4.59771454e-01 1.15134573e+00 -2.21391916e-01
-1.07803261e+00 -9.18123543e-01 -1.17291629e+00 -2.07175940e-01
-9.18800175e-01 3.89004618e-01 7.88129687e-01 6.68943465e-01
-7.56516039e-01 5.26169360e-01 -2.95150161e-01 -6.23855472e-01
4.90216941e-01 5.10719895e-01 -3.82960141e-02 3.08085859e-01
-1.74130392e+00 9.59194481e-01 1.59454972e-01 2.31886342e-01
-6.47326708e-01 -5.26137531e-01 -8.54972541e-01 2.31102541e-01
-2.13403553e-01 -8.15467298e-01 5.74117064e-01 -1.27256429e+00
-9.72445488e-01 1.17918336e+00 -3.45104069e-01 2.78810263e-01
5.58384717e-01 2.20234290e-01 -3.06104034e-01 -4.49218124e-01
1.71968579e-01 9.50034022e-01 8.41720045e-01 -9.88784075e-01
-6.07417285e-01 -7.96239913e-01 -2.05713436e-02 3.56417060e-01
-8.00758600e-01 1.73319057e-01 -3.09515148e-01 -6.10014081e-01
3.08065534e-01 -5.48616588e-01 -1.33502439e-01 -4.59629416e-01
-4.26584989e-01 -4.43865567e-01 6.20938718e-01 -1.00311601e+00
1.61602223e+00 -2.15757680e+00 1.67401448e-01 3.58487308e-01
4.48915750e-01 -2.94744194e-01 2.07389086e-01 1.30307049e-01
-2.66860694e-01 2.08630770e-01 -6.90481067e-02 6.98741823e-02
2.38805979e-01 6.08098507e-03 -5.92624620e-02 3.28350842e-01
9.92318392e-02 9.53179240e-01 -8.14478338e-01 -9.10604835e-01
-1.28824696e-01 6.63747564e-02 -8.25084150e-01 -8.06496888e-02
5.46252251e-01 2.52097487e-01 -3.56030434e-01 5.42355895e-01
1.00982320e+00 1.65157393e-01 2.54222453e-01 -5.12152374e-01
-1.08994447e-01 3.46767187e-01 -1.26495099e+00 1.13499439e+00
-1.75142840e-01 6.09340370e-01 6.10500686e-02 -8.04109693e-01
1.23800933e+00 -1.55161381e-01 -5.28774075e-02 -9.20873284e-01
5.29317081e-01 1.24691010e-01 3.34948182e-01 -5.11594832e-01
8.08744013e-01 6.30864874e-02 -2.95668691e-01 3.64004999e-01
4.24943715e-01 4.75340486e-01 2.42495045e-01 1.66948125e-01
3.52498800e-01 -1.71738222e-01 -1.56321138e-01 -8.79873335e-01
8.04078698e-01 -5.95170140e-01 6.31967962e-01 2.85715967e-01
-4.08159375e-01 1.73182622e-01 1.16358149e+00 -4.18130666e-01
-7.82960474e-01 -7.14067400e-01 -2.36512080e-01 1.49147439e+00
4.80380505e-01 -9.58155617e-02 -1.00997865e+00 -9.34707880e-01
3.27455401e-01 4.59574878e-01 -1.29025924e+00 -7.06860900e-01
-6.08063161e-01 -1.22376418e+00 4.79581743e-01 8.09369147e-01
1.14551580e+00 -1.07401037e+00 -7.48301595e-02 -3.06685776e-01
-3.44105870e-01 -2.74194926e-01 -6.10244870e-01 1.36705801e-01
-9.94745076e-01 -9.76751029e-01 -7.10628688e-01 -1.08931136e+00
1.07984269e+00 -4.53724414e-01 1.21011591e+00 1.98710546e-01
7.46197104e-02 -7.63374418e-02 4.28721339e-01 -6.32025599e-01
1.25596344e-01 5.46426475e-01 -1.03139788e-01 1.21260412e-01
6.91075146e-01 -5.25341868e-01 -7.47500598e-01 4.39069837e-01
-1.58843309e-01 -1.24901310e-02 5.00272989e-01 6.49038136e-01
1.35365158e-01 -2.39686415e-01 6.79544568e-01 -9.29670990e-01
8.02000463e-01 -3.66789401e-01 -2.41114944e-01 -5.75005561e-02
-1.07161438e+00 -9.09413695e-02 6.97798654e-02 -3.67735773e-01
-1.04268730e+00 -7.79908180e-01 -3.22384350e-02 1.56881526e-01
2.06817269e-01 -2.82843746e-02 -6.73409283e-01 1.69818044e-01
4.16311473e-01 5.70798069e-02 -4.96815965e-02 -5.18534660e-01
1.78145617e-01 6.84161544e-01 3.04417908e-01 -5.14614105e-01
7.77060688e-01 2.57781178e-01 -1.27442852e-01 -1.73552841e-01
-3.60235333e-01 1.74845070e-01 -8.46255243e-01 -3.50135267e-01
9.24722970e-01 -7.92010546e-01 -9.74862397e-01 4.55439895e-01
-1.07241511e+00 1.85403191e-02 -1.95709076e-02 4.98464793e-01
4.26176712e-02 2.51093041e-02 -7.45683014e-01 -7.31193125e-01
-4.32581812e-01 -8.25504303e-01 8.33646357e-01 5.37714958e-01
-3.98325235e-01 -9.57583070e-01 -1.90865904e-01 2.57527493e-02
3.77736777e-01 -2.28855431e-01 1.31811643e+00 -5.11901557e-01
-1.99763533e-02 2.11245716e-01 -7.09625542e-01 1.12630650e-01
-1.35013342e-01 5.10342382e-02 -1.15248322e+00 1.83017552e-02
-3.41821522e-01 1.24982901e-01 1.23835444e+00 4.25122321e-01
1.31769168e+00 -3.20409924e-01 -5.17223895e-01 6.34762168e-01
1.18541837e+00 1.29720584e-01 8.21967661e-01 3.68144125e-01
9.51939583e-01 9.97939289e-01 2.98813164e-01 -1.35927945e-01
8.70358706e-01 2.71538883e-01 3.96531790e-01 -5.69553852e-01
-8.99310410e-02 -5.54741144e-01 2.32434839e-01 1.02910554e+00
-4.90171969e-01 3.75871390e-01 -6.57461345e-01 7.50502765e-01
-1.61806071e+00 -9.48345244e-01 -3.20718467e-01 2.06183076e+00
8.44323277e-01 1.61299244e-01 1.34805337e-01 1.75018162e-01
1.00476825e+00 2.64170170e-01 -4.58933771e-01 -8.26032341e-01
9.10328515e-03 -5.33297323e-02 6.45833492e-01 2.29179680e-01
-9.84655440e-01 8.25688422e-01 6.93889236e+00 7.50136733e-01
-9.26565349e-01 6.71830541e-03 8.27457845e-01 1.69097438e-01
-4.76736486e-01 -3.08251917e-01 -9.09953475e-01 7.65825212e-01
3.19659054e-01 1.02586977e-01 3.88324112e-01 6.37858868e-01
-1.50658309e-01 6.34280145e-02 -1.11676192e+00 1.00033128e+00
9.08520743e-02 -5.20969391e-01 1.48696467e-01 1.54119641e-01
3.85907322e-01 -5.03224730e-01 1.81884870e-01 7.92899311e-01
-5.48718423e-02 -9.93336201e-01 8.69992912e-01 5.98482192e-01
7.79836357e-01 -1.00519538e+00 9.60423052e-01 -7.76712075e-02
-1.11458528e+00 -3.42638522e-01 -6.51732683e-01 -3.21942508e-01
-6.07503235e-01 4.77253199e-01 -2.51292020e-01 4.98519748e-01
1.02606118e+00 7.26383984e-01 -1.18902421e+00 3.71633738e-01
-3.98754478e-01 5.11614203e-01 3.25239718e-01 -1.72670767e-01
-2.40528464e-01 -4.23167676e-01 8.11219886e-02 1.22049975e+00
1.25893608e-01 -5.30481696e-01 -5.62737525e-01 1.16300559e+00
-3.41918290e-01 3.48026246e-01 -3.46718609e-01 1.99905545e-01
4.44799185e-01 1.53447425e+00 -8.21685791e-01 -3.84937644e-01
-2.86156327e-01 5.75994194e-01 9.86727476e-02 4.77708697e-01
-9.44983661e-01 -8.32144976e-01 8.21122229e-01 1.99020967e-01
7.69927800e-02 4.32864338e-01 -7.74636090e-01 -7.87638187e-01
-1.89724326e-01 -5.55762172e-01 4.18429941e-01 -6.68791533e-01
-1.04288769e+00 4.07574996e-02 -1.47460505e-01 -7.76045918e-01
2.49113575e-01 -5.47140419e-01 -7.49506712e-01 1.32733274e+00
-1.36482298e+00 -8.68524611e-01 -4.53746915e-01 3.88910294e-01
-1.12890154e-01 -2.60474354e-01 6.83227062e-01 8.29229534e-01
-1.01038623e+00 9.58389759e-01 -5.06067216e-01 6.28930390e-01
9.61955547e-01 -1.18146527e+00 5.70199452e-02 4.71320182e-01
-6.25786543e-01 1.00457382e+00 4.75979745e-01 -6.04211450e-01
-7.18234241e-01 -9.10521030e-01 1.54452991e+00 -5.64687788e-01
3.69154692e-01 -7.30962634e-01 -5.80322862e-01 6.91671848e-01
4.04940635e-01 -4.46984619e-01 4.95197713e-01 6.00458920e-01
-3.42849255e-01 -5.66646397e-01 -1.17823923e+00 6.24487817e-01
1.31433415e+00 -4.90453720e-01 -8.81317914e-01 -3.28627676e-01
5.25561273e-01 -1.58437967e-01 -6.98187232e-01 4.30595130e-01
9.47784185e-01 -8.42056274e-01 9.38105226e-01 -7.43431151e-01
8.30319285e-01 -2.04379067e-01 8.40781108e-02 -1.20793414e+00
-1.01817381e+00 1.21213816e-01 1.45649672e-01 1.76550174e+00
4.59628552e-01 -8.08200896e-01 5.66936612e-01 2.91364968e-01
2.25092635e-01 -7.74668813e-01 -6.75770640e-01 -1.97644174e-01
3.90052170e-01 2.87710220e-01 1.31969404e+00 1.24466789e+00
8.69127586e-02 6.33353353e-01 -2.77857091e-02 -2.51712412e-01
3.46279055e-01 6.13778606e-02 5.08304775e-01 -1.51453805e+00
3.48282009e-01 -8.67320061e-01 -2.60642380e-01 -5.86212873e-01
-3.94757614e-02 -1.10958207e+00 -4.85378414e-01 -1.66028583e+00
6.77448869e-01 -1.99694887e-01 -6.54503107e-01 4.33370352e-01
-5.46350181e-01 1.88547343e-01 -1.27542675e-01 -3.49475518e-02
-3.12536240e-01 5.58762014e-01 1.56387794e+00 -3.99323285e-01
1.97487786e-01 -2.46794909e-01 -1.32492375e+00 8.86439860e-01
5.53476214e-01 -3.58977407e-01 -3.29428226e-01 -4.83339399e-01
7.14520216e-01 -5.90238273e-01 3.79034847e-01 -4.97057974e-01
6.49527647e-04 -2.19711568e-02 1.15303159e+00 -4.68118548e-01
-2.64022470e-01 -6.06694520e-01 -4.32355911e-01 6.71820521e-01
-1.94386035e-01 2.07026079e-01 1.02111287e-01 -1.67876761e-02
-1.52409911e-01 -1.07922480e-01 6.08171284e-01 -1.31159395e-01
-3.11696976e-01 2.04005763e-01 -3.06329817e-01 1.27414137e-01
4.84740585e-01 -2.05888107e-01 -4.12938863e-01 1.66690528e-01
-5.42318523e-01 4.67449039e-01 2.18462259e-01 5.65115333e-01
-2.41357591e-02 -1.83347058e+00 -5.08137584e-01 5.91159999e-01
1.01254769e-02 -3.85965228e-01 2.98051238e-01 1.06614482e+00
3.45859490e-02 1.60052255e-01 -6.05446160e-01 -2.63278246e-01
-1.19640899e+00 3.68692279e-01 6.03545725e-01 -3.59425724e-01
2.07729459e-01 1.20277357e+00 5.98057330e-01 -8.18225563e-01
3.07446510e-01 -2.91802615e-01 -1.13724661e+00 9.77514982e-01
2.73727119e-01 7.96835244e-01 -2.54171789e-02 -5.35887003e-01
-7.14889109e-01 8.04822206e-01 2.27700055e-01 5.72999641e-02
1.00997615e+00 -2.05951989e-01 -6.06756687e-01 5.19481242e-01
1.00839174e+00 1.23780854e-01 -7.30812848e-01 2.56393015e-01
-3.53496313e-01 -2.94850200e-01 -2.54844069e-01 -7.36422300e-01
-1.39240611e+00 1.11527133e+00 7.60865688e-01 2.76808143e-01
9.77423191e-01 -1.57524452e-01 4.05500829e-01 -1.78511351e-01
1.34607732e-01 -1.68069673e+00 -2.54717976e-01 4.78267193e-01
7.05300927e-01 -8.83061767e-01 9.03437883e-02 -2.55740911e-01
-4.73839104e-01 6.76939189e-01 1.06817627e+00 -1.38774231e-01
7.66472638e-01 -6.00940250e-02 8.99310112e-02 -3.32951128e-01
-5.11344671e-01 -1.61030400e-03 3.71992141e-01 5.24070203e-01
6.29011691e-01 5.75577468e-02 -9.65905666e-01 1.43504047e+00
-5.26631713e-01 -3.84455860e-01 -1.04413353e-01 1.67846993e-01
-2.08064064e-01 -8.23696375e-01 -3.83605033e-01 7.01284051e-01
-2.55252391e-01 -1.07027270e-01 -6.06870174e-01 7.52368391e-01
8.38277400e-01 5.64755499e-01 4.65480655e-01 -7.07555294e-01
6.38516307e-01 3.85256559e-01 6.88004494e-01 -2.11686954e-01
-9.02165473e-01 -2.45301351e-01 1.14387367e-02 -5.46460450e-01
-1.16129853e-01 -6.29234254e-01 -1.10703886e+00 -5.19803703e-01
-1.81831792e-01 8.10876265e-02 5.07618904e-01 7.38091946e-01
2.44268298e-01 1.04578996e+00 5.82513511e-01 -5.48539281e-01
-1.62263513e-01 -1.51538551e+00 -9.32815850e-01 4.48465943e-01
-1.71535447e-01 -1.05532575e+00 -4.43930328e-01 -1.50644153e-01] | [9.403170585632324, 10.213443756103516] |
86d81b65-96ca-4825-b8ca-a48dd384be17 | spherical-transformer-adapting-spherical | 2101.03848 | null | https://arxiv.org/abs/2101.03848v3 | https://arxiv.org/pdf/2101.03848v3.pdf | Spherical Transformer: Adapting Spherical Signal to CNNs | Convolutional neural networks (CNNs) have been widely used in various vision tasks, e.g. image classification, semantic segmentation, etc. Unfortunately, standard 2D CNNs are not well suited for spherical signals such as panorama images or spherical projections, as the sphere is an unstructured grid. In this paper, we present Spherical Transformer which can transform spherical signals into vectors that can be directly processed by standard CNNs such that many well-designed CNNs architectures can be reused across tasks and datasets by pretraining. To this end, the proposed method first uses local structured sampling methods such as HEALPix to construct a transformer grid by using the information of spherical points and its adjacent points, and then transforms the spherical signals to the vectors through the grid. By building the Spherical Transformer module, we can use multiple CNN architectures directly. We evaluate our approach on the tasks of spherical MNIST recognition, 3D object classification and omnidirectional image semantic segmentation. For 3D object classification, we further propose a rendering-based projection method to improve the performance and a rotational-equivariant model to improve the anti-rotation ability. Experimental results on three tasks show that our approach achieves superior performance over state-of-the-art methods. | ['Haikuan Du', 'Yin Wang', 'Yuqi Liu', 'Shen Cai'] | 2021-01-11 | null | null | null | null | ['3d-object-classification'] | ['computer-vision'] | [ 1.61014527e-01 -2.28110738e-02 2.61307180e-01 -6.37515545e-01
-3.42414498e-01 -5.25072217e-01 8.23978782e-01 -5.99565744e-01
-5.19509971e-01 2.51331609e-02 -2.75026672e-02 -2.48103812e-01
1.62008643e-01 -1.05747128e+00 -1.10051548e+00 -6.94820404e-01
3.41391444e-01 5.08221865e-01 5.94085634e-01 -1.18299626e-01
6.82596564e-02 8.55193317e-01 -1.49866533e+00 2.84286857e-01
6.38851941e-01 1.44889688e+00 3.21844786e-01 2.01131672e-01
-2.41158485e-01 5.11217117e-01 -3.54004890e-01 -1.92581430e-01
4.56441134e-01 -4.21757698e-02 -5.59618533e-01 9.00392532e-02
3.25106412e-01 -2.32274622e-01 -3.26440215e-01 1.12874532e+00
4.72709626e-01 7.54371583e-02 6.33644104e-01 -8.90925527e-01
-5.97186565e-01 5.72549105e-01 -5.00794113e-01 -2.68230975e-01
3.16085033e-02 -4.99444842e-01 4.55984145e-01 -1.14847481e+00
5.31512439e-01 1.47863007e+00 8.46886516e-01 4.07988369e-01
-7.21362293e-01 -6.02449834e-01 7.53509104e-02 3.03367883e-01
-1.24057245e+00 -2.31486991e-01 1.06070757e+00 -3.21352690e-01
6.94636643e-01 3.39268386e-01 8.61518383e-01 8.37848604e-01
-7.95773491e-02 1.04982305e+00 1.08120894e+00 -1.24230251e-01
2.50439137e-01 -1.05957106e-01 -2.65016139e-01 6.69498801e-01
-2.24659860e-01 -2.03531370e-01 -1.42901629e-01 2.96717644e-01
1.15700996e+00 3.74213666e-01 -2.76960939e-01 -9.05546069e-01
-1.35715652e+00 9.27415371e-01 1.07106936e+00 1.90884262e-01
-2.98491806e-01 1.32999662e-02 2.96372265e-01 -1.47241196e-02
6.76820099e-01 2.84229040e-01 -4.36291456e-01 3.16756397e-01
-7.52234817e-01 1.76285446e-01 6.75822914e-01 8.96687448e-01
6.47415459e-01 1.61595061e-01 5.18699409e-03 1.14409733e+00
3.07348222e-01 6.28886342e-01 7.15852022e-01 -7.87559688e-01
3.07338089e-01 6.62653148e-01 -2.51169235e-01 -1.16979098e+00
-7.63964236e-01 -5.94249606e-01 -1.11690426e+00 1.42618433e-01
2.44725838e-01 2.76548743e-01 -9.33658719e-01 1.37022233e+00
4.14769322e-01 2.30499193e-01 1.02874912e-01 1.17659783e+00
1.24909794e+00 6.64727986e-01 -7.47926652e-01 3.32733512e-01
1.38257051e+00 -1.11610270e+00 -1.27219766e-01 -4.23516184e-02
3.53234828e-01 -7.70684361e-01 9.45929766e-01 3.16879958e-01
-9.43762064e-01 -6.34703100e-01 -1.07805634e+00 -2.80719668e-01
-4.34049070e-01 3.48453522e-01 4.52481598e-01 4.68965292e-01
-9.05606687e-01 4.03080612e-01 -1.07893741e+00 -3.23734164e-01
6.83845043e-01 2.43399516e-01 -2.87422478e-01 -8.20262805e-02
-7.68306494e-01 5.84390223e-01 2.50506461e-01 2.86419004e-01
-8.17804635e-01 -4.99291152e-01 -1.06199431e+00 7.91334510e-02
5.42669669e-02 -6.84428990e-01 1.20882952e+00 -7.31023848e-01
-1.85813975e+00 9.36933100e-01 2.72542853e-02 -3.94331634e-01
5.32467127e-01 -1.21325620e-01 3.22627574e-02 1.32283568e-01
1.44033134e-01 6.46470964e-01 8.87750804e-01 -1.12985790e+00
-5.15154064e-01 -7.91720629e-01 1.92129433e-01 6.74949646e-01
-2.93698668e-01 -1.24108963e-01 -7.27799833e-01 -6.12663746e-01
8.69824946e-01 -1.12655342e+00 -2.82352388e-01 3.84844607e-03
-5.94016492e-01 -4.17965889e-01 9.44985032e-01 -3.93375069e-01
2.07569912e-01 -2.17416668e+00 2.45968759e-01 3.70011419e-01
1.61811128e-01 1.44952685e-01 3.06008477e-02 -3.73185068e-01
3.56877781e-02 -2.33558223e-01 -2.03088120e-01 -4.50414062e-01
-6.73493668e-02 9.20157880e-02 -1.70607939e-01 6.41357183e-01
-9.22773406e-02 1.02045608e+00 -5.51149428e-01 -1.72658384e-01
5.12719631e-01 7.79766321e-01 -5.14648855e-01 1.49064630e-01
-1.35025664e-04 5.10503352e-01 -4.00499284e-01 4.44789320e-01
9.67201829e-01 -2.29481697e-01 -2.44780749e-01 -4.49668765e-01
-2.46247604e-01 2.98860371e-01 -1.16278338e+00 1.93184519e+00
-6.34808302e-01 4.66635585e-01 -6.75579607e-02 -1.63739944e+00
1.27296543e+00 2.96815988e-02 3.84428203e-01 -8.07369351e-01
2.02928141e-01 3.30644310e-01 -3.83681834e-01 -3.19790453e-01
2.21243307e-01 2.60748595e-01 8.10686722e-02 1.15107767e-01
-2.24000458e-02 -6.89387262e-01 -3.27036083e-01 -2.41681591e-01
4.80909526e-01 1.31386220e-01 -5.06833531e-02 -3.00174028e-01
7.16759920e-01 -9.95579064e-02 4.10340667e-01 3.75930101e-01
4.24445897e-01 1.08194077e+00 2.15094656e-01 -7.66814709e-01
-1.13842654e+00 -1.07565796e+00 -2.80836225e-01 9.40231860e-01
3.39809597e-01 1.20613553e-01 -9.44556534e-01 -4.07941192e-01
-3.15195471e-01 2.78499186e-01 -5.34650743e-01 2.59257909e-02
-6.47694886e-01 -8.07211995e-01 4.40594077e-01 6.88182235e-01
9.65553701e-01 -1.08327079e+00 -6.47911668e-01 7.72898719e-02
-1.74368382e-01 -1.40225911e+00 -2.79387027e-01 1.92315102e-01
-9.25162017e-01 -8.77252221e-01 -1.11718035e+00 -1.20269525e+00
6.73642695e-01 7.15366006e-01 7.01661289e-01 -5.50226510e-01
3.47432159e-02 1.43106058e-01 -2.41356984e-01 -5.08413315e-01
2.38918886e-01 2.25927934e-01 -5.92852794e-02 1.79331928e-01
1.58577457e-01 -7.75483429e-01 -6.29389346e-01 5.21897495e-01
-8.65995765e-01 3.55841428e-01 5.61295748e-01 7.23301053e-01
8.45601737e-01 -1.90240577e-01 1.33703634e-01 -8.04397106e-01
3.55508059e-01 -1.66871011e-01 -6.48904622e-01 -7.62834102e-02
-7.18801618e-02 -4.67152754e-03 8.17856848e-01 -4.00026888e-01
-9.94465470e-01 2.73550123e-01 -4.59262162e-01 -6.25304461e-01
-2.31809407e-01 4.24080521e-01 -1.78319961e-01 -2.73310989e-01
5.45579374e-01 6.85854018e-01 4.46822383e-02 -6.70618057e-01
3.98441374e-01 6.42570376e-01 5.56830883e-01 -2.91091383e-01
7.54929721e-01 9.88455296e-01 2.92884801e-02 -1.01093948e+00
-8.83443236e-01 -3.63130063e-01 -5.89136958e-01 2.40451209e-02
1.13260055e+00 -1.03162527e+00 -6.96401298e-01 8.21111500e-01
-1.32482100e+00 -2.52804041e-01 -2.34856546e-01 6.17074549e-01
-7.48788774e-01 2.29190245e-01 -5.17650306e-01 -3.47123802e-01
-4.15236384e-01 -1.52731431e+00 1.29078031e+00 3.98704022e-01
3.75929981e-01 -7.50467777e-01 -1.14611916e-01 3.43785733e-01
4.47700202e-01 8.26552138e-02 6.36560023e-01 -7.00591028e-01
-5.79247117e-01 -3.61044183e-02 -4.77320075e-01 3.87997657e-01
-1.93324089e-01 -4.70929295e-01 -1.01853657e+00 -8.90020579e-02
1.65157840e-01 -4.21469659e-01 8.52945030e-01 6.97038770e-01
1.80877757e+00 -1.78805068e-01 -2.84328699e-01 1.39457846e+00
1.09732938e+00 1.88383251e-01 6.55713260e-01 6.07346356e-01
1.05585444e+00 5.69060683e-01 1.39832288e-01 1.97462291e-01
7.16087818e-01 7.51012325e-01 6.49227381e-01 -3.07239026e-01
3.27509157e-02 -1.60798535e-01 3.64345759e-02 1.05146492e+00
-2.74177432e-01 1.72128156e-01 -6.43519819e-01 3.99694443e-01
-1.73908997e+00 -6.33855879e-01 -6.57054260e-02 1.96661913e+00
4.78413314e-01 4.76364605e-02 -2.16828719e-01 1.73115283e-01
3.78617376e-01 1.70782059e-01 -6.04341209e-01 -1.63464203e-01
-1.18131697e-01 3.69006723e-01 5.26501477e-01 2.11427405e-01
-1.22613597e+00 9.35512900e-01 5.00658512e+00 9.98558044e-01
-1.54568279e+00 -5.07865148e-03 4.60492790e-01 2.26838917e-01
-2.48720452e-01 -4.50406283e-01 -6.89267933e-01 2.51233757e-01
1.26503676e-01 3.36804718e-01 4.89955187e-01 1.16933656e+00
2.95323767e-02 4.47111696e-01 -9.12530601e-01 1.35141170e+00
3.32693785e-01 -1.40877867e+00 1.67060450e-01 -2.76287824e-01
7.97826767e-01 4.55052018e-01 2.92135596e-01 1.05505802e-01
1.08034626e-01 -1.10362756e+00 9.92822707e-01 3.95992696e-01
6.89234078e-01 -6.70193553e-01 7.80962408e-01 4.83412057e-01
-1.25688958e+00 2.77916372e-01 -7.18277335e-01 1.91526040e-01
-1.17487043e-01 5.93809068e-01 -8.04576039e-01 6.82225466e-01
1.17639697e+00 9.10841107e-01 -4.66132253e-01 1.02350867e+00
-1.57117754e-01 1.05391577e-01 -7.66151845e-01 -2.15227231e-01
4.19740498e-01 -5.12693644e-01 4.15275216e-01 9.20260608e-01
5.76886594e-01 -2.75547188e-02 1.30702347e-01 8.61433446e-01
-7.94457719e-02 2.28611633e-01 -7.05597937e-01 4.99927849e-01
2.05326170e-01 1.42833734e+00 -1.01089597e+00 -3.40425879e-01
-5.05377889e-01 9.23323572e-01 1.93857342e-01 4.66179699e-01
-6.87795222e-01 -5.68093836e-01 3.34018469e-01 -3.01270634e-02
5.74663341e-01 -3.63365114e-01 -4.66548502e-01 -1.34248877e+00
1.70307294e-01 -7.02780426e-01 -1.29810259e-01 -9.99067962e-01
-9.21037018e-01 8.97207379e-01 -2.27019206e-01 -1.28313303e+00
-1.61996353e-02 -8.57469976e-01 -5.93461812e-01 6.20590985e-01
-1.51004064e+00 -1.32395709e+00 -6.38070881e-01 1.00003052e+00
5.86236835e-01 -3.57882082e-01 5.14846742e-01 1.78598315e-01
-1.12745464e-01 3.49379659e-01 2.56428838e-01 3.95354241e-01
3.23665857e-01 -1.22595751e+00 4.24376637e-01 6.07051551e-01
3.27466965e-01 4.25906599e-01 1.79880172e-01 -2.07805075e-02
-1.27080703e+00 -1.31422198e+00 2.66099751e-01 -1.30654443e-02
2.89302409e-01 -5.94798684e-01 -6.02679193e-01 5.89256465e-01
-2.00571522e-01 1.70912832e-01 2.73633599e-01 -1.62523821e-01
-3.49493325e-01 -3.24449241e-01 -9.22520459e-01 6.13215387e-01
1.15785086e+00 -4.64302093e-01 -4.61970240e-01 4.63308424e-01
7.72958934e-01 -8.42952251e-01 -6.53240919e-01 6.47902250e-01
6.14765882e-01 -1.07416701e+00 1.32910430e+00 -1.95844442e-01
4.05570447e-01 -3.57212126e-01 -3.21622014e-01 -1.52675629e+00
-1.39111979e-02 -5.34730516e-02 3.45421195e-01 6.40008032e-01
2.71763712e-01 -9.22639191e-01 9.17675912e-01 -4.62165885e-02
-6.10770822e-01 -9.15789843e-01 -1.01789916e+00 -5.61220646e-01
-7.20885256e-03 -5.77968061e-01 8.28919113e-01 9.19483304e-01
-5.04022002e-01 3.66883427e-01 -2.19022203e-02 2.11621016e-01
7.12451994e-01 5.18085361e-01 9.48696196e-01 -1.29421556e+00
-6.17460795e-02 -5.54526567e-01 -6.05228305e-01 -1.90583217e+00
2.47533061e-02 -1.08609974e+00 1.14653990e-01 -1.54359066e+00
5.39194830e-02 -6.90547049e-01 1.12179786e-01 3.15359175e-01
3.57041925e-01 7.24509358e-01 1.72572136e-01 2.26119906e-01
-4.95374858e-01 8.67430508e-01 1.64024007e+00 -2.39317983e-01
-2.30714396e-01 2.79618055e-01 -4.80851740e-01 1.09726095e+00
6.65288150e-01 -1.47861749e-01 -4.03101414e-01 -8.40483904e-01
2.30469912e-01 -1.75361812e-01 4.89734501e-01 -1.11614621e+00
3.90637487e-01 2.81897813e-01 6.60318613e-01 -9.59727645e-01
5.41439474e-01 -7.81409979e-01 -1.45242333e-01 2.23189130e-01
-1.42048016e-01 -2.16866702e-01 -7.75287971e-02 3.15118045e-01
-3.92666757e-01 -6.06151372e-02 8.99053514e-01 -2.37485141e-01
-4.21984673e-01 6.31862760e-01 -9.20135304e-02 -1.57812670e-01
8.28073323e-01 -2.56593794e-01 4.83481809e-02 -2.85214961e-01
-7.42226541e-01 7.66047686e-02 4.27035719e-01 5.14999628e-01
9.86144304e-01 -1.58279419e+00 -5.17374039e-01 6.24968588e-01
-5.25517166e-02 9.14269984e-01 2.51150522e-02 8.48959863e-01
-8.82463396e-01 5.57983160e-01 -2.91484952e-01 -1.21205854e+00
-9.96079624e-01 2.59692967e-01 5.36419809e-01 1.15763899e-02
-7.56482542e-01 1.00688541e+00 8.13112378e-01 -1.23540723e+00
2.31385574e-01 -7.25772142e-01 -6.03084564e-01 -1.72036543e-01
2.59916365e-01 1.16957508e-01 2.28451222e-01 -5.64792871e-01
-3.04837257e-01 1.09416926e+00 2.17167005e-01 -7.45023787e-02
1.64785790e+00 1.33068204e-01 -2.69682258e-01 3.17651004e-01
1.50909412e+00 -3.65798086e-01 -1.21353126e+00 -4.72733140e-01
-4.30704743e-01 -3.11045259e-01 4.12090905e-02 -8.77660289e-02
-1.45405221e+00 1.14377725e+00 5.57830513e-01 2.65013516e-01
1.06185293e+00 1.52827755e-01 7.45951474e-01 5.09840310e-01
3.56345415e-01 -7.28105187e-01 6.95836917e-03 7.64139712e-01
1.04884434e+00 -1.06416702e+00 -3.02071840e-01 -4.66553599e-01
-3.53321791e-01 1.29741848e+00 4.77593213e-01 -5.12242436e-01
8.42733085e-01 6.74196035e-02 7.51250656e-03 -1.31511688e-01
1.53512629e-02 -3.92069593e-02 4.39913422e-01 6.18953288e-01
1.19984329e-01 1.76193863e-01 2.08802018e-02 6.56812489e-01
-5.91488242e-01 -2.45195895e-01 2.76589245e-01 4.28836823e-01
-3.29025388e-01 -5.95497131e-01 -4.95173126e-01 4.29135978e-01
-2.59177178e-01 2.10775454e-02 -9.06356499e-02 4.57589239e-01
1.70569018e-01 3.61797392e-01 5.14899671e-01 -2.96035647e-01
4.76700187e-01 -3.14131588e-01 4.89034265e-01 -4.72663879e-01
-2.68763993e-02 2.41705149e-01 -2.76936978e-01 -6.74097359e-01
-7.56625772e-01 -5.68076491e-01 -1.14020956e+00 -3.56121175e-02
-1.02898315e-01 -4.81680222e-02 1.06826162e+00 8.94040287e-01
4.66557480e-02 5.69492936e-01 7.55960703e-01 -1.38154280e+00
-5.54892480e-01 -9.52278435e-01 -3.93202990e-01 3.15008044e-01
4.63073254e-02 -7.69296229e-01 -1.56142324e-01 -1.75058879e-02] | [8.081743240356445, -3.342715263366699] |
5e01ae74-5b00-47ca-85af-03de165494e3 | evaluation-of-chatgpt-as-a-question-answering | 2303.07992 | null | https://arxiv.org/abs/2303.07992v1 | https://arxiv.org/pdf/2303.07992v1.pdf | Evaluation of ChatGPT as a Question Answering System for Answering Complex Questions | ChatGPT is a powerful large language model (LLM) that has made remarkable progress in natural language understanding. Nevertheless, the performance and limitations of the model still need to be extensively evaluated. As ChatGPT covers resources such as Wikipedia and supports natural language question answering, it has garnered attention as a potential replacement for traditional knowledge based question answering (KBQA) models. Complex question answering is a challenge task of KBQA, which comprehensively tests the ability of models in semantic parsing and reasoning. To assess the performance of ChatGPT as a question answering system (QAS) using its own knowledge, we present a framework that evaluates its ability to answer complex questions. Our approach involves categorizing the potential features of complex questions and describing each test question with multiple labels to identify combinatorial reasoning. Following the black-box testing specifications of CheckList proposed by Ribeiro et.al, we develop an evaluation method to measure the functionality and reliability of ChatGPT in reasoning for answering complex questions. We use the proposed framework to evaluate the performance of ChatGPT in question answering on 8 real-world KB-based CQA datasets, including 6 English and 2 multilingual datasets, with a total of approximately 190,000 test cases. We compare the evaluation results of ChatGPT, GPT-3.5, GPT-3, and FLAN-T5 to identify common long-term problems in LLMs. The dataset and code are available at https://github.com/tan92hl/Complex-Question-Answering-Evaluation-of-ChatGPT. | ['Guilin Qi', 'Yongrui Chen', 'Nan Hu', 'Wenbo Li', 'Yu Li', 'Dehai Min', 'Yiming Tan'] | 2023-03-14 | null | null | null | null | ['semantic-parsing'] | ['natural-language-processing'] | [-6.28801525e-01 3.23509425e-01 3.17731470e-01 -4.89803880e-01
-1.27291751e+00 -7.57998168e-01 3.35291982e-01 2.67650425e-01
-3.32025290e-01 6.93376184e-01 1.97063029e-01 -5.59935510e-01
-5.32046139e-01 -1.02392936e+00 -4.92041051e-01 3.84010859e-02
3.05453658e-01 1.14839995e+00 6.58606529e-01 -7.76340723e-01
2.49240249e-01 -1.84663102e-01 -1.23111951e+00 8.32069755e-01
1.40814996e+00 8.08592439e-01 2.04306364e-01 9.40767527e-01
-8.45317304e-01 1.67258859e+00 -9.85358715e-01 -9.07633245e-01
-2.85660267e-01 -4.12014067e-01 -1.94954944e+00 -5.42731285e-01
3.48621160e-01 -4.09796722e-02 -2.58486476e-02 6.18583441e-01
5.55294454e-01 -8.96273777e-02 1.91990316e-01 -1.44623756e+00
-8.58784437e-01 5.88731229e-01 4.59469825e-01 2.86736488e-01
9.83310282e-01 2.02912316e-01 1.21834338e+00 -5.38454711e-01
6.62031174e-01 1.51592636e+00 6.12897694e-01 5.67936718e-01
-5.56749284e-01 -3.18703979e-01 -3.73001993e-01 6.64591610e-01
-1.08782399e+00 -2.61582509e-02 9.33236107e-02 -1.08943246e-01
1.51661634e+00 4.97558773e-01 3.47455055e-01 4.84692574e-01
1.66879877e-01 9.86274421e-01 1.22087634e+00 -5.85045934e-01
2.79035985e-01 9.69417840e-02 6.83426559e-01 9.65812027e-01
-2.51144499e-01 -6.55201256e-01 -4.28330064e-01 -4.66838002e-01
1.00229330e-01 -5.77257335e-01 -3.34539488e-02 2.58612841e-01
-7.99158573e-01 9.68051493e-01 3.93363267e-01 4.32799488e-01
-4.93377745e-02 1.15174398e-01 3.72706681e-01 7.93157518e-01
2.64765054e-01 8.43875349e-01 -9.63724852e-01 -4.58000004e-01
-1.48877099e-01 6.25479162e-01 1.34459257e+00 9.20943320e-01
7.77047515e-01 -7.21646249e-01 -6.74280047e-01 1.13013124e+00
2.29844689e-01 5.83907127e-01 4.17019844e-01 -1.35522389e+00
8.64499509e-01 1.26968181e+00 4.84537929e-02 -5.97456455e-01
-4.81748879e-01 3.16520408e-02 -1.01787500e-01 -5.43561280e-01
5.59570670e-01 -1.95043758e-01 -4.04754579e-01 1.51013231e+00
3.07136953e-01 -1.82188928e-01 5.68231940e-01 5.78816831e-01
1.41728199e+00 7.78275430e-01 3.71892571e-01 2.77740061e-01
1.73523998e+00 -1.26452768e+00 -7.79752493e-01 -1.79198742e-01
1.26524222e+00 -7.91710973e-01 1.53777397e+00 1.35877021e-02
-1.04060936e+00 -2.95266360e-01 -2.81564176e-01 -4.81231779e-01
-6.48933947e-01 -1.64264262e-01 4.62931901e-01 6.38421655e-01
-1.23181057e+00 -7.04743490e-02 -2.40596265e-01 -5.87453127e-01
8.94089118e-02 -1.31576378e-02 -1.71416461e-01 -7.08200336e-01
-1.74108374e+00 1.19756222e+00 4.15739179e-01 -8.76064450e-02
-7.34955966e-01 -6.50554538e-01 -7.40213633e-01 2.30235502e-01
7.04739630e-01 -9.80544508e-01 1.76390564e+00 -2.10176200e-01
-1.36270249e+00 6.23618960e-01 -1.78238437e-01 -2.57919759e-01
1.66424304e-01 -3.59610915e-01 -5.31840980e-01 3.72906357e-01
5.96784949e-01 6.48022056e-01 -9.46072116e-02 -7.48475015e-01
-5.47013581e-01 -2.29037851e-01 8.47270668e-01 3.39001507e-01
-1.97515450e-02 3.41658205e-01 -6.99899793e-01 4.38866690e-02
-2.76396424e-01 -6.24580920e-01 -1.01283357e-01 -4.25905377e-01
-2.28564084e-01 -9.04785633e-01 6.17312670e-01 -8.00691843e-01
1.23293233e+00 -1.67191327e+00 -3.15541387e-01 -9.89318639e-02
-3.82954517e-04 3.95166963e-01 -5.15303016e-01 1.02217674e+00
4.85741436e-01 4.04493958e-01 -2.37363800e-01 3.63301188e-02
1.03665158e-01 6.81201994e-01 -7.18046650e-02 -5.12359262e-01
5.53801537e-01 1.39941692e+00 -1.00216043e+00 -7.58147001e-01
-6.72215521e-02 -3.90250385e-02 -7.95438051e-01 4.74087983e-01
-8.80362809e-01 3.04083526e-01 -8.95648658e-01 7.54165709e-01
3.96039188e-01 -4.89276558e-01 2.88314298e-02 3.26523989e-01
1.95688859e-01 7.32255459e-01 -8.94280374e-01 1.70503414e+00
-5.61075330e-01 1.78777412e-01 -2.54168004e-01 -6.65167451e-01
7.73215353e-01 5.98452032e-01 -6.20722026e-02 -1.35045373e+00
-1.49903417e-01 5.10395527e-01 -4.88881245e-02 -1.19126856e+00
4.00657147e-01 8.03162679e-02 -3.95871997e-01 4.08088177e-01
3.38968217e-01 -5.86451530e-01 6.27652287e-01 5.17724395e-01
1.66513991e+00 -3.05071205e-01 1.28119126e-01 -3.92829418e-01
9.51589644e-01 6.08029544e-01 9.18012559e-02 8.50177109e-01
-1.36104941e-01 2.04524308e-01 8.43965590e-01 -2.09235281e-01
-4.08737242e-01 -8.81126344e-01 -4.50038463e-02 1.08175683e+00
-1.34603381e-01 -6.33450568e-01 -9.10782278e-01 -8.98401141e-01
-1.04921512e-01 9.34989095e-01 -3.55476737e-01 -3.46137248e-02
-5.74608743e-01 -4.87684667e-01 1.13750517e+00 1.83391944e-01
1.00252986e+00 -1.55282807e+00 -4.03360963e-01 2.18234584e-01
-9.11795020e-01 -1.34943557e+00 8.28400478e-02 -2.31510520e-01
-7.19056845e-01 -1.62030911e+00 -2.12775826e-01 -8.22147131e-01
2.04102829e-01 -7.50131309e-02 1.79702652e+00 5.24715185e-01
-1.61163598e-01 1.05066156e+00 -9.10677016e-01 -2.83508390e-01
-7.10061371e-01 2.30067000e-01 -6.29916072e-01 -6.79252684e-01
5.11461079e-01 1.01427145e-01 -4.48744416e-01 5.25544703e-01
-9.02999103e-01 -2.70441651e-01 3.65566939e-01 6.53851628e-01
1.50504977e-01 3.66832837e-02 1.01492202e+00 -1.03433692e+00
1.31053841e+00 -7.55155504e-01 -3.92801851e-01 7.95075297e-01
-3.98299783e-01 -2.10963115e-02 5.27400911e-01 1.78255200e-01
-1.30118704e+00 -8.79455030e-01 -8.31328154e-01 3.31180543e-01
-1.78144678e-01 9.81651425e-01 -7.40135312e-02 4.13830727e-02
6.42077446e-01 -7.45981038e-02 -3.57767612e-01 -4.59338516e-01
5.93462288e-01 6.29625797e-01 3.23991925e-01 -1.15193772e+00
4.28022236e-01 -1.11474864e-01 -5.99180639e-01 -5.73814988e-01
-1.13341439e+00 -8.23430240e-01 -6.02856018e-02 -2.22460285e-01
9.96112883e-01 -7.00848222e-01 -1.17775762e+00 3.71086508e-01
-1.25449955e+00 -5.44868052e-01 -3.23038608e-01 6.64705411e-02
-4.63259667e-01 2.93473572e-01 -9.57158029e-01 -6.60348117e-01
-6.20185196e-01 -1.03640199e+00 1.02040398e+00 1.77299947e-01
-2.40887448e-01 -1.14107084e+00 4.38695341e-01 1.32025659e+00
4.87876624e-01 -2.81829327e-01 1.53123605e+00 -8.54249179e-01
-4.82698470e-01 -5.58453947e-02 -2.23160863e-01 5.33892632e-01
-2.50516236e-01 -4.11992460e-01 -9.07722354e-01 4.34345705e-03
1.31460026e-01 -9.69066620e-01 1.79962084e-01 -3.15488167e-02
9.80449021e-01 -2.51167566e-01 3.34964544e-01 -1.70315936e-01
1.54462492e+00 2.09078744e-01 8.68089199e-01 6.50609672e-01
3.61841887e-01 8.23553264e-01 8.03983629e-01 -1.23695776e-01
1.05586839e+00 2.29861259e-01 2.65384853e-01 4.23289686e-01
-5.02127968e-03 -2.08621308e-01 2.88325101e-01 1.20893371e+00
2.16302633e-01 -5.11929929e-01 -1.51686180e+00 5.94942749e-01
-1.73508680e+00 -5.26463449e-01 -5.77510595e-01 1.56407475e+00
8.99388075e-01 -1.63470238e-01 -2.99071133e-01 -3.07323728e-02
1.13360688e-01 -2.41220891e-01 -2.77480274e-01 -7.15476573e-01
-9.99987572e-02 5.75269222e-01 -3.26390058e-01 6.42149448e-01
-3.59660417e-01 1.14045143e+00 6.05048895e+00 9.03422594e-01
-2.80750394e-01 3.39693516e-01 4.34919626e-01 5.88832319e-01
-4.54566836e-01 2.91912347e-01 -6.58213615e-01 1.59390405e-01
1.35613549e+00 -4.83420901e-02 2.49420822e-01 5.10758698e-01
-7.67321512e-02 -4.37130868e-01 -7.57511854e-01 3.29500914e-01
-9.34263691e-02 -1.32878697e+00 3.52259547e-01 -6.09514892e-01
5.20376265e-01 2.01739922e-01 -2.93180674e-01 1.13407457e+00
4.92810339e-01 -9.98523355e-01 4.06939119e-01 4.60716188e-01
2.84461379e-01 -3.85586441e-01 1.25108802e+00 6.02394879e-01
-9.80884373e-01 -2.70601839e-01 -3.66193324e-01 -2.59759605e-01
8.78898650e-02 2.74298668e-01 -1.03849709e+00 9.78754699e-01
1.12019324e+00 6.23759739e-02 -1.03799105e+00 9.56797361e-01
-5.45746446e-01 9.97465134e-01 -1.91277415e-01 -2.37896428e-01
4.29536939e-01 -6.42865822e-02 3.96587066e-02 9.86894131e-01
1.38973743e-01 4.44352150e-01 8.69746320e-03 7.80657947e-01
-1.24214292e-01 6.03912175e-01 -1.02340847e-01 -1.20319933e-01
5.18911302e-01 1.01758015e+00 -3.50837529e-01 -4.91488039e-01
-7.08308697e-01 6.68328881e-01 7.10198641e-01 2.60781437e-01
-5.43356180e-01 -2.37707466e-01 1.74915448e-01 -1.72509104e-01
-1.32875368e-01 2.29619324e-01 2.21206948e-01 -1.04097283e+00
4.27383006e-01 -1.24797857e+00 1.03076077e+00 -1.36867237e+00
-1.51323521e+00 6.94690585e-01 1.49996355e-01 -4.88757879e-01
-3.16760182e-01 -5.95545590e-01 -5.84111810e-01 9.61300910e-01
-1.64406621e+00 -1.03407335e+00 -4.16700810e-01 9.43363428e-01
5.29297113e-01 5.16599081e-02 1.12503099e+00 5.19857049e-01
-2.91983724e-01 3.16607207e-01 -2.52474636e-01 2.27157503e-01
5.71107566e-01 -1.36683607e+00 5.01408458e-01 3.08184624e-01
-1.02425300e-01 5.69611073e-01 3.39674205e-01 -5.80344439e-01
-1.33118165e+00 -9.35440183e-01 1.30777764e+00 -1.05271339e+00
8.12204421e-01 -1.58234611e-01 -1.28773177e+00 6.36342466e-01
3.96605313e-01 -3.65422130e-01 7.21892834e-01 2.74113268e-01
-3.68946284e-01 7.99272060e-02 -1.28914869e+00 2.32773721e-01
7.17310727e-01 -8.52626860e-01 -1.14054918e+00 5.93074322e-01
1.03952432e+00 -3.24463934e-01 -1.20009804e+00 4.02588964e-01
-7.56999925e-02 -9.84233022e-01 7.85325527e-01 -7.04575777e-01
3.91531944e-01 -2.74218738e-01 -2.11507663e-01 -1.09248412e+00
1.62057623e-01 -1.84048831e-01 2.39160717e-01 1.26437223e+00
6.52702928e-01 -9.67571914e-01 5.60743570e-01 7.20680773e-01
-2.60079741e-01 -7.88638711e-01 -1.12866950e+00 -5.67281425e-01
4.91139680e-01 -5.84342301e-01 6.54194891e-01 9.69318032e-01
1.75671820e-02 6.53161228e-01 2.69283086e-01 3.03912491e-01
-4.20097215e-03 -2.75337212e-02 6.59560144e-01 -1.10345066e+00
-2.56523430e-01 -1.24124028e-01 -5.64798340e-02 -8.14527988e-01
9.50255617e-02 -9.44977462e-01 -1.25142574e-01 -2.21710491e+00
1.41591772e-01 -5.95651031e-01 1.63531899e-01 6.61466658e-01
-3.91737282e-01 -2.33883813e-01 1.38226748e-01 -5.68778254e-03
-1.05426383e+00 5.54589987e-01 1.40233076e+00 -6.30778745e-02
1.80320263e-01 -1.66254595e-01 -6.01012051e-01 4.07546580e-01
9.66752470e-01 -5.78674197e-01 -7.76961148e-01 -7.49144316e-01
7.89892197e-01 4.60541040e-01 3.03686291e-01 -9.15211558e-01
3.33658040e-01 5.96953295e-02 -4.61358637e-01 -3.62547249e-01
8.09217468e-02 -5.63997984e-01 -2.40462840e-01 4.46537137e-01
-3.45065325e-01 4.68598366e-01 4.17138964e-01 2.56452151e-02
-7.03037739e-01 -7.37762988e-01 1.76478788e-01 -4.80952710e-01
-9.70644355e-01 -9.12861601e-02 -4.63717699e-01 8.93955290e-01
5.34421265e-01 3.87793958e-01 -1.01236534e+00 -4.56788749e-01
-4.00603414e-01 1.06590366e+00 -1.28012553e-01 5.63640773e-01
6.06053770e-01 -9.35139000e-01 -9.04452741e-01 -3.63792926e-01
5.82049489e-01 1.02558583e-01 6.11168206e-01 5.91323078e-01
-9.69290555e-01 8.93114388e-01 1.43356407e-02 -4.51717347e-01
-9.43006098e-01 2.49164268e-01 6.51806474e-01 -7.71147430e-01
-2.56110460e-01 7.85501122e-01 -1.55362830e-01 -1.45070744e+00
-1.73873797e-01 -5.47670066e-01 -4.76077229e-01 -3.60565484e-01
3.59395385e-01 3.89406741e-01 3.11643064e-01 -1.50175929e-01
-2.58903712e-01 3.55286747e-01 1.38856292e-01 -4.25141267e-02
9.30637777e-01 -5.77194616e-02 -6.61633790e-01 2.47740671e-01
1.05556834e+00 -2.08183125e-01 -1.36614842e-02 -2.25701958e-01
4.54424471e-01 -6.15157075e-02 -4.93869543e-01 -1.44214356e+00
-5.95401406e-01 6.49368882e-01 3.58073533e-01 4.32818592e-01
8.97059798e-01 4.73039329e-01 7.60812581e-01 1.10378814e+00
5.45030534e-01 -8.94368410e-01 3.15953851e-01 1.05590522e+00
9.70768213e-01 -1.21647096e+00 -6.74498498e-01 -4.37069088e-01
-5.41545510e-01 7.87072122e-01 9.25838530e-01 3.46187115e-01
3.57728362e-01 -1.78702652e-01 5.47592938e-01 -8.80544901e-01
-1.03264880e+00 -2.91160047e-01 1.32456124e-01 3.07333857e-01
4.38387156e-01 1.22359348e-02 -3.87718797e-01 5.40414453e-01
-5.22449255e-01 9.45505574e-02 3.41281354e-01 1.15499640e+00
-3.81255060e-01 -1.18968916e+00 -3.67155761e-01 4.05512214e-01
-4.67336059e-01 -2.87063450e-01 -4.57931697e-01 1.02561820e+00
-6.58198521e-02 1.62551415e+00 -3.20621997e-01 -1.15017019e-01
8.52925718e-01 5.17886758e-01 1.12165011e-01 -8.15789878e-01
-9.89403546e-01 -7.90701628e-01 7.64403641e-01 -5.61310172e-01
-5.56036234e-01 -1.75695747e-01 -1.38060701e+00 -2.33120307e-01
-2.16670275e-01 1.01201904e+00 3.12611639e-01 1.23071158e+00
3.23420316e-01 3.96064222e-01 -5.11128418e-02 6.86911345e-01
-3.53803694e-01 -1.16501141e+00 -2.51740851e-02 3.53694707e-01
-1.24804966e-01 -3.33112836e-01 -1.52815536e-01 -3.72636169e-01] | [10.607220649719238, 7.974024772644043] |
c19b2e26-6470-403f-b53e-bf2ab98b9a79 | representation-learning-to-classify-and | 2107.04448 | null | https://arxiv.org/abs/2107.04448v1 | https://arxiv.org/pdf/2107.04448v1.pdf | Representation Learning to Classify and Detect Adversarial Attacks against Speaker and Speech Recognition Systems | Adversarial attacks have become a major threat for machine learning applications. There is a growing interest in studying these attacks in the audio domain, e.g, speech and speaker recognition; and find defenses against them. In this work, we focus on using representation learning to classify/detect attacks w.r.t. the attack algorithm, threat model or signal-to-adversarial-noise ratio. We found that common attacks in the literature can be classified with accuracies as high as 90%. Also, representations trained to classify attacks against speaker identification can be used also to classify attacks against speaker verification and speech recognition. We also tested an attack verification task, where we need to decide whether two speech utterances contain the same attack. We observed that our models did not generalize well to attack algorithms not included in the attack representation model training. Motivated by this, we evaluated an unknown attack detection task. We were able to detect unknown attacks with equal error rates of about 19%, which is promising. | ['Najim Dehak', 'Piotr Żelasko', 'Sonal Joshi', 'Jesús Villalba'] | 2021-07-09 | null | null | null | null | ['speaker-identification'] | ['speech'] | [ 4.68851209e-01 3.97478119e-02 2.19214156e-01 -2.72529632e-01
-1.05056834e+00 -1.10218692e+00 6.96997285e-01 2.73092836e-01
-1.67229310e-01 2.62729466e-01 1.22742960e-02 -6.98282421e-01
6.70415834e-02 -6.96890593e-01 -6.04819894e-01 -6.66465700e-01
-3.59355271e-01 2.03554422e-01 1.85727775e-01 -3.81540716e-01
2.39519492e-01 8.72456670e-01 -1.50865304e+00 5.25366426e-01
3.49777907e-01 8.53593588e-01 -5.99681497e-01 1.03394186e+00
1.97731569e-01 7.76556075e-01 -1.35504484e+00 -3.53823155e-01
2.81274736e-01 -3.27285588e-01 -7.52355754e-01 -2.82593459e-01
4.82454598e-01 -1.23271868e-01 -4.23845440e-01 1.18881774e+00
6.44496858e-01 -9.98651683e-02 6.05124176e-01 -1.50532126e+00
-2.53402553e-02 8.85571361e-01 -1.62007347e-01 5.50051928e-01
7.74500310e-01 4.48795557e-02 7.48723447e-01 -3.83115083e-01
1.02157801e-01 1.35852718e+00 3.19745272e-01 8.53146672e-01
-1.17410243e+00 -1.11294663e+00 4.87842523e-02 3.70685071e-01
-1.39501631e+00 -6.20486438e-01 8.24390531e-01 -2.64370829e-01
7.00677514e-01 6.94375336e-01 -9.31015611e-02 1.58756447e+00
7.08168894e-02 5.20236254e-01 9.86611724e-01 -5.84183514e-01
2.61677921e-01 2.85226941e-01 7.28677884e-02 2.29453772e-01
-4.41992562e-03 6.22394025e-01 -3.12648118e-01 -6.77752137e-01
5.81214204e-02 -4.91427779e-01 -4.33155507e-01 2.57261962e-01
-7.95933962e-01 1.11146033e+00 2.59072423e-01 7.26805747e-01
-1.64589226e-01 2.39536092e-02 7.91879356e-01 8.16489458e-01
3.09627742e-01 7.54468620e-01 -3.92937124e-01 -1.49555773e-01
-7.16505051e-01 6.15841663e-03 1.16265881e+00 3.03278893e-01
2.33676240e-01 5.64118564e-01 1.30782783e-01 6.38206065e-01
-7.90240243e-03 6.22033060e-01 3.75741631e-01 -3.24300319e-01
4.55491275e-01 -1.61391959e-01 -2.23320305e-01 -1.02227211e+00
-3.77146125e-01 -4.30703163e-01 -6.83963478e-01 2.94223726e-01
5.09482920e-01 -3.21165919e-01 -6.16403461e-01 1.82490861e+00
1.51944339e-01 4.54665840e-01 4.00166690e-01 5.62506020e-01
1.70680121e-01 7.02247441e-01 -3.03511079e-02 -3.60248387e-01
1.16994905e+00 -2.50814497e-01 -7.04958856e-01 -1.69470519e-01
8.27115119e-01 -1.14067423e+00 7.18539715e-01 7.98278332e-01
-8.06972802e-01 -3.56732428e-01 -1.28267086e+00 8.20230007e-01
-4.52711314e-01 -2.26550385e-01 3.14247817e-01 1.68271077e+00
-7.15775073e-01 5.16156852e-01 -6.42131388e-01 -1.97863355e-02
3.37907337e-02 4.41738188e-01 -3.35783243e-01 1.57382146e-01
-1.60537946e+00 1.04426551e+00 6.03991076e-02 -2.70841867e-01
-1.29356420e+00 -3.64221573e-01 -7.28230953e-01 1.47462130e-01
2.92845607e-01 2.56411135e-02 1.42703927e+00 -8.59991074e-01
-1.44751525e+00 7.28441238e-01 1.37419924e-01 -1.04661679e+00
2.15116858e-01 -3.81819010e-02 -1.18715072e+00 1.90563604e-01
-4.18149740e-01 -3.22225094e-01 1.35508251e+00 -9.51824307e-01
-3.60259265e-01 -4.00482565e-01 2.75166035e-01 -3.79955620e-01
-6.06507778e-01 6.94020987e-01 5.93365312e-01 -8.50556195e-01
-9.53828637e-03 -1.12192225e+00 5.31418398e-02 -4.28921700e-01
-4.49947387e-01 1.21907227e-01 1.23934364e+00 -5.29540658e-01
1.13662088e+00 -2.31855059e+00 4.00740765e-02 5.43323636e-01
-2.32986256e-01 6.50062799e-01 -2.14216590e-01 5.86779833e-01
-6.35442317e-01 3.00719410e-01 -4.05106843e-02 -9.98063907e-02
8.31939280e-02 2.70342510e-02 -1.12538111e+00 7.26009488e-01
1.50606930e-01 1.60650209e-01 -4.29813474e-01 1.35874301e-01
1.98966652e-01 4.73027557e-01 -4.20317739e-01 3.80840063e-01
4.02590260e-02 3.14830005e-01 -4.08225775e-01 4.98363435e-01
5.19254565e-01 5.04312038e-01 8.19858909e-02 1.15440466e-01
2.70515025e-01 5.50922215e-01 -1.40773821e+00 7.59558678e-01
-8.40657949e-01 7.80731976e-01 3.48691344e-01 -1.27918267e+00
1.07040370e+00 8.02491307e-01 1.34408325e-01 -3.11441302e-01
3.21367025e-01 2.64483660e-01 5.43401897e-01 -3.33001614e-02
-1.00628145e-01 -2.65557855e-01 -3.33441734e-01 5.95227778e-01
1.05604924e-01 -2.84718484e-01 -1.76944658e-01 1.18647046e-01
1.29081821e+00 -8.96183431e-01 2.62055039e-01 -8.70570466e-02
9.78766620e-01 -6.38648748e-01 1.02312215e-01 9.20733631e-01
-9.47508961e-02 2.55300105e-01 5.17972291e-01 -2.25703984e-01
-4.27602500e-01 -1.06059813e+00 -7.42061585e-02 1.19652915e+00
-2.41703123e-01 -5.02277136e-01 -7.96112835e-01 -8.07285905e-01
-3.57451975e-01 8.23764622e-01 -4.81798917e-01 -6.80159450e-01
-5.61646283e-01 -4.67680454e-01 1.38996756e+00 4.58080232e-01
5.10969311e-02 -9.14577901e-01 -2.15977415e-01 2.70725042e-01
1.73817307e-01 -1.36698437e+00 -2.95774519e-01 4.62383598e-01
-4.39648479e-01 -1.21520555e+00 -3.47275853e-01 -5.79019248e-01
2.06088409e-01 -6.72041699e-02 7.70916522e-01 5.78553230e-02
-4.13540423e-01 5.25423825e-01 -5.77950001e-01 -8.31523001e-01
-1.20637214e+00 -4.48415056e-02 6.44529045e-01 2.64091462e-01
1.08427264e-01 -5.95128179e-01 6.78739995e-02 7.73904800e-01
-1.00729597e+00 -1.05395246e+00 1.62302703e-01 6.06729507e-01
-2.31981844e-01 4.71194267e-01 7.07443297e-01 -7.25938797e-01
8.15330386e-01 -3.50168645e-01 -5.90344429e-01 1.30359873e-01
-1.70954932e-02 5.83271682e-02 9.00267720e-01 -1.12502480e+00
-4.91189867e-01 4.61397804e-02 -7.85820067e-01 -3.58745277e-01
-4.84943092e-01 3.19333196e-01 -4.75362986e-01 -5.10412037e-01
1.15792084e+00 1.77090853e-01 -1.49398237e-01 -2.18990967e-01
1.41631976e-01 8.56220007e-01 2.07709804e-01 -5.90276837e-01
1.20576549e+00 9.77487117e-02 -8.95622820e-02 -1.16735518e+00
-5.76663435e-01 -3.39217901e-01 4.07700762e-02 -7.78375864e-02
4.18112457e-01 -5.45982957e-01 -8.65661979e-01 5.91708601e-01
-1.20550549e+00 8.30457956e-02 1.05559357e-01 6.01743639e-01
-4.85692799e-01 6.36324227e-01 -6.81001902e-01 -1.24534106e+00
-3.54148835e-01 -1.20690465e+00 7.00327933e-01 -3.36775303e-01
-3.34634393e-01 -7.82637417e-01 -3.37705053e-02 2.04591572e-01
7.15025544e-01 1.33537769e-01 7.83361197e-01 -1.49535525e+00
9.05235037e-02 -6.83677852e-01 3.84409934e-01 6.43344879e-01
2.53856570e-01 -3.82456556e-02 -1.52695334e+00 -5.42065442e-01
5.32840014e-01 -4.45942581e-01 6.03694916e-01 -6.88064173e-02
1.23951483e+00 -5.30340135e-01 -2.81098820e-02 2.31011912e-01
7.35131085e-01 5.34402132e-01 8.38816345e-01 1.46787916e-03
2.05092117e-01 7.02073574e-01 3.76632541e-01 3.11339319e-01
-7.18428969e-01 1.09110761e+00 5.12237549e-01 2.40427062e-01
2.92023987e-01 -2.56156344e-02 7.25507140e-01 2.73509800e-01
2.58608133e-01 -5.25223792e-01 -1.07342815e+00 1.34635240e-01
-1.12827289e+00 -1.24407995e+00 1.16506867e-01 2.36433911e+00
4.78886604e-01 6.42383814e-01 3.00821871e-01 1.11598158e+00
8.50419581e-01 2.77567089e-01 -8.94878060e-02 -8.49882662e-01
-3.12495679e-02 6.24026775e-01 3.62364143e-01 5.32601833e-01
-1.37933981e+00 9.11561728e-01 6.19867468e+00 8.30204606e-01
-1.39411891e+00 -6.98294863e-02 2.15221867e-01 3.12447309e-01
2.17643097e-01 -1.58884183e-01 -6.36752844e-01 2.16692612e-01
1.48358202e+00 -2.27188021e-01 4.39582139e-01 7.90508389e-01
-2.25887150e-01 6.68310344e-01 -1.12013102e+00 8.58168721e-01
2.59107411e-01 -9.25464332e-01 1.07679620e-01 2.02860877e-01
7.19049871e-02 -3.61376554e-01 2.98889071e-01 4.48756039e-01
2.58691400e-01 -1.15046370e+00 3.79978150e-01 -3.30979943e-01
4.07013476e-01 -1.05168450e+00 7.38203824e-01 5.55813491e-01
-1.04307961e+00 -2.41047040e-01 -1.22082099e-01 -1.20999984e-01
5.90988137e-02 2.87519455e-01 -1.19796133e+00 4.50746804e-01
3.54333311e-01 1.90036163e-01 -2.80404329e-01 7.36148834e-01
-2.55667508e-01 1.10658610e+00 -4.75811630e-01 -7.36299083e-02
1.15415089e-01 5.19549072e-01 7.00214148e-01 1.10464334e+00
2.95892090e-01 -7.03066513e-02 2.67473698e-01 2.15410218e-01
9.54078585e-02 8.85767415e-02 -9.90069389e-01 -3.77631485e-01
3.59533161e-01 7.86893189e-01 -4.55139667e-01 1.35466099e-01
-2.80329809e-02 8.42703819e-01 -2.22133636e-01 4.46379855e-02
-9.76835191e-01 -5.61598599e-01 8.99677336e-01 6.60863072e-02
1.59256309e-01 -1.32920220e-01 2.07629114e-01 -1.05004227e+00
-1.57271996e-01 -1.47539377e+00 5.57065725e-01 -2.62275159e-01
-1.33717883e+00 9.83652592e-01 -5.84339835e-02 -1.28414464e+00
-6.90110445e-01 -7.13816941e-01 -8.69992316e-01 8.11556697e-01
-9.40068662e-01 -8.43471825e-01 4.06111002e-01 8.40246081e-01
3.96517634e-01 -6.08484805e-01 1.28506458e+00 2.52131611e-01
-3.62036794e-01 1.05028534e+00 -3.46661329e-01 4.33559209e-01
5.54284155e-01 -7.76431859e-01 6.52523994e-01 9.33579624e-01
8.30302119e-01 5.65821826e-01 9.45492625e-01 -1.61320791e-01
-1.26138139e+00 -8.55419993e-01 5.58506548e-01 -4.04493123e-01
1.00838459e+00 -6.45630658e-01 -1.12051189e+00 4.85849500e-01
-1.30300960e-02 -3.99667807e-02 9.43344772e-01 1.66006178e-01
-9.85061705e-01 1.01544924e-01 -1.41059411e+00 4.25763398e-01
6.11440659e-01 -9.49721992e-01 -6.70719981e-01 4.54580277e-01
6.82725430e-01 -2.40255550e-01 -5.23584306e-01 3.89655560e-01
2.48787045e-01 -6.77890122e-01 1.28195822e+00 -9.03794706e-01
-2.49308720e-01 -1.32354677e-01 -5.35035849e-01 -1.32776093e+00
1.79568961e-01 -7.03839958e-01 -2.06367031e-01 1.29191172e+00
5.77877641e-01 -1.00354016e+00 5.40891767e-01 1.64844111e-01
1.56291634e-01 -1.03711694e-01 -1.27972031e+00 -1.16278172e+00
2.12090835e-01 -8.21115673e-01 5.24647117e-01 1.08699906e+00
3.32411639e-02 2.96369970e-01 -5.17137110e-01 9.32024837e-01
4.76643384e-01 -4.09043044e-01 7.80501008e-01 -1.23885477e+00
-6.04467154e-01 -3.94859046e-01 -9.48373199e-01 -5.27975559e-01
6.19610548e-01 -7.55756915e-01 -2.25183547e-01 -4.41420168e-01
-5.51691711e-01 -1.46147236e-01 -3.04724246e-01 4.42268342e-01
1.17450297e-01 2.60166168e-01 2.33580038e-01 -1.68585718e-01
6.55316338e-02 2.45089024e-01 3.05084944e-01 -4.40295875e-01
2.83653066e-02 7.66249299e-01 -5.43782771e-01 7.83029735e-01
1.24016964e+00 -7.14969158e-01 -2.60013074e-01 8.76661837e-02
-1.84689388e-01 1.66883931e-01 1.86856404e-01 -1.25846541e+00
7.86158517e-02 6.09161705e-02 -9.31610763e-02 -3.71061452e-02
4.91038769e-01 -1.02453494e+00 -2.12597325e-01 6.48603022e-01
-6.01558089e-01 -1.01873122e-01 5.08377433e-01 5.22850811e-01
-3.70034277e-01 -3.88693333e-01 9.74487543e-01 3.12356085e-01
-1.96756512e-01 -8.62067267e-02 -7.09993601e-01 -6.25911728e-02
1.08460915e+00 1.22260399e-01 -3.19679230e-01 -7.67054915e-01
-9.13495898e-01 -2.88946569e-01 1.80628747e-02 6.20047748e-01
6.93271160e-01 -8.76285791e-01 -8.82087469e-01 4.34667021e-01
2.35244825e-01 -8.70113134e-01 1.17009290e-01 2.77402818e-01
-2.45454893e-01 3.05328965e-01 -2.87647685e-03 -4.31907773e-01
-1.98981500e+00 9.91732061e-01 4.52893585e-01 -1.11094601e-01
-1.33913547e-01 6.62902832e-01 -1.78411379e-02 -4.15602684e-01
6.12151563e-01 -3.07678804e-02 -9.49429944e-02 -9.63559449e-02
9.85409081e-01 1.27325788e-01 3.91163617e-01 -7.31854498e-01
-6.59801364e-01 3.91533434e-01 -2.44483382e-01 -2.38957360e-01
8.78886223e-01 6.05275571e-01 3.06477040e-01 2.03606829e-01
1.40711343e+00 3.36708575e-01 -2.86534399e-01 -9.96013656e-02
2.14911461e-01 -5.12618661e-01 -7.95752835e-03 -5.67420483e-01
-9.55048680e-01 1.14036238e+00 8.26486707e-01 1.03267395e+00
1.19830084e+00 -2.80406941e-02 4.32414263e-01 5.33883333e-01
6.82734251e-01 -4.76757437e-01 8.01462755e-02 6.22755766e-01
1.10345924e+00 -9.28317487e-01 -2.95795918e-01 -5.34942567e-01
-5.37926078e-01 1.09316802e+00 2.47444078e-01 -1.47194028e-01
6.98610365e-01 5.82138598e-01 2.20939934e-01 1.49403498e-01
-5.97509563e-01 -1.15859592e-02 2.56295264e-01 8.91292572e-01
3.15394431e-01 -2.70631886e-03 7.73293339e-03 3.91745269e-01
-4.02972788e-01 -7.69921660e-01 7.29314864e-01 8.93430591e-01
-3.53415340e-01 -1.49669814e+00 -8.84965479e-01 1.25800550e-01
-7.72405207e-01 -3.70894149e-02 -8.89512122e-01 3.90699178e-01
-4.42794293e-01 1.45901024e+00 -3.41757238e-01 -9.16563749e-01
4.64756399e-01 1.85738176e-01 2.63511151e-01 -7.58531392e-01
-9.59550381e-01 -2.62905627e-01 2.83481777e-01 -4.30494100e-01
-7.92383626e-02 -5.17637849e-01 -8.18328261e-01 -2.78509766e-01
-6.39951885e-01 5.69171488e-01 8.91716957e-01 6.97772801e-01
-4.21413928e-02 3.82149845e-01 1.27245986e+00 -6.03650153e-01
-1.10964370e+00 -9.59462345e-01 -6.53949499e-01 3.15891087e-01
5.47396958e-01 -4.80744481e-01 -1.04027319e+00 -2.90869445e-01] | [13.957962989807129, 5.807216167449951] |
68dc4fb4-1755-4e35-b765-63335f91f8c1 | recognition-of-handwritten-digit-using | 1909.08490 | null | https://arxiv.org/abs/1909.08490v1 | https://arxiv.org/pdf/1909.08490v1.pdf | Recognition of Handwritten Digit using Convolutional Neural Network in Python with Tensorflow and Comparison of Performance for Various Hidden Layers | In recent times, with the increase of Artificial Neural Network (ANN), deep learning has brought a dramatic twist in the field of machine learning by making it more artificially intelligent. Deep learning is remarkably used in vast ranges of fields because of its diverse range of applications such as surveillance, health, medicine, sports, robotics, drones, etc. In deep learning, Convolutional Neural Network (CNN) is at the center of spectacular advances that mixes Artificial Neural Network (ANN) and up to date deep learning strategies. It has been used broadly in pattern recognition, sentence classification, speech recognition, face recognition, text categorization, document analysis, scene, and handwritten digit recognition. The goal of this paper is to observe the variation of accuracies of CNN to classify handwritten digits using various numbers of hidden layers and epochs and to make the comparison between the accuracies. For this performance evaluation of CNN, we performed our experiment using Modified National Institute of Standards and Technology (MNIST) dataset. Further, the network is trained using stochastic gradient descent and the backpropagation algorithm. | ['Fathma Siddique', 'Md. Abu Bakr Siddique', 'Shadman Sakib'] | 2019-09-12 | null | null | null | null | ['handwritten-digit-recognition'] | ['computer-vision'] | [ 1.56872153e-01 -4.19384569e-01 -4.59707044e-02 -5.37650764e-01
4.22991306e-01 -1.36139929e-01 8.09825778e-01 1.86684132e-01
-5.61104476e-01 6.07737243e-01 -6.98440615e-03 -4.86061037e-01
-2.16203466e-01 -8.45668018e-01 -4.62298840e-01 -5.60296118e-01
4.00043055e-02 5.70889190e-02 1.78314909e-01 -3.68685603e-01
5.70855677e-01 8.57172787e-01 -1.40864873e+00 4.51644272e-01
2.92828351e-01 1.15696228e+00 -1.57359123e-01 8.62983704e-01
-1.81706220e-01 1.42193127e+00 -6.42809570e-01 -4.50992912e-01
-4.27427609e-03 -3.41600567e-01 -9.07918274e-01 1.71142355e-01
6.76488504e-02 1.89121030e-02 -5.61555088e-01 9.55085278e-01
5.43494523e-01 1.54656783e-01 5.18714428e-01 -8.37352276e-01
-7.25226700e-01 4.02087927e-01 -4.70900714e-01 5.83684862e-01
-3.96308722e-03 7.37811550e-02 3.19676578e-01 -8.08692217e-01
3.48396868e-01 1.10274839e+00 9.26171899e-01 5.94575405e-01
-4.67179179e-01 -5.00291407e-01 -2.56666958e-01 2.53944367e-01
-8.95586371e-01 -2.29518682e-01 8.07075143e-01 -3.76135677e-01
1.01475477e+00 1.52886048e-01 4.03790683e-01 9.49186802e-01
3.62389386e-01 6.50146902e-01 9.26233053e-01 -7.59876132e-01
1.39146239e-01 2.26676628e-01 2.56633490e-01 8.21910739e-01
1.06472615e-02 1.55413345e-01 -4.29165095e-01 2.90635943e-01
5.93907237e-01 3.61237496e-01 1.08907767e-01 2.04999223e-01
-9.41556036e-01 7.81351745e-01 5.07619441e-01 5.49145997e-01
-3.82719159e-01 8.40066001e-02 5.94930112e-01 5.50150454e-01
2.25940570e-01 3.97117674e-01 -5.75379968e-01 -1.99466676e-01
-5.03161550e-01 5.43125197e-02 7.40491986e-01 3.30785543e-01
2.88637906e-01 5.72621226e-01 5.12587316e-02 1.31659806e+00
2.09043577e-01 6.86768368e-02 1.19708335e+00 -4.95942295e-01
3.94049227e-01 9.27443743e-01 -4.22405750e-01 -1.29537642e+00
-5.33559859e-01 -5.56697309e-01 -1.40039873e+00 3.17954123e-01
2.89893836e-01 -3.76839668e-01 -9.99473035e-01 1.03910804e+00
7.71313980e-02 -3.86990868e-02 2.44994000e-01 7.40893126e-01
8.04522336e-01 8.21407735e-01 1.18214972e-02 2.63994634e-01
1.10017216e+00 -8.35001469e-01 -3.76910895e-01 -4.23839778e-01
2.87843585e-01 -8.02749097e-01 6.54031813e-01 6.59147739e-01
-6.54534876e-01 -8.50724101e-01 -1.12640452e+00 1.28477588e-01
-7.90536821e-01 1.69578835e-01 8.96273315e-01 7.07103312e-01
-5.90057790e-01 8.52334678e-01 -5.73292673e-01 -7.13783145e-01
6.25064552e-01 5.76372981e-01 -2.67240465e-01 -1.16887651e-02
-9.92480516e-01 8.51157248e-01 6.86968505e-01 3.30350190e-01
-4.74330634e-01 -7.26468954e-03 -6.30843580e-01 7.53549635e-02
-1.14900999e-01 -1.67071700e-01 1.05582297e+00 -1.36346877e+00
-1.63891697e+00 9.19737518e-01 4.10636336e-01 -7.46113241e-01
1.64255500e-01 1.19032040e-02 -7.28813887e-01 -3.01930547e-01
-6.92142606e-01 6.81373298e-01 8.28865707e-01 -4.97217804e-01
-4.76369351e-01 -5.48343122e-01 -2.46558368e-01 -5.30986153e-02
-5.93562365e-01 2.15771616e-01 2.82409489e-01 -7.34497488e-01
1.51823685e-01 -7.49577224e-01 1.13656603e-01 -1.96252421e-01
-2.20480174e-01 -4.20393676e-01 1.12460911e+00 -3.72603089e-01
7.31069565e-01 -2.46002197e+00 -1.44097075e-01 1.13124877e-01
-1.94107190e-01 6.13890767e-01 1.49439499e-01 4.73887473e-01
-8.73080119e-02 -3.07300538e-02 -2.91769743e-01 2.88416669e-02
-3.54324847e-01 3.66252691e-01 -1.09586306e-01 2.94847369e-01
4.47923928e-01 7.22202182e-01 -5.25409758e-01 -1.11255690e-01
3.51142257e-01 4.59166080e-01 -3.94505188e-02 1.59514835e-03
-1.13641463e-01 2.89296001e-01 -2.22579464e-01 6.31963789e-01
2.74842262e-01 -1.14870138e-01 -6.34095073e-02 9.52643752e-02
-1.75629690e-01 -2.30083484e-02 -1.00791299e+00 1.30948842e+00
-2.13261917e-01 1.20691574e+00 -3.26914549e-01 -1.44395506e+00
1.29177272e+00 3.07448268e-01 1.01406813e-01 -7.43707597e-01
4.75137293e-01 1.47149235e-01 3.65088224e-01 -8.29542100e-01
3.14287037e-01 -7.97568038e-02 3.77973825e-01 1.63660720e-01
3.49244922e-01 1.33056805e-01 -3.79502773e-04 -4.49731112e-01
8.22587132e-01 -4.80999023e-01 9.05698836e-02 -3.07644289e-02
8.14477682e-01 -5.15548848e-02 2.46592209e-01 3.35933000e-01
-2.70279884e-01 4.46187824e-01 2.99499482e-01 -1.06588924e+00
-1.16099310e+00 -4.55129027e-01 -1.66083813e-01 9.69095290e-01
-4.07312930e-01 3.84941667e-01 -4.33976501e-01 -2.40332574e-01
-1.16914146e-01 2.14303374e-01 -5.14837861e-01 -2.86353827e-01
-6.81829214e-01 -9.68402505e-01 9.01285946e-01 5.43079078e-01
1.32790005e+00 -1.52978885e+00 -6.72191083e-01 1.52014807e-01
6.18922830e-01 -1.18936181e+00 3.71954232e-01 2.47567773e-01
-1.25593483e+00 -1.08010352e+00 -8.56475294e-01 -1.27744293e+00
4.32179511e-01 -3.75533491e-01 8.06519330e-01 1.78168684e-01
-5.51587820e-01 4.30551618e-02 -4.50404584e-01 -7.29267895e-01
-4.97266978e-01 3.64304543e-01 2.00691298e-02 1.18464753e-01
6.03585720e-01 -4.14514929e-01 -3.99043202e-01 -4.82939705e-02
-1.11437535e+00 -1.56640634e-01 7.85783827e-01 9.05330956e-01
-1.35773405e-01 4.69127029e-01 5.99053860e-01 -9.56481695e-01
9.49987054e-01 -3.88170868e-01 -4.53067929e-01 9.40632150e-02
-4.90359157e-01 1.96004380e-02 7.65498102e-01 -3.84988278e-01
-7.80514002e-01 -2.89423525e-01 -4.89098936e-01 2.95720637e-01
-5.64784765e-01 6.86104834e-01 3.79959822e-01 -2.99160719e-01
8.77139211e-01 4.31830257e-01 5.95988967e-02 -5.56242943e-01
-3.75205576e-01 1.19285154e+00 5.99253058e-01 -4.68007363e-02
2.98367351e-01 1.20289750e-01 2.21531130e-02 -1.09721076e+00
-6.61396563e-01 -1.16963819e-01 -5.07177055e-01 -2.92588264e-01
8.37186277e-01 -2.64453739e-01 -7.29217947e-01 1.08630645e+00
-1.20026016e+00 -2.83878356e-01 1.25797257e-01 5.93621850e-01
2.13348806e-01 1.22681387e-01 -6.66315317e-01 -7.89847672e-01
-4.06729728e-01 -8.08849335e-01 3.56859714e-01 5.49103320e-01
5.17248213e-02 -1.11447704e+00 6.64674938e-02 9.04894695e-02
6.63141310e-01 3.97011608e-01 1.12506080e+00 -7.62573957e-01
-3.92648168e-02 -7.58255720e-01 -1.22305281e-01 1.08384442e+00
1.50153056e-01 1.58246160e-01 -1.00557721e+00 8.67680088e-02
1.42138852e-02 -3.87309045e-01 8.00200164e-01 8.29686448e-02
1.61618102e+00 -3.98700118e-01 1.32014841e-01 5.12966394e-01
1.57040989e+00 5.73882282e-01 7.98645854e-01 7.09553599e-01
5.29892981e-01 5.54466188e-01 -3.24255884e-01 3.10928226e-01
-9.29848626e-02 1.26911730e-01 3.62591922e-01 9.78575647e-02
1.30787432e-01 3.11089873e-01 5.00190519e-02 7.89700270e-01
-2.62466282e-01 -1.28063709e-01 -1.28373420e+00 2.08759800e-01
-1.42781663e+00 -9.36964512e-01 9.98682976e-02 1.93878651e+00
5.97042978e-01 3.37752461e-01 -1.63748205e-01 8.80882263e-01
6.15505576e-01 -9.58587043e-03 -3.98171574e-01 -9.76807177e-01
-6.59749880e-02 3.79370570e-01 3.39885920e-01 1.19508870e-01
-1.10899878e+00 7.17473209e-01 5.78228474e+00 4.94640112e-01
-1.93627656e+00 -2.33046189e-01 8.47779930e-01 5.22805274e-01
6.19841456e-01 -8.31638753e-01 -6.80947363e-01 4.47779775e-01
1.01001358e+00 3.53679597e-01 4.76301461e-01 7.99821079e-01
-8.18976387e-02 -1.68862835e-01 -8.76113057e-01 9.95736241e-01
3.36342752e-02 -1.44824386e+00 1.96664736e-01 -3.46515715e-01
7.59027123e-01 1.82409793e-01 4.42560971e-01 5.42693853e-01
-1.88096687e-01 -1.30570436e+00 1.17709629e-01 3.16420972e-01
3.09541196e-01 -9.30804372e-01 1.17845166e+00 6.23073101e-01
-4.42640930e-01 -5.70458710e-01 -4.24558491e-01 -6.50559545e-01
-5.45823276e-01 4.41811800e-01 -7.12544739e-01 1.29629016e-01
7.63864100e-01 5.05428374e-01 -4.83467221e-01 1.02276313e+00
1.35418624e-01 7.51143396e-01 -6.14539944e-02 -5.06490767e-01
6.99376702e-01 2.56300010e-02 1.32674202e-01 1.23034132e+00
-7.02528507e-02 2.25063264e-01 -2.38593191e-01 3.05940688e-01
-5.18733859e-01 1.48730176e-02 -5.67004323e-01 -4.92251992e-01
1.54110745e-01 9.95705903e-01 -9.46689606e-01 -3.48568529e-01
-2.56540149e-01 7.57485986e-01 7.34825805e-02 2.06867710e-01
-4.26186770e-01 -7.99777091e-01 3.22768599e-01 1.12765832e-02
9.29950774e-02 -5.85252881e-01 -6.33682966e-01 -7.65243053e-01
-1.78281382e-01 -7.95374274e-01 1.17048495e-01 -5.46037734e-01
-1.00631726e+00 7.96894968e-01 -3.56039882e-01 -8.66894782e-01
-2.67911732e-01 -1.19383693e+00 -8.53511393e-01 5.39813161e-01
-1.39249468e+00 -5.13492763e-01 -5.88216841e-01 5.30038953e-01
8.44953597e-01 -7.88765848e-01 7.82458007e-01 4.73031998e-01
-8.51780474e-01 4.05538827e-01 4.18996811e-01 9.22834933e-01
2.36156628e-01 -8.87667060e-01 2.78592706e-01 6.98426723e-01
1.50626093e-01 2.69140244e-01 5.43287814e-01 -8.73412192e-02
-1.25107634e+00 -1.01107085e+00 7.97036707e-01 2.00580150e-01
3.46034855e-01 -1.95953965e-01 -7.02046156e-01 4.30764645e-01
3.02324504e-01 9.81528386e-02 5.50866544e-01 -1.76919907e-01
-1.91812962e-02 -4.77288842e-01 -1.19092023e+00 3.13253343e-01
1.52440056e-01 -2.78803021e-01 -3.85712802e-01 3.72634262e-01
2.03681856e-01 -4.19270635e-01 -4.94140744e-01 5.96794128e-01
6.14513636e-01 -1.19691956e+00 7.27972269e-01 -7.99172878e-01
8.05956662e-01 1.99169561e-01 -1.16380509e-02 -1.02509296e+00
-1.28250867e-01 -1.02215968e-01 -8.75825733e-02 9.23723459e-01
4.20916587e-01 -7.38044083e-01 1.04089785e+00 3.39347064e-01
-1.02157257e-01 -1.27998805e+00 -6.71258450e-01 -5.79938948e-01
9.86952484e-02 -2.59933025e-01 4.09896702e-01 1.08592474e+00
-3.99241060e-01 3.30153584e-01 -2.22275004e-01 -1.85878202e-01
2.70273268e-01 -4.32734638e-01 3.97734612e-01 -1.47014165e+00
-1.81165799e-01 -5.66985190e-01 -1.00326025e+00 -7.25595593e-01
-2.77556479e-01 -5.17497182e-01 -3.04541260e-01 -1.41197491e+00
-2.91688085e-01 -3.62432122e-01 -3.12806070e-01 3.57244074e-01
3.90407622e-01 4.12024349e-01 6.26535155e-03 1.14626542e-01
-2.44385079e-01 3.30539912e-01 9.92689431e-01 -3.22858661e-01
-1.67486921e-01 2.85959423e-01 -3.72236520e-01 9.64466214e-01
1.13320875e+00 -3.66305560e-01 7.96837173e-03 -7.86059558e-01
3.12350899e-01 -4.90259945e-01 2.47162372e-01 -1.50453866e+00
3.98428589e-01 1.85880408e-01 1.02510774e+00 -1.13215603e-01
1.48863435e-01 -8.41586769e-01 -3.66432548e-01 8.00684869e-01
-5.19570589e-01 1.22524768e-01 3.63984823e-01 1.57343760e-01
-6.30892277e-01 -4.88091081e-01 9.47928727e-01 -2.73060471e-01
-1.02661347e+00 2.13203043e-01 -6.53852701e-01 -3.29235673e-01
9.32291329e-01 -5.10430813e-01 -1.08556926e-01 9.88454279e-03
-5.16215920e-01 -9.92373452e-02 -1.90334707e-01 5.17019868e-01
7.83038795e-01 -1.02027810e+00 -5.36227524e-01 3.57185423e-01
-2.85390317e-01 3.60827036e-02 -3.45660411e-02 6.47390842e-01
-1.11276233e+00 5.70561647e-01 -4.52732623e-01 -4.57056284e-01
-1.27903044e+00 2.35974669e-01 5.53212345e-01 -1.16047878e-02
-4.34535623e-01 1.00037241e+00 -7.09958971e-01 -5.55820942e-01
6.94005191e-01 -4.75948691e-01 -6.37051046e-01 -4.20098543e-01
5.99055409e-01 6.18102491e-01 2.88328111e-01 -1.66805089e-01
-3.06565523e-01 4.66156363e-01 -2.07607910e-01 2.74240136e-01
1.46847606e+00 5.11135101e-01 -1.81343958e-01 5.79255700e-01
1.20996165e+00 -5.39597929e-01 -7.37759888e-01 8.90188888e-02
3.90198022e-01 3.92313972e-02 1.65708959e-01 -9.72682714e-01
-1.02235115e+00 1.19197941e+00 7.49908030e-01 5.73000729e-01
1.01324534e+00 -7.14148819e-01 7.94518948e-01 1.07821548e+00
8.83847326e-02 -1.13857245e+00 1.79720104e-01 9.18165207e-01
5.31115770e-01 -1.45576096e+00 -1.38794050e-01 2.98339903e-01
-5.25589526e-01 1.70157826e+00 4.54745054e-01 -4.55535531e-01
8.98681104e-01 2.56675571e-01 4.43688361e-03 -1.17704682e-01
-5.17649949e-01 2.44309247e-01 9.39237028e-02 4.02379364e-01
7.60735571e-01 -2.71654487e-01 -7.64215142e-02 1.51268959e-01
-1.36740401e-01 4.57558692e-01 4.39007312e-01 1.20881629e+00
-8.58785987e-01 -8.70459676e-01 -3.24981689e-01 6.04231954e-01
-9.32598412e-01 -7.11641312e-02 -4.56739008e-01 6.35733128e-01
1.87623009e-01 8.03614199e-01 1.37618706e-01 -4.43407893e-01
1.35063469e-01 1.71341538e-01 5.15462399e-01 -3.96675944e-01
-8.07098269e-01 -8.85230184e-01 -2.30615541e-01 1.74232662e-01
-4.54948634e-01 -8.43231156e-02 -1.16069889e+00 -5.46136081e-01
-1.05632640e-01 2.13737100e-01 1.26890957e+00 1.22829866e+00
1.12086155e-01 6.51215315e-01 6.86264038e-01 -6.41554534e-01
-5.49144983e-01 -1.32802975e+00 -3.76621187e-01 4.21988196e-04
2.08377555e-01 -1.86376125e-01 -1.47028029e-01 1.62521854e-01] | [11.600064277648926, 2.7044730186462402] |
e0d1ea22-2bd1-472f-8569-84bce7914f1a | bonsai-diverse-and-shallow-trees-for-extreme | 1904.08249 | null | https://arxiv.org/abs/1904.08249v2 | https://arxiv.org/pdf/1904.08249v2.pdf | Bonsai -- Diverse and Shallow Trees for Extreme Multi-label Classification | Extreme multi-label classification (XMC) refers to supervised multi-label learning involving hundreds of thousand or even millions of labels. In this paper, we develop a suite of algorithms, called Bonsai, which generalizes the notion of label representation in XMC, and partitions the labels in the representation space to learn shallow trees. We show three concrete realizations of this label representation space including : (i) the input space which is spanned by the input features, (ii) the output space spanned by label vectors based on their co-occurrence with other labels, and (iii) the joint space by combining the input and output representations. Furthermore, the constraint-free multi-way partitions learnt iteratively in these spaces lead to shallow trees. By combining the effect of shallow trees and generalized label representation, Bonsai achieves the best of both worlds - fast training which is comparable to state-of-the-art tree-based methods in XMC, and much better prediction accuracy, particularly on tail-labels. On a benchmark Amazon-3M dataset with 3 million labels, \bonsai outperforms a state-of-the-art one-vs-rest method in terms of prediction accuracy, while being approximately 200 times faster to train. The code for Bonsai is available at \url{https://github.com/xmc-aalto/bonsai} | ['Sujay Khandagale', 'Rohit Babbar', 'Han Xiao'] | 2019-04-17 | null | null | null | null | ['extreme-multi-label-classification'] | ['methodology'] | [ 2.84694880e-01 3.34796379e-04 -3.97889823e-01 -4.85116482e-01
-1.19181669e+00 -7.38121569e-01 5.11605322e-01 2.53029704e-01
-1.22869268e-01 5.63296556e-01 -1.98897552e-02 -4.58227664e-01
-1.70781255e-01 -6.37776494e-01 -4.11691785e-01 -8.00979614e-01
1.13054834e-01 1.04814076e+00 2.76429169e-02 1.24880694e-01
1.16421543e-01 -3.64282131e-02 -1.87950122e+00 7.31920779e-01
3.15153956e-01 1.39977181e+00 -2.65607923e-01 5.65776467e-01
-2.84014553e-01 8.51680934e-01 -2.91931421e-01 -3.96942914e-01
2.32860088e-01 -1.84153199e-01 -1.10582340e+00 1.00812286e-01
4.88142759e-01 2.48196051e-01 2.48617455e-01 6.60191894e-01
2.16261134e-01 4.46698442e-03 1.06334591e+00 -1.42036235e+00
-3.88655365e-01 6.11280024e-01 -8.48149896e-01 -3.12448889e-01
2.93468595e-01 -3.18414986e-01 1.54838681e+00 -1.11578214e+00
4.37137932e-01 1.24985456e+00 8.11636508e-01 6.83330953e-01
-1.54524481e+00 -7.64212966e-01 9.30777416e-02 -7.43137971e-02
-1.43044806e+00 -1.08495392e-01 2.58633405e-01 -7.22809672e-01
8.65978181e-01 3.92212778e-01 1.39013171e-01 8.93238246e-01
-8.08568820e-02 9.30289090e-01 1.56895518e+00 -8.00680876e-01
2.55548894e-01 2.09671721e-01 8.21595728e-01 8.53277862e-01
-6.52615577e-02 1.93855278e-02 -3.45973521e-01 -7.59319305e-01
6.51900694e-02 2.25036249e-01 2.05166772e-01 -3.30916435e-01
-1.16603744e+00 1.13750279e+00 2.46931538e-01 1.75482988e-01
-3.09158675e-02 4.11778629e-01 6.59079254e-01 1.91893727e-01
7.66809642e-01 1.94178998e-01 -7.86100447e-01 1.02255233e-01
-9.14514899e-01 1.34205014e-01 8.94017160e-01 1.01347315e+00
9.90026295e-01 -3.89249176e-01 -2.63392121e-01 1.18115175e+00
4.37855631e-01 3.23081821e-01 6.14496410e-01 -1.12605357e+00
4.62875903e-01 7.52626240e-01 -4.85937744e-02 -2.72150010e-01
-7.09749103e-01 -4.93030667e-01 -7.86062598e-01 1.59501374e-01
4.34462696e-01 -2.94113532e-02 -9.79312062e-01 1.62740695e+00
4.77808267e-01 1.36117488e-01 -1.50951862e-01 3.86558563e-01
7.53341138e-01 6.69623077e-01 2.68897325e-01 -1.50269642e-01
1.67548764e+00 -1.27549481e+00 -2.83674091e-01 -1.92128912e-01
1.33913946e+00 -5.16932368e-01 9.14516509e-01 3.74170750e-01
-6.41069114e-01 -4.84267712e-01 -7.72607505e-01 6.17710724e-02
-7.84942508e-01 1.47790894e-01 8.53383183e-01 7.56538928e-01
-1.04459655e+00 6.04960799e-01 -3.59111756e-01 -2.34703228e-01
3.37581754e-01 3.69935513e-01 -4.51295018e-01 -2.86806017e-01
-1.14469612e+00 8.26857328e-01 5.56162179e-01 -4.74310935e-01
-7.69141316e-01 -5.04739404e-01 -8.14498484e-01 -3.12138870e-02
4.00132567e-01 -3.62146765e-01 1.42622995e+00 -5.28060377e-01
-1.11656857e+00 1.28368819e+00 -2.62495041e-01 -6.98307157e-02
1.54689342e-01 -3.28533724e-02 -2.44236812e-01 -2.23185644e-01
4.03090596e-01 9.35504377e-01 4.90108848e-01 -1.47153115e+00
-1.03529322e+00 -4.10361558e-01 -6.96393624e-02 6.56422451e-02
-1.73324704e-01 1.18686661e-01 -2.32861280e-01 -5.36770999e-01
2.84577906e-01 -1.28756666e+00 -2.06898808e-01 -3.41284722e-01
-5.76660097e-01 -9.42067385e-01 5.62163770e-01 -6.41834959e-02
1.36502206e+00 -2.07072663e+00 -4.91653420e-02 1.90965876e-01
3.50571215e-01 -2.93646026e-02 -3.24940354e-01 5.23843944e-01
-4.02005136e-01 3.88821006e-01 -7.31975399e-03 -8.78955185e-01
2.47542694e-01 2.94636220e-01 -2.25556910e-01 5.18127501e-01
-2.36817956e-01 9.37050581e-01 -8.53252828e-01 -6.71851993e-01
1.71240065e-02 5.01996949e-02 -2.13730201e-01 -4.65934072e-03
-3.66600692e-01 3.94716591e-01 -3.22195888e-01 1.00347388e+00
4.78991956e-01 -7.73391485e-01 1.90682977e-01 3.16867471e-01
1.33651659e-01 2.63491958e-01 -1.18508708e+00 1.44174778e+00
-5.80358028e-01 4.77255285e-02 -3.72616708e-01 -1.00238490e+00
8.11578870e-01 6.46085262e-01 7.46860087e-01 -2.70120740e-01
1.49869666e-01 5.29083014e-01 -5.80198050e-01 2.84743514e-02
7.33563378e-02 -3.13940346e-01 -5.65814912e-01 1.01240790e+00
2.48135865e-01 1.31535217e-01 3.84392530e-01 1.82324976e-01
9.63716507e-01 1.44279554e-01 5.90372741e-01 -1.68116942e-01
3.95255148e-01 -2.59215146e-01 7.38497674e-01 8.31901312e-01
-5.26760286e-03 5.11104763e-01 5.12872219e-01 -6.51637435e-01
-7.27010012e-01 -8.59959245e-01 -3.83183420e-01 1.91103041e+00
-3.44285458e-01 -7.90068448e-01 -5.64939797e-01 -1.12600756e+00
3.20996433e-01 8.42330933e-01 -9.33899581e-01 3.01836044e-01
-2.71532863e-01 -6.23067856e-01 4.66276377e-01 4.42707360e-01
1.70406222e-01 -1.01416910e+00 -3.29164833e-01 1.72610730e-01
-2.47944817e-01 -8.55820954e-01 -3.37867141e-01 9.22901273e-01
-7.98715770e-01 -9.98885691e-01 -5.66948950e-01 -7.02953935e-01
4.07874644e-01 1.29120111e-01 1.39884508e+00 5.65626286e-02
-1.22410797e-01 1.68180004e-01 -6.27010286e-01 -2.67478526e-01
-4.47412431e-01 2.89476544e-01 -3.18048149e-02 4.35584635e-02
4.65608180e-01 -5.73370993e-01 -2.04126179e-01 5.32058537e-01
-5.62336445e-01 2.29769751e-01 3.49418104e-01 8.68107080e-01
7.59370685e-01 8.77885371e-02 6.02209091e-01 -1.62934101e+00
1.23196403e-02 -8.83439898e-01 -3.81597430e-01 5.83689451e-01
-8.99643898e-01 8.46516639e-02 4.70672399e-01 -4.06118870e-01
-4.75476295e-01 4.42624092e-01 -4.72324863e-02 -2.31354535e-01
-2.74981946e-01 3.08658361e-01 3.50119732e-02 2.03356951e-01
6.95951581e-01 -7.62470216e-02 -3.29147995e-01 -7.43896067e-01
8.08247268e-01 9.90532279e-01 1.05152085e-01 -8.39079022e-01
3.95441502e-01 2.19114929e-01 1.28954068e-01 -7.49559402e-02
-1.71335554e+00 -1.04858983e+00 -9.62956369e-01 -2.76902646e-01
8.62289548e-01 -8.38035941e-01 -6.48856759e-01 4.02204245e-01
-8.75764489e-01 -3.21022123e-01 -3.31256151e-01 1.17741965e-01
-7.18838871e-01 -8.72451521e-04 -8.31339717e-01 -7.85740733e-01
-3.80886525e-01 -1.15476394e+00 1.42099214e+00 -4.69067916e-02
-3.64283979e-01 -1.01252782e+00 1.44156054e-01 6.70155227e-01
8.38579461e-02 1.82452872e-01 1.33415723e+00 -1.03292716e+00
-1.57990202e-01 -3.52060348e-01 -2.76998788e-01 3.39021564e-01
-1.32762596e-01 -3.37106079e-01 -1.21723521e+00 -5.10783494e-01
-3.89379799e-01 -8.75381589e-01 1.16734791e+00 1.85790300e-01
1.18440771e+00 -6.99803308e-02 -6.95936203e-01 3.18806350e-01
1.58096468e+00 -1.31261140e-01 1.89736992e-01 2.97588885e-01
7.13738978e-01 6.53814077e-01 5.96101999e-01 5.57564259e-01
5.82495630e-01 8.11582208e-01 4.47283864e-01 1.29585741e-02
-4.27994728e-02 -1.26108393e-01 4.44336608e-02 7.94212282e-01
1.04269408e-01 -2.31003344e-01 -9.95331466e-01 2.98777968e-01
-2.03927302e+00 -6.22255206e-01 -2.62130409e-01 2.13688207e+00
1.02510357e+00 8.81633237e-02 2.98227042e-01 2.92289466e-01
7.18559206e-01 8.34102705e-02 -5.28639257e-01 -4.13835108e-01
1.78412974e-01 2.55704075e-01 4.97000039e-01 4.02182996e-01
-1.60872054e+00 8.96312714e-01 6.29249716e+00 1.22852528e+00
-5.24691582e-01 6.19450450e-01 8.09218585e-01 -1.19112626e-01
-1.10985175e-01 1.76604733e-01 -1.33196950e+00 4.39390302e-01
1.31390440e+00 3.45737576e-01 2.66001135e-01 9.62582231e-01
-5.91595232e-01 -1.28887281e-01 -1.34599531e+00 9.20674801e-01
5.85951768e-02 -9.99526858e-01 -1.94238856e-01 3.69870782e-01
8.94708335e-01 1.34553358e-01 1.13022983e-01 8.59496653e-01
8.54406357e-01 -1.05643773e+00 7.47417450e-01 3.23771596e-01
1.28082001e+00 -4.44180340e-01 7.08217084e-01 6.36559784e-01
-1.35702169e+00 -4.57797170e-01 -3.32963169e-01 -1.08897351e-02
-7.96913207e-02 8.88046026e-01 -4.90308911e-01 4.06531334e-01
5.51148653e-01 6.22947097e-01 -8.47502232e-01 7.43427515e-01
-1.86196223e-01 8.70522678e-01 -3.68847728e-01 6.82352707e-02
3.86329174e-01 -4.64541391e-02 -1.36634007e-01 1.41491616e+00
1.05355166e-01 1.69318803e-02 7.12217093e-01 3.61799210e-01
-1.29605606e-01 2.43834063e-01 -5.91315210e-01 3.65920752e-01
6.26163602e-01 1.49081635e+00 -8.76184702e-01 -5.18473923e-01
-6.04309261e-01 7.21270323e-01 7.53817558e-01 1.90786675e-01
-8.77736211e-01 6.27367496e-02 2.19974577e-01 -9.49309245e-02
2.45231926e-01 2.92020440e-01 -4.09280121e-01 -8.59516263e-01
-2.72884607e-01 -6.98827624e-01 8.41052413e-01 -4.21147853e-01
-1.47638023e+00 5.13290823e-01 -2.57848091e-02 -1.20305908e+00
-3.12831849e-01 -7.23697782e-01 -2.25534379e-01 8.27155828e-01
-1.19737709e+00 -1.55019927e+00 8.24374631e-02 2.86037475e-01
4.82011288e-01 -2.24587023e-01 1.50965047e+00 1.73897639e-01
-5.59534431e-01 7.35376716e-01 5.79482198e-01 -1.41033500e-01
6.27429605e-01 -1.53735924e+00 3.67661923e-01 -1.28753975e-01
4.99012560e-01 2.37824872e-01 3.05282883e-02 -3.98461401e-01
-5.05367875e-01 -1.33227479e+00 1.18928838e+00 -8.68817091e-01
5.22098541e-01 -5.84897757e-01 -5.93184769e-01 9.37075555e-01
-1.74471051e-01 3.05862457e-01 1.28004611e+00 6.41563177e-01
-1.00618744e+00 4.17619981e-02 -1.00570381e+00 5.01413010e-02
8.40161085e-01 -5.35447001e-01 -3.03178698e-01 8.25809836e-01
6.62370443e-01 -6.92769662e-02 -1.02996826e+00 5.37626922e-01
6.97421074e-01 -9.02318120e-01 8.90345097e-01 -6.09991193e-01
3.54878694e-01 6.71378616e-03 -6.33518457e-01 -1.06206059e+00
-7.74777234e-01 -4.04414386e-02 -1.89878896e-01 1.39605582e+00
8.04035723e-01 -6.99510932e-01 8.13996136e-01 2.68317103e-01
8.82456377e-02 -1.35061026e+00 -8.87468338e-01 -7.67994285e-01
5.33647537e-01 -5.02867579e-01 5.39453566e-01 9.47266281e-01
1.62008889e-02 6.87174737e-01 -2.94360459e-01 -3.02509427e-01
7.56634116e-01 4.72172767e-01 4.31112289e-01 -1.70582247e+00
-3.36645007e-01 -3.23046237e-01 -9.06542763e-02 -9.37936425e-01
5.06928504e-01 -1.50352359e+00 -4.74792384e-02 -1.53075206e+00
6.03776395e-01 -1.05935431e+00 -5.96232116e-01 1.08618164e+00
-1.67209908e-01 4.98737365e-01 3.27312589e-01 6.65853977e-01
-1.14560258e+00 1.15520291e-01 8.90511215e-01 -9.37082544e-02
3.01900655e-01 2.68761367e-01 -7.34460592e-01 6.87091410e-01
8.16739678e-01 -9.27087307e-01 -1.63187549e-01 -2.33942866e-02
2.35686317e-01 1.11089051e-01 -7.50866085e-02 -9.37871099e-01
5.88998124e-02 -3.94522846e-02 2.26099670e-01 -8.27148795e-01
3.89712989e-01 -6.07683122e-01 1.96898058e-01 2.70234227e-01
-7.42828608e-01 -1.99318632e-01 -3.09261709e-01 4.64864880e-01
1.10850394e-01 -6.68828607e-01 8.88654470e-01 -3.91155869e-01
-5.44132233e-01 8.66615921e-02 -2.90277094e-01 2.00719699e-01
1.18287480e+00 -6.14231415e-02 -4.39551413e-01 -6.64770529e-02
-9.04509544e-01 3.95762593e-01 2.94766247e-01 1.14232481e-01
1.48251146e-01 -1.52814436e+00 -7.66441166e-01 1.10358581e-01
4.92974401e-01 -8.28924328e-02 -1.27313593e-02 5.20509005e-01
4.03309278e-02 5.36224365e-01 3.29388052e-01 -6.12116098e-01
-1.32754183e+00 6.78220868e-01 2.17428192e-01 -9.49131548e-01
-4.23575759e-01 9.89803910e-01 2.84235299e-01 -1.08101821e+00
3.80469263e-01 5.45130745e-02 -2.78703004e-01 3.63474905e-01
4.00762826e-01 3.91450971e-01 1.03882235e-02 -7.27874756e-01
-4.22457010e-01 8.22299361e-01 -2.56206900e-01 -9.50854495e-02
9.85023856e-01 5.64136356e-02 -3.16641688e-01 9.39376950e-01
1.50075018e+00 -4.22574580e-01 -8.70377123e-01 -5.81249774e-01
4.94162619e-01 -1.70996919e-01 -2.58724123e-01 -1.18111801e+00
-8.60588670e-01 7.93761492e-01 5.29942155e-01 2.89478421e-01
8.31544220e-01 4.66101706e-01 5.81259429e-01 2.82585472e-01
7.12231815e-01 -9.54639792e-01 1.75320372e-01 7.62320936e-01
5.40710866e-01 -1.36742508e+00 1.43422661e-02 -4.30387586e-01
-6.58151031e-01 7.99859881e-01 3.74578118e-01 2.08443567e-01
9.77974653e-01 1.36292174e-01 3.06690276e-01 -3.64732862e-01
-1.20006049e+00 -3.99403065e-01 1.36139110e-01 3.86911184e-01
5.40873349e-01 5.52174687e-01 -6.34884611e-02 5.41472852e-01
-9.15688351e-02 -2.96703637e-01 3.61227281e-02 8.96415353e-01
-4.97409582e-01 -1.31605864e+00 -4.44528967e-01 8.99788797e-01
-5.17800987e-01 -5.83371371e-02 -1.43762439e-01 4.75666225e-01
5.61892450e-01 1.21105731e+00 -2.07454681e-01 -4.97569263e-01
1.37407571e-01 7.71837771e-01 1.24937885e-01 -1.07315624e+00
-5.89112759e-01 -4.52579319e-04 2.08555803e-01 -4.13288414e-01
-2.51401991e-01 -7.32285559e-01 -1.31056952e+00 -1.36900544e-01
-7.12068141e-01 4.08132344e-01 6.51651740e-01 8.44731569e-01
8.68617892e-02 1.96458936e-01 9.37243879e-01 -8.78494024e-01
-8.06496263e-01 -1.34414840e+00 -1.06086183e+00 5.40281475e-01
3.54822390e-02 -9.14440334e-01 -5.38477004e-01 -1.66536778e-01] | [9.53431224822998, 4.387480735778809] |
77335cd8-0053-4596-b99c-a032f53fc977 | simsc-a-simple-framework-for-semantic | 2305.02385 | null | https://arxiv.org/abs/2305.02385v1 | https://arxiv.org/pdf/2305.02385v1.pdf | SimSC: A Simple Framework for Semantic Correspondence with Temperature Learning | We propose SimSC, a remarkably simple framework, to address the problem of semantic matching only based on the feature backbone. We discover that when fine-tuning ImageNet pre-trained backbone on the semantic matching task, L2 normalization of the feature map, a standard procedure in feature matching, produces an overly smooth matching distribution and significantly hinders the fine-tuning process. By setting an appropriate temperature to the softmax, this over-smoothness can be alleviated and the quality of features can be substantially improved. We employ a learning module to predict the optimal temperature for fine-tuning feature backbones. This module is trained together with the backbone and the temperature is updated online. We evaluate our method on three public datasets and demonstrate that we can achieve accuracy on par with state-of-the-art methods under the same backbone without using a learned matching head. Our method is versatile and works on various types of backbones. We show that the accuracy of our framework can be easily improved by coupling it with more powerful backbones. | ['Victor Adrian Prisacariu', 'Xingchen Wan', 'Kai Han', 'Xinghui Li'] | 2023-05-03 | null | null | null | null | ['semantic-correspondence'] | ['computer-vision'] | [ 3.18970799e-01 1.08488694e-01 -3.73690516e-01 -6.09593332e-01
-1.15673518e+00 -7.88662732e-01 5.19863725e-01 -1.19237632e-01
-5.40543139e-01 1.88522309e-01 2.38235295e-01 1.36042148e-01
-5.99512421e-02 -7.46353924e-01 -1.04881239e+00 -5.62477827e-01
2.19013706e-01 4.93478626e-01 4.29076314e-01 -2.68619418e-01
4.93353307e-01 2.96690136e-01 -1.84905565e+00 3.41869652e-01
7.10346520e-01 1.04917622e+00 1.32325172e-01 5.10823309e-01
-5.78657463e-02 2.22021788e-01 -2.20412403e-01 -4.83881176e-01
7.23038375e-01 2.31914204e-02 -1.08374274e+00 -9.53333378e-02
1.07567513e+00 -3.35150868e-01 -3.81456405e-01 1.08474708e+00
4.14079815e-01 1.45804450e-01 3.50435495e-01 -1.18815231e+00
-4.82772678e-01 5.49393415e-01 -3.42064202e-01 -1.16889462e-01
3.06862146e-01 1.50343955e-01 1.43960702e+00 -7.64635324e-01
7.78058767e-01 1.30383968e+00 9.93135571e-01 5.92022419e-01
-1.27273047e+00 -6.81121707e-01 1.44510061e-01 1.26673833e-01
-1.35080159e+00 -6.06235981e-01 4.18719351e-01 -2.97852218e-01
1.04213059e+00 2.07643807e-02 3.25330019e-01 6.81915700e-01
8.44617561e-02 7.21600831e-01 6.73189580e-01 -3.64210397e-01
8.75341222e-02 -6.23957328e-02 2.08454356e-02 8.69037151e-01
-1.66871116e-01 3.73063117e-01 -4.95025963e-01 -3.44732553e-01
7.03877270e-01 4.72338386e-02 -3.91392633e-02 -6.66413367e-01
-1.21196651e+00 9.59630072e-01 8.02090704e-01 1.94856897e-01
1.19545655e-02 5.82604289e-01 3.59854966e-01 6.44250870e-01
2.36753255e-01 6.65750325e-01 -7.45386600e-01 -4.07003053e-02
-1.03050590e+00 5.72192252e-01 8.13600183e-01 1.00430334e+00
1.24315226e+00 -7.51257956e-01 -3.51108819e-01 7.28174210e-01
9.12255645e-02 4.00692403e-01 3.12567949e-01 -1.38783813e+00
3.26305777e-01 5.51263630e-01 4.09520082e-02 -5.10359347e-01
-4.51905549e-01 -4.06180710e-01 -3.35729718e-01 2.85219520e-01
5.05482972e-01 1.26825556e-01 -1.15700090e+00 2.14083672e+00
3.47400427e-01 2.08386466e-01 -3.30989480e-01 8.99512351e-01
5.31115353e-01 1.63210437e-01 1.07541353e-01 5.51320374e-01
1.34231198e+00 -1.23453188e+00 -1.63725913e-01 -2.14124173e-01
9.00353551e-01 -8.74165356e-01 1.15004015e+00 -1.30956052e-02
-1.08447385e+00 -5.64022183e-01 -1.19351351e+00 -3.00539106e-01
-3.57162505e-01 -2.83678561e-01 9.03103232e-01 3.67985666e-01
-1.41149318e+00 1.23390162e+00 -8.38860095e-01 -6.39794171e-01
4.27956641e-01 7.36242473e-01 -5.11144340e-01 -2.08373681e-01
-1.19846344e+00 1.09446156e+00 1.41042739e-01 -3.88524771e-01
-5.10075390e-01 -1.15600741e+00 -8.18506718e-01 2.50664037e-02
1.05495945e-01 -1.27373278e+00 1.50079560e+00 -1.00694001e+00
-1.40078413e+00 1.12720788e+00 -2.91627645e-01 -1.07685305e-01
4.60670441e-01 -1.38664432e-02 -1.65986028e-02 2.18454897e-02
3.52725804e-01 1.12279999e+00 1.01203060e+00 -6.19049430e-01
-7.14673221e-01 -3.67798656e-01 1.17176838e-01 1.35728464e-01
-2.62536734e-01 9.71087348e-03 -8.01283062e-01 -4.59146887e-01
-4.84835915e-03 -1.09926784e+00 -4.47286040e-01 3.40812296e-01
-5.33819348e-02 -3.09031218e-01 3.41112614e-01 -2.75261462e-01
9.56396163e-01 -2.08780909e+00 9.86119583e-02 4.92850572e-01
1.74973235e-01 -1.14672527e-01 -6.35546505e-01 1.51000515e-01
1.26395866e-01 -9.31459293e-02 -2.23434120e-01 -3.59421879e-01
3.63224387e-01 1.42917618e-01 -4.29331362e-02 6.19273961e-01
3.08477789e-01 1.14673495e+00 -8.24461997e-01 -3.56864959e-01
5.16078621e-02 3.50463659e-01 -1.09462249e+00 2.66519010e-01
-1.88545793e-01 2.42145523e-01 -4.52397943e-01 4.56688762e-01
8.68617475e-01 -6.43379390e-01 -7.28704929e-02 -4.13628101e-01
6.34117946e-02 4.28160101e-01 -1.14674783e+00 2.32698870e+00
-4.76650953e-01 2.66614854e-01 1.19966954e-01 -9.88275826e-01
7.36845493e-01 -8.88864771e-02 5.33694267e-01 -8.39762151e-01
-9.50850323e-02 3.42528611e-01 -3.45939010e-01 -2.60451287e-01
4.86507177e-01 -6.07506279e-03 -2.87328154e-01 5.81670105e-01
3.45432013e-01 -9.96892303e-02 -9.29825902e-02 -2.93352548e-03
1.32685709e+00 3.25990885e-01 -7.35126585e-02 -5.51620126e-01
2.57662207e-01 -4.61367927e-02 4.27637190e-01 8.83212507e-01
-8.06611329e-02 6.38066053e-01 4.22397777e-02 -5.59867799e-01
-1.30258811e+00 -1.06667626e+00 -3.13996077e-01 1.66010129e+00
2.37482101e-01 -4.26564544e-01 -9.96487439e-01 -8.03764582e-01
5.51667929e-01 5.16793914e-02 -9.16997313e-01 -3.46455246e-01
-6.76487744e-01 -6.10471904e-01 5.49187720e-01 5.84175467e-01
2.98054814e-01 -8.35256159e-01 -2.32526824e-01 2.29258627e-01
3.78558710e-02 -1.00009882e+00 -8.94798636e-01 3.14174294e-01
-8.14232707e-01 -8.91793370e-01 -5.87314725e-01 -7.71808267e-01
7.09261954e-01 1.11746721e-01 1.47677982e+00 4.69406158e-01
-3.31190407e-01 1.11452959e-01 -1.09867387e-01 2.63595339e-02
-2.77810246e-01 8.06820929e-01 -3.34222078e-01 -2.70335406e-01
5.26897609e-01 -6.78225875e-01 -8.20070028e-01 4.42885011e-01
-8.19928110e-01 -2.27244366e-02 3.60256404e-01 8.83757949e-01
4.13544387e-01 -6.17189109e-01 5.33075631e-01 -9.86613870e-01
1.67100489e-01 -2.36209676e-01 -7.59148538e-01 3.28994572e-01
-8.86805475e-01 8.11477721e-01 3.90873462e-01 -3.26332867e-01
-6.17671788e-01 3.65184724e-01 -3.56264621e-01 -3.42890292e-01
5.13864234e-02 -8.60505328e-02 -2.19885986e-02 -7.31493950e-01
7.39112079e-01 -2.24202991e-01 -4.64314446e-02 -5.92723489e-01
6.73969328e-01 5.36427557e-01 7.08948910e-01 -8.58819485e-01
1.01772499e+00 5.17353773e-01 -6.81983754e-02 3.69819254e-02
-1.07921445e+00 -6.55421674e-01 -5.77939689e-01 3.27058852e-01
5.24287820e-01 -9.38371420e-01 -7.37275422e-01 4.21040148e-01
-9.20037389e-01 -4.70406204e-01 -2.92452037e-01 1.14539519e-01
-8.27479303e-01 1.51009351e-01 -6.44865692e-01 1.30802050e-01
-5.88708162e-01 -1.11753416e+00 1.49360883e+00 2.06299305e-01
-2.90188432e-01 -1.04745865e+00 2.60712296e-01 2.58287609e-01
8.82532656e-01 -1.08792633e-02 8.20509315e-01 -5.01093268e-01
-7.57263064e-01 2.96029868e-03 -4.55761373e-01 -1.68267284e-02
-2.25962996e-02 -2.24599808e-01 -1.21373761e+00 -4.60592717e-01
-4.86736476e-01 -3.67620200e-01 1.29714227e+00 2.18082771e-01
1.19439352e+00 -8.09719414e-02 -5.20299971e-01 1.25422013e+00
1.35902250e+00 -6.35742486e-01 6.09537780e-01 7.40104198e-01
5.81156969e-01 4.03547943e-01 4.54378456e-01 3.62548769e-01
5.89468181e-01 9.70386028e-01 3.78303856e-01 -2.52547204e-01
-2.38860577e-01 -3.55819046e-01 1.47707939e-01 4.39605087e-01
2.69085795e-01 3.36093009e-01 -4.80617344e-01 5.51922858e-01
-2.06861734e+00 -7.16034830e-01 3.95552069e-01 2.17345166e+00
1.13843119e+00 -7.27467090e-02 1.71919867e-01 -3.11839193e-01
6.17818594e-01 3.91057655e-02 -7.72017181e-01 -3.62648070e-01
-9.79923457e-02 6.84878826e-01 8.86070490e-01 7.71411359e-01
-1.26478601e+00 1.32039571e+00 7.36170769e+00 8.23360324e-01
-1.05320883e+00 1.09888636e-01 1.83136374e-01 -1.34887800e-01
-7.21186876e-01 2.54700363e-01 -9.39046443e-01 3.55664551e-01
7.53229558e-01 9.05742198e-02 8.72485280e-01 7.87688553e-01
-2.96014339e-01 3.67757648e-01 -1.47376657e+00 7.83906221e-01
-3.49871635e-01 -1.55301487e+00 -7.72766620e-02 -1.10764734e-01
8.59365940e-01 4.98620421e-01 6.63700104e-02 3.12717050e-01
4.88091648e-01 -1.04578567e+00 6.41173363e-01 4.29883242e-01
1.00570691e+00 -5.85862458e-01 5.16499162e-01 -1.26218766e-01
-1.18535817e+00 -3.24719492e-03 -5.20318151e-01 1.69745639e-01
-8.76829401e-02 5.17055631e-01 -5.24829984e-01 2.66329378e-01
7.59291887e-01 7.27904856e-01 -7.08189487e-01 1.15528834e+00
-1.24726340e-01 1.38587728e-01 -4.90253419e-01 2.79703200e-01
2.31494516e-01 1.22654125e-01 1.62476495e-01 1.42837012e+00
2.33943149e-01 -3.48047495e-01 2.18449026e-01 9.23904061e-01
-4.13431555e-01 7.11862594e-02 -2.59208798e-01 2.84494042e-01
5.72971523e-01 1.30202639e+00 -3.09136271e-01 -2.53515720e-01
-3.93002808e-01 1.17087257e+00 8.68161857e-01 2.53671825e-01
-7.47511864e-01 -5.25114536e-01 9.60284591e-01 -1.58326812e-02
6.03656411e-01 3.60696554e-01 -3.56583089e-01 -1.36079514e+00
4.22039926e-02 -7.21858859e-01 5.45829058e-01 -4.31945443e-01
-1.50642407e+00 3.17040592e-01 -3.61644924e-01 -8.27456415e-01
-4.24998671e-01 -5.30398190e-01 -4.95473385e-01 9.46978271e-01
-1.84775949e+00 -1.10125327e+00 -3.70106369e-01 8.00017297e-01
-2.46685538e-02 2.00457558e-01 9.73365426e-01 5.34981430e-01
-2.04048738e-01 1.38108408e+00 6.22271597e-02 -5.78255728e-02
1.21959257e+00 -1.27788079e+00 8.06930840e-01 4.38009024e-01
-3.20878513e-02 4.96352077e-01 7.12173462e-01 -1.61943078e-01
-1.38637292e+00 -1.07167971e+00 9.82037902e-01 -4.57987994e-01
7.43640959e-01 -4.64249372e-01 -9.21276569e-01 6.52709663e-01
-1.03072770e-01 2.64037490e-01 4.12461668e-01 4.06973302e-01
-9.68978941e-01 -1.57630846e-01 -1.39897513e+00 3.82923305e-01
1.45728207e+00 -7.03101218e-01 -5.13582528e-01 3.56648535e-01
9.79213595e-01 -5.22177815e-01 -1.20668900e+00 4.39954937e-01
8.03384960e-01 -6.70655966e-01 1.02171135e+00 -7.69792676e-01
1.51431903e-01 -2.43943319e-01 -2.35160932e-01 -1.23354733e+00
-7.44425058e-01 -7.61799335e-01 3.79277542e-02 9.88908470e-01
4.81664151e-01 -6.71726584e-01 9.41510081e-01 8.22014570e-01
-3.99058871e-02 -6.44459248e-01 -8.70659471e-01 -6.76994205e-01
2.12461889e-01 -2.30300635e-01 9.77432311e-01 7.58074164e-01
1.03384793e-01 1.40039653e-01 -2.83580851e-02 4.02229801e-02
8.97222519e-01 3.86838973e-01 8.77122283e-01 -1.29130054e+00
-4.81790781e-01 -6.32106245e-01 -5.27865410e-01 -1.21058917e+00
4.29412991e-01 -1.18948293e+00 -1.62715111e-02 -1.09998751e+00
5.76625347e-01 -7.71087348e-01 -5.16759872e-01 7.65316188e-01
-4.04991269e-01 3.48166257e-01 2.02889577e-01 2.20042467e-01
-7.91027188e-01 2.71352649e-01 1.26065195e+00 -4.93543781e-02
-8.71115737e-03 -1.30484581e-01 -8.83034229e-01 3.80820781e-01
6.31851137e-01 -6.00042045e-01 2.95210397e-03 -5.43772697e-01
2.94534296e-01 -4.93454844e-01 2.27693006e-01 -7.58462369e-01
3.73395681e-01 -6.59219548e-03 3.00364643e-01 -8.75859037e-02
2.93693524e-02 -6.20068431e-01 1.23985320e-01 2.82937735e-01
-6.13020539e-01 -6.62315171e-03 1.72400787e-01 1.69769228e-01
3.34076956e-03 -1.76570758e-01 9.11691189e-01 -5.73553406e-02
-6.58601522e-01 5.75754225e-01 2.45077074e-01 2.77326256e-01
5.75806558e-01 -5.31201996e-02 -4.57338572e-01 -1.81955218e-01
-4.89454925e-01 4.68678534e-01 1.04068530e+00 5.21213233e-01
1.29989609e-01 -1.68289900e+00 -6.79467797e-01 3.66215259e-01
4.96600479e-01 -2.25749657e-01 5.53188957e-02 7.21162915e-01
-1.87537923e-01 2.64518440e-01 -1.36801064e-01 -6.60558105e-01
-8.87094736e-01 5.99056423e-01 5.67884326e-01 -5.28766513e-01
-5.55955708e-01 8.71747196e-01 2.52257198e-01 -8.59711349e-01
3.71476978e-01 -2.24077359e-01 2.76580453e-01 -3.00493360e-01
5.61486483e-01 5.76627366e-02 3.09967160e-01 -3.89751136e-01
-4.67013299e-01 1.05661917e+00 -2.42725939e-01 -5.11655409e-04
1.40448141e+00 -1.18476249e-01 -1.18131034e-01 -1.56531081e-01
1.78913260e+00 -3.12163204e-01 -1.56264865e+00 -5.46914518e-01
3.15607004e-02 -4.09184366e-01 2.19379812e-02 -6.64861679e-01
-1.22196341e+00 5.91294706e-01 6.43398762e-01 -2.47510970e-01
1.06568563e+00 2.69826680e-01 1.13855875e+00 4.05432135e-01
4.82131124e-01 -1.22077048e+00 -1.77005336e-01 5.35628259e-01
4.64075327e-01 -1.22630060e+00 -2.33519226e-01 -2.79760331e-01
-1.47263780e-01 1.06392825e+00 4.80675280e-01 -5.68788171e-01
7.10537136e-01 5.27471423e-01 6.64662421e-02 -2.42236778e-01
-8.04008543e-01 -2.80290306e-01 5.65280676e-01 4.45285290e-01
3.00255537e-01 -3.34454775e-02 9.66473818e-02 2.92842776e-01
-4.59616542e-01 7.11128265e-02 -1.66102797e-01 8.04573894e-01
-6.09017193e-01 -1.38131070e+00 -2.24267095e-01 4.37747777e-01
-6.57359421e-01 -2.63845205e-01 -7.24121258e-02 4.65067804e-01
1.40302747e-01 6.26343787e-01 2.27985844e-01 -5.90221107e-01
4.36740845e-01 2.87355538e-02 8.74458432e-01 -3.25009555e-01
-9.23044503e-01 -3.40847522e-01 -1.37056122e-02 -1.36344635e+00
-3.18124592e-01 -5.02186477e-01 -1.29861259e+00 -6.11197054e-01
-3.00213337e-01 -1.81805640e-02 6.33352280e-01 9.62459266e-01
6.94680333e-01 1.70091428e-02 7.58723199e-01 -9.59414303e-01
-8.63918602e-01 -6.29696667e-01 -3.06566298e-01 9.40866947e-01
4.37222481e-01 -5.84977269e-01 -4.63484466e-01 -1.79972157e-01] | [8.262397766113281, -1.8707923889160156] |
0af73d7a-6bc4-424f-a8de-791912cc592a | robust-image-stitching-with-multiple-1 | 2011.11784 | null | https://arxiv.org/abs/2011.11784v1 | https://arxiv.org/pdf/2011.11784v1.pdf | Robust image stitching with multiple registrations | Panorama creation is one of the most widely deployed techniques in computer vision. In addition to industry applications such as Google Street View, it is also used by millions of consumers in smartphones and other cameras. Traditionally, the problem is decomposed into three phases: registration, which picks a single transformation of each source image to align it to the other inputs, seam finding, which selects a source image for each pixel in the final result, and blending, which fixes minor visual artifacts. Here, we observe that the use of a single registration often leads to errors, especially in scenes with significant depth variation or object motion. We propose instead the use of multiple registrations, permitting regions of the image at different depths to be captured with greater accuracy. MRF inference techniques naturally extend to seam finding over multiple registrations, and we show here that their energy functions can be readily modified with new terms that discourage duplication and tearing, common problems that are exacerbated by the use of multiple registrations. Our techniques are closely related to layer-based stereo, and move image stitching closer to explicit scene modeling. Experimental evidence demonstrates that our techniques often generate significantly better panoramas when there is substantial motion or parallax. | ['Ramin Zabih', 'Ce Liu', 'Michael Krainin', 'Emil Keyder', 'Richard Strong Bowen', 'Chen Wang', 'Charles Herrmann'] | 2020-11-23 | robust-image-stitching-with-multiple | http://openaccess.thecvf.com/content_ECCV_2018/html/Charles_Herrmann_Robust_image_stitching_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Charles_Herrmann_Robust_image_stitching_ECCV_2018_paper.pdf | eccv-2018-9 | ['image-stitching'] | ['computer-vision'] | [ 8.28441441e-01 -2.01882794e-01 3.61366905e-02 -1.97738007e-01
-5.92184782e-01 -7.04390764e-01 7.13234007e-01 -1.86661541e-01
-3.10518652e-01 4.27353829e-01 4.56188731e-02 -1.84163228e-01
1.17599726e-01 -6.85805857e-01 -8.42640221e-01 -6.78152621e-01
2.37434000e-01 1.17281131e-01 7.28399515e-01 -2.43483588e-01
4.34773803e-01 4.02719706e-01 -1.55333734e+00 5.39400913e-02
7.47972608e-01 8.28646302e-01 3.95399332e-01 6.62396491e-01
1.51044950e-02 5.56969762e-01 -2.93764293e-01 -5.34021556e-01
6.78454161e-01 -2.64153779e-01 -6.89064980e-01 6.52663291e-01
1.19026208e+00 -6.23623133e-01 -1.39730826e-01 1.21053398e+00
1.34533972e-01 2.52905399e-01 1.81868732e-01 -1.12683523e+00
-4.25890796e-02 6.83371425e-02 -1.22478223e+00 -8.78686458e-02
4.56502289e-01 1.57077923e-01 7.36512959e-01 -7.29787707e-01
8.16643298e-01 1.15173185e+00 7.96442568e-01 2.06368104e-01
-1.36547768e+00 -5.37216723e-01 -6.86137378e-02 -4.01955172e-02
-1.14092004e+00 -7.70957828e-01 8.50816071e-01 -3.39822590e-01
5.63352883e-01 3.98467958e-01 7.72881806e-01 5.88460922e-01
3.20234895e-01 3.88789207e-01 1.02306914e+00 -6.21370196e-01
3.45915370e-02 -5.41799888e-02 -4.98419702e-02 6.19329274e-01
3.32000226e-01 1.33265331e-01 -5.55753052e-01 -2.10069522e-01
1.20375776e+00 2.68380284e-01 -6.20235145e-01 -7.30583847e-01
-1.39081371e+00 6.21219635e-01 2.94484943e-01 5.71281724e-02
-2.45603710e-01 4.02039975e-01 -9.86353233e-02 2.33871862e-01
3.13555390e-01 4.21101868e-01 -1.30927935e-01 1.00051604e-01
-1.14326632e+00 2.49710664e-01 6.04455531e-01 9.42983210e-01
1.12053907e+00 -1.35146633e-01 4.51345593e-01 9.53382850e-01
1.45544544e-01 3.02399784e-01 2.08104596e-01 -1.59799731e+00
3.81926775e-01 4.16037709e-01 8.31780210e-02 -1.00937378e+00
3.92203517e-02 5.45973107e-02 -5.78679800e-01 6.52138889e-01
4.97110963e-01 5.58603108e-02 -9.07448649e-01 1.56906402e+00
5.04411221e-01 5.11955559e-01 -2.11244762e-01 6.81100547e-01
2.68826604e-01 5.09125054e-01 -4.89589512e-01 -1.91527814e-01
1.36428642e+00 -9.80017662e-01 -4.73861337e-01 -4.45197940e-01
-1.92189571e-02 -1.19465899e+00 8.56238425e-01 4.88874197e-01
-1.35589540e+00 -3.53175610e-01 -1.20088291e+00 -3.62262458e-01
3.85235921e-02 -5.29095471e-01 6.74718499e-01 6.03421628e-01
-1.13348532e+00 7.15731680e-01 -6.95860803e-01 -4.83972430e-01
2.38015905e-01 3.15053433e-01 -4.60944235e-01 -4.19940740e-01
-5.93794286e-01 9.26941395e-01 5.26198409e-02 -1.50860682e-01
-3.90539289e-01 -6.82008147e-01 -8.82717907e-01 -2.90844053e-01
4.98788714e-01 -8.53561103e-01 1.17338014e+00 -1.27745485e+00
-1.32283771e+00 9.78396297e-01 -4.61756378e-01 -1.89198822e-01
5.70395470e-01 -2.78144121e-01 -1.43357560e-01 1.09600343e-01
7.75603056e-02 8.99134636e-01 1.18397665e+00 -1.44069707e+00
-6.91685557e-01 -3.11667442e-01 6.78165257e-02 4.23134357e-01
-5.76023273e-02 -3.17681730e-02 -1.00636160e+00 -6.06650054e-01
4.26147878e-01 -1.01747882e+00 -3.72646809e-01 4.00332749e-01
-3.20599854e-01 4.27735239e-01 8.41143131e-01 -6.15101039e-01
9.27824497e-01 -2.12496638e+00 2.76206702e-01 3.73972327e-01
2.53926456e-01 -9.07678083e-02 -1.46022448e-02 1.78846464e-01
4.44370173e-02 -3.12461555e-01 -5.26883662e-01 -5.04031539e-01
-4.70808089e-01 2.19501182e-01 -3.03518772e-01 5.46080947e-01
-1.92218378e-01 5.86380064e-01 -8.72104943e-01 -5.02470613e-01
5.20088136e-01 3.83947879e-01 -5.78625083e-01 -1.37448549e-01
9.03200824e-03 2.98652530e-01 4.44363430e-02 4.76663411e-01
8.39607954e-01 -2.71417171e-01 1.76192507e-01 -3.42477083e-01
-3.70230734e-01 1.38433337e-01 -1.47397757e+00 1.88726461e+00
-3.54839027e-01 9.32521224e-01 1.64225683e-01 -3.89088899e-01
4.44771826e-01 -1.01962157e-01 5.37390947e-01 -4.93027300e-01
-7.36783221e-02 1.07689589e-01 -2.18452737e-01 -2.56959140e-01
8.52935672e-01 1.13986626e-01 2.80266821e-01 4.74026203e-01
-3.66247028e-01 -6.46795511e-01 1.89810470e-02 2.08970308e-01
9.34887767e-01 1.19279131e-01 3.84918243e-01 -6.92964122e-02
1.92489788e-01 1.41416222e-01 5.02513230e-01 6.88510239e-01
1.46346807e-01 1.15633786e+00 3.84264290e-02 -3.05428207e-01
-1.24676025e+00 -1.02711618e+00 -2.29424968e-01 6.20637536e-01
6.80856287e-01 -3.89406711e-01 -8.36596727e-01 -2.32672885e-01
-1.88255623e-01 4.81178552e-01 -3.54126334e-01 1.73194915e-01
-7.05749273e-01 -5.39702177e-01 1.90157503e-01 2.61926234e-01
7.21267700e-01 -6.46892250e-01 -1.03867567e+00 1.06511861e-01
-2.35179424e-01 -1.15328944e+00 -1.00005579e+00 5.28958179e-02
-1.09587312e+00 -1.18210018e+00 -8.26726556e-01 -7.47846723e-01
8.99581194e-01 9.50110853e-01 1.03085315e+00 2.94911325e-01
-2.35838160e-01 5.68355441e-01 -6.27256632e-02 -4.03818600e-02
-3.39849323e-01 -4.18548584e-01 -2.17163920e-01 2.22667065e-02
-2.78470274e-02 -7.46460676e-01 -7.87441075e-01 4.11515266e-01
-1.19535470e+00 5.41488111e-01 3.68604362e-01 7.00299621e-01
5.62052488e-01 -5.19719273e-02 -3.94542426e-01 -8.36751044e-01
1.79814965e-01 -1.53604403e-01 -6.92798734e-01 1.55398190e-01
-4.54174519e-01 -6.88502863e-02 6.55174628e-02 -5.32206833e-01
-9.25231814e-01 3.16683948e-01 2.59322345e-01 -5.93945861e-01
7.27844909e-02 -4.27946039e-02 -2.54273787e-02 -3.19713324e-01
5.60831010e-01 1.88324213e-01 2.58664846e-01 -3.25028956e-01
4.42625523e-01 3.08856547e-01 8.37498069e-01 -2.48535156e-01
9.83446240e-01 8.68750274e-01 -7.57892989e-03 -1.12923205e+00
-5.30575156e-01 -4.84461904e-01 -5.88901281e-01 -2.62002707e-01
8.55811596e-01 -8.20093989e-01 -3.40076208e-01 7.10924387e-01
-1.21466088e+00 -3.86337906e-01 -2.79235959e-01 2.32001379e-01
-3.99428040e-01 6.57325029e-01 -3.69345486e-01 -4.53004956e-01
-3.42802405e-02 -1.40226078e+00 1.16875744e+00 3.41013223e-01
-2.71281272e-01 -9.64711010e-01 1.65900558e-01 3.02367747e-01
2.18394950e-01 1.51131138e-01 6.71864510e-01 1.86271086e-01
-1.00743032e+00 8.27222914e-02 -1.04111180e-01 3.12652588e-01
4.09145117e-01 1.21071734e-01 -9.97815728e-01 -2.02163726e-01
1.75698429e-01 1.45786628e-01 7.93249428e-01 5.39628446e-01
8.42218876e-01 -2.60355473e-01 -4.34275866e-01 9.01709378e-01
1.50108397e+00 3.75492364e-01 1.05685806e+00 5.41061938e-01
9.52961564e-01 7.23593116e-01 4.03658181e-01 -4.43359539e-02
3.39412272e-01 1.09494126e+00 4.73548859e-01 -2.98787773e-01
-3.32995802e-01 -1.52007148e-01 3.34205747e-01 5.32079577e-01
-1.31425947e-01 9.72844958e-02 -6.51560903e-01 4.07446921e-01
-1.70988202e+00 -1.05249369e+00 -2.52551049e-01 2.71599221e+00
8.53075504e-01 -1.03688039e-01 -8.28459784e-02 1.19529344e-01
6.96824849e-01 3.68626297e-01 -4.63670969e-01 -2.16966450e-01
-1.51029721e-01 1.85773268e-01 8.38288784e-01 8.81084085e-01
-9.31205869e-01 8.18349600e-01 6.62924337e+00 6.58212662e-01
-1.10108376e+00 4.95850183e-02 6.25356913e-01 -9.22483131e-02
-5.54776013e-01 4.92702544e-01 -4.77248937e-01 5.13156891e-01
1.22253180e-01 2.21493766e-01 6.87194705e-01 5.30998349e-01
7.08390325e-02 -7.18345225e-01 -1.20054460e+00 1.18540657e+00
2.49652773e-01 -1.54553854e+00 -1.14349782e-01 2.89522380e-01
1.05694330e+00 -2.35369038e-02 -4.74477634e-02 -6.50937021e-01
3.36351037e-01 -7.89929807e-01 8.44631910e-01 3.04346234e-01
6.18819475e-01 -4.11957741e-01 2.07582742e-01 1.35154337e-01
-1.05763912e+00 1.86635539e-01 -2.57873327e-01 1.49504021e-01
3.31158996e-01 7.00267017e-01 -5.21345019e-01 3.68103206e-01
7.23920107e-01 6.26530528e-01 -4.45795268e-01 1.13924503e+00
4.83983681e-02 8.31235722e-02 -5.70685804e-01 4.93551910e-01
-1.33163944e-01 -5.66964746e-01 6.60239816e-01 8.38342428e-01
5.07875621e-01 -1.74539074e-01 -1.16404109e-02 6.14323020e-01
8.65501836e-02 -2.53417730e-01 -7.73302436e-01 6.96125567e-01
5.42830229e-01 1.04580235e+00 -9.10086155e-01 -3.45938444e-01
-5.49175322e-01 1.51989639e+00 -7.30684996e-02 3.78299832e-01
-6.71167850e-01 -1.37894690e-01 8.17689240e-01 3.77476484e-01
3.12584668e-01 -2.89728492e-01 -3.76556814e-01 -1.21943462e+00
9.39005315e-02 -9.49114859e-01 9.90611985e-02 -9.70999897e-01
-9.34320271e-01 1.85376450e-01 1.48095727e-01 -1.56071174e+00
-3.27468246e-01 -2.63685226e-01 -7.78496921e-01 6.43682778e-01
-1.45571423e+00 -8.85009766e-01 -4.90750164e-01 7.95062900e-01
7.94742703e-01 3.31569284e-01 3.89468938e-01 3.08318734e-01
-2.92563081e-01 2.49381334e-01 6.30710348e-02 -1.63084596e-01
8.08682382e-01 -1.03421533e+00 4.82597262e-01 1.20284522e+00
4.16438222e-01 6.83170319e-01 8.16159904e-01 -5.53153574e-01
-1.35114205e+00 -7.32364178e-01 6.75520003e-01 -4.21718091e-01
3.57915193e-01 -1.25375301e-01 -7.39030361e-01 7.66198337e-01
2.48506695e-01 -1.88385651e-01 5.97306304e-02 -3.17642629e-01
-3.64119053e-01 -1.97200730e-01 -1.00621462e+00 9.12978411e-01
1.01512110e+00 -3.81066978e-01 -3.67091596e-01 9.67035890e-02
5.09180486e-01 -7.26637363e-01 -4.45454121e-01 1.77896187e-01
7.42657781e-01 -1.39993083e+00 1.23211217e+00 1.66124955e-01
4.70461011e-01 -6.21967435e-01 -1.97276160e-01 -1.17870855e+00
-1.71916839e-03 -9.34500337e-01 8.32692906e-02 1.13270593e+00
1.09758399e-01 -6.32197440e-01 5.61058104e-01 7.34813690e-01
3.96774784e-02 -2.99967319e-01 -8.35690022e-01 -5.49483478e-01
-3.36350888e-01 -3.97687346e-01 4.77407664e-01 1.13989282e+00
-3.51157248e-01 7.03275530e-03 -4.64329004e-01 1.07070863e-01
9.87818837e-01 1.95101857e-01 1.10299098e+00 -1.08034635e+00
-4.65177983e-01 -4.68652010e-01 -3.20563436e-01 -1.46267343e+00
-3.97116393e-01 -3.67503464e-01 2.29748800e-01 -1.22964835e+00
1.73384175e-01 -6.35216534e-01 3.74848902e-01 2.25129351e-01
-1.83212578e-01 6.09404981e-01 2.33787358e-01 4.67018217e-01
-1.15343921e-01 2.85207257e-02 1.24280524e+00 -1.90207362e-02
-4.19293493e-01 -1.60925254e-01 -7.57539570e-01 1.03323352e+00
4.22745436e-01 -4.01330888e-01 -5.04296780e-01 -6.95727885e-01
1.48697630e-01 3.72954980e-02 5.75366795e-01 -9.95396018e-01
3.30314755e-01 -3.84856880e-01 4.55110520e-01 -3.36480945e-01
6.46055520e-01 -1.00008357e+00 7.43335307e-01 3.09899569e-01
-1.06650754e-03 2.45109409e-01 5.53210527e-02 3.79077047e-01
-3.32480744e-02 -3.02606553e-01 9.69158053e-01 -2.22277015e-01
-7.48927474e-01 1.67943224e-01 -1.55901715e-01 -3.08154255e-01
1.03613436e+00 -8.79730582e-01 -2.66557276e-01 -5.57509482e-01
-1.41579866e-01 -4.09815507e-03 1.12725008e+00 3.27831060e-01
6.37204885e-01 -1.18327296e+00 -3.31330568e-01 3.51208001e-01
-1.44254521e-01 3.98247540e-01 1.70528755e-01 9.61520374e-01
-7.43455768e-01 -1.78768188e-01 -5.14080636e-02 -9.06789362e-01
-1.49607503e+00 1.31696552e-01 2.80118763e-01 8.83701593e-02
-9.47274208e-01 7.14072227e-01 4.23543602e-01 1.80610567e-01
8.38172957e-02 -3.15640450e-01 2.24861354e-01 -4.85352054e-02
4.90349412e-01 4.61685210e-01 -8.77961703e-03 -6.90085232e-01
-2.16682136e-01 1.10607779e+00 -1.13821402e-01 -5.51734090e-01
1.09594023e+00 -4.40243810e-01 -2.03532487e-01 2.92780370e-01
9.55239058e-01 5.42048931e-01 -1.73325169e+00 -1.40567049e-01
-2.67159313e-01 -8.72249305e-01 8.10641125e-02 -2.19828323e-01
-1.24781656e+00 5.44209480e-01 4.47781950e-01 7.72698894e-02
1.22528839e+00 -3.19842286e-02 9.83011842e-01 8.61525610e-02
5.72567761e-01 -8.31030548e-01 3.05263922e-02 2.33565301e-01
6.46613777e-01 -1.10692787e+00 3.29366028e-01 -5.04049242e-01
-5.43577731e-01 1.02558434e+00 4.20694411e-01 -1.52905419e-01
4.68792051e-01 4.08830643e-01 7.59869665e-02 -1.28555074e-01
-3.98562044e-01 6.68782042e-03 1.87213644e-01 5.55960417e-01
-3.29230018e-02 -3.10752153e-01 1.10955186e-01 -3.34604412e-01
-2.01216981e-01 -1.22723028e-01 6.89958334e-01 9.78030920e-01
-4.25653726e-01 -1.30434072e+00 -7.20895946e-01 3.43359351e-01
-2.48219088e-01 -2.64321744e-01 -1.95089400e-01 7.64645636e-01
1.14210963e-01 8.30950499e-01 3.98979694e-01 -2.43089646e-01
1.57719076e-01 -1.57641441e-01 8.03230882e-01 -4.91729796e-01
-4.10511732e-01 2.55514681e-01 -1.01866182e-02 -8.38424861e-01
-7.49919534e-01 -8.10247183e-01 -1.04594302e+00 -4.50836122e-01
-4.55955595e-01 -3.70865613e-01 6.86878979e-01 7.72986591e-01
1.30153313e-01 1.22367144e-01 6.30893826e-01 -1.28995109e+00
-7.91286211e-03 -4.27386701e-01 -4.15303349e-01 5.60341597e-01
5.93806028e-01 -5.37221968e-01 -5.08735061e-01 5.89869559e-01] | [9.303654670715332, -2.4975831508636475] |
957ba3d8-a865-4480-a965-9b3828678e57 | voice2series-reprogramming-acoustic-models | 2106.09296 | null | https://arxiv.org/abs/2106.09296v3 | https://arxiv.org/pdf/2106.09296v3.pdf | Voice2Series: Reprogramming Acoustic Models for Time Series Classification | Learning to classify time series with limited data is a practical yet challenging problem. Current methods are primarily based on hand-designed feature extraction rules or domain-specific data augmentation. Motivated by the advances in deep speech processing models and the fact that voice data are univariate temporal signals, in this paper, we propose Voice2Series (V2S), a novel end-to-end approach that reprograms acoustic models for time series classification, through input transformation learning and output label mapping. Leveraging the representation learning power of a large-scale pre-trained speech processing model, on 30 different time series tasks we show that V2S performs competitive results on 19 time series classification tasks. We further provide a theoretical justification of V2S by proving its population risk is upper bounded by the source risk and a Wasserstein distance accounting for feature alignment via reprogramming. Our results offer new and effective means to time series classification. | ['Pin-Yu Chen', 'Yun-Yun Tsai', 'Chao-Han Huck Yang'] | 2021-06-17 | null | null | null | null | ['ecg-classification'] | ['medical'] | [ 3.98892015e-01 -6.62744939e-02 7.95130953e-02 -5.13608992e-01
-1.07057953e+00 -7.57241786e-01 7.18112767e-01 1.30948022e-01
-4.24100131e-01 2.54816443e-01 1.97455212e-01 -6.93975389e-01
-3.01797569e-01 -3.09127897e-01 -5.94806790e-01 -5.54269314e-01
-6.80344224e-01 6.19895346e-02 -7.24394396e-02 -1.27163127e-01
-1.49177745e-01 5.28122604e-01 -1.43711245e+00 -9.92001072e-02
6.39423966e-01 1.32325125e+00 -1.72659203e-01 8.35328996e-01
-9.12050083e-02 4.08321470e-01 -6.12316668e-01 2.29886975e-02
4.09125566e-01 -5.28041899e-01 -4.87276614e-01 -2.05330104e-01
5.50964773e-02 -6.16618246e-02 -3.28468829e-01 7.55209029e-01
5.33459365e-01 6.86059356e-01 7.11690545e-01 -1.38557315e+00
-4.74084884e-01 7.99846292e-01 -9.15379524e-02 4.39573288e-01
7.81439394e-02 -1.34058625e-01 1.05079031e+00 -8.47347260e-01
2.77745634e-01 1.06982112e+00 9.65214610e-01 5.54571152e-01
-1.39580142e+00 -5.58166027e-01 1.10479146e-01 1.90585181e-01
-9.43223774e-01 -5.53878009e-01 1.06836736e+00 -4.36544448e-01
1.23985827e+00 5.66308677e-01 4.42029417e-01 1.38516080e+00
1.18126124e-01 7.15034008e-01 1.01131284e+00 -6.48430705e-01
5.89149237e-01 -1.24065086e-01 2.61316955e-01 3.31585616e-01
-6.21760011e-01 3.74114066e-01 -5.56816101e-01 -2.48562604e-01
5.02577245e-01 -1.43387631e-01 -1.25664175e-02 -2.65547127e-01
-1.24350500e+00 7.14989543e-01 1.12827480e-01 5.73442936e-01
-3.49718124e-01 2.00552553e-01 7.28992522e-01 8.92435133e-01
7.37444520e-01 4.66601163e-01 -7.91811764e-01 -4.92059737e-01
-8.04732263e-01 -1.16734825e-01 6.28754199e-01 5.46130240e-01
1.15508527e-01 4.78003085e-01 -1.02234304e-01 9.62247789e-01
-4.87646125e-02 5.72292566e-01 1.01445091e+00 -6.21460378e-01
2.99480140e-01 6.51075765e-02 -2.76941806e-01 -5.11836290e-01
-6.49285138e-01 -5.41853607e-01 -6.77613318e-01 -4.14221585e-02
4.70761180e-01 -2.04329714e-01 -8.94909561e-01 2.17803001e+00
1.16104461e-01 2.43233845e-01 1.13509998e-01 6.07621551e-01
-4.40440280e-03 8.01655948e-01 8.26685876e-03 -4.79498416e-01
1.03327668e+00 -6.78813577e-01 -7.62115717e-01 1.34030297e-01
6.41129673e-01 -5.36839426e-01 1.33029568e+00 4.40740496e-01
-7.13483393e-01 -5.84891915e-01 -1.01506293e+00 6.74775690e-02
-5.05918860e-01 1.69943087e-02 5.28060138e-01 7.40494430e-01
-8.50059032e-01 8.84050608e-01 -1.20314753e+00 -4.06340688e-01
1.20349325e-01 3.64465296e-01 -2.85377890e-01 5.90611279e-01
-1.19351590e+00 6.84472859e-01 2.02239454e-01 -2.67285220e-02
-7.39579380e-01 -9.01465595e-01 -7.60211110e-01 -2.94311568e-02
1.04746684e-01 -3.39122385e-01 1.49090230e+00 -7.70999908e-01
-1.89398062e+00 3.56646508e-01 -1.24630801e-01 -8.54255617e-01
5.79771698e-01 -2.46646404e-01 -7.47470677e-01 4.91840008e-04
-8.02788734e-02 1.92174330e-01 1.13780379e+00 -5.35833001e-01
-3.26935738e-01 -2.95722812e-01 -5.79740167e-01 -2.45119840e-01
-8.78609776e-01 1.56032100e-01 2.85672814e-01 -1.09105337e+00
4.54200134e-02 -7.77803004e-01 -3.86847466e-01 -2.61033684e-01
-2.66018122e-01 -5.84115982e-01 7.37930298e-01 -7.69768536e-01
1.36748493e+00 -2.39223409e+00 8.29325840e-02 4.47451353e-01
-1.76694393e-01 1.27484784e-01 -5.28583467e-01 6.29830301e-01
-3.70532036e-01 -1.83773916e-02 -4.35722679e-01 -5.75534344e-01
4.60066915e-01 2.22206004e-02 -1.04807055e+00 5.00836253e-01
3.98654759e-01 5.97118437e-01 -9.72788095e-01 8.88283327e-02
3.85864079e-01 5.38071275e-01 -3.23829472e-01 2.56365508e-01
-9.32704359e-02 4.18258756e-01 -2.27456093e-01 3.13669980e-01
1.89156011e-01 1.40542224e-01 -8.25869218e-02 -1.18138254e-01
-1.14268541e-01 6.69008613e-01 -7.62332439e-01 1.68389392e+00
-7.27983713e-01 6.42118990e-01 -3.38357836e-01 -1.14183581e+00
1.06361842e+00 5.18650234e-01 8.21663141e-01 -5.42416692e-01
2.76089638e-01 4.99303520e-01 -3.94509360e-02 -3.48411053e-01
1.09928615e-01 -3.57723743e-01 -2.11816117e-01 5.85808039e-01
2.70027518e-01 -2.51025319e-01 -1.22833543e-01 -3.99625599e-01
1.32899940e+00 1.41499266e-01 1.57095760e-01 -1.19298227e-01
3.02828699e-01 -2.99546659e-01 2.93822914e-01 5.93166709e-01
-2.70788938e-01 5.63399255e-01 3.68591607e-01 -3.53325605e-01
-1.17351818e+00 -1.23194981e+00 -1.10792108e-01 1.41954172e+00
-7.01615155e-01 -3.50359142e-01 -6.62809908e-01 -5.83362460e-01
-5.03302105e-02 8.59228611e-01 -7.21560538e-01 -4.36896354e-01
-6.57741487e-01 -4.23170328e-01 8.09922278e-01 7.47825563e-01
-2.67267883e-01 -1.04402912e+00 -4.95795548e-01 4.83015150e-01
1.62474990e-01 -1.16454983e+00 -7.59905457e-01 7.17433870e-01
-9.57265079e-01 -6.46862864e-01 -7.50192702e-01 -5.84827542e-01
2.79703677e-01 3.80660929e-02 5.48307002e-01 -7.17921615e-01
-1.20489947e-01 5.36985338e-01 -4.82633680e-01 -5.83426714e-01
-5.29714942e-01 2.22915649e-01 6.63766503e-01 2.59413093e-01
3.52272272e-01 -1.01869810e+00 -1.59449428e-01 7.14756995e-02
-7.80719519e-01 -3.53484660e-01 2.42074534e-01 9.00192201e-01
4.72763985e-01 -1.82813760e-02 9.73210871e-01 -2.64547884e-01
8.76188338e-01 -4.08162653e-01 -6.22112334e-01 1.53422385e-01
-5.40237606e-01 2.14312091e-01 9.19296861e-01 -8.22248101e-01
-4.79982704e-01 1.89918563e-01 -2.33099833e-01 -8.76600504e-01
-5.34907356e-02 6.66382730e-01 1.68721035e-01 3.01502258e-01
7.42336333e-01 4.07648593e-01 2.69946009e-01 -6.48652673e-01
5.10380268e-01 5.95377922e-01 5.06958187e-01 -5.20351291e-01
9.45017457e-01 3.43930960e-01 -2.08072573e-01 -1.05713618e+00
-6.66052878e-01 -3.73535365e-01 -7.64434814e-01 -1.70504302e-01
6.25012100e-01 -5.66785693e-01 -5.54040074e-01 2.88387924e-01
-1.01608789e+00 -4.75514084e-01 -6.33370697e-01 8.34769189e-01
-8.66620600e-01 2.00297445e-01 -4.16715801e-01 -1.23280132e+00
-4.18601602e-01 -4.88707960e-01 1.01377857e+00 -4.24091905e-01
-4.28453237e-01 -1.13883889e+00 2.52902985e-01 -1.29631847e-01
6.06717467e-01 3.18434954e-01 9.71780539e-01 -1.18367290e+00
1.19223699e-01 -3.12102169e-01 3.34563553e-01 7.89126694e-01
2.83891439e-01 -1.36826783e-01 -1.26643491e+00 -2.76522934e-01
3.67539376e-01 -1.16385646e-01 7.01486647e-01 3.04465652e-01
1.32981586e+00 -2.32714638e-01 5.94200864e-02 5.32206714e-01
8.73373091e-01 4.60686564e-01 1.22264653e-01 1.24519579e-01
5.28697491e-01 7.86289215e-01 3.09649765e-01 4.82106328e-01
3.22020613e-02 5.17717659e-01 -3.35959904e-02 1.73783869e-01
-1.31912176e-02 -3.46668601e-01 8.11117828e-01 1.54741538e+00
1.77100807e-01 5.84930703e-02 -8.69782269e-01 6.68461978e-01
-1.67624080e+00 -8.87301147e-01 2.53663033e-01 2.25488114e+00
8.17957997e-01 2.65534431e-01 4.55360085e-01 7.39021957e-01
4.77122277e-01 7.34424144e-02 -7.95931756e-01 -3.52406323e-01
-1.08962141e-01 6.03876352e-01 2.54628688e-01 2.81606287e-01
-1.24668324e+00 6.20849133e-01 6.59496880e+00 8.26854229e-01
-1.45585728e+00 2.03025907e-01 2.91095823e-01 -1.87962756e-01
-3.23852837e-01 -4.09994125e-01 -1.27739847e-01 2.28768885e-01
1.61223674e+00 -4.83249664e-01 6.13739371e-01 6.51917160e-01
3.87693405e-01 7.69705415e-01 -1.51091647e+00 9.97103155e-01
-2.39427149e-01 -8.98503482e-01 -1.36858165e-01 2.68126838e-02
2.38450378e-01 1.79336339e-01 4.51635242e-01 5.76665163e-01
7.50118047e-02 -8.81312490e-01 9.70202625e-01 1.39632985e-01
8.72685134e-01 -5.47215939e-01 2.99610257e-01 9.78147537e-02
-1.22590339e+00 -3.15266639e-01 5.71851954e-02 -4.50829118e-02
2.46816143e-01 4.84092832e-01 -9.38212514e-01 5.11533141e-01
3.63954782e-01 7.47977316e-01 -1.35744885e-01 7.57055223e-01
1.57161757e-01 1.11347950e+00 -6.20608747e-01 -5.61538376e-02
2.59565651e-01 3.84420808e-03 6.53963149e-01 1.15258968e+00
6.42202616e-01 -1.02851182e-01 -2.23261993e-02 5.48650563e-01
1.56710804e-01 1.89377725e-01 -7.17374027e-01 -5.58132529e-01
5.25928438e-01 9.60637569e-01 -4.73518252e-01 -6.32372126e-02
-3.14565331e-01 7.14698553e-01 1.13253161e-01 3.54212940e-01
-7.07232773e-01 -6.47549450e-01 7.85205841e-01 -1.60682589e-01
2.97229052e-01 -4.92195696e-01 -1.71204552e-01 -1.07489932e+00
3.18889052e-01 -8.01908076e-01 2.88271666e-01 -4.09780920e-01
-1.54185581e+00 8.72463286e-01 -1.62500530e-01 -1.54674757e+00
-6.48895741e-01 -7.97939122e-01 -4.51729685e-01 6.82194769e-01
-1.28496397e+00 -1.05560648e+00 2.97695786e-01 5.70293248e-01
6.24989748e-01 -4.30248499e-01 1.00500178e+00 3.36794466e-01
-3.83904994e-01 8.15482914e-01 2.65037090e-01 1.45563722e-01
4.90504354e-01 -1.31585050e+00 7.73138463e-01 8.22285175e-01
3.33811939e-01 6.25167489e-01 7.41957486e-01 -7.64568821e-02
-1.56029201e+00 -1.08898771e+00 8.27708364e-01 -5.41352570e-01
1.35790908e+00 -7.16973841e-01 -1.05038714e+00 4.41006035e-01
-1.13168925e-01 1.05990702e-02 8.98561835e-01 2.15826407e-01
-6.93608105e-01 -2.72653490e-01 -7.93793201e-01 5.50513983e-01
1.11700606e+00 -1.21264613e+00 -9.08331573e-01 2.73452103e-01
1.00986016e+00 -6.49781749e-02 -9.04511154e-01 3.64177257e-01
5.51831365e-01 -4.14816737e-01 8.19946766e-01 -1.01190555e+00
-1.18083460e-02 1.11633427e-01 -4.99720693e-01 -1.47205901e+00
1.18212916e-01 -1.22564662e+00 -2.11703256e-01 1.10483050e+00
3.69013250e-01 -8.06090474e-01 3.12398881e-01 3.33699495e-01
-4.64233220e-01 -4.09244120e-01 -1.39880681e+00 -1.36928427e+00
2.84584016e-01 -1.00941563e+00 6.11719966e-01 1.04547918e+00
2.88299680e-01 1.66722819e-01 -1.72493890e-01 4.36955690e-02
3.31849247e-01 -9.73590389e-02 3.85348201e-01 -1.37060475e+00
-2.88241267e-01 -7.28278637e-01 -4.35774297e-01 -7.08446085e-01
4.26324546e-01 -1.00637460e+00 8.84058997e-02 -8.51295412e-01
-6.17141247e-01 -4.13833529e-01 -8.38534474e-01 4.71234113e-01
2.31166929e-01 -2.15010360e-01 9.87492725e-02 -1.81137510e-02
-1.26535833e-01 8.88332188e-01 7.13867664e-01 2.87915990e-02
-3.73060644e-01 2.78012931e-01 -1.72877520e-01 4.70783979e-01
7.80570388e-01 -2.64052778e-01 -6.79926872e-01 -2.37250790e-01
-1.60661284e-02 1.68863147e-01 2.65157670e-01 -1.05371940e+00
1.18542962e-01 1.14903614e-01 -1.37473777e-01 -4.45840627e-01
5.56656778e-01 -8.79386723e-01 -1.95290357e-01 4.78974223e-01
-7.45669484e-01 9.22243074e-02 3.04971606e-01 7.09052205e-01
-2.38339633e-01 2.09374011e-01 3.96127045e-01 3.95063818e-01
-5.38192630e-01 2.70087421e-01 -3.90497208e-01 -1.00818448e-01
7.75435925e-01 1.21848099e-01 -7.80238165e-03 -3.99633825e-01
-1.09937322e+00 -1.07974425e-01 -4.01656598e-01 9.40780282e-01
4.36698079e-01 -1.57023013e+00 -6.52198613e-01 5.17986298e-01
9.60994214e-02 -6.04361176e-01 -5.27749546e-02 8.32590699e-01
2.61421204e-01 4.80506688e-01 -8.05404633e-02 -5.63904047e-01
-1.04501092e+00 5.75457513e-01 3.83911639e-01 1.84317961e-01
-7.63591409e-01 6.35788023e-01 -3.79841626e-02 -6.62572086e-01
5.30222297e-01 -9.86134589e-01 1.12411544e-01 2.36869231e-01
4.58474874e-01 3.12404126e-01 3.77858132e-01 -3.00228059e-01
-4.45907265e-01 3.43681484e-01 1.95902705e-01 -6.74936950e-01
1.62290490e+00 2.72465013e-02 2.05883995e-01 1.09445906e+00
1.50661469e+00 -6.33807108e-02 -1.12036872e+00 -4.26354975e-01
3.82237256e-01 2.41905041e-02 -8.28780606e-02 -5.53370535e-01
-6.24336064e-01 1.04700172e+00 7.72526026e-01 8.43509495e-01
1.19327283e+00 -1.69381514e-01 9.14999008e-01 4.50319827e-01
3.56943339e-01 -1.07116055e+00 8.34304467e-02 7.00848103e-01
1.05129230e+00 -9.90244567e-01 -7.68946171e-01 -5.82917966e-02
-3.96000952e-01 1.19754016e+00 2.21410878e-02 -1.47650391e-01
9.10900593e-01 2.48860270e-01 2.21122324e-01 2.03907385e-01
-1.04227364e+00 -1.58828706e-01 4.68474567e-01 6.91837192e-01
4.79604751e-01 1.90448433e-01 -3.89066478e-03 9.62276816e-01
-6.66800320e-01 -1.21867992e-01 2.42980286e-01 7.04896927e-01
-1.87875435e-01 -1.08104587e+00 -2.03231484e-01 4.14090544e-01
-3.51749808e-01 8.51442367e-02 -1.82032153e-01 4.83769655e-01
-5.82240641e-01 1.01670241e+00 1.30337596e-01 -6.67550325e-01
4.30217564e-01 4.86753136e-01 3.43794495e-01 -3.27530742e-01
-6.13993168e-01 1.98848888e-01 -2.69244332e-02 -4.06372249e-01
-1.44532904e-01 -8.66221607e-01 -1.28683782e+00 1.11385077e-01
-9.50870290e-02 2.59623945e-01 9.54412699e-01 9.26751912e-01
4.74668115e-01 7.70376980e-01 1.08045316e+00 -8.06101441e-01
-1.33890522e+00 -1.04079020e+00 -5.15452862e-01 2.98559755e-01
9.08725083e-01 -4.70522255e-01 -6.83861315e-01 1.10597901e-01] | [15.090948104858398, 5.68483304977417] |
815de829-e5c5-40d7-a900-77a9ccc47586 | virus2vec-viral-sequence-classification-using | 2304.12328 | null | https://arxiv.org/abs/2304.12328v1 | https://arxiv.org/pdf/2304.12328v1.pdf | Virus2Vec: Viral Sequence Classification Using Machine Learning | Understanding the host-specificity of different families of viruses sheds light on the origin of, e.g., SARS-CoV-2, rabies, and other such zoonotic pathogens in humans. It enables epidemiologists, medical professionals, and policymakers to curb existing epidemics and prevent future ones promptly. In the family Coronaviridae (of which SARS-CoV-2 is a member), it is well-known that the spike protein is the point of contact between the virus and the host cell membrane. On the other hand, the two traditional mammalian orders, Carnivora (carnivores) and Chiroptera (bats) are recognized to be responsible for maintaining and spreading the Rabies Lyssavirus (RABV). We propose Virus2Vec, a feature-vector representation for viral (nucleotide or amino acid) sequences that enable vector-space-based machine learning models to identify viral hosts. Virus2Vec generates numerical feature vectors for unaligned sequences, allowing us to forego the computationally expensive sequence alignment step from the pipeline. Virus2Vec leverages the power of both the \emph{minimizer} and position weight matrix (PWM) to generate compact feature vectors. Using several classifiers, we empirically evaluate Virus2Vec on real-world spike sequences of Coronaviridae and rabies virus sequence data to predict the host (identifying the reservoirs of infection). Our results demonstrate that Virus2Vec outperforms the predictive accuracies of baseline and state-of-the-art methods. | ['Murray Patterson', 'Imdad Ullah Khan', 'Pin-Yu Chen', 'Ria Thazhe Punathil', 'Prakash Chourasia', 'Babatunde Bello', 'Sarwan Ali'] | 2023-04-24 | null | null | null | null | ['specificity'] | ['natural-language-processing'] | [ 2.36234367e-01 -6.45357788e-01 -3.04039627e-01 -2.71872580e-01
-5.01104221e-02 -9.14148390e-01 5.84632933e-01 3.05508077e-01
-5.57381749e-01 6.08292580e-01 1.27835125e-01 -5.61039448e-01
2.63414472e-01 -5.70843279e-01 -4.62124735e-01 -8.41032267e-01
-5.22393346e-01 8.14701080e-01 -2.60222614e-01 -3.49250019e-01
-2.12253537e-02 8.85397077e-01 -1.24079561e+00 2.27957234e-01
8.36908937e-01 4.46066111e-01 4.36696291e-01 7.22843468e-01
-2.93182343e-01 -4.81627844e-02 -4.22241509e-01 -3.18464749e-02
1.48133039e-01 -2.93186605e-01 -1.89602897e-01 -6.32814586e-01
-7.82903284e-02 -4.50916260e-01 -1.35655984e-01 6.07179582e-01
7.69872218e-02 -4.35622543e-01 1.05692816e+00 -1.40695941e+00
-5.95134676e-01 -1.28028095e-01 -6.23012424e-01 4.26718771e-01
2.23357510e-02 6.41787052e-01 9.54286337e-01 -5.61031997e-01
1.31827199e+00 1.29771817e+00 8.13260198e-01 6.16716802e-01
-1.43392456e+00 -5.75786948e-01 -3.93645823e-01 5.89914739e-01
-1.25759172e+00 1.39213791e-02 2.52196670e-01 -1.12624109e+00
1.18374217e+00 3.67779464e-01 9.18694317e-01 1.20743883e+00
6.91341698e-01 5.74209869e-01 8.77579689e-01 3.96777689e-01
1.40004098e-01 -1.14498593e-01 3.75242203e-01 6.46759510e-01
6.57814860e-01 4.14515465e-01 2.13328034e-01 -1.02022612e+00
5.33069015e-01 7.07824111e-01 -5.86926639e-01 -4.49817717e-01
-9.42177355e-01 1.40741479e+00 3.40941995e-01 3.59834224e-01
-7.77858257e-01 -1.67913824e-01 4.53273565e-01 1.36113957e-01
5.80003001e-02 2.27019131e-01 -6.02316499e-01 1.75092608e-01
-5.83012879e-01 3.14078093e-01 6.11983120e-01 3.14915329e-01
6.80866182e-01 7.27969361e-03 -1.35529861e-01 6.50100708e-01
3.72166574e-01 1.00621831e+00 1.94432959e-01 7.32597858e-02
-2.93468297e-01 8.06572795e-01 1.27347603e-01 -8.28124404e-01
-5.90084732e-01 -4.15591076e-02 -7.74130404e-01 -9.22469422e-03
1.67466983e-01 6.52346667e-03 -9.90487158e-01 1.70420909e+00
6.56801879e-01 5.97233534e-01 -1.00176387e-01 9.00869489e-01
5.60208738e-01 9.79813457e-01 2.08318830e-01 -3.95925969e-01
1.68685997e+00 -2.60949939e-01 -4.00986463e-01 5.11951894e-02
4.98835117e-01 -4.15606588e-01 5.76794267e-01 -1.64601952e-01
-1.03259400e-01 -9.07402560e-02 -9.75120246e-01 4.41166222e-01
-4.83928263e-01 -4.43268180e-01 6.05812550e-01 5.75859070e-01
-6.39513254e-01 5.77560008e-01 -9.82679069e-01 -7.87145793e-01
3.41326177e-01 2.28425488e-01 -5.03116429e-01 -6.99152946e-02
-9.74390745e-01 1.03547680e+00 8.19511935e-02 -4.07119505e-02
-9.31943059e-01 -9.29547369e-01 -6.62733674e-01 -1.64758023e-02
-2.42659837e-01 -7.54445612e-01 5.60578644e-01 -1.90083295e-01
-8.53991866e-01 8.54030967e-01 -2.43708327e-01 -6.82052851e-01
-8.31911787e-02 -7.08075687e-02 -4.80903566e-01 3.09860036e-02
-2.39531070e-01 4.41620797e-01 6.20959640e-01 -1.08255267e+00
-2.68943667e-01 -6.67700589e-01 -6.34877801e-01 -4.28358167e-01
4.98446405e-01 3.30088735e-01 8.44578519e-02 -6.00236535e-01
-4.83251393e-01 -1.25489104e+00 -4.14826959e-01 -2.91895941e-02
-1.16152450e-01 -2.80966520e-01 1.02256942e+00 -7.27788210e-01
8.05617571e-01 -2.04791403e+00 2.88471282e-01 1.99826747e-01
4.16609883e-01 8.00664246e-01 -5.13081849e-01 8.22748542e-01
1.42180622e-01 3.30087356e-02 -4.78756100e-01 6.78765774e-01
-2.51148134e-01 4.42266226e-01 -4.44179118e-01 1.04089880e+00
8.83707628e-02 1.17587161e+00 -9.81205285e-01 -1.02067620e-01
2.84452319e-01 1.29736197e+00 -6.50575995e-01 4.15253162e-01
-4.51864660e-01 3.70234638e-01 -4.95801926e-01 5.36997795e-01
9.79018033e-01 -4.18484509e-01 5.88447630e-01 -2.76497364e-01
1.97244674e-01 -8.76028184e-03 -3.76180619e-01 8.32582831e-01
6.59931302e-02 5.28059185e-01 2.62466729e-01 -5.37436306e-01
5.45916200e-01 6.87183887e-02 7.17992842e-01 -2.05671668e-01
4.42759216e-01 -9.06550214e-02 2.27768674e-01 -5.58367789e-01
-7.82626122e-02 -1.18152045e-01 2.28673071e-01 4.65994149e-01
-1.78259254e-01 3.53439897e-01 3.24327350e-02 1.30135447e-01
1.07268035e+00 -2.76402801e-01 6.56321526e-01 -1.96729600e-01
4.09522563e-01 3.80125433e-01 7.51665473e-01 2.30535716e-01
-4.04560357e-01 -2.12921984e-02 2.02825367e-01 -6.34052873e-01
-9.52198267e-01 -1.39291024e+00 -2.29860231e-01 1.12877512e+00
-1.76632747e-01 -2.72257924e-01 -4.42772627e-01 -4.18713301e-01
7.25549996e-01 7.40846932e-01 -7.63795018e-01 -2.80090105e-02
-7.55483270e-01 -7.08043516e-01 7.43632019e-01 1.56547904e-01
-4.85414267e-01 -1.07184541e+00 -7.68286586e-01 2.15216517e-01
-1.98769450e-01 -6.75955236e-01 -5.91274977e-01 1.04828700e-01
-6.62413716e-01 -1.68318558e+00 -3.63606334e-01 -6.03028893e-01
3.64920080e-01 2.65241861e-01 6.84343815e-01 1.82933554e-01
-7.39941835e-01 5.57605587e-02 -1.98525414e-01 -2.09455237e-01
-7.24816799e-01 -5.89459777e-01 5.53991854e-01 -2.15511009e-01
9.80626762e-01 -2.68495798e-01 -7.05935597e-01 2.29702860e-01
-8.99172544e-01 -3.53926182e-01 3.37829530e-01 8.68810534e-01
6.36373520e-01 -5.92519164e-01 4.53611016e-01 -6.32672369e-01
4.66513872e-01 -9.44457591e-01 -7.60071695e-01 2.51598567e-01
-3.03572714e-01 -5.46569079e-02 6.03299677e-01 -5.07796586e-01
-3.98560315e-01 6.08518673e-03 -2.17741594e-01 -4.26913530e-01
-9.61939543e-02 2.11000547e-01 2.48692408e-01 2.28028134e-01
4.04975504e-01 4.83824909e-01 3.90821159e-01 -4.28904355e-01
5.06018758e-01 7.28107750e-01 4.43916440e-01 1.58408597e-01
7.76477277e-01 6.03192866e-01 -1.46916613e-01 -1.33070791e+00
2.65881140e-02 -8.14053535e-01 -1.43038407e-01 -4.61030304e-02
1.31871402e+00 -6.56512678e-01 -1.22795510e+00 2.34422833e-01
-1.32171440e+00 7.10574910e-02 4.18723434e-01 5.87071836e-01
-3.51521492e-01 3.11474681e-01 -9.85826671e-01 -5.91372490e-01
-7.42919028e-01 -1.11486018e+00 9.30630028e-01 -1.46094397e-01
-4.55974996e-01 -7.52531171e-01 7.37899363e-01 2.42112383e-01
5.15339196e-01 4.21607107e-01 1.37629962e+00 -1.18223667e+00
-4.55244660e-01 1.01593666e-01 -1.10338464e-01 8.34908336e-02
3.45982492e-01 1.04007706e-01 -5.56475163e-01 -5.59559107e-01
-8.13044235e-03 5.94526157e-02 9.16791022e-01 4.79837596e-01
1.90722197e-01 -6.20072365e-01 -8.78909767e-01 8.07866871e-01
1.37644231e+00 4.49194014e-01 3.87697726e-01 2.54453048e-02
4.80577171e-01 7.36096203e-01 3.44220310e-01 3.19330484e-01
1.40658095e-01 7.01807916e-01 4.54411447e-01 3.38422090e-01
4.40476984e-01 -1.60560951e-01 2.69159645e-01 7.68790781e-01
9.67904702e-02 -2.33399421e-01 -7.19807446e-01 4.13506597e-01
-1.38204420e+00 -1.22234797e+00 -4.78945598e-02 1.96392739e+00
6.50245488e-01 -6.86976850e-01 1.36778891e-01 -6.29341543e-01
6.50045037e-01 2.04278037e-01 -7.85555720e-01 -5.57003558e-01
-7.22076595e-02 1.73005089e-01 5.95539451e-01 3.17009240e-01
-9.36068416e-01 8.02481651e-01 6.42085171e+00 3.81581396e-01
-1.26527274e+00 -1.25989348e-01 4.43508662e-03 4.19703033e-03
-5.20620704e-01 -3.57070625e-01 -6.34179294e-01 5.21443367e-01
9.21834171e-01 -2.05309615e-01 7.14890361e-01 7.41600156e-01
5.61336987e-02 4.98488158e-01 -9.87658620e-01 1.10921896e+00
7.92680606e-02 -1.53647161e+00 -1.97184309e-02 3.19754839e-01
3.12754303e-01 8.53122890e-01 -3.31234753e-01 1.45247087e-01
3.60636145e-01 -1.07591820e+00 -3.18160951e-01 1.04599416e-01
6.56199276e-01 -8.31273139e-01 7.89596379e-01 1.57021374e-01
-1.16892898e+00 1.86251193e-01 -3.97490978e-01 5.65932810e-01
7.87670553e-01 2.92522013e-01 -1.26583672e+00 -9.47527140e-02
3.95955175e-01 4.05691504e-01 -2.37756521e-01 8.10272932e-01
1.46926805e-01 5.19905925e-01 -4.24308807e-01 -5.61050847e-02
2.71164000e-01 -5.85525095e-01 6.81327105e-01 1.25587201e+00
1.02694720e-01 3.90932083e-01 7.45193139e-02 8.11292171e-01
1.85962394e-01 2.22677201e-01 -8.86385322e-01 -6.36878610e-01
4.78134245e-01 1.21781027e+00 -5.41887403e-01 -2.90657699e-01
-1.24343470e-01 9.00479555e-01 2.36995313e-02 2.80361414e-01
-8.38781834e-01 -1.91890225e-01 1.94160521e+00 1.71343014e-01
9.03283417e-01 -1.34604603e-01 4.59426880e-01 -9.02639508e-01
-6.62328780e-01 -9.74457741e-01 3.16390932e-01 -3.33837569e-01
-1.57102787e+00 7.02620804e-01 -2.37459049e-01 -9.75439906e-01
-6.68390036e-01 -7.87304342e-01 -5.51580012e-01 7.47827828e-01
-1.10978544e+00 -1.06966937e+00 2.18641520e-01 4.54381675e-01
1.56940565e-01 -1.32873937e-01 8.57815266e-01 -1.82594247e-02
-1.81224719e-01 1.53573826e-01 6.76226258e-01 1.01901859e-03
4.41372842e-01 -5.47791958e-01 7.94360280e-01 2.34532833e-01
-6.38644025e-02 1.15419304e+00 9.90883768e-01 -1.22419608e+00
-2.04581761e+00 -1.31764984e+00 1.03595316e+00 -3.27385008e-01
8.01477551e-01 -5.41149557e-01 -9.38992441e-01 7.47621179e-01
2.18100473e-01 -4.17914391e-01 1.14873242e+00 -3.37658882e-01
-8.68603468e-01 4.17096704e-01 -1.56358409e+00 6.78311348e-01
6.03136897e-01 -5.43637872e-01 -7.35280037e-01 3.62201542e-01
8.59828532e-01 2.55039364e-01 -5.57849050e-01 3.73354107e-01
1.17941809e+00 -8.34435880e-01 1.26324272e+00 -1.19540131e+00
-2.04427719e-01 -5.60885251e-01 -2.56334513e-01 -1.48777783e+00
-3.86259109e-01 -2.44055644e-01 4.80605736e-02 5.41323543e-01
-4.44992147e-02 -8.78309429e-01 3.01291585e-01 -4.11584198e-01
2.42651403e-01 -7.26829529e-01 -8.27772737e-01 -8.67223203e-01
-3.38731110e-02 -1.09339610e-01 8.42800021e-01 9.44651306e-01
-2.85661489e-01 4.57269609e-01 -5.69044471e-01 8.29951763e-02
5.91408670e-01 4.18321431e-01 7.42446601e-01 -1.45778358e+00
-2.72885650e-01 -3.31078678e-01 -6.54880822e-01 -6.37362957e-01
7.67278969e-02 -1.26470792e+00 -1.06411070e-01 -1.45140946e+00
2.03378275e-01 -1.01056218e-01 -3.65073115e-01 2.52827525e-01
-5.07031158e-02 2.33318657e-01 2.44357407e-01 1.45501122e-01
1.76196396e-01 2.91890442e-01 8.46059322e-01 -2.65205175e-01
-3.39095294e-01 -1.28697336e-01 -3.82371962e-01 6.57124877e-01
7.91495442e-01 -5.34025848e-01 -1.64910927e-01 -1.83677822e-01
3.72529328e-02 -1.05639599e-01 2.49519661e-01 -1.58771992e-01
-2.85691142e-01 -5.75679064e-01 9.77494717e-02 -1.19092441e+00
4.17637318e-01 -9.51224685e-01 6.55450344e-01 1.25456405e+00
2.63721645e-01 3.49164695e-01 2.05094934e-01 6.88443601e-01
3.99904341e-01 2.82526672e-01 8.30974817e-01 1.95268065e-01
-4.53664213e-01 5.27227163e-01 -9.85649288e-01 7.89083466e-02
1.19004321e+00 -2.92496663e-03 -6.22048795e-01 -7.04903305e-02
-2.10109726e-02 1.78739414e-01 7.44512141e-01 7.29201317e-01
8.06419909e-01 -1.12078226e+00 -1.10185742e+00 5.36237538e-01
3.38914275e-01 -9.36090112e-01 3.60005498e-01 7.87546456e-01
-8.88784945e-01 7.74695814e-01 -5.61529100e-01 -8.06643009e-01
-1.63396966e+00 1.17575920e+00 -1.59955323e-01 -5.28591424e-02
-5.41276515e-01 6.66373491e-01 3.53289843e-01 -6.93404555e-01
-7.81780183e-02 -1.74985930e-01 -4.24694240e-01 1.54757440e-01
8.39389026e-01 4.18490797e-01 -3.66453320e-01 -1.16833782e+00
-9.42213833e-01 4.39547181e-01 -1.88630491e-01 6.80275500e-01
1.66918612e+00 4.64248419e-01 -5.51854253e-01 -7.84950622e-04
1.44306922e+00 1.58991769e-03 -5.19741952e-01 1.83873191e-01
-9.88132283e-02 -2.07117409e-01 -5.57158172e-01 -5.61955035e-01
-7.81086147e-01 6.11415684e-01 9.55152094e-01 -2.57310599e-01
6.70731425e-01 1.55973479e-01 1.37776315e+00 2.66157120e-01
5.42613506e-01 -3.17953348e-01 -7.09719896e-01 5.12700737e-01
7.25705981e-01 -9.03395653e-01 -1.79608047e-01 -1.91665545e-01
-6.04348719e-01 4.78230864e-01 1.96167871e-01 -1.28103748e-01
5.25973380e-01 4.11484033e-01 2.53793210e-01 -3.32558185e-01
-1.05589056e+00 -5.96278869e-02 1.45243943e-01 8.95482123e-01
2.41421759e-01 5.71629941e-01 -6.47743523e-01 3.58597696e-01
-5.40270098e-02 -9.37857106e-02 1.19491272e-01 8.58807206e-01
-6.79575920e-01 -1.19494486e+00 -3.66051972e-01 8.73236835e-01
-1.78857684e-01 -3.68002146e-01 -5.59775412e-01 4.54722315e-01
6.16159514e-02 6.68824017e-01 3.81841213e-01 -4.25589114e-01
-2.97044013e-02 1.72005638e-01 2.70209134e-01 -2.96503723e-01
-8.15787077e-01 3.22741307e-02 -1.71634734e-01 -6.02388144e-01
-3.57244983e-02 -4.29399163e-01 -1.27367616e+00 -7.09264219e-01
-1.33121789e-01 1.80256888e-01 7.34178841e-01 5.88472843e-01
6.78143322e-01 -1.28622487e-01 5.23944676e-01 -5.10858655e-01
-7.08247900e-01 -8.59482348e-01 -5.16235650e-01 6.43778622e-01
4.91185367e-01 -5.49813807e-01 -3.89302135e-01 -1.14878677e-01] | [4.950372219085693, 5.278842926025391] |
e540419c-9324-409d-8ce7-2b3f35f13db6 | are-you-stealing-my-model-sample-correlation | 2210.15427 | null | https://arxiv.org/abs/2210.15427v1 | https://arxiv.org/pdf/2210.15427v1.pdf | Are You Stealing My Model? Sample Correlation for Fingerprinting Deep Neural Networks | An off-the-shelf model as a commercial service could be stolen by model stealing attacks, posing great threats to the rights of the model owner. Model fingerprinting aims to verify whether a suspect model is stolen from the victim model, which gains more and more attention nowadays. Previous methods always leverage the transferable adversarial examples as the model fingerprint, which is sensitive to adversarial defense or transfer learning scenarios. To address this issue, we consider the pairwise relationship between samples instead and propose a novel yet simple model stealing detection method based on SAmple Correlation (SAC). Specifically, we present SAC-w that selects wrongly classified normal samples as model inputs and calculates the mean correlation among their model outputs. To reduce the training time, we further develop SAC-m that selects CutMix Augmented samples as model inputs, without the need for training the surrogate models or generating adversarial examples. Extensive results validate that SAC successfully defends against various model stealing attacks, even including adversarial training or transfer learning, and detects the stolen models with the best performance in terms of AUC across different datasets and model architectures. The codes are available at https://github.com/guanjiyang/SAC. | ['Ran He', 'Jian Liang', 'Jiyang Guan'] | 2022-10-21 | null | null | null | null | ['adversarial-defense'] | ['adversarial'] | [ 4.68152642e-01 -2.07099125e-01 -3.92548531e-01 -3.93384755e-01
-9.75718141e-01 -1.15504670e+00 5.76590478e-01 -3.57094914e-01
-9.49742720e-02 7.02735841e-01 -6.95562780e-01 -5.84674656e-01
1.94936782e-01 -8.44379723e-01 -1.03917503e+00 -7.03375041e-01
8.10248107e-02 4.33685005e-01 1.55002698e-01 1.29613891e-01
1.94886476e-01 6.30230248e-01 -7.61341929e-01 3.40305686e-01
9.62530077e-01 1.15530849e+00 -4.72133100e-01 4.46604460e-01
2.28322059e-01 1.89286068e-01 -7.38577962e-01 -9.81559932e-01
6.32202864e-01 -4.06266123e-01 -1.83316290e-01 -5.94848335e-01
4.92761075e-01 -3.81362319e-01 -3.46350789e-01 1.27141404e+00
3.83683830e-01 -4.81118262e-01 4.84811783e-01 -1.92922521e+00
-4.77068007e-01 4.91609901e-01 -4.24952954e-01 -1.14499792e-01
5.58696948e-02 4.93593335e-01 5.40123641e-01 -8.94115210e-01
4.87011164e-01 1.16826463e+00 9.54284787e-01 6.73920572e-01
-1.33037233e+00 -1.63519216e+00 -3.62472177e-01 1.31505638e-01
-1.59228110e+00 -4.19881940e-01 7.66334355e-01 -4.42186832e-01
1.56852111e-01 6.87605917e-01 2.57577807e-01 1.73648918e+00
3.70531201e-01 7.76911676e-01 1.30506742e+00 3.66866812e-02
1.09641671e-01 5.43699920e-01 2.38286164e-02 4.97028500e-01
6.38159633e-01 7.77772248e-01 -2.91154802e-01 -8.63854110e-01
4.60448444e-01 7.93682262e-02 -2.71879047e-01 -5.09266973e-01
-9.07780290e-01 8.73656034e-01 3.63930702e-01 -1.64758950e-01
7.81592056e-02 -6.77913874e-02 4.85971451e-01 3.73431355e-01
1.62305474e-01 5.27339399e-01 -5.29877603e-01 4.85235631e-01
-9.73326623e-01 1.30448550e-01 7.81454444e-01 8.43057215e-01
4.08694595e-01 1.40461503e-02 -7.34942593e-03 3.62624556e-01
3.36553186e-01 8.77744675e-01 3.52458149e-01 -4.01583999e-01
5.96913695e-01 4.01324272e-01 2.55675882e-01 -1.03761387e+00
1.64371386e-01 -6.31772220e-01 -7.95902610e-01 1.05837427e-01
4.58073109e-01 -1.44739822e-01 -6.64278209e-01 1.53444338e+00
4.63692158e-01 6.78587377e-01 -4.60994728e-02 5.26873410e-01
3.75293851e-01 3.38373423e-01 4.87245768e-02 9.35093090e-02
1.22983575e+00 -8.31100225e-01 -2.80874610e-01 -1.78409040e-01
6.68671608e-01 -7.19517827e-01 8.23015928e-01 3.10362786e-01
-5.09572804e-01 -3.84767532e-01 -1.15464330e+00 5.41431725e-01
-5.62252164e-01 -5.45075051e-02 4.41842556e-01 1.26169825e+00
-3.66021395e-01 7.03672767e-01 -8.08255434e-01 3.64241228e-02
8.52955639e-01 4.90234792e-01 -5.62434673e-01 -1.63849801e-01
-1.61228335e+00 5.23964345e-01 2.06505522e-01 1.44287139e-01
-1.14443076e+00 -9.39850807e-01 -7.62298465e-01 -6.49307594e-02
3.12005699e-01 -3.76885772e-01 1.02746499e+00 -8.33784461e-01
-9.49722052e-01 8.02419543e-01 1.61883086e-01 -7.22718239e-01
1.23261285e+00 -3.49704735e-02 -9.14907038e-01 -1.60484761e-01
-6.46247640e-02 2.79391021e-01 1.21996701e+00 -1.29815602e+00
-1.20400146e-01 -3.40828449e-01 -3.80014241e-01 -6.46324694e-01
-2.57719040e-01 1.34156272e-01 -2.50724424e-02 -8.80910337e-01
-9.67656299e-02 -1.25908279e+00 3.25520821e-02 3.11836749e-01
-7.94504225e-01 4.09930915e-01 1.12297654e+00 -8.01852703e-01
1.15876448e+00 -2.17019987e+00 -6.99364007e-01 6.71034932e-01
1.55864552e-01 9.33697581e-01 -1.31083235e-01 3.67804110e-01
-8.41392279e-02 4.78507966e-01 -4.65575069e-01 -5.41884750e-02
7.76354671e-02 -3.32447179e-02 -6.64046407e-01 6.11298323e-01
1.75282180e-01 1.09239089e+00 -8.55769813e-01 -4.42412853e-01
-5.25274016e-02 5.20570338e-01 -1.54697716e-01 2.10320130e-01
1.00542918e-01 5.03514826e-01 -7.03693449e-01 7.30044663e-01
1.41268122e+00 -2.71525621e-01 1.99907064e-01 -1.80656482e-02
7.00612307e-01 -6.21005520e-02 -1.09496188e+00 8.07036698e-01
-1.86872289e-01 3.17110002e-01 1.59909925e-03 -5.18120587e-01
1.24391592e+00 1.09297596e-01 1.02987252e-01 -3.84899288e-01
8.64805803e-02 4.76650178e-01 1.62648469e-01 -1.70079410e-01
-6.22209273e-02 -1.86533436e-01 -1.51143610e-01 4.55992669e-01
-3.56773496e-01 2.88539439e-01 -4.60497737e-01 4.12779339e-02
9.26074803e-01 1.00998692e-01 8.87974203e-02 7.15604350e-02
7.47689307e-01 -1.96444467e-01 7.05221057e-01 7.49033451e-01
-3.77188802e-01 3.06852072e-01 5.64219534e-01 -1.43341497e-01
-7.10393846e-01 -1.36224711e+00 -2.62168378e-01 6.68976963e-01
2.79620737e-01 5.78151178e-03 -7.58200765e-01 -1.22647524e+00
6.42939210e-01 7.28447378e-01 -7.61622429e-01 -6.46237850e-01
-4.01395559e-01 -5.26060581e-01 1.30273283e+00 3.59025806e-01
4.99169439e-01 -8.58852625e-01 -7.35063851e-02 -7.45924860e-02
3.69654894e-02 -8.91916692e-01 -7.42366731e-01 -2.21581832e-01
-6.87854052e-01 -1.49390900e+00 -6.18909240e-01 -4.45953339e-01
7.54307985e-01 8.00611451e-02 6.46593213e-01 1.11926287e-01
-1.80219457e-01 -2.75698990e-01 -2.37759613e-02 -5.23168564e-01
-9.53068614e-01 6.88015968e-02 1.45228922e-01 4.97546941e-01
3.10559392e-01 -2.03384086e-01 -4.99930501e-01 7.95158565e-01
-9.19955254e-01 -1.47512048e-01 5.50685287e-01 8.37127686e-01
4.90539551e-01 -2.05861315e-01 7.65934944e-01 -1.44500756e+00
4.09426898e-01 -7.31710553e-01 -8.14875007e-01 5.94893873e-01
-8.01908553e-01 -4.99017239e-02 9.19152677e-01 -6.69369519e-01
-8.19741189e-01 7.14340732e-02 1.22667782e-01 -8.78341734e-01
1.83348469e-02 4.07682993e-02 -5.62994182e-01 -4.42106366e-01
6.81125641e-01 3.28971416e-01 2.68837601e-01 -5.36253929e-01
1.84762664e-02 6.49615169e-01 4.10059750e-01 -3.31562638e-01
1.37160385e+00 3.85104477e-01 4.08667922e-02 -2.36392125e-01
-4.25006986e-01 -1.93661153e-01 -3.25386554e-01 -3.07429507e-02
1.91118687e-01 -5.85812509e-01 -6.56660676e-01 8.92006993e-01
-9.43284512e-01 -8.16693455e-02 3.38098168e-01 4.68590170e-01
-1.50654495e-01 5.30874372e-01 -4.76366073e-01 -6.58452272e-01
-5.44305623e-01 -1.10357547e+00 6.36231840e-01 -7.62204677e-02
-9.72971916e-02 -8.75844300e-01 -1.75369844e-01 5.87390840e-01
5.39713144e-01 7.79449522e-01 6.66245222e-01 -1.34048986e+00
-5.30437648e-01 -1.05255508e+00 -1.44682169e-01 6.39613509e-01
3.15367915e-02 4.62117791e-02 -1.11267340e+00 -5.59598684e-01
-8.05038866e-03 -1.21589378e-01 5.80552578e-01 -1.72216315e-02
1.22800887e+00 -5.09819329e-01 -6.86031997e-01 7.16419876e-01
1.17225790e+00 1.90921530e-01 6.08413279e-01 9.85614657e-02
7.12069869e-01 4.03353959e-01 6.23592496e-01 2.75195360e-01
-2.40558043e-01 6.45410597e-01 4.35365856e-01 -1.45019412e-01
3.87847722e-01 -7.03467190e-01 3.32655340e-01 1.28637880e-01
6.21783376e-01 -2.81524301e-01 -9.27661538e-01 -2.54748669e-02
-1.49727702e+00 -1.01370001e+00 -1.33267224e-01 2.57023907e+00
6.26092255e-01 3.40860158e-01 -3.81383151e-02 -6.44226074e-02
1.17417812e+00 6.62849322e-02 -1.06204462e+00 -3.27367991e-01
-5.07334471e-02 2.38047227e-01 8.03216934e-01 2.94093192e-01
-1.17516482e+00 7.84426272e-01 5.44783020e+00 9.55717504e-01
-1.18238688e+00 1.86381772e-01 8.38356495e-01 -3.38781849e-02
-2.59430051e-01 -7.36286715e-02 -8.06955218e-01 9.44114447e-01
9.38138902e-01 -4.04224724e-01 2.57628232e-01 1.09191823e+00
1.49332911e-01 4.72973615e-01 -1.14921010e+00 6.73898757e-01
-1.60738453e-02 -1.22915637e+00 4.17623855e-02 3.31254154e-01
6.04640365e-01 -9.94659364e-02 3.24862778e-01 4.27636236e-01
1.35782272e-01 -1.11021698e+00 4.99018341e-01 5.72492719e-01
7.55494714e-01 -7.88923323e-01 9.07407701e-01 4.42114711e-01
-9.50991750e-01 1.58787012e-01 -1.23406999e-01 5.89508414e-01
-7.78105780e-02 1.97319239e-01 -1.01763916e+00 5.69278300e-01
3.28167945e-01 4.40283567e-01 -8.00332904e-01 1.07514167e+00
-1.78316981e-01 8.71855855e-01 -3.50284308e-01 1.95434198e-01
-4.64089662e-02 5.39951697e-02 4.55544949e-01 9.43297684e-01
2.96365529e-01 -5.24827480e-01 -1.16900600e-01 1.08170283e+00
-2.88510352e-01 -7.64604360e-02 -6.54844940e-01 -1.51646838e-01
8.30001652e-01 1.06281710e+00 -3.70173514e-01 -2.30653331e-01
3.45421061e-02 8.63975823e-01 -2.49843746e-01 1.58400849e-01
-1.16146052e+00 -3.70944679e-01 6.41078472e-01 2.38249645e-01
2.35962823e-01 3.14626187e-01 -6.83785975e-02 -1.01969028e+00
2.55869895e-01 -1.14205253e+00 4.96027142e-01 -3.00005615e-01
-1.59331930e+00 4.52639669e-01 -2.06261009e-01 -1.66608131e+00
-2.48037845e-01 -4.75616097e-01 -7.53745914e-01 1.01151145e+00
-1.33226228e+00 -1.28347433e+00 -1.69326961e-01 4.79444891e-01
-1.50586203e-01 -3.34037513e-01 7.94025660e-01 2.56254345e-01
-5.93433976e-01 1.41388237e+00 5.19907236e-01 4.31835055e-01
9.31599200e-01 -6.24162078e-01 8.74751389e-01 8.41759980e-01
-6.46756515e-02 7.34609604e-01 4.59278733e-01 -7.26330101e-01
-1.30851972e+00 -1.26615405e+00 7.00014710e-01 -6.06171906e-01
7.46449530e-01 -5.36306024e-01 -1.21592844e+00 6.30584121e-01
-5.24028122e-01 5.39716899e-01 1.03469443e+00 -5.42428493e-01
-9.60066974e-01 -2.32291624e-01 -1.78337252e+00 1.75049141e-01
6.56286597e-01 -7.07330287e-01 -4.25281525e-02 1.82322159e-01
5.28150439e-01 -2.50300586e-01 -8.53517473e-01 3.99307519e-01
9.34222877e-01 -7.99202561e-01 1.08053219e+00 -9.48139191e-01
8.51173997e-02 -3.46487641e-01 -2.93555669e-02 -8.63830566e-01
2.05191270e-01 -6.69520974e-01 -3.73558849e-01 1.31990445e+00
5.74180245e-01 -1.13878918e+00 1.08508420e+00 7.14308262e-01
3.89524817e-01 -7.07659006e-01 -9.17643309e-01 -1.30011475e+00
2.31422618e-01 -2.91191220e-01 1.20887625e+00 1.31233406e+00
-5.61111748e-01 -3.37014705e-01 -7.04121649e-01 4.73280698e-01
1.03673887e+00 3.00913639e-02 9.68647420e-01 -1.22865975e+00
-4.11127120e-01 -2.33593553e-01 -2.84067780e-01 -4.52092677e-01
2.88070768e-01 -1.13644171e+00 -3.06353837e-01 -5.81523478e-01
-1.41625464e-01 -7.14694202e-01 -6.25086248e-01 4.47450101e-01
-1.00177020e-01 4.42647725e-01 3.46037358e-01 4.23173487e-01
-2.37393290e-01 1.97995290e-01 1.19817436e+00 -2.65020311e-01
6.34396225e-02 6.17856443e-01 -6.43432021e-01 4.58364636e-01
1.01007330e+00 -1.01348352e+00 -3.19184393e-01 2.03818351e-01
-2.35860229e-01 1.41679198e-01 7.33965635e-01 -8.58568609e-01
-3.19455080e-02 -2.26425573e-01 4.50308800e-01 -3.66330117e-01
1.52156845e-01 -1.04537284e+00 7.81537890e-01 1.02997124e+00
-3.74084413e-01 -2.11799443e-01 1.93101808e-01 6.84750736e-01
9.17401314e-02 -2.27264598e-01 9.89527106e-01 1.37850568e-01
-3.10786605e-01 5.90372622e-01 2.07958177e-01 -1.22692294e-01
1.27121365e+00 -2.34168634e-01 -5.25337696e-01 8.14016387e-02
-6.80661619e-01 2.59565413e-01 5.58042288e-01 5.11697531e-01
4.35973972e-01 -1.34431207e+00 -6.64421260e-01 7.12295055e-01
3.69080454e-01 -5.58802187e-01 2.97555834e-01 5.71108282e-01
-4.05332893e-01 2.71572798e-01 -2.36467049e-01 -3.58610541e-01
-1.54043078e+00 7.02683687e-01 6.40815914e-01 -2.69204855e-01
-1.15336500e-01 4.95511532e-01 9.05157477e-02 -6.87464952e-01
-1.28341541e-02 1.39597982e-01 4.19138283e-01 -2.10763589e-01
4.43740994e-01 5.46595454e-01 -5.21491282e-02 -7.27951944e-01
-6.08502805e-01 8.31537023e-02 -3.72836262e-01 4.18894023e-01
8.18189323e-01 3.77272457e-01 2.07851633e-01 -6.96661323e-02
1.49200046e+00 -7.85419866e-02 -1.29167128e+00 -2.63958126e-01
1.01992330e-02 -9.07498240e-01 -4.16765422e-01 -9.47323322e-01
-1.04990172e+00 8.52779746e-01 8.23997378e-01 1.92610294e-01
5.80797553e-01 -3.00144166e-01 1.04522645e+00 1.25335783e-01
4.80599135e-01 -4.88456875e-01 -2.34364420e-01 -6.39309436e-02
7.03605354e-01 -1.34000337e+00 -1.46378845e-01 -4.71016258e-01
-6.09775424e-01 7.17777252e-01 6.03657007e-01 -7.32420245e-03
6.92785442e-01 1.90014411e-02 1.47040650e-01 2.62881488e-01
-2.32586354e-01 6.73752844e-01 2.25388750e-01 7.81090796e-01
-2.80319631e-01 4.33356792e-01 2.58626007e-02 8.58374894e-01
-1.64476112e-01 8.70024562e-02 1.62006751e-01 7.34975755e-01
5.42091914e-02 -1.49182296e+00 -5.81661105e-01 7.34568119e-01
-5.10953903e-01 -4.05874364e-02 -6.05327189e-01 7.46321380e-01
5.41283973e-02 6.27518952e-01 -3.45017135e-01 -7.05749512e-01
2.67957687e-01 2.32049868e-01 -1.84951033e-02 -1.25764921e-01
-8.51259887e-01 -3.54946166e-01 -1.84627205e-01 -3.63431573e-01
1.84143901e-01 -4.71328497e-01 -7.75693834e-01 -6.84800744e-01
-6.20405555e-01 1.81861937e-01 4.99136448e-01 5.44029534e-01
6.38780296e-01 -1.12617895e-01 1.21779358e+00 -4.54568416e-01
-1.19800675e+00 -7.84582555e-01 -6.93350971e-01 5.46205163e-01
3.74246001e-01 -5.82114339e-01 -5.27372062e-01 -2.27065340e-01] | [5.769282817840576, 7.444635391235352] |
717e10f9-2192-4237-945b-29be058506e6 | document-based-recommender-system-for-job | null | null | https://aclanthology.org/N18-3027 | https://aclanthology.org/N18-3027.pdf | Document-based Recommender System for Job Postings using Dense Representations | Job boards and professional social networks heavily use recommender systems in order to better support users in exploring job advertisements. Detecting the similarity between job advertisements is important for job recommendation systems as it allows, for example, the application of item-to-item based recommendations. In this work, we research the usage of dense vector representations to enhance a large-scale job recommendation system and to rank German job advertisements regarding their similarity. We follow a two-folded evaluation scheme: (1) we exploit historic user interactions to automatically create a dataset of similar jobs that enables an offline evaluation. (2) In addition, we conduct an online A/B test and evaluate the best performing method on our platform reaching more than 1 million users. We achieve the best results by combining job titles with full-text job descriptions. In particular, this method builds dense document representation using words of the titles to weigh the importance of words of the full-text description. In the online evaluation, this approach allows us to increase the click-through rate on job recommendations for active users by 8.0{\%}. | ['Martin Riedl', 'Chris Biemann', 'Ahmed Elsafty'] | 2018-06-01 | null | null | null | naacl-2018-6 | ['document-embedding'] | ['methodology'] | [-4.94443029e-02 8.29230249e-02 -2.71535784e-01 -4.96438712e-01
-3.49386185e-01 -3.80022556e-01 8.16884100e-01 5.13211071e-01
-7.33310521e-01 3.92347217e-01 4.97832566e-01 -2.89250582e-01
-5.27720094e-01 -8.16433132e-01 -2.81902313e-01 -2.47233853e-01
1.09387875e-01 9.10479844e-01 3.00107628e-01 -4.90217626e-01
4.77217376e-01 4.52486217e-01 -1.72740054e+00 2.60570794e-01
4.95454878e-01 7.54366338e-01 4.79751259e-01 6.36893928e-01
-4.61819828e-01 3.04041535e-01 -5.87936163e-01 -7.37072587e-01
1.50463998e-01 -1.90380707e-01 -7.21370339e-01 8.44625607e-02
4.71747398e-01 -1.95349380e-02 -4.26295459e-01 7.91111708e-01
4.80940670e-01 8.01106095e-01 5.78191102e-01 -5.56553602e-01
-5.82916200e-01 8.98617685e-01 -3.51261586e-01 4.49482530e-01
4.20627326e-01 -3.34438592e-01 1.16309321e+00 -8.97792041e-01
7.27365851e-01 1.03199530e+00 3.98778766e-01 3.47525179e-01
-1.05332196e+00 -4.68212157e-01 2.47716531e-01 1.77424133e-01
-1.18788850e+00 -3.01867723e-01 6.40944898e-01 -6.39851034e-01
8.95246804e-01 5.23978651e-01 4.93294418e-01 1.03130352e+00
-4.53939289e-02 4.98762310e-01 8.06049228e-01 -4.22743440e-01
1.16842370e-02 6.16303027e-01 7.24522233e-01 4.64830846e-01
1.57025799e-01 -1.91179723e-01 -3.38679492e-01 -3.18086952e-01
4.85524356e-01 4.89587724e-01 2.08369404e-01 -9.13716704e-02
-9.71703231e-01 9.34678733e-01 4.10015851e-01 8.07609379e-01
-5.28776467e-01 -1.56826243e-01 3.67577076e-01 2.14812681e-01
4.16772872e-01 9.22360241e-01 1.29803985e-01 -4.52521965e-02
-8.44386280e-01 4.17556822e-01 7.16709197e-01 4.06607538e-01
7.78143466e-01 -4.81244594e-01 -6.37545645e-01 1.20131481e+00
1.65078029e-01 1.72495335e-01 8.47716749e-01 -6.71887457e-01
2.32179686e-01 4.27702129e-01 1.94956618e-03 -1.17336214e+00
-3.64416867e-01 -9.92489219e-01 -6.52944088e-01 -1.99902073e-01
2.54440933e-01 8.56097266e-02 -4.50239658e-01 1.09555840e+00
3.18822980e-01 1.14416234e-01 -1.97420821e-01 9.34918523e-01
7.39471972e-01 4.48310494e-01 1.15420267e-01 -2.35992149e-01
1.58214080e+00 -9.46049392e-01 -5.74597299e-01 -2.26467833e-01
4.45646524e-01 -8.71965230e-01 1.09659195e+00 2.92480677e-01
-8.23401093e-01 -9.81115103e-01 -7.32322037e-01 1.40803620e-01
-4.12704736e-01 3.24669540e-01 8.07523549e-01 7.50921965e-01
-6.05484903e-01 9.60148215e-01 -4.35127378e-01 -3.84631842e-01
-3.85770947e-02 3.33341837e-01 -2.51523554e-01 -5.19373901e-02
-1.22960734e+00 6.23300493e-01 1.25229463e-01 -2.43584096e-01
-4.51157093e-01 -3.56478244e-01 -7.08202899e-01 3.77579927e-01
5.24677336e-01 -4.73526567e-01 1.14754879e+00 -5.26205480e-01
-1.20194864e+00 6.69639766e-01 -3.96444835e-02 -4.13795233e-01
2.24014342e-01 -1.45937666e-01 -5.12062371e-01 -2.28908554e-01
1.31865099e-01 4.98828962e-02 6.29529297e-01 -8.97970140e-01
-6.41506612e-01 -4.29690301e-01 2.21260339e-01 1.64298475e-01
-9.53895330e-01 -3.83546315e-02 -9.38389957e-01 -5.05522013e-01
-3.34108174e-02 -1.04515374e+00 -5.10003567e-01 -9.18457031e-01
-2.63105690e-01 -6.70780540e-01 9.97901633e-02 -4.13177341e-01
1.45215321e+00 -2.11361384e+00 -6.64226115e-02 6.13145709e-01
3.74295503e-01 1.90246716e-01 -1.54859394e-01 6.12166286e-01
2.78620005e-01 -7.96491206e-02 6.35800004e-01 -5.86920738e-01
9.25979242e-02 -9.14522074e-03 -1.91003770e-01 1.00108087e-01
-3.45019549e-01 6.49373651e-01 -8.55181813e-01 -3.44817430e-01
4.67670560e-02 2.95299828e-01 -5.72615623e-01 1.93662181e-01
1.41846180e-01 7.49176443e-02 -6.91769481e-01 1.70820892e-01
1.75274476e-01 -3.20292711e-01 5.78624725e-01 -7.88536146e-02
-5.20576686e-02 4.70628351e-01 -1.09883213e+00 1.39633000e+00
-6.37780547e-01 4.89091456e-01 -1.96775571e-01 -7.17825115e-01
1.10836232e+00 -9.24732536e-02 4.43465918e-01 -8.61737967e-01
-2.24840138e-02 4.88145389e-02 -1.54213399e-01 -2.71171331e-01
1.05249465e+00 1.81617245e-01 -1.67053461e-01 5.29842615e-01
-2.23356113e-01 5.55568397e-01 6.74223483e-01 5.25518656e-01
1.24357152e+00 -3.26672703e-01 2.69223660e-01 -1.18787967e-01
6.69106662e-01 -1.69077650e-01 2.15893790e-01 1.02546716e+00
2.45402947e-01 2.93662399e-01 3.68423492e-01 -4.86132026e-01
-8.30321968e-01 -6.47161186e-01 -2.91955248e-02 1.70152187e+00
-2.31310222e-02 -7.59286284e-01 -2.50994712e-01 -9.49666560e-01
2.90215284e-01 7.35295057e-01 -5.75015426e-01 1.50472131e-02
-5.05379617e-01 -2.21515968e-01 1.40629500e-01 4.83620852e-01
-1.20786116e-01 -1.15812898e+00 -9.19442028e-02 2.35469654e-01
7.23395646e-02 -7.08238006e-01 -5.54741263e-01 5.55103421e-02
-5.48680604e-01 -7.53651261e-01 -9.11717534e-01 -5.88781714e-01
6.40281856e-01 4.27123904e-01 1.10719252e+00 1.40335619e-01
-2.35936210e-01 2.66826361e-01 -5.18626273e-01 1.26137793e-01
-2.59390771e-01 4.23458487e-01 2.66751826e-01 -5.26281260e-02
3.24648857e-01 -4.30030793e-01 -5.88244319e-01 7.12328017e-01
-7.98049688e-01 -4.29335922e-01 8.14268410e-01 7.34070122e-01
5.02125084e-01 3.10217086e-02 4.92606878e-01 -1.45537329e+00
1.16453791e+00 -7.09398925e-01 -4.12476987e-01 7.07269236e-02
-1.23866367e+00 5.97337186e-02 5.06844997e-01 -6.54937446e-01
-7.89561391e-01 -3.59631726e-03 -5.98238051e-01 -3.86098564e-01
-5.73523492e-02 7.19946444e-01 4.79307562e-01 3.26223493e-01
9.09289837e-01 7.47174397e-02 -3.32296014e-01 -9.71767902e-01
3.82438689e-01 9.35617387e-01 4.59083527e-01 -3.74148250e-01
6.45443380e-01 -5.40516302e-02 -3.02832276e-01 -7.59591997e-01
-7.37240493e-01 -1.22024941e+00 -2.20387295e-01 -1.95612729e-01
5.30985594e-01 -6.03592634e-01 -1.06582189e+00 -5.16585171e-01
-8.16208303e-01 7.95647278e-02 -2.46953577e-01 7.80345917e-01
-7.82779157e-02 5.53103030e-01 -5.54309309e-01 -6.68314993e-01
-1.63076982e-01 -9.48059916e-01 8.90815854e-01 2.35479072e-01
-3.62968594e-01 -6.10873163e-01 4.11852866e-01 7.13449299e-01
5.13219476e-01 -5.57155132e-01 6.75347149e-01 -1.23210287e+00
-1.48020976e-03 -5.48015237e-01 -1.32237837e-01 1.98106468e-01
-1.07867263e-01 -5.29241085e-01 -8.60182226e-01 -3.98369223e-01
-5.05146682e-01 -6.47368059e-02 1.02132905e+00 1.30247802e-01
1.16308105e+00 -1.46839887e-01 -4.01944965e-01 5.21993972e-02
7.93829441e-01 -1.14259169e-01 4.79021370e-01 2.59609252e-01
4.26528156e-01 6.62545204e-01 7.45356321e-01 6.34967566e-01
1.04984296e-02 1.19718337e+00 1.08878531e-01 1.92063704e-01
1.12734223e-02 -3.53793085e-01 2.00474218e-01 7.39193797e-01
-7.94416890e-02 -3.21161509e-01 -6.46283686e-01 4.01386499e-01
-1.73473883e+00 -1.01682281e+00 -1.18874591e-02 2.52467179e+00
5.27809799e-01 5.25697947e-01 3.42932045e-01 4.30419594e-02
5.83749354e-01 1.33137360e-01 -8.58902857e-02 -4.28009182e-01
6.79143846e-01 1.59088865e-01 3.34641069e-01 3.67809325e-01
-9.32078302e-01 6.69089258e-01 5.51030064e+00 9.27205384e-01
-7.46027291e-01 2.31772050e-01 4.18894172e-01 -1.78328529e-01
-4.44713265e-01 -2.87344813e-01 -1.24212635e+00 3.80224764e-01
9.46883261e-01 -3.07203680e-01 4.10959810e-01 1.22917128e+00
-3.74165028e-02 1.60113022e-01 -8.11603010e-01 8.63631010e-01
1.54077545e-01 -1.48935509e+00 -1.61263853e-01 4.83814925e-01
6.05219424e-01 -1.58074573e-01 5.83988726e-02 9.27290797e-01
1.02211349e-01 -6.56952679e-01 2.24370584e-01 5.48576951e-01
5.02944171e-01 -7.43774295e-01 9.66237962e-01 3.75320017e-01
-1.00852895e+00 -2.27786928e-01 -5.99673808e-01 -3.66898850e-02
5.48226126e-02 9.55326855e-01 -1.22307253e+00 4.38069880e-01
4.83525038e-01 3.88950199e-01 -5.92794716e-01 9.53674316e-01
2.23945484e-01 4.81971204e-01 -9.23207700e-02 -3.63830507e-01
1.65474653e-01 -1.75195336e-01 4.25179064e-01 1.33786440e+00
3.49647641e-01 -2.50299871e-01 5.30706286e-01 3.72820228e-01
-4.23922986e-01 6.11554146e-01 -5.88320792e-01 -4.36211377e-01
2.63810307e-01 1.66420996e+00 -8.37063074e-01 -2.95231193e-01
-2.64495969e-01 9.50881481e-01 4.66756403e-01 4.52914201e-02
-3.81376952e-01 -4.60815161e-01 7.21585512e-01 5.82225144e-01
2.53474027e-01 -2.09584713e-01 4.04828489e-01 -5.54239690e-01
-2.49501720e-01 -5.90103984e-01 4.56205666e-01 -3.53756636e-01
-1.42205119e+00 8.26129377e-01 -1.58194661e-01 -7.97785640e-01
-5.29868186e-01 -4.50368434e-01 -4.22321558e-01 7.89734662e-01
-9.49804783e-01 -6.45033121e-01 -3.28337014e-01 3.49912405e-01
6.68136358e-01 -3.98772836e-01 7.08450615e-01 7.84559488e-01
-4.38699812e-01 6.98644757e-01 2.08281070e-01 3.26378606e-02
7.63666153e-01 -1.20023346e+00 4.91341442e-01 2.47199595e-01
6.41251504e-01 1.03927696e+00 7.16077626e-01 -9.66091931e-01
-1.24217749e+00 -8.75918150e-01 1.26530313e+00 -4.26433325e-01
4.71188337e-01 -1.57131329e-01 -5.86863577e-01 4.22787786e-01
4.28328232e-04 -2.81735301e-01 8.62245202e-01 1.04884458e+00
-2.15737447e-01 -2.18470633e-01 -7.80727088e-01 3.73844534e-01
8.92326653e-01 -4.24052387e-01 -4.72229213e-01 6.35639131e-01
4.72065300e-01 -7.25761279e-02 -1.01421571e+00 -9.16922912e-02
7.70157337e-01 -8.73342097e-01 1.11196184e+00 -7.29608417e-01
4.60045934e-01 -1.85161270e-02 2.14953925e-02 -1.23533416e+00
-8.16935599e-01 -4.16451991e-01 2.46577375e-02 1.17766809e+00
4.81598526e-01 -3.54931802e-01 8.96619558e-01 2.19472662e-01
-8.28275606e-02 -7.13283598e-01 -5.30379593e-01 -6.37333333e-01
-5.97980261e-01 -2.41550639e-01 4.87023801e-01 7.90415168e-01
1.37143195e-01 5.84017098e-01 -5.51314592e-01 -2.03374609e-01
2.13543579e-01 3.82145822e-01 7.09093332e-01 -1.58603501e+00
-7.83753812e-01 -5.75088382e-01 -3.62730473e-01 -1.07367277e+00
4.87776920e-02 -1.12940097e+00 -1.20556943e-01 -1.40124619e+00
4.20608789e-01 -6.30763829e-01 -7.07957983e-01 2.09540308e-01
-1.36802375e-01 4.97217000e-01 2.32275259e-02 3.13137650e-01
-8.72903764e-01 4.11761440e-02 9.78777409e-01 -1.56362489e-01
-2.38736317e-01 5.86375475e-01 -7.92654634e-01 2.21929595e-01
5.53999305e-01 -5.70636272e-01 -2.90779889e-01 -1.15033902e-01
6.60328269e-01 1.13676097e-02 -4.74496596e-02 -8.90066326e-01
2.39497632e-01 -2.56122816e-02 3.69032532e-01 -3.72527510e-01
3.32365066e-01 -6.00614369e-01 2.09680647e-01 2.60038555e-01
-8.25599551e-01 -1.34233451e-02 -2.09639072e-01 8.29410911e-01
-2.94835735e-02 -6.69897199e-01 3.51475835e-01 -1.94398150e-01
-5.43532789e-01 -1.16233518e-02 -3.67790848e-01 -4.15736079e-01
6.47570789e-01 2.59511948e-01 -1.28689289e-01 -3.95536304e-01
-9.72695947e-01 -7.23421946e-02 1.45995393e-01 6.21312022e-01
3.83467585e-01 -1.26797652e+00 -4.78791863e-01 2.36535028e-01
4.83669609e-01 -8.89602959e-01 1.98555648e-01 7.10376978e-01
-7.51404241e-02 4.50300217e-01 -7.70456865e-02 -1.35901228e-01
-1.51377332e+00 5.36826193e-01 -2.44763836e-01 -6.27205372e-01
-5.57613373e-01 8.62673938e-01 -1.79149024e-02 -3.52169365e-01
3.94822836e-01 2.62283057e-01 -8.63594353e-01 3.01870883e-01
7.63260841e-01 5.87638974e-01 4.08681571e-01 -3.83493364e-01
-1.84562489e-01 5.78700565e-02 -4.77209955e-01 -5.76092526e-02
1.16775942e+00 3.59828882e-02 2.18159780e-01 1.78924128e-01
9.65397954e-01 6.71863139e-01 -4.81871873e-01 -3.91485870e-01
1.54760107e-03 -7.31408596e-01 3.01550955e-01 -6.56850040e-01
-9.95298386e-01 4.11340445e-01 6.86478317e-01 6.35882080e-01
5.85917354e-01 2.09069148e-01 8.55154216e-01 7.67646015e-01
5.97211272e-02 -1.05035710e+00 1.73675254e-01 3.34335804e-01
4.48953718e-01 -1.13281035e+00 1.91708848e-01 -1.32069498e-01
-8.59863698e-01 9.66948509e-01 2.27369636e-01 1.33967116e-01
4.55432624e-01 -1.73589766e-01 -6.96572885e-02 -4.42929685e-01
-6.70275688e-01 -5.16688645e-01 8.15555930e-01 5.09021841e-02
7.34272659e-01 1.82184994e-01 -9.11637604e-01 8.04917872e-01
-2.22855178e-03 -1.08064242e-01 1.86330304e-01 4.99083877e-01
-7.39339471e-01 -1.50029933e+00 -3.73805426e-02 8.58626246e-01
-4.93838936e-01 6.61075115e-02 -3.92036885e-01 1.37981832e-01
-1.70992747e-01 1.00348532e+00 -3.87365296e-02 -8.06836009e-01
4.68343318e-01 6.49089217e-02 -1.24148745e-02 -8.54234099e-01
-9.92914259e-01 2.79984642e-02 4.35381383e-01 -5.61436474e-01
2.24624127e-01 -5.51869929e-01 -8.91376793e-01 -2.54675388e-01
-2.44236633e-01 6.97096705e-01 1.10716760e+00 8.12703609e-01
4.69561517e-01 6.53723478e-01 7.49355137e-01 -8.52324545e-01
-8.62415314e-01 -1.00026870e+00 -7.74039328e-01 7.61108398e-01
-2.52650023e-01 -8.03296924e-01 -9.53567699e-02 -4.25560385e-01] | [10.207293510437012, 5.778411865234375] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.